American Journal of Physical Medicine and …lib.ajaums.ac.ir/booklist/47437.pdfAmerican Journal of Physical Medicine and Rehabilitation April 2007, Volume 86, Issue 4,pp.253-334 253

American Journal of Physical Medicine and Rehabilitation April 2007, Volume 86, Issue 4,pp.253-334 253 About the Journal and Scientific Publishing.

Walter R. Frontera, MD, PhD

Research Article Robotics

255 Electromyography-Controlled Exoskeletal Upper-Limb-Powered Orthosis for Exercise Training After Stroke. Joel Stein, MD; Kailas Narendran, M.Eng; John McBean, MS; Kathryn Krebs, OTR/L; Richard Hughes, PT, MS, NCS

262 Scanning Laser-Doppler Imaging of Leg- and Foot-Skin Perfusion in Normal Subjects: Analysis of Age, Gender, Site, and Laser-Type Effects. Charles F. Kunkel, MD; Stephen F. Figoni, PhD; Jana M. Baumgarten, MD; Claudio M. Carvalho, DO; Hyung S. Kim, MD; Resa L. Oshiro, MD; Milena D. Zirovich, MD; Oscar U. Scremin, MD, PhD; A M. Erika Scremin, MD

272 Recovery Preference Exploration: Analysis of Patient Feedback After Imagined Scenarios. Margaret G. Stineman, MD, FACRM; Richard N. Ross, MS; Greg Maislin, MS, MA; Nicholas Marchuk, BS; Serina Hijirida, OTD; Mark G. Weiner, MD

282 Pathological Changes of Human Ligament After Complete Mechanical Unloading. Chia-Hsin Chen, MD; Xuhui Liu, MD; Ming-Long Yeh, PhD; Mao-Hsiung Huang, MD; Qihui Zhai, MD; Walter R. Lowe, MD; David M. Lintner, MD; Zong-Ping Luo, PhD

290 Effects of Glossopharyngeal Pistoning for Lung Insufflation on Vital Capacity in Healthy Women. Malin Nygren-Bonnier, PT, MSc; Peter Lindholm, MD, PhD; Agneta Markström, MD, PhD; Maria Skedinger, MD, PhD; Eva Mattsson, PT, PhD; Brita Klefbeck, PT, PhD

295 Lung Inflation by Glossopharyngeal Breathing and "Air Stacking" in Duchenne Muscular Dystrophy. John R. Bach, MD; Carlo Bianchi, MD; Mauro Vidigal-Lopes, MD; Sandra Turi, MD; Giorgio Felisari, MD

Commentary Pulmonary

301 Air Doping: An Expose on "Frog" Insufflation in Competitive Sports. John R. Bach, MD; George Tewfik, BS

Brief Report Imaging

304 High-Resolution Magnetoencephalographic Functional Mapping of the Cortical Network Mediating Intentional Movement. Carlos Amo, MD, PhD; Robert A. Boyajian, MD, MS; John S. Romine, MD; Shirley M. Otis, MD

Invited Review Imaging

310 Introduction to Musculoskeletal Diagnostic Ultrasound: Examination of the Upper Limb. Henry L. Lew, MD, PhD; Carl P. C. Chen, MD; Tyng-Guey Wang, MD; Kelvin T. L. Chew, MD

Visual Vignette 322 Synovitis of the Sternoclavicular Joint: The Role of Ultrasound.

Steve J. Wisniewski, MD; Jay Smith, MD

AAP Annual Meeting Abstracts PDF Only

324 Abstracts of Scientific Papers and Posters Presented at the Annual Meeting of the Association of Academic Physiatrists: San Juan, Puerto Rico April 10-14, 2007.

About the Journal and Scientific Publishing

The publication of this issue of the American Journal ofPhysical Medicine and Rehabilitation (AJPM&R) coincideswith the celebration of the Annual Meeting of the Associ-ation of Academic Physiatrists in San Juan, Puerto Rico. Ata time when the Association meets to share new informa-tion and deliberate about the future of the field, I wouldlike to use this editorial page to comment on several thingsrelated to our Journal that may be of interest to thereadership.

Some new events and features of the AJPM&R areworth mentioning. In terms of content, a supplementincluding the papers presented at the Musculoskeletal Re-search Conference organized by the National Center forMedical Rehabilitation Research, National Institutes ofHealth, was published together with the first issue of theAJPM&R in 2007. As the field of physical medicine andrehabilitation (PM&R) makes a concerted effort to strengthenits scientific basis, the main objective of the conference was todevelop an agenda for musculoskeletal research. The manu-scripts included in the supplement summarize the state ofthe science and identify important gaps in our knowledgeas it relates to the diagnosis, treatment, and rehabilitationof musculoskeletal pain and injury. Another importantpublication is a commentary on the Asian Perspective ofPM&R that will be published in the May issue of theAJPM&R in time for the Fourth World Congress of theInternational Society for Physical and Rehabilitation Med-icine. The congress will be held in Seoul, Korea on June10–14, 2007. Many countries in Asia have a long-standingtradition in PM&R and have contributed significantly tothe expansion of the field in that part of the world. Also,many of our colleagues in Asia are submitting their workfor consideration by the editorial board of the AJPM&R.Finally, a four-part special series on poststroke and braininjury rehabilitation will be published beginning with theJune issue. These papers, two per issue, address clinicalproblems that may not be getting enough attention fromclinicians in this area and discuss novel approaches to theacute treatment and rehabilitation of these conditions. Dowe need to increase blood pressure after an acute stroke?

In terms of presentation and process, we have recentlystarted to add more color figures in selected articles tomake the publication more visually attractive and to showdetails in color that may not be seen in black and whiteimages. With the objective of making new informationavailable more quickly to the reader and to clinicians,

selected papers are being published early online beforeprint at www.AJPMR.com. In an attempt to simplify thesubmission process, an email submission option for send-ing new manuscripts to [email protected] is nowavailable as described in the new Instructions for Authors.Also, we are offering the option for authors to publishvideo clips online as support material to printed manu-scripts. This new video feature has been launched with thepaper by Stein et al. in this issue of the Journal.

Finally, we have made several changes in the compo-sition of the Editorial Board. In 2006, three Members of theBoard have been “promoted” to Associate Editors, includ-ing James Atchison (University of Florida, Gainsville),Denise Campagnolo (Barrow Neurological Institute), andJames Richardson (University of Michigan). Paul Pasquina(Walter Reed Medical Center) was appointed Member of theEditorial Board, and Moon Suk Bang (Department ofPM&R, Seoul National University, South Korea) was cho-sen as International Editorial Consultant.

With the purpose of increasing the number of non-physiatrists participating in the editorial process and aug-menting the expertise in the rehabilitation sciences, wehave appointed James Ashton-Miller (Professor of Mechan-ical Engineering and Director of the Biomechanics Re-search Laboratories, University of Michigan) and Rory A.Cooper (Paralyzed Veterans of America (PVA) Chair andDistinguished Professor, Department of Rehabilitation Sci-ence and Technology, and Director of the Human Engi-neering Research Laboratories, University of Pittsburgh) asAssociate Editors. We are also very happy to announce theappointment of Scott Ward (Professor and Chair, Divisionof Physical Therapy at the University of Utah, and Presidentof the American Physical Therapy Association) as a Mem-ber of the Editorial Board. Scientific diversity is needed onthe board to make the Journal even better.

In 2006, the very important discussion about scientificmisconduct resurfaced. Misconduct is defined at 42 CFR50.102 as “fabrication, falsification, plagiarism or otherpractices that seriously deviate from those that are com-monly accepted within the scientific community for pro-posing, conducting or reporting research. It does not in-clude honest error or honest differences in interpretationor judgments of data.” In a recent editorial, DonaldKennedy, editor of Science, summarized an experience bythat very prestigious journal that had resulted from thepublication and subsequent retraction of two manuscripts

April 2007 Editorial 253

EDITORIAL

with fraudulent data.1 In his opinion, “the environment forscience now presents increased incentives for the produc-tion of work that is intentionally misleading or distorted byself-interest.” He recommends that special attention begiven to the papers that, although small in number, arelikely to be visible and influential. Other more clinicallyoriented journals, such as the New England Journal ofMedicine, also have had to deal with the consequences ofscientific misconduct.2 Clearly, not everyone plays by therules and follows the scientific code of ethics. Althoughcases of scientific misconduct have not been an issue inour field, we should maintain a high level of suspicionduring the review process to make sure it does not happen.Our vigilance will also prevent scholarly ethical dilemmassuch as simultaneous submission of the same manuscriptto multiple journals.

The topic of conflict of interest in scientific publishingalso received some publicity in 2006.3,4 The editor-in-chiefof one journal resigned because a published manuscript onwhich he was an author did not disclosed his ties to the

manufacturer of a device that was highlighted in the re-view paper. Further, all other authors of the paper are paidadvisers to the company, and the first draft had beenwritten by a professional writer hired by the company. Theeditor called it an oversight, but the perception of a con-flict is enough to raise concerns. The trust of our col-leagues, researchers, and clinicians as well as that of thepublic on the information published by clinical and scien-tific journals will depend on the quality of peer-review andon the ability of editorial teams to monitor the process.

See you in San Juan!

Walter R. Frontera, MD, PhDEditor-in-Chief

REFERENCES1. Kennedy D: Responding to fraud. Science 2006;314:1353

2. Couzin J: Breakdown of the year: scientific fraud. Science 2006;314:1853

3. Holden C: The undisclosed background of a paper on a depressiontreatment. Science 2006;313:598–9

4. Holden C: Flap claims journal editor. Science 2006;313:1219

254 Frontera Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 4

Authors:Joel Stein, MDKailas Narendran, M.EngJohn McBean, MSKathryn Krebs, OTR/LRichard Hughes, PT, MS, NCS

Affiliations:From the Department of PhysicalMedicine and Rehabilitation, HarvardMedical School, Boston,Massachusetts (JS, KK, RH);Spaulding Rehabilitation Hospital,Boston, Massachusetts (JS, KK, RH);Active Joint Brace Research Lab,Department of MechanicalEngineering, Massachusetts Instituteof Technology, Boston, Massachusetts(KN, JM); and Myomo, Inc., Boston,Massachusetts (KN, JM).

Correspondence:All correspondence and requests forreprints should be addressed to JoelStein, MD, Spaulding RehabilitationHospital, 125 Nashua St., Boston, MA02114.

Disclosures:Mr. Narendran and Mr. McBean bothown equity stakes in Myomo, Inc.,which licenses the technologydescribed in this article from MIT.

Editor’s Note: Video demonstrationof this device is available atwww.AJPMR.com.

0894-9115/07/8604-0255/0American Journal of PhysicalMedicine & RehabilitationCopyright © 2007 by LippincottWilliams & Wilkins

DOI: 10.1097/PHM.0b013e3180383cc5

Electromyography-ControlledExoskeletal Upper-Limb–PoweredOrthosis for Exercise Training AfterStrokeABSTRACT

Stein J, Narendran K, McBean J, Krebs K, Hughes R: Electromyography-controlled exoskeletal upper-limb–powered orthosis for exercise training afterstroke. Am J Phys Med Rehabil 2007;86:255–261.

Objective: Robot-assisted exercise shows promise as a means ofproviding exercise therapy for weakness that results from stroke or otherneurological conditions. Exoskeletal or “wearable” robots can, in principle,provide therapeutic exercise and/or function as powered orthoses to helpcompensate for chronic weakness. We describe a novel electromyogra-phy (EMG)-controlled exoskeletal robotic brace for the elbow (the activejoint brace) and the results of a pilot study conducted using this brace forexercise training in individuals with chronic hemiparesis after stroke.

Design: Eight stroke survivors with severe chronic hemiparesis wereenrolled in this pilot study. One subject withdrew from the study becauseof scheduling conflicts. A second subject was unable to participate in thetraining protocol because of insufficient surface EMG activity to controlthe active joint brace. The six remaining subjects each underwent 18 hrsof exercise training using the device for a period of 6 wks. Outcomemeasures included the upper-extremity component of the Fugl-Meyerscale and the modified Ashworth scale of muscle hypertonicity.

Results: Analysis revealed that the mean upper-extremity component ofthe Fugl-Meyer scale increased from 15.5 (SD 3.88) to 19 (SD 3.95)(P � 0.04) at the conclusion of training for the six subjects who completedtraining. Combined (summated) modified Ashworth scale for the elbowflexors and extensors improved from 4.67 (�1.2 SD) to 2.33 (�0.653 SD)(P � 0.009) and improved for the entire upper limb as well. All subjectstolerated the device, and no complications occurred.

Conclusion: EMG-controlled powered elbow orthoses can be success-fully controlled by severely impaired hemiparetic stroke survivors. This techniqueshows promise as a new modality for assisted exercise training after stroke.

Key Words: Robots, EMG, Stroke, Exercise, Hemiparesis

April 2007 EMG-Controlled Orthosis 255

RESEARCH ARTICLE

Robotics

More than 700,000 strokes occur each year inthe United States, and there are more than 4.5million stroke survivors in the population at anygiven time. Of these, approximately 1 million havesubstantial disability as a result of stroke.1 Im-paired motor function after stroke has been dem-onstrated to be a primary contributor to poststrokedisability.2 Although existing rehabilitation pro-grams provide significant benefits with regard tofunctional independence, these programs have hadlimited impact on the underlying motor impair-ment. Outcomes for upper-limb function in hemi-paresis caused by stroke are particularly poor.3

Rehabilitation programs for hemiparesis afterstroke provide therapeutic exercises designed toincrease functional independence, attempting tostimulate neurologic recovery and teach compen-satory techniques. Although most patients regaintheir walking ability, 30–66% are no longer able touse the affected arm functionally.4 It has beenfound that only 5% of individuals who receiveintensive therapy for severe upper-extremity weak-ness after stroke regain functional use of the upperextremity during the course of rehabilitation.3 Theneed for effective rehabilitation for the paretic up-per limb after stroke remains largely unmet.

In view of the limited success of traditional re-habilitation programs in restoring upper-limb func-tion after stroke, research has focused on facilitatingmotor recovery through a variety of novel ap-proaches. Both human and animal studies have es-tablished that cerebral plasticity plays an importantrole in motor recovery.5–8 Studies of exercise to in-fluence motor recovery have demonstrated improvedmotor abilities and associated changes in metabolicactivity and excitability in the cerebral cortex.9–11

For people with stroke, it has been demonstratedthat highly repetitive, task-specific exercise trainingcan facilitate cortical plasticity in the brain, withconcomitantly improved motor abilities and en-hanced functional activity performance.9,12 A numberof techniques have been proposed to facilitate thedelivery of this type of exercise therapy after stroke,including constraint-induced movement therapy androbot-aided rehabilitation.13–15

Robotic devices are of particular interest be-cause they can provide exercise therapies in accu-rate, reproducible dosages, and they can provideassistance during task training for individuals withmore severe weakness after stroke. Furthermore,these devices have the potential to ultimately be-come an economic complement to traditional, la-bor- and time-intensive neurological rehabilita-tion. The best-studied robotic devices for providingupper-extremity exercise after stroke are the MIT-Manus robot and the MIME robot.14–21 Improvedmotor function has been shown with the MIT-

Manus robot in patients early after stroke.17,18

These improvements were maintained at the timeof a 3-yr follow-up study.19 The MIT-Manus robotalso has been shown to provide modest improve-ments in motor function in people with chronichemiparesis after stroke.14,20,21 Both the MIT-Ma-nus and the MIME robots are stationary trainingdevices, and they cannot be directly incorporatedinto the performance of activities of daily living ina community setting.

Electromyographic (EMG) signals have beenused to trigger robot-assisted exercises and to con-trol powered exoskeletal devices to provide assis-tance with movement.22–25

Limitations of these systems include controlalgorithms that have difficulty correcting for co-contraction of antagonist muscles,23 fixed (station-ary) devices that are not inherently mobile,22,23,25

and bulky actuators that are not wearable.24

Device DescriptionWe have developed a novel device, the active

joint brace (AJB), which effectively combines anexoskeletal robotic brace with EMG control algo-rithms (see Fig. 1). The AJB is a noninvasive, light-weight, wearable system that uses surface EMGsignals from affected muscle groups to control apowered orthosis to assist with the movement of aparetic limb. The AJB functions by continuouslymonitoring the surface EMG signals of the user’sflexor and extensor muscles of the elbow joint.These signals are filtered and processed to infer adesired joint torque. The signal processing of themeasured surface EMG is accomplished through asystem that comprises off-the-shelf EMG sensors,analog signal–processing components, and digitalsignal–processing components. All digital signalprocessing is implemented on a PIC microcon-troller, with code written in C.

The signal-processing algorithm enables bidi-rectional control, from a single degree of input,through the use of a unilateral active assist, com-bined with a competing passive force.

FIGURE 1 Subject with active joint brace in use.

256 Stein et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 4

The EMG sensors used were the BL-AE-W fromB&L Engineering. The sensors have a gain of 300,bandwidth of 10 Hz to 3.12 kHz, and �100 M� ofinput impedance. The averaged output of three iden-tical sensor channels was used to control the motor.Analog signal processing is accomplished through acombination of linear and nonlinear filtering ele-ments that act as a signal-conditioning front end to aPIC microcontroller. The analog signal conditioningis similar to a power calculation that selects low-order frequency components of the input signal, inaddition to performing antialiasing signal condition-ing. The digital signal–processing algorithms imple-mented in the PIC allow adjustment of the bandwidthof the software system and adjustment of the passiveextension force.

The parameter of system gain (amount of as-sistance in the active assist direction) generallyvaries during the course of a session as the subjectfatigues. The base unit for software gain corre-sponds to 12 V of motor voltage per volt of surfaceEMG voltage.

The parameter for the passive opposing forcewas generally constant throughout a study session,usually changing slightly from session to session toaccount for changes in muscle tone.

The parameters were set by study staff eachsession, using the following procedure. The bracewas placed on the arm, in a neutral position (90-degree bend in arm). If the device was in bicepassist mode, the spring was increased until the armreached close to full extension. At that point, thesubject was asked to bring a hand to his or her face.While the subject flexed his or her arm, the gainwas increased to the lowest level that would enablecontrolled movement. If the brace was in tricepsmode, the procedure was the same, but with thedirections reversed.

These data are used to externally apply a pro-portional torque in parallel with the existing mus-cle force (via the actuator and drive train) to assistthe user in achieving the desired motion. An elec-tric motor is used to power the device through a

system of cables attached to the brace (Fig. 2). Abrushed 12-V DC motor was used to move thebrace. Torque was transmitted through a throttlecable system of steel cables and flexible cable hous-ing. Separate power supplies were used for motorcontrol and signal processing to eliminate noise.

The device provides active assistance with el-bow flexion and extension and allows the user toalternate between these movements without theneed to change any device settings. The user’scentral nervous system is incorporated into thecontrol loop through a combination of kinesthetic,proprioceptive, tactile, and visual sensory feedback.This allows the user to accomplish position con-trol, with the AJB acting as a forward-loop strengthamplifier. The treating therapist (typically the pa-tient’s occupational or physical therapist) can ad-just the system parameters to alter the amount ofassistance that the device provides. Control algo-rithms are supplemented with hardware and soft-ware safety interlocks to ensure that an excessiveamount of force is never exerted on the user’s limb.These include physical limitations in range of mo-tion inherent in the modified off-the-shelf elbowbrace used (Bledsoe Brace Systems, Grand Prairie,TX), a motor with a stall torque such that themaximum force that can applied at the hand is 10lbs, and a back-drivable system.

We hypothesized that stroke survivors withpersistent weakness after stroke would be able tosafely and effectively control the AJB, allowingthem to perform exercise training with the device.We report the results of a pilot feasibility study ofexercise training with the AJB for individuals withchronic hemiparesis after stroke.

METHODSEight subjects with chronic weakness at least 6

mos after stroke were enrolled in the study. Sub-jects were required to have weakness and loss ofmotor control sufficient to interfere with func-tional use of the upper extremity. Individuals witheither ischemic stroke or hemorrhagic stroke were

FIGURE 2 Photo and computer-aided design model of the active joint brace system.


eligible. Subjects were required to have completedtheir formal physical and occupational therapy pro-grams before study entry. Exclusion criteria in-cluded history of more than one stroke, other neu-rologic conditions that might affect motor abilities(e.g., Parkinson disease), other medical conditionsthat were likely to interfere with the ability tocomplete the protocol, active participation in an-other research study, and cognitive or perceptualdeficits or aphasia sufficiently severe to interferewith the ability to follow instructions and completethe protocol. Subjects were asked to refrain frombeginning any new interventions for motor reha-bilitation (such as physical or occupational therapyfor the arm) during the course of their participa-tion in the study.

Electrodes were placed over the center of thebiceps and triceps muscles (location determined byvisual estimation). One subject dropped out of thestudy because of scheduling conflicts. A secondsubject was found to have insufficient surface EMGactivity in the target muscles (biceps and triceps)to control the device. This subject had no clinicallyevident movement at the elbow (manual muscletesting score of 0/5 at elbow flexors and extensors)and an upper-extremity Fugl-Meyer scale of 6. Thedata on the six subjects who completed the trainingprotocol were analyzed and are reported here.

The subjects’ descriptive data are provided inTable 1. Subjects underwent a baseline assessmentof motor function by a physical or occupationaltherapist before initiating treatment with the de-vice, and at the conclusion of training. Assessmentsincluded measurement of upper-extremity motorfunction using the Fugl-Meyer scale, measure-

ments of active and passive range of motion at theshoulder, elbow, wrist, and hand, and measure-ments of muscle hypertonicity using the modifiedAshworth scale (MAS).26,27 MAS measurementswere made at 14 muscle groups in the affectedupper limb. For the purposes of statistical analysis,the MAS values were converted from a 0–4 scale(which includes a value of 1� between scores of 1and 2) to a 0–5 scale; the resulting values weresummated to obtain an overall score ranging from0 to 70. Clinical scales were assessed by one of twoexperienced research therapists; the same exam-iner performed all measurements for each individ-ual subject.

Training consisted of 2–3 hrs of training perweek for a total of 18 hrs of exercise trainingduring a 6- to 9-wk period. Each session consistedof approximately 60 mins of exercise training usingthe AJB. Exercises consisted of a defined set offunctionally oriented upper-extremity tasks tai-lored to each subject’s motor abilities, such asmoving blocks from one area to another or turninga light switch on and off. Subjects performed thesemotor tasks while using the AJB. Tasks were tai-lored to the subjects’ distal motor abilities becauseall subjects had significant weakness and impairedmotor control in the hand and wrist. The study wasconducted in an open-label fashion.

This study was approved by the institutionalreview boards of both Spaulding RehabilitationHospital and the Massachusetts Institute of Tech-nology. All subjects provided informed consent be-fore participation.

STATISTICAL ANALYSISGiven the preliminary nature of this feasibility

study, no a priori power analysis was performed.Outcome measures were found to have a normaldistribution and were analyzed using two-tailedpaired t tests.

RESULTSTraining with the AJB was well tolerated by all

participants, and no complications of treatmentoccurred. All users were able to successfully con-trol the device to assist them in elbow flexion andextension movements, with the exception of theone subject who was dropped from the study be-cause of undetectable elbow-flexion and elbow-ex-tension motor control.

Motor control, as measured by the upper-ex-tremity portion of the Fugl-Meyer scale, showedstatistically significant gains, from a mean of 15.5(�3.88 SD) before treatment to a mean of 19(�3.95 SD) at the end of treatment (P � 0.04)(Fig. 3).

Measurements of muscle hypertonicity, asmeasured by the MAS, also demonstrate a decrease

TABLE 1 Demographic and clinical data onstudy participants (data providedonly for subjects completing thestudy protocol)

Number of subjects 6

Mean age 53Gender

Male 5Female 1

Mean (range) duration sincestroke, yrs

3.67 (1.5–10)

Mechanism of strokeInfarct 4Hemorrhage 2

Side of strokeRight 4Left 2

Anatomy of strokeMCA distribution (infarct) 4Internal capsule/basal ganglia(hemorrhage)

2


in the summated MAS for the upper limb as aresult of the intervention, from a mean of 19(�5.21 SD) to a mean of 15.17 (�6.31 SD) (P �0.04). The largest decrease in MAS was in theelbow, with the sum of elbow flexor and elbowextensor tone decreasing from a mean of 4.67(�1.2 SD) to 2.33 (�0.65 SD) (P � 0.009) (Fig. 4).MAS values have been converted to a 0–5 scale.

Subjective impressions of the participantswere not formally collected, but anecdotally, virtu-ally all participants found training with the deviceenjoyable and felt that they experienced meaning-

ful improvement in elbow control and use of theupper limb as a result of participation.

DISCUSSIONThis pilot study demonstrates the feasibility of

using an EMG-controlled powered exoskeletal or-thosis for exercise training in stroke survivors. Italso provides some preliminary evidence suggest-ing that the AJB system may be efficacious forimproving upper-limb motor function in strokesurvivors with chronic hemiparesis.

FIGURE 3 Upper-extremity Fugl-Meyer scores.

FIGURE 4 Summated modified Ashworth scale by upper-extremity joint before and after AJB treatment pro-gram. The number of muscle groups tested for each joint is shown in parentheses.


The size of the improvement found in the upper-extremity component of the Fugl-Meyer scale (3.5points) is modest in absolute terms, although it rep-resents a 23% increase in this severely impaired sam-ple. Although the minimal clinically important differ-ence for the upper-extremity component of the Fugl-Meyer scale has not been established, this degree ofimprovement is comparable with that found in otherstudies of robot-assisted exercise for the upper limbafter stroke.14 Similarly, the magnitude of reductionin muscle hypertonicity at the elbow, as measured bythe Ashworth scale, seems to be clinically relevantand to be comparable with the degree of improve-ment stipulated as significant in studies of othertreatments for spasticity.28

The data obtained regarding efficacy should beinterpreted cautiously. This study was an open-label, uncontrolled trial with a small number ofsubjects; it was not designed to test efficacy. Po-tential confounders include placebo/nonspecific ef-fects of medical attention and encouragement, andparticipation in a formal upper-limb exercise pro-gram for individuals who were not previously ac-tive in such a program. A larger study with blindedassessors and appropriate controls is needed tofurther assess efficacy of this training device andprogram.

The ability of severely hemiplegic stroke sur-vivors to effectively control a powered exoskeletalorthosis using EMG-based control is itself an in-triguing finding. Hemiparesis after stroke is typi-cally considered more of an impairment of motorcontrol than of muscle-force generation. Thus, onemight assume that the ability to control the pareticlimb would not be enhanced using this technique(which relies on the subject’s own motor control)and provide amplification of force generation. Wehypothesize that the role of weakness in impairingmotor recovery may be underappreciated.29 Ourpopulation comprised a group of individuals withfairly severe motor impairments of the upper limb,and they were unable to generate much force at theelbow at baseline. Such individuals are unable topractice the use of the limb because of the severityof their baseline impairments. This, in turn, mayinhibit their motor recovery. The ability of thisdevice to provide a “power assist” to the hemipa-retic muscle groups may help close the feedbackloop of brain intention and actual limb movementthat is believed to be a key component of cerebralplasticity in motor recovery.

Potential concerns regarding the use of a pow-ered EMG-controlled orthosis for patients withspasticity are addressed only partially in this pilotstudy. We excluded individuals with marked spas-ticity in order to minimize the risk of amplifyingany involuntary movements. Further studies ofsubjects with greater degrees of hypertonicity are

needed, as are studies of individuals who are at riskof flexor spasms at the elbow (e.g., those withcervical spinal cord injuries).

This type of device has the potential for twodistinct therapeutic uses. One use is as an aid forexercise training, as demonstrated in this study.This application could be further subdivided intocontrol and training algorithms targeted at train-ing motor control for stroke, traumatic brain in-jury, and other central nervous system disorders.The device also could be programmed for resis-tance training for strengthening exercises, asmight be appropriate in patients after injury ororthopedic surgery who do not have upper–motorneuron damage.

A second potential application of this type ofdevice is as a powered brace for functional applica-tions. Individuals with chronic weakness (such asthose caused by spinal cord injury, peripheral neu-ropathy, or myopathy) could potentially use thistype of device during their daily activities as apower assist. The goal of this use would be tofacilitate performance of daily tasks, rather than asexercise therapy. Further research in this applica-tion is needed, as is further development of thedevice to address issues of portability, ease of ap-plication, and comfort.

ACKNOWLEDGMENTSSupport for this research was provided by the

MIT Deshpande Center for Technological Innova-tion. The assistance of Kate Zebrose, Mira Sahney,Mina Hsang, Jenny Hu, Chris Possinger, and PaulPetrone is gratefully acknowledged. This work andtechnology would not have been possible withoutthe unwavering and gracious support of ProfessorWoodie Flowers at MIT.

REFERENCES1. American Heart Association: Heart disease and stroke sta-

tistics—2006 update. Circulation 2006;113:e85–151

2. Shelton FD, Volpe BT, Reding M: Motor impairment as apredictor of functional recovery and guide to rehabilitationtreatment after stroke. Neurorehabil Neural Repair 2001;15:229–37

3. Gowland C: Recovery of motor function following stroke:profile and predictors. Physiotherapy 1982;34:77–84

4. Kwakkel G, Kollen BJ, Wagenaar RC: Therapy impact onfunctional recovery in stroke rehabilitation: a critical reviewof the literature. Physiotherapy 1999;13:457–70

5. Cramer SC, Nelles G, Benson RR, et al: A functional MRIstudy of subjects recovered from hemiparetic stroke. Stroke1997;28:2518–27

6. Nudo RJ, Milliken GW: Reorganization of movement repre-sentations in primary motor cortex following focal ischemicinfarcts in adult squirrel monkeys. J Neurophysiol 1996;75:2144–9

7. Nudo RJ, Milliken GW, Jenkins WM, Merzenich MM: Use-dependent alterations of movement representations in pri-mary motor cortex of adult squirrel monkeys. J Neurosci1996;16:785–807

8. Nudo RJ, Wise BM, SiFuentes F, Milliken GW: Neural sub-


strates for the effects of rehabilitative training on motorrecovery after ischemic infarct. Science 1996;272:1791–4

9. Liepert J, Bauder H, Wolfgang HR, Miltner WH, Taub E,Weiller C: Treatment-induced cortical reorganization afterstroke in humans. Stroke 2000;31:1210–6

10. Nelles G: Cortical reorganization—effects of intensive ther-apy. Restor Neurol Neurosci 22;2004:239–44

11. Schaechter JD: Motor rehabilitation and brain plasticityafter hemiparetic stroke. Prog Neurobiol 2004;73:61–72

12. Levy CE, Nichols DS, Schmalbrock PM, Keller P, ChakeresDW: Functional MRI evidence of cortical reorganization inupper-limb stroke hemiplegia treated with constraint-induced movement therapy. Am J Phys Med Rehabil 2001;80:4–12

13. Taub E, Miller NE, Novack TA, et al: Technique to improvechronic motor deficit after stroke. Arch Phys Med Rehabil1993;74:347–54

14. Fasoli SE, Krebs HI, Stein J, et al: Effects of robotic therapyon motor impairment and recovery in chronic stroke. ArchPhys Med Rehabil 2003;84:477–82

15. Lum PS, Burgar CG, Shor PC, Majmundar M, Van der LoosM: Robot-assisted movement training compared with con-ventional therapy techniques for the rehabilitation of up-per-limb motor function after stroke. Arch Phys Med Re-habil 2002;83:952–9

16. Hogan N, Krebs HI, Sharon A, Charnnarong J: InteractiveRobot Therapist. US Patent 5,466,213, issued November 14,1995

17. Aisen FL, Krebs HI, Hogan N, McDowell F, Volpe BT: Theeffect of robot-assisted therapy and rehabilitative trainingon motor recovery following stroke. Arch Neurol 1997;54:443–46

18. Volpe BT, Krebs HI, Hogan N, Edelstein OTR L, Diels C, AisenM: A novel approach to stroke rehabilitation: robot-aided sen-sorimotor stimulation. Neurology 2000;54:1938–44

19. Volpe BT, Krebs HI, Hogan N, Edelsteinn L, Diels CM, Aisen

ML: Robot training enhanced motor outcome in patientswith stroke maintained over 3 years. Neurology 1999;53:1874–6

20. Stein J, Krebs HI, Frontera WR, Fasoli SE, Hughes R,Hogan N: A comparison of two techniques of robot-aidedupper limb exercise training after stroke. Am J Phys MedRehabil 2004;83:720–8

21. Ferraro M, Palazzolo JJ, Krol J, Krebs HI, Hogan N, VolpeBT: Robot-aided sensorimotor arm training improves out-come in patients with chronic stroke. Neurology 2003;61:1604–7

22. Dipietro L, Ferraro M, Palazzolo JJ, Krebs HI, Volpe BT,Hogan N: Customized interactive robotic treatment forstroke: EMG-triggered therapy. IEEE Trans Neural SystRehabil Eng 2005;13:325–34

23. Cheng HS, Ju MS, Lin CC: Improving elbow torque outputof stroke patients with assistive torque controlled by EMGsignals. J Biomech Eng 2003;125:881–6

24. Ferris DP, Czerniecki JM, Hannaford B: An ankle-foot or-thosis powered by artificial pneumatic muscles. J Appl Bio-mech 2005;21:189–97

25. Rosen J, Brand M, Fuchs MB, Arcan M: A myosignal-basedpowered exoskeletal system. IEEE Trans Syst Man Cybern2001;31:210–22

26. Fugl-Meyer AR, Jaasko L, Leyman I, Olsson S, Steglind S:The post stroke hemiplegic patient. A method for evaluationof physical performance. Scand J Rehabil Med 1975;7:13–31

27. Bohannon RW, Smith MD: Interrater reliability of a modi-fied Ashworth scale of muscle spasticity. Phys Ther 1987;67:206–7

28. Brashear A, Gordon MF, Elovic E, et al: Intramuscularinjection of botulinum toxin for the treatment of wrist andfinger spasticity after a stroke. N Engl J Med 2002;347:395–400

29. Patten C, Lexell J, Brown HE: Weakness and strength train-ing in persons with poststroke hemiplegia. Rationale,method, and efficacy. J Rehabil Res Dev 2004;41:293–312


Authors:Charles F. Kunkel, MDStephen F. Figoni, PhDJana M. Baumgarten, MDClaudio M. Carvalho, DOHyung S. Kim, MDResa L. Oshiro, MDMilena D. Zirovich, MDOscar U. Scremin, MD, PhDA.M. Erika Scremin, MD

Affiliations:From the Department of PhysicalMedicine and Rehabilitation (CFK,SFF, AMES) and Research Services(OUS), VA Greater Los AngelesHealthcare System, Los Angeles,California; Departments of Physiology(OUS) and Medicine (CFK, AMES),David Geffen School of Medicine atUCLA, Los Angeles, California; andthe UCLA/VA Greater Los AngelesPhysical Medicine and RehabilitationResidency Program, Los Angeles,California (CFK, JMB, CMC, HSK,RLO, MDZ, AMES).

Correspondence:All correspondence and requests forreprints should be addressed toCharles F. Kunkel, MD, MS, PM&R(117), VA West Los AngelesHealthcare Center, 11301 WilshireBlvd., Los Angeles, CA 90073.

Disclosures:Supported by VA Rehabilitation R&DMerit Review Project A2860R and aSenior Research Career ScientistAward (B2541SA) to Dr. O. Scremin.None of the authors have anycommercial associations,consultancies, equity interests, orpatent-licensing arrangements thatmight pose a conflict of interest inconnection with the work.

The manuscript was represented intwo posters at the 2005 AAP Meetingin Tucson, AZ.


DOI: 10.1097/PHM.0b013e3180383526

Scanning Laser-Doppler Imaging ofLeg- and Foot-Skin Perfusion inNormal SubjectsAnalysis of Age, Gender, Site, and Laser-Type Effects

ABSTRACT

Kunkel CF, Figoni SF, Baumgarten JM, Carvalho CM, Kim HS, Oshiro RL,Zirovich MD, Scremin OU, Scremin AME: Scanning laser-Doppler imaging ofleg- and foot-skin perfusion in normal subjects: analysis of age, gender, site, andlaser-type effects. Am J Phys Med Rehabil 2007;86:262–271.

Objectives: To report normal values of skin perfusion in healthysubjects in three age groups using a laser Doppler imager; to determinedifferences attributable to gender, age, site, and use of red or near-infrared lasers; and to correlate transcutaneous oxygen with laser fluxvalues

Design: Flux and transcutaneous oxygen were measured at ten sites inthe lower extremity in 60 subjects from three age groups. Heated andunheated sites were scanned with red and near-infrared lasers.

Results: Heat hyperemia was prominent at all sites. Small, statisticallysignificant mean � SD differences were found between heated andnonheated sites for the red and near-infrared lasers (P � 0.02). All fluxratios were independent of gender but were higher in the oldest group.Plantar sites demonstrated higher flux in unheated areas and lower fluxratios compared with leg sites. Transcutaneous oxygen did not correlatesignificantly with flux for either laser type.

Conclusions: Scanning laser-Doppler imaging flux values provide areference for identifying patients at risk for tissue ischemia and poorhealing potential caused by impaired circulatory reserve in the legs anddistal feet. The lack of correlation between flux and transcutaneous oxygenin healthy individuals suggests that they measure different physiologicprocesses.

Key Words: Microcirculation, Laser Doppler Flowmetry, Rehabilitation, Wound Healing

262 Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 4

RESEARCH ARTICLE

Ischemia

More than 100,000 amputations are performedin the United States per year.1 Many of these areconsequences of critical limb ischemia, especiallyin patients with arteriosclerosis and diabetes. Oneof the challenges of amputation surgery is theaccurate limb-level selection. Because of poor as-sessment of blood flow before procedures, surgicalwounds fail to heal, and patients may need toundergo revisions at a more proximal site. There-fore, the most successful procedures in these pa-tients require an accurate initial assessment of skinperfusion.

Radioisotope clearance has been used to eval-uate skin blood flow.2–4 This technique can beexpensive, invasive, and time consuming. LaserDoppler flowmetry uses fiberoptic probes to studytissue microperfusion of small single sites at restand in response to a local thermal challenge. It hasbeen used to assess lower-extremity blood flow,amputation level, and wound healing.5–9

The scanning laser-Doppler imaging (LDI)technique uses the same principles as laser Dopplerflowmetry, but it has the advantages of samplinglarge areas rather than single points without directskin contact. This technique has been used in der-matology to measure burn depth10 and the effectsof drugs,11,12 inflammation,13 and other skin con-ditions.14 However, despite having clear advantagesover other methods of measurement, LDI has notbeen systematically used for the evaluation of pe-ripheral arterial disease (PAD). There is a need tofurther investigate the usefulness, effectiveness,and validity of scanning LDI to assess blood flowand healing potential in patients with PAD and inpatients requiring amputation.

TcPO2 values have a demonstrated use foridentifying sites with a good prognosis of healing,sites associated with healing complications, andsites associated with postoperative “failure.”15–17

Some investigators have reported both TcPO2 dataand laser Doppler data18,19; however, none havecorrelated the results of the methods in the samesites over extensive sampling of leg and foot sites.The TcPO2 method can be time consuming for theclinician and the patient. LDI scanning is a prom-ising technique for determining prognosis for post-operative wound healing because it is relativelyrapid, noninvasive, and painless.

One of the current limitations of using the LDItechnique is the lack of reported normal values andmethod standardization. The purposes of this studyare to (1) determine normal values of skin bloodflow (flux) in response to skin warming in healthysubjects in three age groups using LDI, (2) deter-mine differences attributable to gender, age, site,and use of red or near-infrared (NIR) lasers, (3)estimate test–retest reliability, and (4) correlate

LDI flux with TcPO2 values. The use of LDI as arapid, noninvasive, and accurate method of assess-ing skin perfusion would help guide surgeons tomake accurate initial decisions regarding amputa-tion level and avoid postsurgical complications andthe need for reamputation brought about by poorwound healing.

METHODSSubjects

A convenience sample of 60 healthy adults wasperformed in three age groups of 20 subjects each(20–39, 40–59, and 60–79 yrs old). The 20- to79-yr-old age range (with three subgroups at 20-yrintervals) was chosen to represent the averageadult life span.20 Written informed consent, ap-proved by the medical center’s institutional reviewboard, was obtained from subjects before their par-ticipation in this study. Prospective subjects com-pleted an extensive self-report health survey (seeAppendix 1). On the health survey, if subjects re-ported a history of diabetes mellitus, Raynaud’s orother PAD, leg amputation, skin disease, coronaryartery disease, cerebral vascular accident, dyslipi-demia, hypertension, smoking, edema, or rest/ex-ercise leg pain, they were excluded from the study.Subjects also were asked to report any medicationsthey were currently taking and were excluded fromentering the study if they listed any medicationstypically used to treat the above conditions, such ashypoglycemic and antiplatelet agents, lipid-lower-ing drugs, antihypertensives, or antiarrhythmicsand cardiovascular medications; they also were ex-cluded if they were on any cardiovascular medica-tions such as antihypertensives or antiarrhythmicsor if they reported a history of diabetes mellitus,Raynaud’s or other PAD, leg amputation, skin dis-ease, coronary artery disease, cerebral vascular accident, dyslipidemia, hypertension, smoking, edema,or rest/exercise leg pain. Resting blood pressure wasassessed before testing.

InstrumentationA Moor laser Doppler imager (Moor Instru-

ments, Ltd., Devon, England) with red (633 nm)and NIR (830 nm) laser beams was used for mea-surements of skin perfusion. With this instrument,laser light is projected onto the skin by a computer-controlled mirror in a raster pattern. A portion ofthe back-scattered laser light is detected as “flux,”representing the product of red cell velocity (cal-culated from the Doppler shift of the back-scat-tered light) and the red cell concentration (oftenreferred to as blood volume).21,22 Flux is expressedin arbitrary perfusion units and is calculated usingthe first moment of the power spectral density. Fiveidentical TcPO2 monitors were used to heat skin

April 2007 Scanning Laser-Doppler Imaging 263

and monitor TcPO2 (TCO2M, model 860, Novame-trix Medical Systems, Wallingford, CT).

Testing ProceduresEach subject was brought to an environmen-

tally controlled testing room. Subjects were offeredsafety goggles before laser scanning to preventdamage to their eyes from stray laser radiation. Tensites in either the right or the left lower extremity(leg randomly selected by a coin toss) were desig-nated as follows:

1. plantar surface over the first metatarsophalan-geal (MTP) joint,

2. plantar surface over the fifth MTP joint,3. dorsal surface over the first MTP joint,4. dorsal surface over the fifth MTP joint,5. lateral surface over the leg at a site two thirds

of the distance from the fibular head to lateralmalleolus (lateral distal leg),

6. lateral surface over the leg at midpoint be-tween the fibular head and the lateral malleo-lus (lateral middle leg),

7. lateral surface over the leg at one third of thedistance from the fibular head to lateral mal-leolus (lateral proximal leg),

8. anterior surface over the leg at a site two thirdsof the distance from the fibular head to lateralmalleolus (anterior distal leg),

9. anterior surface over the leg at midpoint be-tween the fibular head to lateral malleolus(anterior middle leg), and

10. anterior surface over the leg at a site one thirdof the distance from the fibular head to lateralmalleolus (anterior proximal leg).

Probes on leg sites were placed anteriorly overthe tibialis anterior and laterally over the peronealand lateral gastrocnemius muscles. On the foot,probes were placed on flat areas from which probeswould not easily detach. The sites were arbitrarilychosen for accessibility to TcPO2 and LDI evalua-tion and for their broad topographic distribution(proximal–distal leg; medial–lateral and dorsal–plantar foot).

Initial skin temperatures (before probe wasapplied) and final skin temperatures (immediatelyafter probe removal) were measured at each sitewith a handheld infrared thermometer, becauseskin perfusion is known to increase nonlinearlywith an increase in skin temperature, especiallywith temperatures above 33°C.23

Skin sites were cleaned with alcohol wipes,skin hair was shaved with a razor when necessary,and the TcPO2 probes were placed tightly on theskin site to provide an airtight seal using speciallydesigned adhesive rings. Subjects rested in the semi-

recumbent posture on a hospital bed for foot andanterior leg measurements and side-lying posture forlateral leg measurements. Then, each skin site waswarmed with TcPO2 probes to 44°C (111°F). Whilewaiting for the TcPO2 probe to warm up the skinsurface, the LDI laser head was aligned perpendicu-larly to the skin sites. The area to be scanned wasoutlined by the laser beam before actual laser Dopplerscanning to be certain that all the probes were con-tained in the scanning area.

TcPO2 values were recorded every 2 mins. Im-mediately after TcPO2 readings reached a steadystate (defined as unchanged values for two consec-utive measurements), the probes were removed.The surface temperature measured and the siteincluding the heated and surrounding unheatedareas was scanned using the red and NIR lasers of theLDI. The length of time for TcPO2 to reach steadystate ranged from 11 to 117 min, varying by site andregion. The region over plantar surfaces took signif-icantly longer to reach steady state in comparisonwith all other regions on the leg and foot. The timelapse between the removal of the probes and thecompletion of the LDI scanning averaged 1 min. Thetypical area scanned in the foot measured 8 � 4 cmand, in the anterior and lateral leg, 14 � 4 cm. Thedistance between the LDI laser head and the skin siteswas adjusted from 30 to 50 cm to provide the areanecessary for scanning.

RetestingTo assess the test–retest reliability of LDI mea-

surements, 15 subjects (five from each age group)were retested. Retest subjects were randomly cho-sen by paper lots. Each lot identified the name ofthe subject, and, together, they formed a pool of allstudy participants. As each lot was drawn blindlyfrom the pool by an investigator, it was assigned anumber from 1 to 60 in order of its draw and wasthen placed into one of the study’s three age groupsreflecting the age of the selected subject. After alllots had been drawn and assigned, from each agegroup, the first five subjects determined by theorder of their draw were selected for retesting. If,for any reason, a subject was unavailable for retest-ing, the next subject in order of assignment fromthe same age group was selected, until five subjectsfrom each age group were retested. Time elapsedbetween test and retest ranged from 2 wks to 6mos. Skin perfusion was assumed to be a stablephysiologic phenomenon in our healthy subjects.

Data AnalysisEach site was scanned with the heating probe

on the skin and immediately after the heatingprobe had been removed from the skin. To deter-mine flux values, the image produced by LDI wasviewed on a computer monitor. LDI software pro-

264 Kunkel et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 4

duced a color-coded image of skin perfusion and ablack and white image of optical density that wasused to identify the limits and position of theheating probes. A polygonal region of interest(ROI) (greatest diameter � 10 mm) was manuallydrawn around the probe image, saved, and thenused on the perfusion image to collect perfusiondata (mean of all pixels’ flux values within the ROI)at the heated site (location of the heating probe)and at three adjacent nonheated sites (Fig. 1).Images with streaks of high flux values withinnonheated areas, presumed to be blood vessels,were avoided. Flux ratios were calculated as themean flux in the heated ROI divided by the meanflux in the three adjacent unheated ROIs. As such,the flux ratios represent the number of multiples ofperfusion resulting from maximal vasodilation ofskin above the nonheated condition.

In Figure 1, the measurement of flux in theheated zone may seem to be inaccurate as theimage is clearly saturated over the heated region.The Moor LDI system frequently displays the pixelswithin the heated area as “white,” representing thehighest range of flux values. However, the heatedarea consists of numerous pixels of varying fluxmagnitudes; therefore, white areas are not neces-sarily fully saturated.

Because this study aimed to determine normalvalues of skin flux in response to skin warming,TcPO2 electrodes were used to topically heat theskin to 44°C (111°F). Heating to 44°C causes atemperature-dependent microstructural changefrom solid to liquid of the lipid phase of the stra-tum corneum,24 thereby optimizing skin perme-ability to O2 and enhancing vasodilation. LDI fluxvalues of nonheated areas do not benefit from theeffect of heat on skin permeability to O2 and vaso-dilation. The flux ratio of heated to nonheatedareas measures the reserve capacity of the skincapillaries to vasodilate when challenged by themetabolic stress of heating. We anticipated that theskin of PAD subjects would exhibit impaired abilityto vasodilate under such a stressful circumstance.

Alternative stressors might also include exercise orulceration that increase the metabolic and circula-tory demand on the skin tissue, but topical heatingis the safest and most feasible.

StatisticsThe statistical software NCSS (Number Cruncher

Statistical System 2001, NCSS, Inc., Kaysville, UT)was used for all calculations, and alpha was preset at0.05 for all statistical hypothesis testing. Descriptivestatistics (mean � SD) were calculated for flux ofheated and nonheated sites, flux ratios, and TcPO2.For the dependent variables of mean flux of heatedand nonheated areas and flux ratio, a four-factormixed-model analysis of variance (ANOVA) and posthoc Tukey–Kramer multiple comparison tests wereused to determine the main effects and interactionsfor skin site (ten sites), laser wavelength (red vs.infrared), age group (ages 20–39, 40–59, and 60–79)and gender (male vs. female). Additionally, separatethree-factor mixed-model ANOVAs for each laser (redvs. NIR wavelength) were used to determine maineffects of skin site, age, and gender, with Tukey–Kramer tests. For the dependent variable of TcPO2, aone-factor repeated-measures ANOVA was used todetermine differences among sites.

Two approaches were used for assessing test–retest reliability of flux ratios and TcPO2 values: (a)linear regression and Pearson correlation of pairedvalues of tests 1 and 2, and (b) Bland–Altman plotsof the difference between values obtained in test 2and test 1 as a function of the average of bothtests.25 Group values for each site were used in theregression analysis and individual values of testsfor the Bland–Altman approach. In addition, theNIR laser mean flux ratios from both tests werecompared, using paired t tests.

RESULTSDemographics

Subjects’ mean age and age ranges for each ofthe three groups and the breakdown of subjects by

FIGURE 1 Laser Doppler image showing a heated region of interest (center) and three nonheated regions ofinterest.


age group, gender, and ethnicity are shown in Table1. More female participants were found in group 3 incomparison with the other two, younger-agedgroups. Caucasian females comprised the largest sub-ject pool (35%; n � 21); the second largest were Asianmales (17%; n � 10).

Flux DataANOVA indicated significant effects of the skin

site and laser wavelength on the mean flux ofunheated and heated skin areas and on flux ratios.Age was also a significant factor, with mean � SEflux ratio values of the 60- to 79-yr-old group(8.9 � 0.17) significantly higher than the 20- to39-yr-old (8.1 � 0.17) and 40- to 59-yr-old (8.1 �0.17) groups. The mean flux ratio for all sites andgroups measured with the NIR laser (8.6 � 0.14)was significantly higher than with the red laser(8.1 � 0.14). No significant effect of gender onmean flux of unheated areas or flux ratios wasdetected. However, mean flux for heated areas forfemales was significantly higher (P � 0.004) thanfor males (males: 300.4 � 5.9 for males vs. 324.3 �5.4 for females).

Mean flux and flux ratio values from all sub-jects in each of the ten measured sites are depictedin Table 2 for the red laser and in Table 3 for theNIR laser types. For both laser types, plantar firstMTP and fifth MTP sites showed significantlyhigher unheated means in comparison with all legsites. Because the heated flux values of plantar siteswere not significantly greater than for any of theleg sites, plantar flux ratios were significantly lowerthan the ratios for the leg sites. No significantdifferences were found among flux ratios of the sixleg sites. Mean � SE flux ratios measured with theNIR laser for the three age groups are depicted inFigure 2.

Initial and final skin temperatures at test sitesshowed significant (P � 0.05) differences among theten skin sites. Before and after heating, foot temper-atures (unheated means 29.3–30.0°C; heated means34.9–36.4°C) were slightly lower than leg tempera-tures (unheated means 30.1–31.1°C; heated means37.3–38.0°C). Unheated flux and skin temperaturebefore heating correlated significantly but weakly(r � 0.226, P � 0.000), supporting previous find-ings23 that skin temperature is nonlinearly related to

TABLE 1 Subject demographics

Age Group, yrs 20–39 40–59 60–79

Mean Age, yrs 32.4 49.8 66.7

Numbers of subjects per gender, age group, and ethnic group

Gender Male Female Male Female Male Female Total

Caucasian 2 5 7 6 2 10 32African American 1 1 1 2 0 0 5Asian 5 2 1 2 4 4 18Pacific Islander 2 1 0 0 0 0 3Hispanic (non-Caucasian) 1 0 1 0 0 0 2Total per gender 11 9 10 10 6 14 60Total per age group 20 20 20 60

TABLE 2 Mean (�SD) red laser fluxes and ratios by sites for all subjects

Site Unheated Mean Flux Heated Mean Flux Mean Flux Ratios

Plantar first MTP 80.6 (79.0) 320.1 (197.7) 5.1 (2.54)Plantar fifth MTP 75.4 (67.4) 273.1 (187.2) 4.6 (2.46)Dorsal first MTP 59.0 (32.9) 418.0 (166.2) 8.5 (4.0)Dorsal fifth MTP 53.1 (30.2) 345.1 (156.0) 7.6 (3.7)Lateral distal leg 37.8 (18.1) 324.3 (122.5) 9.5 (3.5)Lateral middle leg 44.1 (21.0) 333.7 (120.6) 8.4 (3.2)Lateral proximal leg 48.6 (22.5) 370.9 (153.7) 8.5 (3.7)Anterior distal leg 41.0 (28.5) 356.8 (156.2) 10.3 (3.8)Anterior middle leg 40.8 (25.3) 320.0 (121.2) 9.1 (3.4)Anterior proximal leg 44.6 (31.1) 320.6 (115.5) 8.6 (3.3)

Unit of measurement for flux is arbitrary perfusion unit (PU). MTP, metatarsophalangeal joint.


perfusion. Heated flux and skin temperature afterheating did not significantly correlate (r � �0.017,P � 0.684). Hence, flux was not dependent on skintemperature after heating.

TcPO2 DataThe mean values of the steady-state TcPO2

before LDI scanning for each site are listed in Table3. No differences were found in these values amongleg sites, but mean values of the two plantar sites inthe foot were significantly lower than any othersite, and below those values given as a reference fordetection of critical limb ischemia. Moreover, 14and 18 out of 60 subjects had a TcPO2 of zero overthe first MTP and the fifth MTP sites, respectively.Linear regression comparison of the last recordedmean TcPO2 to mean heated flux magnitudeshowed no significant correlations for either laser

type at any site, except for the dorsal first MTP site(r � 0.268, P � 0.044).

ReliabilityFive subjects (25%) of each group (n � 15)

were selected randomly for reliability analysis. Onesubject’s retest data were invalid because of tech-nical difficulties, so the retest sample size was n �14. Regression analysis indicated high Pearson cor-relation coefficients between the first and secondtests (flux ratio � 0.96; TcPO2 � 0.98). The regres-sion slope was not significantly different from one,and the regression ordinate intercept was not sig-nificantly different from zero. The Bland–Altmanapproach indicated a lack of dependence of differ-ences between tests 2 and 1 on the average of tests2 and 1, with differences clustering around zero forboth variables (Figs. 3 and 4). Paired t tests to

FIGURE 2 Mean � SE flux ratios at each age group at each skin site. MTP, metatarsophalangeal joint; Ant,anterior; lat, lateral; prox., proximal; mid., middle; dist., distal.

TABLE 3 Mean (�SD) near-infrared laser fluxes and ratios and steady-state TcPO2 by site for allsubjects

Site Unheated Flu Heated Flux Flux Ratios TcPO2 (mm Hg)

Plantar first MTP 63.6 (48.2) 308.0 (195.8) 5.6 (2.9) 13.7 (18.1)Plantar fifth MTP 63.8 (54.4) 271.7 (189.5) 4.8 (2.1) 9.8 (13.1)Dorsal first MTP 50.9 (26.1) 358.6 (131.2) 8.1 (3.7) 56.7 (13.8)Dorsal fifth MTP 42.1 (19.6) 285.3 (111.8) 7.6 (3.5) 51.1 (11.4)Lateral distal leg 30.8 (11.7) 283.5 (85.3) 10.1 (3.7) 57.9 (12.1)Lateral middle leg 35.7 (13.7) 294.5 (78.1) 9.3 (3.6) 58.1 (12.2)Lateral proximal leg 37.6 (12.9) 309.9 (85.8) 9.0 (3.3) 55 (10.5)Anterior distal leg 29.4 (11.3) 297.5 (94.7) 11.0 (3.6) 55.8 (16.2)Anterior middle leg 29.4 (11.7) 263.1 (68.4) 9.7 (2.8) 51.7 (14.0)Anterior proximal leg 29.4 (10.1) 262.6 (80.8) 9.5 (3.1) 50.1 (14.5)

Unit of measurement for flux is arbitrary perfusion unit (PU). MTP, metatarsophalangeal joint.


compare means of the first and second tests foreach skin site indicated no significant differencesfor any of the sites or variables.

DISCUSSIONTo date, normal value parameters for the lower

extremity using the scanning LDI have not beenpublished. In our study, no consistent differencesbetween leg sites in heated or unheated skin flux orratios were found, but lower flux ratios were foundfor the plantar and dorsal foot sites compared withthe leg sites. This was attributed to higher meanflux values of unheated areas obtained on the plan-tar and dorsal sites coupled in a ratio calculation

with mean flux values after heating that were com-parable with those for the leg sites. This observa-tion could be explained by known pressure-inducedvasodilatation that is higher at the plantar areas ofthe foot; this has been recently reported as a pro-tective response to mechanically induced pressuresin the human skin.26

Few studies comparing laser wavelengths andperfusion values have been published to date. TheMoor LDI is capable of measuring flux using redand NIR simultaneously. Bray et al.27 report thatred and NIR wavelength photons provided similarblood-flow information in 20 patients with hyper-trophic burn scars. Their study reportedly repre-

FIGURE 3 The differences in flux ratios (near-infrared laser, NIR) between tests 1 and 2 are plotted against themeans of both tests for each site in 14 subjects in the reliability study. The slope and ordinateintercept of the linear regression of differences on means for all data points (solid line) were notsignificantly different from zero. The two dotted lines represent two SDs above and below the meandifference for all data points.

FIGURE 4 The differences in TcPO2 between tests 1 and 2 are plotted against the means of both tests for eachsite in all 14 subjects in the reliability study. The slope and ordinate intercept of the linear regressionof differences on means for all data points (solid line) were not significantly different from zero. Thetwo dotted lines represent two SDs above and below the mean difference for all data points.


sents multiple ethnicities and quantifies color dif-ferences in burn scars.

Our study shows a significant overall differ-ence between measured flux values for red and NIRwavelengths. We did not include subjects with anyskin diseases, scarring, edema, or lesions in ourROI. The NIR laser consistently showed lowermean flux values at unheated and heated areas.This gave overall higher values for flux ratios inmost sites except for the dorsal foot. However,within each laser type, the flux ratio trends weresustained for each site. Even though subjects fromseveral ethnic groups were represented in ourstudy, there was a preponderance of Caucasiansand Asians. Previous studies on laser Doppler flow-metry and race have shown no significant differ-ences when comparing Caucasians with Hispanics,or Caucasians with blacks.28 Skin pigment was notquantified in our study, but most of the partici-pants had fair skin. Our data suggest that eitherlaser would be adequate for the evaluation of skinperfusion, provided that the corresponding normalvalues are used for comparison. We have thus pro-vided normal values for red and NIR laser types.Indications for a particular laser type for a clinicalsituation have not been determined, and furtherstudies evaluating effects of skin pigmentation onflux values are warranted.

We anticipated a decrease in heat vasodila-tion, as measured by flux ratio, with advancingage. However, only some sites showed statisti-cally significant differences in flux ratios amongthe three age groups. Although the differenceswere small, the older group in each case had thehighest values. These results suggest thathealthy aging per se may not be a contributingfactor to impaired cutaneous circulation. Previ-ous studies also support this finding.28 Age mayhave an effect in slowing vasodilation, but thiswas not evaluated in our study.28

Previous research regarding gender has shownmen to have slightly higher skin microcirculationthan women, but this has been controversial andincompletely studied.12,29 In this study, no significantdifferences in flux ratios between genders were ob-served. There were more women in our study thanmen because of the predominance of women in thehighest age group of 60–79 yrs. The younger two agegroups were closely matched for gender.

Repeat testing on 25% of our total subjectsample showed no significant difference in meanflux ratios or TcPO2 values between two tests. Al-though repeat tests were completed weeks tomonths after initial testing and by different testers,values were consistent across all ten sites. Thisprovides confidence that the data presented areindependent of intertester variation. The results weobtained with TcPO2 evaluated with the Bland–

Altman approach are remarkably similar to thosepreviously obtained by de Graaff et al.30

In this study, attempts were made to controlfor circulatory variations that could have beencaused by physical and mental stressors that dis-turb metabolic and circulatory steady states. It isconceivable that subjects had subclinical or undi-agnosed diseases that were not reported on theirintake questionnaires. However, no compensationwas associated with participation, making it un-likely that information would have been manipu-lated for secondary gain.

Other limitations to this study include thesmall subject sample, which was relatively morehomogeneous than the population at large. Most ofour subjects were women of Caucasian and Asianancestry and were under the age of 60. If prospec-tive subjects were not aware of or honest abouttheir health status, it was also possible that somesubjects with subclinical PAD may have been in-cluded in the sample.

It is interesting to note that in many subjects,TcPO2 levels approached zero at the two plantarfoot sites, and the means of TcPO2 in these siteswere below the level generally considered an indi-cation of severe ischemia. This observation invali-dates the use of TcPO2 for the study of ischemia onthe plantar surface of the foot, the most frequentsite of lower-extremity ulceration in PAD. In con-trast, LDI flux of unheated areas was highest in theplantar sites, and the ratios were always positive.LDI flux determination could replace TcPO2 in thestudy of skin blood flow of the feet.

LDI values did not correlate with TcPO2 valuesin healthy subjects at any of the ten tested sites.This contradicts findings in studies of subjects withPAD,31 and it may be explained by the fact thatlocal blood flow correlates with TcPO2 only at lowlevels of skin perfusion.32 The variations observedin TcPO2 values in our results could be attributableto diffusion factors rather than perfusion levels.The low TcPO2 levels recorded at the plantar footsurface could be explained by the greater thicknessof the epidermis in this area, averaging 0.5–1.4 mmcompared with 0.04–0.1 mm for the leg.33

The perfusion values reported in this studyfrom healthy adults indicate that the variabilityassociated with each mean value is large, withcoefficients of variability ranging from 29 to 52%for LDI flux ratios and from 29 to 134% for TcPO2.Therefore, comparison of these values againstthose from a specific patient with compromisedlower-limb perfusion may be problematic. Moremeaningful normal values may be those measuredon the clinical patient near heart level. Within thesame patient, these values should only be com-pared with more distal compromised values, pro-vided that the same type of laser was used in both


measurements. Although our results do not specifythe normal perfusion values for PAD patients, sta-tistical norms for healthy adults are presented. Ourresults suggest that LDI flux ratios �2 SDs belowthe mean may indicate statistical abnormality. Fur-ther research is necessary to correlate clinically thesigns/symptoms and severity of PAD with the mag-nitude of LDI flux ratio.

CONCLUSIONNormative values for hyperemic skin blood-

flow responses to thermal stress derived from 60healthy adults in three age groups were presentedas mean flux and mean flux ratios for ten lower-extremity sites. Flux ratios at all sites were inde-pendent of gender, but they were higher in the 60-to 79-yr-old age group. The leg presented higherflux ratios and lower baseline flux values than theplantar and dorsal foot sites. Consistent and smalldifferences between red and NIR wavelength lasersat the leg sites indicate that either laser may beclinically acceptable if contrasted to respective nor-mative values. The results from this study providea reference of normal values that can be comparedand contrasted with findings from future studies ofpatients with known PAD in the legs and feet. Thisstudy also shows that there is no correlation be-tween flux and TcPO2 values in healthy individuals,indicating that the source of variance for the twomethods may be based on different processes.Thus, TcPO2 cannot be used as an estimate ofperfusion in subjects without ischemia. Becausethe values of flux of heated areas recorded by LDI inthe group with a maximal heating time of 15 minswas not different from longer times, a period of 15mins is considered appropriate for evaluation ofskin perfusion with this methodology. More re-search is needed to determine LDI values in pa-tients with ischemia, if such data can help deter-mine levels of amputation consistent withsuccessful wound healing and rehabilitation out-come.

ACKNOWLEDGMENTSWe gratefully acknowledge the Physical Medi-

cine and Rehabilitation Service at the VA West LosAngeles Healthcare Center for use of facilities, staff,and supplies.

REFERENCES1. Collins TC, Beyth RJ: Process of care and outcomes in

peripheral arterial disease. Am J Med Sci 2003;325:125–34

2. Lassen NA, Holstein P: Use of radioisotopes in assessmentof distal blood flow and distal blood pressure in arterialinsufficiency. Surg Clin North Am 1974;54:39–55

3. Ohta T: Noninvasive technique using thallium-201 forpredicting ischaemic ulcer healing of the foot. Br J Surg1985;72:892–5

4. Moore WS, Henry RE, Malone JM, Daly MJ, Patton D,Childers SJ: Prospective use of Xenon [Xe-133] clearancefor amputation level selection. Arch Surg 1981;116:86–8

5. Karanfilian RG, Lynch TG, Zirul VT, Padberg FT, Jamil Z,Hobson RW: The value of laser Doppler velocimetry andtranscutaneous oxygen tension determination in predict-ing healing of ischemic forefoot ulcerations and amputa-tions in diabetic and nondiabetic patients. J Vasc Surg1986;4:511–6

6. Fairs SLE, Ham RO, Conway BA, Roberts VC: Limb perfu-sion in the lower limb amputee—a comparative studyusing a laser Doppler flowmeter and a transcutaneousoxygen electrode. Prosthet Orthot Int 1987;11:80–4

7. Gebuhr P, Jorgensen JP, Vollmer-Larsen B, Nielsen SL,Alsbjorn B: Estimation of amputation level with a laserDoppler flowmeter. J Bone Joint Surg Br 1989;71:514–7

8. Lantsberg L, Goldman M: Laser Doppler flowmetry, trans-cutaneous oxygen tension measurements and Dopplerpressure compared in patients undergoing amputation.Eur J Vasc Surg 1991;5:195–7

9. Padberg FT Jr, Back TL, Hart LC, Franco CD: Comparison ofheated-probe laser Doppler and transcutaneous oxygen mea-surements for predicting outcome of ischemic wounds. J Car-diovasc Surg (Torino) 1992;33:715–22

10. Pape SA, Skouras CA, Byrne PO: An audit of the use of laserDoppler imaging (LDI) in the assessment of burns of inter-mediate depth. Burns 2001;27:233–9

11. Stucker M, Heese A, Hoffmann K, Rochling A, Altmeyer P:Precision of laser Doppler scanning in clinical use. Clin ExpDermatol 1995;20:371–6

12. Algotsson A, Nordberg A, Winblad B: Influence of age andgender on skin vessel reactivity to endothelium-dependentand endothelium-independent vasodilators tested with ion-tophoresis and a laser Doppler perfusion imager. J GerontolA Biol Sci Med Sci 1995;50:M121–7

13. Ferrell WR, Balint PV, Sturrock RD: Novel use of laserDoppler imaging for investigating epicondylitis. Rheuma-tology (Oxford) 2000;39:1214–7

14. Fullerton A, Stucker M, Wilhelm KP, et al: Guidelines forvisualization of cutaneous blood flow by laser Doppler per-fusion imaging. A report from the Standardization Group ofthe European Society of Contact Dermatitis based upon theHIRELADO European community project. Contact Derma-titis 2002;46:129–40

15. Malone JM, Anderson G, Lalka SG, et al: Prospective com-parison of noninvasive techniques for amputation level se-lection. Am J Surg 1987;154:179–84

16. Rhodes GR: Uses of transcutaneous oxygen monitoring inthe management of below-knee amputations and skin en-velope injuries (SKI). Am Surg 1985;51:701–7

17. Lalka SG, Malone JM, Anderson GG, Hagaman RM, McIn-tyre KE, Bernhard VM: Transcutaneous oxygen and carbondioxide pressure monitoring to determine severity of limbischemia and to predict surgical outcome. J Vasc Surg1988;7:507–14

18. Caspary L, Creutzig A, Alexander K: Comparison of laser-Doppler-flux and tcPO2 in healthy probands and patientswith arterial ischemia. Adv Exp Med Biol 1987;220:235–40

19. Allen PI, Goldman M: Skin blood flow: a comparison oftranscutaneous oximetry and laser Doppler flowmetry. EurJ Vasc Surg 1987;1:315–8

20. Ogrin R, Darzins P Khalil Z: Age-related changes in micro-vascular blood flow and transcutaneous oxygen tensionunder basal and stimulated conditions. J Gerontol 2005;60:200–6

21. Abbot NC, Ferrell WR, Lockhart JC, Lowe JG: LaserDoppler perfusion imaging of skin blood flow using redand near-infrared sources. J Invest Dermatol 1996;107:882–6


22. Essex TJ, Byrne PO: A laser Doppler scanner for imagingblood flow in skin. J Biomed Eng 1991;13:189–94

23. Sparks HV: Skin and muscle, in Johnson PC (ed): PeripheralCirculation. New York, NY, John Wiley & Sons, 1978, pp193–230

24. Beran AV, Tolle CD, Huxtable RF: Cutaneous blood flow andits relationship to transcutaneous O2/CO2 measurements.Crit Care Med 1981;9:736–41

25. Lexell JE, Downham DY: How to assess the reliability ofmeasurements in rehabilitation. Am J Phys Med Rehabil2005;84:719–23

26. Koitka A, Abraham P, Bouhanick B, Sigaudo-Roussel D,Demiot C, Saumet JL: Impaired pressure-induced vasodila-tion at the foot in young adults with type 1 diabetes.Diabetes 2004;53:721–5

27. Bray R, Forrester K, Leonard C, McArthur R, Tulip J, LindsayR: Laser Doppler imaging of burn scars: a comparison ofwavelength and scanning methods. Burns 2003;29:199–206

28. Eun HC: Evaluation of skin blood flow by laser Dopplerflowmetry. Clin Dermatol 1995;13:337–47

29. Stucker M, Steinberg J, Memmel U, Avermaete A, HoffmannK, Altmeyer P: Differences in the two-dimensionally mea-sured laser Doppler flow at different skin localisations. SkinPharmacol Appl Skin Physiol 2001;14:44–51

30. de Graaff JC, Ubbink DT, Legemate DA, de Haan RJ, JacobsMJ: Interobserver and intraobserver reproducibility of pe-ripheral blood and oxygen pressure measurements in theassessment of lower extremity arterial disease. J Vasc Surg2001;33:1033–40

31. Krishnan STM, Rayman G: The LDIflare: a novel test ofC-fiber function demonstrates early neuropathy in type 2diabetes. Diabetes Care 2004;27:2930–5

32. Huch R, Huch A, Lubbers DW: Transcutaneous PO2. NewYork, NY, Thieme-Stratton Inc., 1981, pp 41–50

33. Southwood WFW: The thickness of the skin. Plast ReconstrSurg 1956;15:423–9

APPENDIX 1.


Authors:Margaret G. Stineman, MD, FACRMRichard N. Ross, MSGreg Maislin, MS, MANicholas Marchuk, BSSerina Hijirida, OTDMark G. Weiner, MD

Affiliations:From the Departments of PhysicalMedicine and Rehabilitation (MGS,RNR, NM), Medicine (GM), GeneralInternal Medicine (MW), LeonardDavis Institute of Health Economics(MGS), Clinical Epidemiology Unit,Center for Epidemiology andBiostatistics (MGS), Institute onAging (MGS), School of Nursing(GM), and University of PennsylvaniaHealth System (SH), University ofPennsylvania, Philadelphia,Pennsylvania; and BiomedicalStatistical Consulting, Wynnewood,Pennsylvania (GM).

Correspondence:All correspondence and requests forreprints should be addressed toMargaret G. Stineman, MD, 101Ralston Center, 3615 Chestnut Street,Philadelphia, PA 19104-2676.


DOI: 10.1097/PHM.0b013e3180383acb

Recovery Preference ExplorationAnalysis of Patient Feedback After Imagined Scenarios

ABSTRACT

Stineman MG, Ross RN, Maislin G, Marchuk N, Hijirida S, Weiner M: Recoverypreference exploration: analysis of patient feedback after imagined scenarios.Am J Phys Med Rehabil 2007;86:272–281.

Objective: To present a new assessment approach, referred to asrecovery preference exploration (RPE), for exploring the personal mean-ing of functional loss and recovery. RPE determines how people wouldchoose to recover from profound disability if they could control thatrecovery.

Design: Twenty-six patients with a variety of medical conditions andone or more limitations in the functions being addressed were recruitedfrom an inner-city ambulatory care clinic. The patients imagined recoveryfrom 15 functional limitations, beginning with severe problems in allfunctions and ending with complete independence. Individual-specificpreferences for recovery in each function were calculated relative to theother 14 and were submitted for principal components analyses.

Results: Imagined difficulty in toileting and with depression were mosttroubling. Principal components analyses identified trade-off choicesamong domains of physical, psychological, and social functioning. Somepeople valued physical independence above psychological well-being orsocial abilities. Others had opposite patterns. Patients’ narrative explana-tions, when triangulated, were consistent with their preferences.

Conclusion: RPE makes visible the highly personal nature of feelingsabout ability and disability. Our results may help guide the selection ofrehabilitation interventions in ambulatory care.

Key Words: Rehabilitation, Quality of Life, Decision Support Techniques, Physician–Patient Relations


RESEARCH ARTICLE

Quality of Life

The medical anthropologist Gay Becker1 writes,“Discontinuities in life force individuals to recon-struct their biographies . . . a health biography in-cludes highly personal and particularized memo-ries, conceptions and meanings of events, pain, anddiscomfort, experiences of past illnesses, andhealth-care experiences. . . . Identity is fluid, notstatic . . . different illnesses have different effectson identity.” The biomedical model places its high-est value on those scientifically reproducible out-comes that are perceived as most useful for guidingpractice. However, complete reliance on such datamay disconnect care from patients’ perspectivesand values. There is insufficient time during thetypical patient encounter to extract health biogra-phies. Consequently, standardized measures of func-tion and health-related quality of life (HRQL) haveemerged to quantify the influence of illness onoverall well-being. Such measures typically expressa patient’s level of physical, mental, and socialfunctioning.2–5

Measuring physical, mental, and social func-tioning6,7 alone is insufficient when attempting tounderstand implications to well-being. A man candescribe moderate difficulties moving inside hishome, but this says nothing about how he seessuch an experience influencing his quality of life.Theoretically, each HRQL domain has both objec-tive (status related) and subjective (value laden)qualities.3,6–9 Although objective status is com-monly addressed in rehabilitation practices andresearch, the subjective meaning of that status istypically not addressed because it is intrinsicallymore abstract. Subjective meaning is described,but it is left unmeasured.

We introduce an innovative procedure calledrecovery preference exploration (RPE) as a processfor visualizing the subjective meaning of functionalstatus. RPE works by inducing a state akin todecentering. Decentering applied in social psychol-ogy and family therapy is a technique throughwhich a person is encouraged to set aside his or herown point of view and take the point of view ofsomeone else in a completely different circum-stance. In family therapy, decentering is intendedto help one member understand and empathizewith another member of the family. In contrast,with RPE, the patient is asked to keep his or herown point of view but to imagine what life would belike if he or she had complete disability accordingto a specified set of functional activities. This pro-cess of partial decentering is intended to inspireempathy—not necessarily for others, but for one-self.10 After imagining complete disability, the pa-tient ranks each activity according to how he or shewould want to recover and provides rationale forthe rankings.

We address the following questions throughRPE: do the value-laden constructs of meaningful-ness reflect the classical physical, psychological, andsocial status–related domains of HRQL? To what de-gree are concepts of meaningfulness unique to eachindividual? Finally, we discuss how RPE might beapplied to enhance the depth of discussion abouttherapeutic decision making in general ambulatorycare and rehabilitation practices.

METHODOLOGYRPE ProceduresTheoretical Framework

Reality is constituted by attributing meaningto life experiences, partly through interacting withothers. As authoritative figures, physiatrists andother rehabilitation professionals are, rightly orwrongly, in a position to influence a patient’s re-action to the functional sequelae of illness. In-formed by clinical training and practice, the clini-cian’s viewpoint may be at odds with the patient’sview. By increasing the depth of discourse betweenpatients and clinicians,11 the exploration of recov-ery preferences moves outside professional con-cepts to extract the implications of functional lossand recovery from the patient’s vantage point.

RPE is an emerging therapeutic tool with a foun-dation in concepts of phenomenology.12,13 It is in-tended to clarify the meaning of disability and symp-toms as phenomena experienced or observed in thelived and highly personal state. RPE is derived fromthe first part of the two-part Features ResourceTrade-Off game, which was developed originally as aresearch tool.14 The original game adopted conceptsfrom economic theory and applied them to discoverhow groups of people sharing some trait would wantto recover, assuming complete disability and the abil-ity to control the trajectory of their recoveries. Theset amount of recovery allowed at any point in theprocedure is analogous to the number of dollars avail-able for spending, assuming limited resource avail-ability and the need to make choices between items.The alternative functions or symptoms listed for im-provement are analogous to alternative products onemight choose to buy. RPE forces explicit choices ofimagined recovery among functions as diverse astoileting and making friends, yielding utilities foralternative functional states. In contrast to the Fea-tures Resource Trade-Off game, RPE is being de-signed for application by single individuals and forclinical applications.15 Through the clinician-admin-istered RPE procedure, the patient displays his or herpreferred recovery choices. It is intended for applica-tion with any set of health status, functional status, orsymptom-related questions.

April 2007 Recovery Preference Exploration 273

Measures of Functional StatusApplicable to any ordinal functional status or

health status measure, RPE is applied here to 15questions derived from the National Health InterviewSurvey.16,17 These questions were selected becausethey are self-reported, applicable to a variety ofchronic health conditions, intended to be relevant forperson dwelling in the community (appropriate tooutpatient care), and because they describe func-tional concepts that are generally accepted as impor-tant. They include a series of six physical, five psy-chological, and four social functions fundamental toHRQL (see Fig. 1 for questions).

Each function was addressed according to thefollowing levels of difficulty: total assistance/severe,a lot of difficulty/frequent problem, some difficulty/occasional problem, and no difficulty/no problem.Originally dichotomous, response levels for someof the psychological and social functioning do-mains were expanded so that all measures wouldinclude four levels.

SubjectsWe enrolled an ambulatory care population

with disabilities related to chronic illness in aneffort to gain insight into the meaning of disabilityamong those living in the community. Subjects

were approached during routine visits to an inner-city ambulatory care clinic and were screened forinclusion. Inclusion criteria were chronic healthcondition(s) causing at least one functional limita-tion or symptom (see Table 1) to be addressed andthe ability to read and understand simple writteninstructions. The project was approved by the Uni-versity of Pennsylvania institutional review board.All patients signed and dated an informed consentdocument.

FIGURE 1 Game board: patients dem-onstrate their preferred pat-terns of recovery by sequen-tially numbering the opensquares.

TABLE 1 Perceived functional limitationsamong study patients

n (%)

ADL difficulties 20 (76.9)IADL difficulties 26 (100)Physical limitations 22 (84.6)Difficulty with hand functions 3 (11.5)Difficulty reaching overhead 8 (30.8)Difficulty with social functions 9 (34.6)Psychological difficulties 25 (96.2)Communicating so family and friends can

understand9 (34.6)

ADL, activities of daily living; IADL, instrumental activ-ities of daily living.

274 Stineman Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 4

The RPE ProcedurePatients were given verbal and written defini-

tions of the 15 limitations and were asked to recordtheir status according to each function. They werethen provided an RPE board (Fig. 1). The board listedthe 15 functions along its left side, with decreasingseverity levels moving from left to right across thetop. The RPE procedure was administered by a phy-sician and several assistants to groups of five to eightsubjects. It took about 30 mins.

Patients were asked to “imagine that you havea serious health condition that makes it impossiblefor you to get to and use the toilet, get in and outof a chair and bed, move inside the house, etc.” (Allitems were read.) “Recovery will be complete butwill take many years. How would you choose torecover if you could control how you recover?”Essentially, patients were instructed to pretend ini-tially that they were completely disabled in all 15National Health Interview Survey questions. Whendeciding on what recovery choices to make, pa-tients were asked to “make moves in a way thatwould maximize your ability to perform the activ-ities that you see as most essential and reduce thesymptoms that are most troubling.”

After this introduction, patients were instructed:“starting on the left-hand side, select that limita-tion or symptom you would most like to reduce.Place a ‘1’ in the first column next to that item inthe ‘moderate’ difficulty column. This indicatesimagined improvement from ‘severe to moderatedifficulty.’” After the patient had selected the firstmove, the investigator stated, “now, imagine thatyou still have moderate difficulty with (the selectedfunction), and you still need total assistance in allthe other functions. Select the function you wouldnext like to recover. You may choose to imaginemore recovery in (the first function), or select adifferent function. Place a ‘2’ in the correspondingbox . . .” This continued until all functions wereimagined as being no problem. The patient ex-pressed his or her desired sequence of recovery bynumbering the items and levels on the board. Atthe conclusion of the ranking procedure, each pa-tient was asked to explain why he or she hadchosen that particular recovery sequence.

Preference WeightsA preference weight is calculated from the move

numbers and represents the relative value the indi-vidual places on recovering a particular functioncompared with the other 14 functions. For each func-tion, the unit of measurement was improvement inone level (i.e., one move). There were 45 moves orchoices. Preference for each function was the inversesum of the move numbers for that item, multipliedby 100. This yielded a set of 15 weights corresponding

to each item for each individual. An attribute forwhich points were spent early in the procedure isassumed to have more value than one left for later.For example, an item with move numbers 1, 7, and 8would be more valuable than one with move numbers6, 12, and 23.

Exploring the Individual Nature of MeaningWe plotted the 15 preference weights for each

function. The distribution of each person’s prefer-ence weights across the 15 conditions is referred tohis or her “value ruler.”15

Physical, Psychological, and Social Domainsof Functional Meaning

Principal components analysis using SAS wasapplied to see whether simplified summary indicesof preferences could be derived from the 15 indi-vidual preference weights. Principal componentsanalysis was applied to discover the broad under-lying dimensions of meaningfulness. We expectedmeaningful and interpretable principal compo-nents reflecting the physical, psychological, andsocial constructs of HRQL. Principal componentsanalysis18,19 is a multivariable statistical techniquethat linearly transforms an original set of variablesinto a substantially smaller set of uncorrelated in-dices that capture most of the variance (and co-variance) present in the original items. Its goal is toreduce the dimensionality of the original data18

without substantial loss of information, making iteasier to discern important patterns in the ob-served data. A set of three indices would provide asimpler summary of preferences that might be eas-ier to understand than the set of 15 individualutilities. The first principal component of a multi-

TABLE 2 Patient characteristics

n (%)

Age, yrs18–44 3 (12)45–65 13 (52)65–69 2 (8)70–74 2 (8)75� 5 (20)

GenderFemale 16 (61.5)

RaceWhite 10 (38.5)Black 15 (59.7)Other 1 (3.9)

Perceived healthExcellent 1 (3.9)Very good 4 (15.4)Good 12 (46.2)Fair 4 (15.4)Poor 5 (19.2)


variate vector is simply the linear combinationwith maximum variance. The second principalcomponent is the linear combination, orthogonal(i.e., uncorrelated) to the first that maximizes re-sidual variance, and so forth. Because principalcomponents reflect maximum variance in a multi-dimensional space, they are ideally suited to displaylocation in a high (e.g., 15 dimensional) spaceusing a reduced number of dimensions (e.g., theanticipated three dimensions). This analysis wasused to develop the typology trade-off space thattheoretically makes it possible to see how the in-dividual patient’s preferences are balanced acrossthe physical, psychological, and social domains.

RESULTSThe majority of patients were adults younger

than 65 yr of age. More than half were African Amer-ican (Table 2). Diagnoses listed by the patients in-cluded heart failure, short-bowel syndrome, stroke,multiple myeloma, diabetes, thyroid disorders, osteo-porosis, end-stage renal disease, lower-extremity am-putation, Parkinson disease, legal blindness, manicdepression, paranoid schizophrenia, depression, bloodclots, bowel resection, asthma, chronic obstructive pul-monary disease, and others. All had difficulties with theinstrumental activities of daily living. Most had physicallimitations, and many described psychological difficul-ties (Table 1).

The Individual Nature of MeaningFigure 2 superimposes all 26 patient value

rulers connecting each patient’s pattern of desiredrecovery as a series of confusing interconnectedpoints. It illustrates complex and wide scatter, high-lighting the individual nature of meaning. Functionswith relative values above ten represent the mostvaluable, and those below five represent the leastvaluable. Discordance between patients’ preferred re-covery sequences and physiologically plausible recov-ery patterns highlights the distinctions between sta-tus and the value placed on achieving that status.Independence in toileting, absence of depression, andcommunication with family members were the mostvalued, and learning new things, making new friends,and getting along with people were the least val-ued among the physical, psychological, and so-cial functions. Nearly all patients considered com-munication with family members more importantthan communication with people outside the fam-ily. In sharp contrast to Figure 2, Figure 3 shows“value rulers” that represent two separate patients’utilities for each of the items.

Physical, Psychological, and SocialDomains of Functional Meaning

When subjected to principal components anal-ysis, the 15 preference weights expressed two com-ponents (Table 3). The first component reflected

FIGURE 2 The personal meaning ofimagined functional loss andrecovery is extremely vari-able. Each series of pointsconnected by its line showsthe relative value of eachfunction to the individualshowing the highly personalnature of preference. Toilet-ing, depression reduction,concentration, and fear reso-lution were of greatest relativeimportance to the largestnumber of people. In contrast,preference for learning newthings, making new friends,getting along with people, andcommunicating outside thefamily were the least strong.


trade-offs between social and physical function.Social function items had large negative weights,and physical functions had large positive weights.

The second component reflected trade-offs betweenpsychological and social function recovery. Psycho-logical function items had predominantly negativeweights, and social function items had large positiveweights. Together, these two linear combinationsproduce scores that express individuals’ preferencesrelative to their desires to be able to perform physical,psychological, and social functions.

We computed summary scores from the prin-cipal components analysis for each patient. As aconsequence of the forced trade-offs, each principalcomponent index reflected the degree of preferencefor recovery in one domain vs. another. Each pa-tient’s scores were plotted producing a “preferencetrade-off space” that showed the relative degree towhich he or she would value one type of recoveryover a different type (Fig. 4). This space is a two-dimensional display translating the 15 functionalitems into two axes representing trade-offs (andthus relative preferences) between social (�) andphysical functions (�) along the y-axis; and trade-offs between psychological (�) and social functions(�) along the x-axis. This translates into recoverypreference across the three dominant dimensions

TABLE 3 Principal components analysis onutility weights

Principal Components

Attribute One Two

Toileting 0.341 �0.090Transferring 0.342 0.007Moving 0.343 �0.094Dressing 0.323 0.229Eating 0.258 0.251Bathing 0.353 0.187Depression �0.053 �0.433Concentration �0.097 �0.357Confusion �0.194 �0.319Fears �0.125 �0.250Learning �0.194 0.148Making friends �0.295 0.194Getting along with others �0.210 0.325Communication with family �0.261 0.210Communication with others �0.226 0.380

FIGURE 3 Value rulers demonstrating extreme patterns. The two patients’ value rulers demonstrate oppositepreferences. Patient U would choose to recover the physical functions before the psychological andsocial, whereas patient P would prefer psychological and social recovery.


in the following way: physical preference is highestat the point in the trade-off space where the y-axisis �3 and the x-axis is 0. Psychological preferenceis highest at the point where the x-axis is �3 andthe y-axis is 0. Social preference is highest at thepoint where the y-axis is �3 and the x-axis is �3.

A patient’s location in this space reflects his orher individual recovery strategy anchored on com-plete disability. The closer the plotted value is to theorigin (0, 0) of the coordinate axes, the more balancedthe preferred recovery. Cases d, s, n, and j within theinner circle show the most balanced patterns. Pa-tients whose plotted preferences are beyond the outercircle are outliers demonstrating extreme prefer-ences. The space suggests a continuum of recoverypreference across the three dimensions. Patients’ ver-bal descriptions of strategies for imagined recoverywere compared against their plotted position. Thenarrative content explaining choices provided by pa-tients with the most divergent preference patterns(those furthest from the origin) were consistent withspatial position. For example:

Patient u, a white female in her 60s who ratedher health as good, described having some difficultwith heavy work around the house, washing win-dows, walking a quarter of a mile, climbing tensteps, or standing for 20 mins. She showed themost dominant physical preference within the pref-erence trade-off space (Fig. 4) desiring the recoveryof physical activities of daily living consistentlyover all types of function. This individual consis-tently traded imagined recovery of social and psy-chological well-being for physical independence,

stating, “if you are physically independent, you canthen go to work on other things. Independence andquality of life are most important.”

Patient p, a woman with glaucoma and breastcancer in her 70s who rated her health as very good,described having difficulty doing heavy housework,lifting 10 lbs, reaching with her arms, and occasionalforgetfulness. She showed a pattern opposite that ofpatient u (Fig. 4), valuing psychological and socialfunctioning almost equally and placing relatively lessvalue on physical functioning. On the basis of expe-rience with relatives, this woman stated, “if I cancommunicate, I can tell them the physical things Ineed. Moving is no good if you can’t have a purpose.My brother . . . with Alzheimer disease can move fine.Doesn’t know where he’s going.” On the basis of herexperience with relatives, she was impressed withhow little physical abilities matter when psychologi-cal capacity “is adrift.” She stated, “my mother, aunt,and mother-in-law went to nursing homes. Mymother was physically fit with dementia. When myaunt went in, she had physical problems but wascompletely alert. My mother-in-law was physically fitand alert but had attitude problems. She didn’t wantto be there. My aunt is the one I would like to model.If I can communicate, I can tell them the physicalthings I need.”

Patient h, a woman in her 50s with diabetes,arthritis, hypothyroid, manic depression, panic at-tacks, and renal damage, described her health aspoor. She noted some difficulty with getting in andout of bed, preparing meals, managing money, shop-ping, doing housework, washing windows, doing

FIGURE 4 A plot of each individual’s re-covery preferences across thephysical, psychological, andsocial dimensions of function.The letters indicate patients’position relative to their indi-vidual recover strategy ac-cording to their forced trade-offs within the space. Physicalpreference is highest at y � 3and x � 0. Psychological pref-erence is highest at y � 0 andx � �3. Social preference ishighest at y � �3 and x � 3.The space is a continuumwhere individuals’ preferencesfor independence in one typeof function are rarely absoluterelative to the other type(s).


light cleaning, lifting 10 lbs, climbing ten steps,standing for 20 mins, and bending down to pick up anobject. She also noted some difficulty with concen-tration, stress, forgetfulness, confusion, fears, andcommunicating with others. She showed a dominantpsychological pattern (Fig. 4). She stated, “clearthinking is essential. Before you can do anythingabout these basic tasks, confusion has to be straight-ened out. Those phobias and fears keep you in, un-dermine everything.”

Patient z, a man in his 60s who had quadruplebypass surgery, operations on his knees, a numberof small strokes, uvula extraction, congestive heartfailure, carotid endarterectomy, a hearing aid, andrenal failure, valued physical and social functionsmore than psychological functions. He describedhaving some difficulty eating, getting in and out ofhis bed, managing money, doing heavy housework,washing windows, and picking up an object fromthe floor. He also noted some difficulty with con-centration, stress, confusion, and forgetfulness. Hisnarrative discussed the social implications of beingable to function physically. He stated, “if you can’tmove, how can you let friends in? When you arevery sick, it is your family that is most important.”

DISCUSSIONWe present preliminary evidence that con-

structs of meaningfulness reflect the classical physi-cal, psychological, and social domains of HRQL. Thissuggests that metrics of value can be reported along-side metrics of status.6 Within the physical, psycho-logical, and social functioning domains, patients’preferences for recovery from limitations in activitiesare highly personal and are relevant to individual lifecontexts and experiences. The differences in relativemeaning assigned to the same functional activities byindividuals suggest why some people with devastatingdisabilities perceive a high quality of life, whereasothers fall into states of hopeless despair. The mean-ing a person assigns to particular patterns of func-tional loss or recovery is distinct from his or hermeasured status. As value-laden quantities, recoverypreferences are driven by the contexts of a person’slife. Personal feelings, such as need for privacy andpersonal pride, often drive the value of being able touse the toilet without help. These drivers of meaningand utility are distinct from the physiologic driversthat determine return of the physical movements,coordination, and cognitive responses necessary toaccomplish independent toilet use. Consequently, attimes, the sequence of recovery preferences selectedby patients within or across domains will seem clin-ically incoherent and physiologically impossible.

The RPE procedure yields metrics distinctfrom functional status, questions about subjectivequality of life,20 and preference-based scores suchas those obtained from the Health Utility Index.21

Functional status includes directly observable ex-teriorized signs of the individual.22 Global ques-tions about quality of life ask how the individualviews life as a whole. The Health Utility Index,currently being translated into many languages,embodies community preferences, under the as-sumption that preference-based scoring functionsare robust across populations. In contrast to re-cording observable, exteriorized signs of functionalstatus, RPE assays the internalized representationof those exteriorized signs. Rather than addressinghow the individual views life as a whole, RPE fo-cuses on feelings about specific functional orhealth states relative to others. Instead of embod-ying community preferences, RPE operates underthe assumption that credible estimates of person-level subjective HRQL must reflect the individualviewpoint rather than population summaries.

RPE also differs from and complements othermeasures and procedures that are designed to in-clude the patient’s perspective in the rehabilitationprocess, such as the Participation Objective, Par-ticipation Subjective and the Canadian Occupa-tional Performance Measure. In performing Partic-ipation Objective, Participation Subjective, thepatient rates 26 activities according to these activ-ities’ importance to the patient’s perceived well-being.23 The value of each separate item is ratedindependently of the value of the others. Ratherthan being tied to specific activities, the patientRPE is a procedure that can be applied to any set ofquestions or functional status measures of rele-vance to the population being studied. In the Ca-nadian Occupational Performance Measure, thetherapist performs a semistructured interview witha patient to identify the problems in daily activitiesthat are most troubling to the patient, and then thepatient ranks these activities according to how im-portant they are to him or her.24 In contrast, RPEranks named activities taken from any available func-tional measure to yield a hierarchy of activity prefer-ences instead of ranking patient-generated concepts.

Developing standard approaches to incorporat-ing the patient’s perspective in care is important.There is tension in medicine between the pursuit offacts about disease and its resulting disabilities, andan understanding of the individual’s subjective per-sonal responses. Physicians and other cliniciansrecord the patients’ state of pathology and its func-tional consequences according to empirically delin-eated concepts. Facts about functional difficulty aredevoid of meaning. For people with chronic disabili-ties, the normal condition may no longer be a mean-ingful or realistic standard. Consequently, loss of“normalcy” typically requires a shift in self concept.

To reestablish a sense of well-being after dis-ability, the individual must experience a process ofembodiment25 by which different facets of his or


her altered identity are synthesized into a newwhole.26 The experience of partial decentering inthe performance of RPE could facilitate embodi-ment. As the patient thinks about alternative re-covery choices from the point of view of pretendingcomplete disability, he or she weighs the implica-tions of the disabilities he or she has, along withthose disabilities he or she might have had if thecondition were much more severe. As a standard-ized procedure, we believe RPE uncovers value-laden ideas and attitudes that could accelerate thepatient’s process of reconstructing his or her iden-tity. Additionally, the process of partial decenteringtends to reduce embarrassment by having the pa-tient pretend he or she has disabilities he or she doesnot have. Not having to admit particular disabilities,the patient can talk more freely about how he or shefeels about different aspects of functioning.

The structured patient–clinician discoursethat accompanies RPE encourages deep thoughtabout the consequences of functional loss and re-covery. The content essence is taken to representthe essential meaning of functional abilities anddisabilities within the patient’s life contexts at aparticular moment in time.13

The approach to treating disease and disabilitycan either be salutogenic (seeking the origins ofhealth) or pathogenic (seeking explanations for caus-es).27 We hope that exploring recovering preferenceswill encourage a salutogenic approach to rehabilita-tion and medicine by providing a cogent explanationof personal hopes for the avoidance of particular dis-abilities and desired patterns of recovery. RPE is be-ing designed as an interdisciplinary tool that we be-lieve will have applications in multiple settings.

RPE has many potential clinical applications.Patients’ exploratory recovery preferences may beexpressed either as separate utilities for each activ-ity, or broadly, through the summary physical,psychological, and social indices. Graphic displaysof individual utilities on a value ruler (Fig. 2) showhow a patient valued each activity compared with allothers. Alternatively, utilities expressed as physical,psychological, and social indices plotted within thetypology trade-off space (Fig. 4) provide a way tocompare a patient’s values against those of a largerpopulation. A highly ranked activity on the patient’svalue ruler for an activity he or she has difficulty withcan identify an important therapeutic target. Suchvalue ruler patterns might be applied through a per-son-centered planning approach28 to better under-stand the patient’s aspirations and attitudes aboutdisability recovery. Patient p, for example, reportedoccasional forgetfulness. According to her value ruler(Fig. 3), she identified reducing confusion as the mostimportant priority if she were to become completelydisabled. Discussing strategies for enhancing organi-zation and memory might prove meaningful.

In contrast to the value ruler, a patient’s positionwithin the typology trade-off space identifies thebroader domains of function he or she most valuesand provides insights into his or her reasoning. Pa-tient u (Fig. 3) valued physical ability over other typesof abilities. She saw the maintenance of physicalindependence as essential to achieving well-being inother areas. Although she was able to perform all thebasic physical activities included in this RPE proce-dure, she described difficulty with more challengingactivities, such as climbing steps. Formal functionalevaluation with the development of a home exerciseprogram to enhance endurance and improve currentfunction would represent a reasonable approach tolifestyle planning and enhanced well-being.

Utilities from RPE are not intended to super-sede professional judgment. However, they canhelp with the process of individualizing recoveryplans by revealing those aspects of functioning thatare most important to the patient. For example, aman with a spinal cord injury might have walkingas his highest utility. Although the physiatristknows that walking may be an impractical recoverygoal, he or she can use the knowledge of howimportant mobility is to this patient to set otherrehabilitation goals. For example, the patient’s re-habilitation plan may focus on teaching him to usemobility aids early in the rehabilitation process oron helping him identify home environment modi-fications that will allow him to regain mobilitywithin his environment. By using utilities fromRPE in combination with professional judgment, itmay be possible to create rehabilitation plans thatfocus on those activities or skills that are mostmeaningful to the patient.

Summary physical, psychological, and socialindices and individual item utilities establishedthrough RPE can also be used to assess patientprogress by weighting the importance of improve-ments in various functional skills. For example, ingoal attainment scaling, a goal attainment scalingscore is calculated using the weights assigned toeach goal area and the attained score for each goalarea.29 Recovery in an area with a higher weightwould have a greater effect on the goal attainmentscaling score than an equal amount of recovery inan area with a lower weight.

Because only a small number of patients haveexperienced RPE to date, findings must be consid-ered exploratory. Future work should test the sta-bility of the physical, psychological, and social do-mains of functional meaningfulness as well as thedegree to which recovery preferences are relativelystable vs. fluctuating.30 It will be key to address thedegree to which meaningfulness constructs varyacross different health and functional status mea-sures applied in various rehabilitation settingssuch as the Functional Independence Measure, the


standard assessment used in the inpatient rehabil-itation setting.31 Future work might also explorehow patient circumstances and status, includingvariables such as age, diagnoses, and disability,influence preferences.

We believe that RPE will prove particularly ther-apeutic for people being rehabilitated in the inpatientsetting after the sudden onset of severe disabilities.The capacity to imagine greater disability and to setpriorities depends on the ability to think abstractly.Consequently, it may be necessary to derive a lessdemanding procedure for those with strokes or otherconditions that cause cognitive impairment.

RPE, a procedure in its infancy, acknowledgesthat people with similar functional manifestationsof illness may experience them differently. Opti-mally, the practice of medicine, like qualitativeresearch, is individual and context rich.32 There is aplethora of quantitative tools to aid the practice ofmedicine, but few of these tools assay the innerworlds of patients. We believe that RPE could emergeas a clinical tool with the potential to help patientsmake more sense of their loss of function and toguide them toward new positive life trajectories, thusinstilling the rehabilitation process with even greatermeaning and enhanced outcomes.

ACKNOWLEDGMENTSThis research was supported in part by the Na-

tional Institutes of Health grant R21 HD045881 fromthe National Institute of Child Health and HumanDevelopment. The opinions of the authors are notnecessarily those of the supporting agencies. The firstauthor expresses sincere gratitude to Pamela M. Ristfor outstanding contributions to the revision of thispaper.

REFERENCES1. Becker G: Disrupted Lives: How People Create Meaning in

a Chaotic World. Los Angeles, CA, University of CaliforniaPress, 1997

2. Ware JE, Sherbourne CD: The MOS 36-item short-formhealth survey (SF-36). I. Conceptual framework and itemselection. Med Care 1992;30:473–83

3. Gill TM, Feinstein AR: A critical appraisal of the quality oflife measurements. JAMA 1994;272:619–26

4. Guyatt GH, Cook DJ: Health status, quality of life, and theindividual. JAMA 1994;272:630–1

5. Leplege A, Hunt S: The problem of quality of life in medi-cine. JAMA 1997;278:47–50

6. Testa MA, Simonson DC: Assessment of quality-of-life out-comes. N Engl J Med 1996;334:835–40

7. Covinsky KE, Wu AW, Landefeld CS, et al: Health statusversus quality of life in older patients: does the distinctionmatter? Am J Med 1999;106:435–40

8. Stineman MG: A model of health environmental integra-tion. Top Stroke Rehabil 2001;8:34–45

9. Dijkers M: Measuring quality of life: methodological issues.Am J Phys Med Rehabil 1999;78:286–300

10. Kogan SM, Gale JE: Decentering therapy: textual analysis ofa narrative therapy session. Fam Process 1997;36:101–26

11. Clark FA, Ennevor BL, Richardson PLA: Grounded theory ofoccupational storytelling and occupational story making, inZemke R, Clark FA (eds): Occupational Science the EvolvingDiscipline. Philadelphia, PA, F.A. Davis Company, 1996, pp 373–2

12. Redmond MV: A multidimensional theory and measure ofsocial decentering. J Res Pers 1995;29:35–58

13. Giorgi A: The phenomenological movement and research inthe human sciences. Nurs Sci Q 2005;18:75–82

14. Stineman MG, Maislin G, Nosek M, Fiedler R, Granger CV:Comparing consumer and clinician values for alternative func-tional states: application of a new feature trade-off consensusbuilding tool. Arch Phys Med Rehabil 1998;79:1522–9

15. Stineman MG, Wechsler B, Ross R, Maislin G: A method formeasuring quality of life through subjective weighting offunctional status. Arch Phys Med Rehabil 2003;84:S15–22

16. National Center for Health Statistics. Data File Documen-tation, National Health Interview Survey on Disability,Phase II, Adult File 1995. Hayattsville, MD, National Centerfor Health Statistics, 1998

17. National Center for Health Statistics. Data File Documen-tation, National Health Interview Survey on Disability,Phase I, Adult File 1994. Hayattsville, MD, National Centerfor Health Statistics, 1996

18. Dunteman GH. Principal components, in Lewis-Beck MS(ed): Quantitative Applications in the Social Sciences, vol69. Newbury Park, CA, Sage Publications, 1989, pp 7–8

19. Morrison DF: Applied Statistical Methods, chapter 8. Engle-wood Cliffs, NJ, Prentice-Hall, Inc., 1983

20. Brown M, Gordon WA, Haddad L: Models for predictingsubjective quality of life in individuals with traumatic braininjury. Brain Inj 2000;14:5–19

21. Shimoda S, de Camargo B, Horsman J, et al: Translationand cultural adaptation of Health Utilities Index (HUI) Mark2 (HUI2) and Mark 3 (HUI3) with application to survivors ofchildhood cancer in Brazil. Qual Life Res 2005;14:1407–12

22. World Health Organization. International Classification ofFunctioning, Disability and Health: ICF. Geneva, Switzer-land, World Health Organization, 2001

23. Brown M, Dijkers MP, Gordon WA, Ashman T, Charatz H,Cheng Z: Participation objective, participation subjective: ameasure of participation combining outsider and insiderperspectives. J Head Trauma Rehabil 2004;19:459–81

24. Eyssen ICJM, Beelen A Dedding C, Cardol M, Dekker J: Thereproducibility of the Canadian Occuptional PerformanceMeasure. Clin Rehabil 2005;19:888–94

25. Merleau-Ponty M: Phenomenology of Perception, 21 ed.Smith C, trans. New York, NY, Humanities Press, 1962

26. Iwakuma M: The body as embodiment: an investigation ofthe body by Merleau-Ponte, in Shakespear MC (ed):Disability/Postmodernity: Embodying Disability Theory.London, UK, Continuum Press, 2002, pp 76–87

27. Nilsson B, Westman G: The impact of patients’ influence onrecovery in a group of patients with dyspepsia. Fam Pract1999;16:515–21

28. Holburn S, Jacobson JW, Vietze PM, Schwartz AA, Sersen E:Quantifying the process and outcomes of person-centeredplanning. Am J Ment Retard 2000;105:402–16

29. Rockwood K, Howlett S, Stadnyk K, Carver D, Powell C,Stolee P: Responsiveness of goal attainment scaling in arandomized controlled trial of comprehensive geriatric as-sessment. J Clin Epidemiol 2003;56:736–43

30. Kraemer HC, Gullion CM, Rush AJ, Frank E, Kupfer DJ: Canstate and trait variables be disentangled? A methodological frame-work for psychiatric disorders. Psychiatry Res 1994;52:55–69

31. Hamilton BB, Granger CV, Sherwin FS, Zielezny M, TashmanJSA: Uniform National Data System for Medical Rehabilitation,in Fuhrer MJ (ed): Rehabilitation Outcomes: Analysis andMeasurement. Baltimore, MD, Paul H. Brookes, 1987

32. Gilchrist VJ, Engel JD: Qualitative research and clinicalcare. J Fam Pract 1995;41:229–30


Authors:Chia-Hsin Chen, MDXuhui Liu, MDMing-Long Yeh, PhDMao-Hsiung Huang, MDQihui Zhai, MDWalter R. Lowe, MDDavid M. Lintner, MDZong-Ping Luo, PhD

Affiliations:From the Department of PhysicalMedicine and Rehabilitation,Kaohsiung Medical UniversityHospital, College of Medicine,Kaohsiung Medical University,Kaohsiung, Taiwan (C-HC, M-HH);Department of Orthopedic Surgery,Baylor College of Medicine, Houston,Texas (C-HC, XL, M-LY, WRL, Z-PL);Biomedical Engineering Institute,Cheng Kung University, Tainan,Taiwan (M-LY); Department ofPathology, the Methodist Hospital,Houston, Texas (QZ); and Departmentof Orthopedic Surgery, The MethodistHospital, Houston, Texas (DML).

Correspondence:All correspondence and requests forreprints should be addressed to Zong-Ping Luo, PhD, Department ofOrthopedic Surgery, Baylor College ofMedicine, Physical Address: 6550Fannin Street, Suite 451, Houston,TX 77030.


DOI: 10.1097/PHM.0b013e31803215dc

Pathological Changes of HumanLigament After Complete MechanicalUnloading

ABSTRACT

Chen C-H, Liu X, Yeh M-L, Huang M-H, Zhai Q, Lowe WR, Lintner DM, Luo Z-P.Pathological changes of human ligament after complete mechanical unloading.Am J Phys Med Rehabil 2007;86:282–289.

Objective: To investigate the pathologic changes with a time sequenceamong patients with injured ligaments after complete mechanical unload-ing, based on a human anterior cruciate ligament (ACL) model.

Design: Pathologic examinations were done on remnants of com-pletely ruptured ACLs at various times up to 14 wks after injury on 31patients and on normal ACLs from five cadaver donors. Testing variablesincluded fibroblast density, crimp amplitude, and crimp nuclear shape.

Results: Sequential changes were observed: Fibroblast density signif-icantly increased within 5–6 wks of unloading. By 7–8 wks, crimpamplitude significantly decreased, accompanied by formation of irregularfiber patterns and fragments. This was followed by crimp wavelength andnuclear shape change within 9–14 wks.

Conclusions: These pathologic findings suggest that the ACL under-goes significantly deleterious changes from 5 to 6 wks after mechanicalunloading. This study may emphasize the important concept of earlyimplementation of mechanical force in rehabilitation programs for patientswith injured ligaments.

Key Words: ACL, Mechanical Unloading, Dynamic Change, Basic Science

A challenging yet controversial question in the management of completemechanical unloading of joints after injury concerns the optimal timing forphysical rehabilitation activity. Joint immobilization promotes healing of frac-tured bones or injured soft tissues. However, prolonged immobilization, such asjoint disuse, may have deleterious effects on normal tissues because of stressdeprivation, leading to degeneration, weakening, or disorganization of ligamentstructures.1 On the other hand, early rehabilitation may interfere with the


RESEARCH ARTICLE

Histology

healing process.2 It has been shown that acceler-ated rehabilitation protocols result in increasedincidence of synovitis after ligament injury.3,4

To date, the best knowledge to address this ques-tion has been obtained from histologic and biome-chanical studies conducted in animal models.5–7 Theresults show that the degradation of collagen wasincreased in canine ligament after mechanical un-loading for 12 wks.8 However, rates of collagen syn-thesis and degradation were affected by mechanicalunloading, and results varied depending on the typeof animal model used, the type of ligament, and theduration of mechanical unloading.9 On the otherhand, mechanical unloading, resulting in decreasedload force on the ligaments, varied in different mod-els. The patterns of mechanical unloading—such aswith the knee fixed in full flexion,6 internal fixation,10

or external fixation11—and the exact amount of re-duced force also differed. In animal models, investi-gations reported that mechanical unloading couldinduce synthesis of the collagen and collagencrosslinks and could result in random arrangementsor fixation of the fibers that further inhibited move-ment.9 Degenerative changes with necrosis of cells atthe central area of rabbit anterior cruciate ligaments(ACLs) were also noted after 6 wks of immobiliza-tion.12 The clear limitation in these animal studies isthe species difference between human and animalmodels; the information from these studies cannot betranslated directly to human beings. However, thesestudies still offer important information regardingthe effects of mechanical unloading.

This study presented a human ACL model toevaluate the mechanical unloading effects on thehistologic changes of ligament tissue over time.The objective was to provide pathologic informa-tion for investigating the dynamic effects for peoplewith ligament injury after mechanical unloading.

METHODSExperimental Model

The human in situ ligament mechanical un-loading model was designed using remnants ofcompletely ruptured human ACLs with a large

sample size at various times after injury.13 Rup-tured ACLs were obtained from 35 athletes under-going ACL reconstruction using the protocol ap-proved by the institutional review board at BaylorCollege of Medicine. The ruptured sites often oc-curred in the middle parts of ACLs. Remnant sitesthat were more than 6 mm from the rupture endswere included.14 The more distal remnant ACLs re-semble the histology of intact ACLs, and the ex-planted samples were referred to as having comefrom the normal zones.15 The ACLs were transectedat their insertion area with a scalpel by a surgeon, andall specimens were obtained from the normal zones.Thirty-one samples were included on which thecrimp morphology was measurable. Four sampleswere excluded because of unreadable crimp arrange-ment. Five intact ligaments were obtained from fivecadaver donors. Informed consent was obtained fromeach subject with a ruptured ligament, and the pro-cedures followed were in accordance with the ethicalstandards of the institutional review board.

To observe the histologic changes across differ-ent time periods, the samples were divided into fivegroups according to the mechanical unloading time(or the time between ACL rupture and reconstruc-tion): 1–2, 3–4, 5–6, 7–8, and 9–14 wks. Control ACLligaments were taken from five normal, physicallyfunctional cadaver donors using the same proceduresas for the ruptured ligaments. These ACLs were har-vested within 48 hrs after death, and they were clas-sified as the 0-wk group. The patient characteristicsincluding age range, mean age, gender, and timefrom rupture are listed in Table 1.

Tissue Preparation and StainingTissue samples for histologic examination

were taken from normal human ACLs. They werefixed in 10% buffered formalin at room tempera-ture for 48 hrs and were then embedded in paraffin.The specimens were cut into five-micrometer-thicklongitudinal sections and stained with hematoxylinand eosin stain. Histologic slides of longitudinalsections were evaluated under microscopy withnormal and polarized light.

TABLE 1 Patient characteristics for control and rupture human anterior cruciate ligaments

Groups Sample Size Age

Gender

M F

Control 5 32.0 (20–41) 3 21–2 wks after injury 7 22.7 (15–57) 5 23–4 wks after injury 5 17.6 (16–20) 2 35–6 wks after injury 7 31.8 (16–53) 3 47–8 wks after injury 5 36.4 (20–53) 1 49–14 wks after injury 7 23.7 (13–35) 1 6

April 2007 Ligament Changes After Mechanical Unloading 283

Measurements of VariablesCrimp amplitude, crimp wavelength, nuclear

density, and nuclear shape of fibroblasts were usedfor measurement assays.

Crimp Wavelength and AmplitudeThe morphological variables of ACL fibers were

observed according to histologic changes. First, thecollagen fascicle morphology and architecture to de-scribe the crimp of the wavelength and amplitude wasbased on Yahia’s model.12 In brief, changes of thecollagen fibers exhibited a sinusoid wave periodicallyinto two dimensions and ran continuously along theligament (Fig. 1). They played an important role inmaintaining ligament morphology and strength toresist tensile force.16 Crimp wavelength and ampli-tude decreased in an immobilized ligament status.17

In this study, measurement of crimp wavelength andamplitude were calculated with a calibrated scale un-der light microscopy. Ten locations along the lengthof each specimen were randomly evaluated, and meancrimp wavelength and amplitude were determinedand used for analysis.

Fibroblast Density and Nuclear ShapeParameters of fibroblasts were also included and

analyzed by a second researcher. The fibroblast den-sity was calculated at four independent areas under amicroscopic field. Ten nuclear shapes in each specimenwere included in the calculation of ratios. The fibroblas-

tic nucleus form was described according to the ratio ofnuclear length and width. A nuclear ratio �4 was clas-sified as spindle. A ratio �4 was defined as ovoid.13

Characteristics in Atomic Force MicroscopyThe characteristics of the single-fibril micro-

structure of ACLs were viewed, and atomic forcemicroscopy (AFM) (Synergy ESPM, Novascan, Ames,IA) was employed.18 The AFM generates images by aninteraction force between the cantilever tip and themolecules of the specimen rather than the photon orelectron for optical microscopy or electron micros-copy. This approach can generate magnificationnearly similar to that of electron microscopy; how-ever, the specimen can be viewed under more naturalconditions with the AFM.19 The specimen process foroptical microscopy can be directly used for AFM ob-servation after the specimen is deparaffined. It is anideal instrument to investigate the detailed fibril ul-trastructure of the ACL. Therefore, the same slideobserved by optical microscopy was processed forAFM evaluation. The fixed paraffin slides were heatedto 60°C on a hot plate, then they were deparaffinedwith xylene and ethanol alcohol and allowed to air dryfor AFM observation. A morphology of a single fibrilwas taken using the tapping mode of the AFM. Asilicone tip with a cantilever length of 130 �m and aresonance frequency of 145 KHz was used (NSC12,Mikromasch USA, Portland, OR). The image was re-corded with a 1-Hz scan rate and 512 � 512 resolu-tion. The measured test dimension was calculatedusing Image Analysis software.

Data Analysis and StatisticsA Kruskal–Wallis test was used for statistical

analysis in time sequence. Bonferroni correctiontesting was performed to compare each studygroup with the control group. Differences wereconsidered significant at the P � 0.05 level.

RESULTSHistologic changes in the ACLs were signifi-

cantly noted after 5–6 wks of mechanical unloadingcompared with the control group, and a sequence ofdevelopment was found to be continuous (Table 2).

FIGURE 1 Schematic diagram of the crimp patternsshows the waveform model. Crimps runalong the longitudinal axis. w, wave-length; a, amplitude.

TABLE 2 Ultrastructural measurements of the control and rupture human anterior cruciate ligaments

GroupsCell Density,

no./mm2Fibril Amplitude,

�mFibril Distance,

�mNuclear Aspect

Ratio

Control 303.8 � 45.3 5.1 � 0.4 25.0 � 1.8 4.4 � 1.51–2 wks after injury 322.3 � 19.7 5.2 � 0.2 26.0 � 1.8 4.6 � 1.33–4 wks after injury 342.8 � 32.8 5.0 � 0.2 25.3 � 1.2 4.4 � 1.15–6 wks after injury 472.8 � 80.3 4.7 � 0.3 25.6 � 2.0 4.4 � 1.57–8 wks after injury 590.8 � 103.2 4.0 � 0.3 25.4 � 1.6 4.2 � 1.49–14 wks after injury 826.3 � 187.4 3.2 � 0.2 22.6 � 1.0 3.8 � 1.2

284 Chen et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 4

The cell density was first significantly changed, fol-lowed by the crimp amplitude, nuclear shape, andcrimp wavelength.

Fiber MorphologyIrregular fiber arrangements and decreased

crimp amplitude were markedly apparent after 7–8wks (Fig. 2A–F). Fibers were closely packed andwell arranged longitudinally, and the crimp pat-terns were clearly present in the first 4 wks ofmechanical unloading. The AFM image from lon-gitudinally cut normal ACLs from the controlgroup showed regular arrangement of fibrils andtheir periods (Fig. 2G). In some areas, the fibrils

from the study group were not well aligned in thesame direction, and some smaller fibrils orientedvertically to larger parallel fibrils after 7–8 wks ofmechanical unloading (Fig. 2H).

Fibroblast DensityFibroblast density increased, especially from 5 to

6 wks after mechanical unloading, compared with thecontrol group (P � 0.01). The cellularity also signif-icantly increased at 7–8 and 9–14 wks (Fig. 3). Thefibroblast density presented a tendency to increasefrom 322.3 to 342.8 fibroblasts per squared millime-ter during the first 4 wks of mechanical unloading,but there was no significant change (P � 0.05).

FIGURE 2 Photomicrographs (original magnification �1000) of the control group (A) of anterior cruciate ligamentsshow that crimp patterns are clearly evident and that fibrils are closely packed. The longitudinallyoriented fibrils with a period crimp pattern were still present after 1 wk of complete mechanical unloading(B). There was no marked change observed in the morphology after 3 wks of complete mechanicalunloading (C). The difference was that intervals between crimp bundles were wider, and the widening ofthe intervals between fibril bundles was more common and obvious after 5 wks of complete mechanicalunloading. Some increased fibroblasts were also noted (D). Dramatic changes of irregular fibrillararrangement and decreased density were noted after 7 wks of complete mechanical unloading (E). Crimpamplitude further decreased, and the arrangement tended to be irregular. Some fragments and loss offibril density were also found after 9 wks of complete mechanical unloading (F). Atomic force microscopyimages of longitudinally cut anterior cruciate ligament in the normal group and after 7 wks of completemechanical unloading show that the fibrils were well arranged and that interfibrillar architecture wasalso uniform in the control group (G), but loss of collagen fibril arrangement, more irregular course,and arbitrary organization of the single fibril were noted after 7 wks of complete mechanicalunloading (H).


Crimp Amplitude and WavelengthFiber amplitude was found to decrease with

time, showing significant change from 7 to 8 wksonward, compared with the control group (P �0.01) (Fig. 4). Fiber wavelength slightly increasedduring the first 8 wks but significantly decreasedduring weeks 9–14 compared with the controlgroup (P � 0.01) (Fig. 5).

Nuclear ShapeThe nuclear ratio of the fibroblast was signifi-

cantly decreased at 9–14 wks after mechanical un-loading compared with the control group (P �0.05). The results showed a spindle form in the first

8 wks and significantly changed into ovoid formfrom 9 to 14 wks onward (Fig. 6).

DISCUSSIONThis study showed that the significantly disor-

dered changes of human ligament occurred from thefifth week of complete mechanical unloading and thatchanges in the fibroblast and fibers were continuous.After 5–6 wks, the fibroblast density increased re-markably compared with the initial 4 wks, as noted inFigure 3. Although an ACL study showed that fibro-blast density was increased after 6 wks,13 samples forthe study were harvested from the rupture ends of theligament. Living tissues demonstrate an ability to

FIGURE 3 Histogram showing the changes in fibroblast density across time after complete mechanical unload-ing of the anterior cruciate ligament compared with the fibroblast density of the control ligament.The error bar represents the mean and standard deviation. *P � 0.01.

FIGURE 4 Histogram showing the changes in crimp amplitude across time after complete mechanical unload-ing of the anterior cruciate ligament compared with the fiber amplitude of the control ligament. Theerror bar represents the mean and standard deviation. *P � 0.01.


respond to mechanical loading,20 and this conditionis very important for living tissues to maintain ten-don structures.21 Fibroblast transformation withinthe ligaments after mechanical unloading offers sig-nals to the homeostatic response of ligaments.22

Therefore, the fibroblasts are first responsible for ex-periencing the changes of the external mechanicalforce or motion and transferring signals into a cellu-lar response to adjust the synthesis or regeneration ofthe ligament.

The dynamic change that took place in theextracellular matrix was progressive deterioration,as seen in Figure 4. Crimp amplitude significantlydecreased in complete mechanical unloading at

7–8 wks. The mechanism of decreased crimp am-plitude was related to the biomechanical phenom-enon.9 After the proliferation of fibroblasts in acompletely immobilized status, this signal couldtrigger the production of collagen precursors inthe extracellular matrix.23 Therefore, an increasein the amount of newly synthesized collagen oc-curs, and less mature fibers may decrease the re-sistance to tensile load.24 The crimp wavelengthwas also significantly changed at 9–14 wks, as seenin Figure 5. This meant that the deterioration ofthe fibroblasts and fibers might not occur simul-taneously. Another finding in this study was incon-sistent with those in other animal model studies.

FIGURE 6 Histogram showing the changes in nuclear ratio across time after complete mechanical unloading ofthe anterior cruciate ligament compared with the nuclear ratio of the control ligament. The error barrepresents the mean and standard deviation. *P � 0.05.

FIGURE 5 Histogram showing the changes in crimp wavelength across time after complete mechanical unload-ing of the anterior cruciate ligament compared with the fiber wavelength of the control ligament. Theerror bar represents the mean and standard deviation. *P � 0.01.


The fragmentation of collagen fibers appeared after6 wks of mechanical unloading in a canine study.25

These significant differences implied that thechanges after complete mechanical unloading inhuman beings were not similar to those of animalmodels. However, they presented common dy-namic and continuous changes in ligaments aftermechanical unloading.

The changes of the nuclear shapes indicatedthat fibroblasts underwent a new configuration.Further changes in the amounts and the functionof intracellular organelles might occur.6 The fibro-blasts in the control group in the early stageswere spindle shaped. They became ovoid shapedafter 9 –14 wks of complete mechanical unload-ing, as seen in Figure 6. Fibroblasts took on theconfiguration from ovoid shaped to spindleshaped in a rabbit ACL study after 9 –14 wks ofcomplete mechanical unloading.23 Nuclear adap-tation to the influence from the extracellularmatrix seemed dissimilar between human andanimal environments.

Fibers were found in an irregular arrangementafter a period of complete mechanical unloading inthis study. Widening of the intervals betweencrimp bundles was markedly obvious after 7–8 wksof complete mechanical unloading (Fig. 2). Thefibrils were found to be arbitrarily organized (Fig.2H). Collagen degeneration and significant de-creases in collagen-fiber density after a period ofmechanical unloading resulted in decreased stiff-ness of ligament.9,10 Collagen fibrils have beenknown to provide resistance to mechanical load-ing.24 If this resistance is decreased, the ligament isimpaired.

The concepts regarding the effect of completemechanical unloading on morphological changesof the ligament are clinically important. Accordingto the literature, physical loadings provide an im-portant stimulant source for maintaining the nor-mal structure and function of ligament tissue. Pro-longed mechanical unloading has adverse effectson ligaments. According to one study, disusedmuscle atrophy started immediately, and the char-acteristics were significant within 1 wk.26 In an-other study of internally immobilized knees, devel-opment of surface irregularity and degeneration ofarticular cartilage was found.27 Ligament sub-stance weakness was also found in the insertionsite and was attributed to osteoclastic resorption ofbone after mechanical unloading.28 However, thecriteria of the complete mechanical unloading du-ration have not been determined. Future studiesexploring these criteria could provide importantinformation about ligament histologic change un-der complete mechanical unloading.

Some limitations in this study need to be ad-dressed. First, ruptured ACL tissues obtained dur-

ing ACL reconstruction were provided as an idealhuman model because of complete mechanical un-loading. ACL properties may not represent those ofother ligaments under complete mechanical un-loading. Second, histologic changes may not be thesole indicator for clinical applications. For exam-ple, biochemical activities of fibroblast synthesismay occur at different points of time. These aretopics to be further investigated in the near future.

CONCLUSIONThis study introduced a human ACL model

with which to evaluate complete mechanical un-loading. The result provides morphological evi-dence explaining the altered structure–time periodrelationship in ACLs under mechanical unloading.It also emphasized that implementation of a reha-bilitation plan for patients with ligament injuryshould take mechanical force into consideration inthe early stages of rehabilitation.

ACKNOWLEDGMENTSThe authors thank Carolyn Adams for her as-

sistance with the preparation of this manuscript.

REFERENCES1. Schollmeier G, Uhthoff HK, Sarkar K, Fukuhara K: Effects

of immobilization on the capsule of the canine glenohu-meral joint. A structural functional study. Clin Orthop1994;37–42

2. Akeson WH, Amiel D, Abel MF, et al: Effects of immobili-zation on joints. Clin Orthop 1987;28–37

3. Shelbourne KD, Nitz P: Accelerated rehabilitation after an-terior cruciate ligament reconstruction. Am J Sports Med1990;18:292–9

4. Shelbourne KD, Klootwyk TE, Wilckens JH, et al: Ligamentstability two to six years after anterior cruciate ligamentreconstruction with autogenous patellar tendon graft andparticipation in accelerated rehabilitation program.Am J Sports Med 1995;23:575–9

5. Goldstein WM, Barmada R: Early mobilization of rabbitmedial collateral ligament repairs: Biomechanic and histo-logic study. Arch Phys Med Rehabil 1984;65:239–42

6. Newton PO, Woo SL, MacKenna DA, et al: Immobilizationof the knee joint alters the mechanical and ultrastructuralproperties of the rabbit anterior cruciate ligament. J OrthopRes 1995;13:191–200

7. Trudel G, Jabi M, Uhthoff HK: Localized and adaptive sy-noviocyte proliferation characteristics in rat knee joint con-tractures secondary to immobility. Arch Phys Med Rehabil2003;84:1350–6

8. Klein L, Player JS, Heiple KG, et al: Isotopic evidence forresorption of soft tissues and bone in immobilized dogs.J Bone Joint Surg Am 1982;64:225–30

9. Amiel D, Akeson WH, Harwood FL, et al: Stress deprivationeffect on metabolic turnover of the medial collateral liga-ment collagen. A comparison between nine- and 12-weekimmobilization. Clin Orthop 1983;265–70

10. Binkley JM, Peat M: The effects of immobilization on theultrastructure and mechanical properties of the medial col-lateral ligament of rats. Clin Orthop 1986;301–8

11. Roth JH, Mendenhall HV, McPherson GK: The effect ofimmobilization on goat knees following reconstruction ofthe anterior cruciate ligament. Clin Orthop 1988;278–82

12. Kanda T, Ochi M, Ikuta Y: Adverse effects on rabbit anterior


cruciate ligament after knee immobilization: changes inpermeability of horseradish peroxidase. Arch OrthopTrauma Surg 1998;117:307–11

13. Murray MM, Martin SD, Martin TL, et al: Histologicalchanges in the human anterior cruciate ligament afterrupture. J Bone Joint Surg Am 2000;82:1387–97

14. Murray MM, Spector M: Fibroblast distribution in the an-teromedial bundle of the human anterior cruciate ligament:the presence of alpha-smooth muscle actin-positive cells.J Orthop Res 1999;17:18–27

15. Murray MM, Spector M: The migration of cells from theruptured human anterior cruciate ligament into collagen-glycosaminoglycan regeneration templates in vitro. Bioma-terials 2001;22:2393–402

16. Magnusson SP, Qvortrup K, Larsen JO, et al: Collagen fibrilsize and crimp morphology in ruptured and intact Achillestendons. Matrix Biol 2002;21:369–77

17. Hurschler C, Provenzano PP, Vanderby R Jr: Scanningelectron microscopic characterization of healing and nor-mal rat ligament microstructure under slack and loadedconditions. Connect Tissue Res 2003;44:59–68

18. Raspanti M, Alessandrini A, Ottani V, et al: Direct visualiza-tion of collagen-bound proteoglycans by tapping-modeatomic force microscopy. J Struct Biol 1997;119:118–22

19. Gutsmann T, Fantner GE, Kindt JH, et al: Force spectros-copy of collagen fibers to investigate their mechanical prop-erties and structural organization. Biophys J 2004;86:3186–93

20. Robbins JR, Evanko SP, Vogel KG: Mechanical loading andTGF-beta regulate proteoglycan synthesis in tendon. ArchBiochem Biophys 1997;342:203–11

21. Screen HR, Shelton JC, Bader DL, et al: Cyclic tensile strainupregulates collagen synthesis in isolated tendon fascicles.Biochem Biophys Res Commun 2005;336:424–9

22. Newton PO, Woo SL, Kitabayashi LR, et al: Ultrastructuralchanges in knee ligaments following immobilization. Ma-trix 1990;10:314–9

23. Hayashi K: Biomechanical studies of the remodeling of kneejoint tendons and ligaments. J Biomech 1996;29:707–16

24. Provenzano PP, Martinez DA, Grindeland RE, et al: Hind-limb unloading alters ligament healing. J Appl Physiol2003;94:314–24

25. Anderson AF, Lipscomb AB: Analysis of rehabilitation tech-niques after anterior cruciate reconstruction. Am J SportsMed 1989;17:154–60

26. Arvidsson I, Arvidsson H, Eriksson E, et al: Prevention ofquadriceps wasting after immobilization: an evaluation ofthe effect of electrical stimulation. Orthopedics 1986;9:1519–28

27. Trudel G, Himori K, Uhthoff HK: Contrasting alterations ofapposed and unapposed articular cartilage during joint con-tracture formation. Arch Phys Med Rehabil 2005;86:90–7

28. Laros GS, Tipton CM, Cooper RR: Influence of physicalactivity on ligament insertions in the knees of dogs. J BoneJoint Surg Am 1971;53:275–86


Authors:Malin Nygren-Bonnier, PT, MScPeter Lindholm, MD, PhDAgneta Markstrom, MD, PhDMaria Skedinger, MD, PhDEva Mattsson, PT, PhDBrita Klefbeck, PT, PhD

Affiliations:From the Division of Physiotherapy,Department of Neurobiology, CareScience and Society, KarolinskaInstitutet, and the Department ofPhysical Therapy, KarolinskaUniversity Hospital, Stockholm,Sweden (MNB, EM, BK); the Sectionof Environmental Physiology,Department of Physiology andPharmacology, Karolinska Institutet,Stockholm, Sweden (PL); NationalRespiratory Centre, KarolinskaInstitutet, Department of ClinicalSciences, Danderyd Hospital,Stockholm Sweden (AM); and theDepartment of Respiratory Medicineand Allergy, Karolinska UniversityHospital, Huddinge, Sweden (MS).

Correspondence:All correspondence and requests forreprints should be addressed to MalinNygren-Bonnier, Division ofPhysiotherapy, Department ofNeurobiology, Care Science andSociety, Karolinska Institutet, 23100,SE-141 83 Huddinge, Sweden.


DOI: 10.1097/PHM.0b013e3180383367

Effects of GlossopharyngealPistoning for Lung Insufflation onVital Capacity in Healthy Women

ABSTRACT

Nygren-Bonnier M, Lindholm P, Markstrom A, Skedinger M, Mattsson E,Klefbeck B: Effects of glossopharyngeal pistoning for lung insufflation on vitalcapacity in healthy women. Am J Phys Med Rehabil 2007;86:290–294.

Objectives: To determine whether healthy women could be trained toperform glossopharyngeal pistoning (GP) to insufflate the lungs to vol-umes exceeding maximum inspiratory capacity (IC), whether such insuf-flation caused discomfort, and the immediate and long-term effects onvital capacity (VC).

Design: A randomized controlled trial. Twenty-six healthy women wererandomly assigned to a training group (TG, n � 17) or to a control group(CG, n � 9). The TG performed 15–30 deep inspiratory efforts supple-mented by GP to lung volumes exceeding IC, three times per week for 6wks. Pulmonary function and chest expansion were measured before andafter the 6-wk period. The TG was retested again 12 wks after the end ofthe training period.

Results: One of 17 women had difficulty performing GP and wasexcluded. Temporary symptoms (while performing GP) were reported in44% of subjects in the TG. After 6 wks of training, subjects in the TG hadsignificantly increased their VC (P � 0.001). VC did not change in theCG. The increase in vital capacity of the TG was still evident after 12 wkswithout performing GP. Chest expansion increased significantly with GP.

Conclusion: The women in the TG were able to perform the technique,and it did not cause major discomfort. VC increased significantly in theTG, and the increase was still present after 12 wks without GP.

Key Words: Chest Expansion, Glossopharyngeal Breathing, Pulmonary Function,Respiratory Physiotherapy


RESEARCH ARTICLE

Pulmonary

Glossopharyngeal breathing (GPB) is a proce-dure in which the glossopharyngeal muscles areused to piston boluses of air past the vocal cords intothe lungs to lung volumes exceeding inspiratory ca-pacity; this has been called gulping. GPB can increasealveolar ventilation for patients with respiratory mus-cle paralysis, and it can be used for greater coughvolumes.1–8 GPB was first described by Dail in 19519

in patients with poliomyelitic paralysis. Patients havereported that the ability to cough increases, voicevolume increases, and the chest wall becomes moremobile, when the amount of air in the lungs is in-creased by glossopharyngeal pistoning (GP) to lungvolumes exceeding inspiratory capacity (IC).1,5,6,8,10

The technique, however, is not frequently taught topatients by physiotherapists or by respiratory thera-pists.

The lung volume added to IC by GP can bedenoted the “glossopharyngeal pistoning lung insuf-flation volume.” The volume exhaled by a subject in acomplete exhalation after performing GP is the glos-sopharyngeal pistoning lung insufflation volume �vital capacity (VC), or VCGP. This is similar to the VCsupplemented by GP, called the glossopharyngealmaximum single-breath capacity,8 described for pa-tients with reduced VC.8 Breath-hold divers use theextra volume obtained by GP to boost diving perfor-mance.11,12 These divers also use GP on dry land as astretching maneuver to improve the flexibility of thechest and diaphragm (personal communication be-tween divers and Lindholm 2002). Large lung volumeshave been reported in competitive breath-hold divers,11

but no one has studied whether this is a result of ge-netically gifted individuals being overrepresented in thepopulation of competitive breath-hold divers or whetherit is a result of performing GP for many years.

It is unclear whether the performance of GPhas any long-term effects on VC. Most studies havefocused on patients with reduced VC, who try toachieve a volume closer to their predicted VC usingGPB.1,2,3,10,13–15 To our knowledge, no study hasinvestigated the long-term effects of GP on pulmo-nary function in the general population.

We hypothesized that 6 wks of performing GPwould increase VC in healthy women. The aims ofthis study were to determine whether the subjectscould be trained to perform the technique, whethersuch training would cause discomfort, and the im-mediate and long-term effects of GP on VC.

MATERIALS AND METHODSSubjects

Subjects were recruited among female physio-therapists working at a university hospital. Criteriafor inclusion were adults younger than 60 yrs,healthy, and nonsmoking. Twenty-six subjects agreed

to participate. Subjects were randomly assigned ei-ther to a training group (TG, n � l7) or to a controlgroup (CG, n � 9). Table 1 summarizes the subjectcharacteristics. The research ethics committee atKarolinska Institutet in Stockholm, Sweden approvedthe study, and it was conducted in accordance withthe Helsinki Declaration. All subjects gave their writteninformed consent before participation.

Performing Glossopharyngeal PistoningSubjects were trained to carry out GP. The sub-

jects first carried out a maximal inhalation and thenperformed GP using as many gulps of air as possiblewithout discomfort from lung hyperexpansion. Fi-nally, they relaxed the larynx, and the air (volume �VCGP) was passively expelled. Subjects performed GPvia the mouth; some subjects used a nasal clip toavoid air leakage.2,5,8

Each subject in the TG received individual in-struction in GP from a physiotherapist. The subjectswatched an instructional video and then learned thetechnique together with the instructor. Subjects alsoreceived written information.

The TG performed GP three times a week for 6wks. They carried out a short warm-up programwith stretching exercises for the chest, and thenthey performed 15–30 repetitions of GP. Each ses-sion took about 20 mins. The instructor supervisedat least one session per subject per week. The CGdid not perform GP.

Pulmonary Function TestsStatic and dynamic spirometry was performed by

a blinded independent technician using a body ple-thysmograph (Master Screen Body, Jaeger) in accor-

TABLE 1 Physical characteristics of subjectsat baseline

Training Group Control Group

Subjects, n 16 9Age, yrs 37.25 � 8.99 37.11 � 10.65Height, cm 167.06 � 5.25 168.44 � 5.10Weight, kg 60.75 � 5.00 61.67 � 7.26BMI, kg/m2 21.78 � 1.7 21.73 � 2.40TLC, liters 6.13 � 0.68 5.92 � 0.55TLC, % pred 116.75 � 11.12 111.33 � 9.68VC, liters 4.36 � 0.55 4.19 � 0.43VC, % pred 123.68 � 16.07 117.44 � 16.50FEV1, liters 3.36 � 0.47 3.43 � 0.46FEV1, % pred 109.69 � 16.70 110.11 � 11.70RV, liters 1.77 � 0.29 1.73 � 0.30RV, % pred 110.31 � 11.40 105.11 � 14.51

Data are presented as means � SD. BMI, body massindex; TLC, total lung capacity; % pred, percentage of thepredicted value; VC, vital capacity; FEV1, forced expiratoryvolume in 1 sec; RV, residual volume. There was no differ-ence between groups at baseline.

April 2007 Glossopharyngeal Pistoning in Women 291

dance with American Thoracic Society standards.16

All subjects of both groups were tested at baseline andafter 6 wks. Reference values from Quanjer et al.17

were used. The coefficient of variance for test–retestintrareliability of the measurement of VC (n � 50)was 1.1%. The TG was reexamined after another 12wks without performing GP.

A portable spirometer (infrared interruption flowsensor; SpirobankG, MIR, Rome, Italy) was used tomeasure VC in both groups every week during the6-wk period. In the TG, the VCGP and VC after per-forming GP were also measured each week duringthe weekly supervised session. We ensured that thesubjects in the TG carried out GP correctly at eachsession.

Chest ExpansionChest circumference was measured at baseline

and after 6 wks, at the fourth costae and at the levelof the xiphoid process. It was measured with a tapemeasure, and both the upper and the lower rib cagewere measured. The subjects were instructed toperform a maximal exhalation and then a maximalinhalation, with chest expansion being the differ-ence between the two.18 Chest expansion was alsodetermined using GP. The coefficient of variancefor test–retest intrareliability of the tape measure-ments at the level of the xiphoid process was 0.8%,and it was 0.4% for measurements at the fourthcostae (n � 50).

Other AssessmentsThe subjects in the TG graded the perceived

tension in the chest using the Borg CR-10 scale19

each time VCGP was measured during the 6-wk pe-riod. These subjects also recorded symptoms and de-tails (the number of cycles per session and the num-ber of gulps per GP cycle) after each training session.This information was used to assess compliance withthe prescribed protocol.

Statistical AnalysisDescriptive statistics are presented as mean �

SD and range. The results were analyzed using

parametric statistics. The significance of differ-ences between baseline and after training was as-sessed by Student’s paired t test. The significanceof differences between the TG and the CG wasdetermined by repeated-measures ANOVA. A statis-tical significance level of P � 0.05 was accepted.

All analyses were performed using Statistica7.0 (Stat Soft, Inc, Tulsa, OK).

RESULTSOne subject in the TG was not able to perform

the GP with or without a nasal clip. This subjectwas excluded from the study, and the number ofsubjects in the TG was thus 16. Compliance withthe protocol was 84%, as determined by examiningthe subjects’ records. Subjects performed an aver-age of 19 GP cycles (range 15–30), with an averageof 10 gulps per GP cycle (range 2–42). Subjectsgraded the perceived tension in their chest duringthe GP maneuvers at a median score of 4 (range1.5–8) on the Borg CR-10 scale. Five subjects usednasal clips during GP, and no subjects brought inair via the nose. Seven subjects (44%) experiencedoften temporary symptoms, including light-headedness, hacking cough, headache, and yawn-ing. None of the subjects fainted during the GPmaneuvers.

There were no significant differences betweenthe two groups in any of the measured parametersat baseline (Table 1).

Pulmonary Function TestsTable 2 shows the results obtained from the

body plethysmograph: VC increased after the 6-wkperiod by 0.13 liters (P � 0.001) in the TG, whereasit did not change in the CG (Fig. 1). The statisticalcomparison between the groups indicated thatthere was a significant increase in the TG com-pared with the CG (P � 0.01). The increase in theVC in the TG persisted for at least 12 wks after thestudy (4.47 � 0.6 liters).

Table 3 shows VCGP during the 6 wks, and thecorresponding values of VC before and after per-forming GP.

TABLE 2 Lung volumes at baseline and after 6 wks

BaselineTG

After 6 wks PBetween-Groups

P BaselineCG

After 6 wks P

Subjects, n 16 9VC, liters 4.36 � 0.55 4.49 � 0.60 �0.001 �0.01 4.19 � 0.43 4.19 � 0.48 NSTLC, liters 6.13 � 0.68 6.21 � 0.69 NS NS 5.92 � 0.55 5.94 � 0.59 NSRV, liters 1.77 � 0.29 1.72 � 0.34 NS NS 1.73 � 0.30 1.75 � 0.31 NS

Values are presented in liters as mean � SD. TG, training group; CG, control group; VC, vital capacity; TLC, total lungcapacity; RV, residual volume. P values are given for differences between baseline and after 6 wks and between groups.Measurements were made with a body plethysmograph.

292 Nygren-Bonnier et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 4

Chest ExpansionChest expansion was greater at the level of the

xiphoid process (1.0 � 0.7 cm, P � 0.001) and atthe level of the fourth costae (1.0 � 0.7 cm, P �0.001) with GP than it was during normal maxi-mum inhalation. The mean chest expansion with-out GP did not increase significantly after the 6-wkperiod.

DISCUSSIONAll women in the TG were able to perform GP

for 6 wks without any major discomfort. The mainfindings were that the VCGP was 14.6% (signifi-

cantly) greater than VC and that chest expansionalso increased significantly with GP.

Some subjects learned GP immediately, whereasit took a few training sessions for others. Properinstruction and motivation are important for learn-ing.20 The portable spirometer was useful as a teach-ing tool, giving direct quantitation of efficacy.

From our data, and from an extensive review ofthe literature, it does not seem that GPB increasesthe risk of pulmonary trauma.4,5,8,10,15,21,22 Theincrease in intrathoracic pressure by GP may causeorthostatic syncope23 attributable to the reductionin venous return, especially if the subject is stand-ing upright. Forceful straining (e.g., in labor) hasbeen shown to cause pneumomediastinum withsubcutaneous emphysema in some individuals,24

and it is possible that GP may cause pneumothoraxin susceptible individuals from the increased stresson the pulmonary tissue.

Our data show that the chest may expand to alarger volume than the volume achieved usingordinary inspiratory muscles. The exact mecha-nism for the increase in VC is not known, althoughthis may be a result of an increase in pulmonarycompliance that results from stretching, enablingthe inspiratory muscles to inhale to a greater lungvolume.14

The mean VC of the subjects in the TG in-creased by 3% after the 6-wk period. Other authorshave also shown a small increase in the mean VC ofhealthy subjects25,26 after inspiratory muscle train-ing. However, the training in our study differs fromthat described by other authors in that our trainingconsisted of conventional inhalation to maximuminspiratory capacity, followed by GP. This may re-sult in a stretching effect rather than an increase inmuscle power. With lobectomy, the remaining pul-monary tissue expands.27 This suggests that lungvolume in healthy subjects is limited either by themagnitude of chest-wall expansion or by resistanceto diaphragm movement. The period of GP sessionsin this study was brief, and the question remainswhether a longer period might affect VC more.Some of the breath-hold divers that have beenstudied11 have been using this technique for manyyears. For these divers, VCGP can be 30–40%greater than VC, whereas this figure was 14.6% forour subjects.

ACKNOWLEDGMENTSThis study was supported by the Swedish Na-

tional Centre for Research in Sports and the HealthCare Sciences Postgraduate School.

REFERENCES1. Metcalf VA: Vital capacity and glossopharyngeal breathing

in traumatic quadriplegia. Phys Ther 1966;46:835–8

2. Warren VC: Glossopharyngeal and neck accessory muscle

FIGURE 1 Measurements of vital capacity (VC) bybody plethysmograph at baseline and af-ter a 6-wk period of performing glosso-pharyngeal pistoning for lung insufflation(GP). Values are presented as means (SD).�, training group (TG, n � 16) that per-formed sessions of GP for a period of 6wks. The VC increased in the TG (*P �0.001). e , control group (CG, n � 9).There was significant difference in effectbetween groups (P � 0.01).

TABLE 3 Subjects’ highest vital capacitysupplemented by glossopharyngealpistoning (VCGP) during the sixtraining weeks, and the correspondingVC measurements before and after,measured with the portable spirometer

TG P

Subjects, n 16VC before the training

session, liters4.17 � 0.52

VCGP, liters 4.78 � 0.60 �0.001VC after the training

session, liters4.39 � 0.57 �0.05

Values are presented as means � SD. TG, training group;VC, vital capacity. Significance (P � 0.05) represents anincrease in VC after vs. before the VCGP maneuvers.

April 2007 Glossopharyngeal Pistoning in Women 293

breathing in a young adult with C2 complete tetraplegiaresulting in ventilator dependency. Phys Ther 2002;82:590–600

3. Affeldt JE, Dail CW, Collier CR, et al: Glossopharyngealbreathing: ventilation studies. J Appl Physiol 1955;8:111–3

4. Bach JR: Update and perspectives on noninvasive respira-tory muscle aids. Part 1: the inspiratory aids. Chest 1994;105:1230–40

5. Dail CW, Affeldt JE, Collier CR: Clinical aspects of glosso-pharyngeal breathing; report of use by one hundred post-poliomyelitic patients. J Am Med Assoc 1955;158:445–9

6. Montero JC, Feldman DJ, Montero D: Effects of glossopha-ryngeal breathing on respiratory function after cervicalcord transection. Arch Phys Med Rehabil 1967;48:650–3

7. Bach JR: Management of neuromuscular ventilatory failureby 24 hour non-invasive intermittent positive pressure ven-tilation. Eur Respir Rev 1993;3:284–91

8. Bach JR, Alba AS, Bodofsky E, et al: Glossopharyngealbreathing and noninvasive aids in the management of post-polio respiratory insufficiency. Birth Defects Orig Artic Ser1987;23:99–113

9. Dail CW: “Glossopharyngeal breathing” by paralyzed patients;a preliminary report. Calif Med 1951;75:217–8

10. Bach JR, Smith WH, Michaels J, et al: Airway secretionclearance by mechanical exsufflation for post-poliomyelitisventilator-assisted individuals. Arch Phys Med Rehabil1993;74:170–7

11. Lindholm P, Nyren S: Studies on inspiratory and expiratoryglossopharyngeal breathing in breath-hold divers employ-ing magnetic resonance imaging and spirometry. Eur J ApplPhysiol 2005;94:646–51

12. Muth CM, Radermacher P, Pittner A, et al: Arterial bloodgases during diving in elite apnea divers. Int J Sports Med2003;24:104–7

13. Collier CR, Dail CW, Affeldt JE: Mechanics of glossopharyn-geal breathing. J Appl Physiol 1956;8:580–4

14. Kang SW, Bach JR: Maximum insufflation capacity. Chest2000;118:61–5

15. Bach JR: New approaches in the rehabilitation of the trau-matic high level quadriplegic. Am J Phys Med Rehabil1991;70:13–9

16. Standardization of Spirometry, 1994 Update. AmericanThoracic Society. Am J Respir Crit Care Med 1995;152:1107–36

17. Quanjer PH, Tammeling GJ, Cotes JE, et al: Lung volumesand forced ventilatory flows. Report Working Party Stan-dardization of Lung Function Tests, European Communityfor Steel and Coal. Official Statement of the EuropeanRespiratory Society. Eur Respir J Suppl 1993;16:5–40

18. Carlson B: Normal chest excursion. Phys Ther 1973;53:10–4

19. Borg GA: Psychophysical bases of perceived exertion. MedSci Sports Exerc 1982;14:377–81

20. Dickinson DG, Murphy AJ, Talner NS: Glossopharyngealbreathing in the management of the chronic poliomyeliticrespirator patient. Arch Phys Med Rehabil 1956;37:631–6

21. Bach JR, Alba AS, Shin D: Management alternatives forpost-polio respiratory insufficiency. Assisted ventilation bynasal or oral-nasal interface. Am J Phys Med Rehabil 1989;68:264–71

22. Dail CW, Wendland LV, Affeldt JE: Usefulness of glossopha-ryngeal breathing from the viewpoint of the patient. ArchPhys Med Rehabil 1956;37:7–11

23. Lempert T, Bauer M, Schmidt D: Syncope: a videometricanalysis of 56 episodes of transient cerebral hypoxia. AnnNeurol 1994;36:233–7

24. Maunder RJ, Pierson DJ, Hudson LD: Subcutaneous andmediastinal emphysema. Pathophysiology, diagnosis, andmanagement. Arch Intern Med 1984;144:1447–53

25. Leith DE, Bradley M: Ventilatory muscle strength and en-durance training. J Appl Physiol 1976;41:508–16

26. Fanta CH, Leith DE, Brown R: Maximal shortening of in-spiratory muscles: effect of training. J Appl Physiol 1983;54:1618–23

27. Ravikumar P, Yilmaz C, Dane DM, et al: Regional lunggrowth following pneumonectomy assessed by computedtomography. J Appl Physiol 2004;97:1567–74

294 Nygren-Bonnier et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 4

Authors:John R. Bach, MDCarlo Bianchi, MDMauro Vidigal-Lopes, MDSandra Turi, MDGiorgio Felisari, MD

Affiliations:From the Department of PhysicalMedicine and Rehabilitation,University of Medicine and Dentistryof New Jersey–New Jersey MedicalSchool, Newark, New Jersey (JRB);Divisione di Riabilitazione,Fondazione Istituto Sacra Famiglia,Cesano Boscone, Milan, Italy (CB, ST,GF); and Vent-Lar MuscularDystrophy Program–Julia KubitschekHospital and the Servico dePneumologia e Cirurgia Toracica ofthe Madre Teresa Hospital, BeloHorizonte, Minas Gerais, Brazil (MV-L).

Correspondence:All correspondence and requests forreprints should be address to John R.Bach, MD, Professor of PhysicalMedicine and Rehabilitation,Professor of Neurosciences, ViceChairman, Department of PhysicalMedicine and Rehabilitation,University Hospital B-403, 150Bergen St., Newark, NJ 07103.


DOI: 10.1097/PHM.0b013e318038d1ce

Lung Inflation by GlossopharyngealBreathing and “Air Stacking” inDuchenne Muscular Dystrophy

ABSTRACT

Bach JR, Bianchi C, Vidigal-Lopes M, Turi S, Felisari G: Lung inflation byglossopharyngeal breathing and “air stacking” in Duchenne muscular dystrophy.Am J Phys Med Rehabil 2007;86:295–300.

Objective: To compare the use of glossopharyngeal breathing (GPB)and air stacking to increase lung volumes and cough peak flows (CPF),and GPB to increase ventilator-free breathing ability (VFBA), for patientswith Duchenne muscular dystrophy.

Design: A case series of all referred patients with declining vital capacity (VC).Seventy-eight patients underwent training in and monitoring of the efficacy of airstacking (retaining consecutively delivered volumes of air delivered via manualresuscitator and held by glottic closure) to maximum insufflation capacity (MIC).GPB also was demonstrated to all 78 patients, and 32 were formally trained andprescribed GPB as their VCs decreased below 400 ml. To be successful, theMIC or GPB maximum single-breath capacity (GPmaxSBC) had to exceed VC.Improvements in VFBA were determined by requiring fewer ventilator-assistedbreaths per minute. CPFs were measured by peak flow meter.

Results: Seventy-four (94.9%) of the patients could air stack (MIC � VC),and, thus far, 21 (27%) are able to GPB. Fifteen could GPB sufficiently to delayonset of daytime ventilator use and, later, to require 1.9 fewer ventilator assistedbreaths per minute. For the 47 patients with multiple data points, as VC deteri-orated from 1080 � 870 to 1001 � 785 ml, MIC increased from 1592 �887 to 1838 � 774 ml. For 21 patients, GPmaxSBC significantly exceededVC (824 � 584 vs. 244 � 151 ml, respectively, P � 0.001). The ability toincrease lung volume by air stacking (MIC) was better retained than was the abilityto increase lung volume by GPB (GPmaxSBC). Air stacking also permittedassisted CPF to exceed unassisted CPF: 289 � 91 and 164 � 76 liters/m,respectively (P � 0.001).

Conclusions: GPB and air stacking can increase lung volumes and,thereby, cough flows. GPB also can be used in many cases to delay anddecrease daytime ventilator use.

Key Words: Glossopharyngeal Breathing, Cough, Duchenne Muscular Dystrophy, Re-spiratory Therapy, Noninvasive Mechanical Ventilation, Life Expectancy

April 2007 Lung Inflation by Glossopharyngeal Breathing 295

RESEARCH ARTICLE

Pulmonary

The key to the successful long-term use of nonin-vasive mechanical ventilatory support is in effectivelyexpelling airway secretions when necessary.1,2 To doso, the augmentation of lung air volumes can becrucial to optimize cough peak flows (CPF).3–6 In1981, it was reported that 47 Duchenne musculardystrophy (DMD) patients reached a maximum (pla-teau) vital capacity (VC) between ages 10 and 12(range 9–16); in the general population, VC plateausat age 19.7 Subsequently, VC decreases by 5–10% peryear in patients with DMD.7,8 This contributes to thedecrease in unassisted CPF. Two methods of lung-inflation therapy that can result in increased coughflows are air stacking and maximum-depth glosso-pharyngeal breathing (GPB).

Air stacking involves the use of a manual re-suscitator or volume-cycle ventilator to deliver vol-umes of air that are consecutively held by glotticclosure until no more air can be retained.9 The max-imum lung volume that can be held by air stacking isthe maximum insufflation capacity (MIC).

Dail first described GPB as gulping air into thelungs for the purpose of providing deep lung vol-umes to increase the cough flows of five postpolio-myelitis patients.10,11 Use of GPB for ventilator-freebreathing has been reported for high-level trau-matic tetraplegia patients and poliomyelitis pa-tients,11–13 but not for DMD patients.13 Indeed,DMD patients who are not taught and equippedwith respiratory muscle aids to avoid respiratoryfailure often undergo tracheotomy before they areable to benefit from GPB for ventilator-free breath-ing ability (VFBA).1

The VFBA is considered limited when patientsneed intermittent positive pressure ventilations tosupplement unassisted minute ventilation to pre-vent distress and to maintain baseline blood-gaslevels. When absent, cessation of ventilator useresults in immediate distress and blood-gas deteri-oration.

Advances made during the last two decades innoninvasive mechanical ventilatory support andmechanically assisted coughing have greatly im-proved survival without resort to tracheotomy.1

This makes air stacking and GPB increasingly impor-tant for the autonomous augmentation of coughflows, voice volume, and VFBA. GPB mastery also caneliminate fear of ventilator dysfunction or disconnec-tion. Thus, we prospectively determined whether theability to air stack and GPB could improve with prac-tice and result in improved VFBA and cough flows.

MATERIALS AND METHODSThis work was approved by the hospital ethics

committee. Seventy-eight consecutive males whohad visited a clinic since 1996 were studied after VCplateau. The initial appointment for 16 patients

had been before 1996. The diagnosis was based onclinical, enzymatic, electromyographic, and biopsyassessments. Gene-deletion studies were positivefor 54 of 70 patients. All except one lost the abilityto walk by age 11. Four of the 78 patients had takenglucocorticoids.

All patients were trained in air stacking andwere given a demonstration of GPB after the pla-teauing of the VC. The age and magnitude of theplateau was documented by pre- and post-VC mea-surements below a maximum (plateau) for 24 pa-tients, or it was assumed for the others because oftheir age and because subsequent data points werelower than the initial VC measurement. All 74self-directed patients mastered air stacking wheninitially introduced to it (MIC � VC) and wereasked to practice it two to three times per day,10–15 maneuvers each time, from then on. Airstacking was performed by the patient receivingconsecutively delivered volumes of air from a man-ual resuscitator or a volume-cycled ventilator via amouthpiece, nasal, or oral–nasal interface and re-taining as much as possible with a closed glottis(MIC).9 The volume was then measured spirometri-cally to determine MIC.

Each patient was asked about his practice ef-forts at every visit. Thirty-one of 47 who returnedfor at least one follow-up visit reported performingit at least twice daily.

GPB was demonstrated after VC plateau atevery clinic visit, and each patient was screened forthe ability to exceed VC by GPB. Once a patientdemonstrated mastery (maximum single-breathcapacity [GPmaxSBC] �VC), he was asked to prac-tice it three times a day. If he had not masteredGPB by the time his VC had decreased to 400 ml(the point at which many patients with DMD beginto use daytime ventilatory assistance), a demon-stration videotape was dispensed, and GPB practicewas formally prescribed.14 GPB practice was con-sidered optional before this point because it couldnot be used for VFBA early on, and air stacking waseasier to master and the ability easier to retain overtime for deep lung expansion, as typified by thepatient record in Figure 1.

GPB was taught by having the patient take adeep breath and hold it, then imitate the clinicianand take 15–20 gulps, and then blow the volumeinto a spirometer. The GPmaxSBC was the maxi-mum volume that could be gulped in. Training wasfacilitated by monitoring gulp efficiency, whichwas defined as (GPmaxSBC � VC) � (number ofgulps to a maximum insufflation) in milliliters pergulp.15 If initially unsuccessful, the patient’s nos-trils were sealed to demonstrate to him the needfor the soft palate to seal off the nasopharynx. Eventhen, nine patients could not master the glotticmovements necessary for successful GPB. On obser-

296 Bach et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 4

vation, they did not have as much rostral–caudalmovement of the glottis as did patients whose GP-maxSBC exceeded VC, nor did they have the glotticclosure gulp (click) that is heard with successful GPB.VC, MIC, and GPmaxSBC were measured spirometri-cally (Mark 14 spirometer; Ferraris Development andEngineering Co, Ltd, London, UK).

Assisted CPF are defined as CPF augmented byan abdominal thrust that is timed to glottic open-ing after air stacking or GPB to deep lung inflation.Unassisted and assisted CPF were measured by Ac-cess Peak Flow Meter (Health Scan Products Inc,Cedar Grove, NJ) at every visit.

The maximum of four or five measurementswas recorded for the initial (post-VC plateau) VC,MIC, GPmaxSBC, and CPF; the maximum observedassisted CPF, MIC, and GPmaxSBC; and the mostrecent data. A standard oral–nasal mask was usedfor spirometry and CPF measurements.

Diminished VFBA was defined by lack of au-tonomous ability to breathe without using GPB, orby a difference in need for mouthpiece intermittentpositive-pressure ventilations during 5-min periodswhen using GPB vs. periods when it was not used,with no change in end-tidal carbon dioxide or oxyhe-moglobin saturation. Patients took these assistedbreaths to avoid dyspnea. Delayed onset of daytimeventilatory assistance was recognized when patientswho normally used GPB throughout waking hoursbecame dyspneic and hypercapnic when breathingwithout it, to the extent that only resumption of GPBor deep, ventilator-assisted breaths could maintainthem.

Statistical AnalysisDescriptive statistics included mean and stan-

dard deviations. Comparisons between mean valuesof MIC and GPmaxSBC with VC at the initial, plateau,

and most recent evaluations were made by t testusing the Bonferroni correction for six comparisons.This warranted a P value �0.008 for statistical sig-nificance.

RESULTSAll 78 patients cooperated for VC measure-

ments, but four were too cognitively impaired tolearn lung-inflation techniques. Twenty-four DMDpatients’ VCs plateaued at mean age 12.7 � 3.1 (range10.5–16.1) yrs, at 2026 � 555 (940–2510) ml.

The 74 remaining self-directed patients (meanage 20.6 � 3.1 yrs) were taught air stacking aftertheir VC was on the decline. Their mean initial VC,and MIC by air stacking, were 987 � 631 and 1501 �618 ml, respectively, and unassisted and assistedCPF were 145 � 112 and 250 � 84 liters/m, re-spectively. The VC, MIC, and CPF changes overtime for the 47 patients with multiple data pointsduring a 7- to 169-mo follow-up are illustrated inFigure 2. For 31 of the 47 patients who reportedpracticing air stacking at least twice a day, the MICincreased over time despite diminishing VC. Forthese 31, MIC plateaued at 21 � 18 (2–51) mosafter initial training. Eight of the 47 patients’ post-plateau VCs increased by 35–70 ml for one or morefollow-up visits after practicing lung-insufflationtherapy. Two patients lost the ability to air stack(MIC � VC) at ages 26 and 38.

Ten patients had mastered GPB on their ownand presented with GPmaxSBC �VC. Twenty-twoothers were formally taught GPB. The VC and GP-maxSBC changes over time for 19 of the 32 pa-tients who mastered GPB and had two or more datapoints are illustrated in Figure 3. Eighteen of the19 reported using GPB daily. For 11 of the 19,GPmaxSBC increased with practice and plateaued13 � 16 (range 3–49) mos after initial mastery

FIGURE 1 Graphic for a patient with 20 data points. VC, vital capacity; MIC, maximum insufflation capacity;GPmaxSBC, glossopharyngeal maximum single-breath capacity.


(Figure 4). Gulp capacities were from 8 � 9 to70 � 34 ml, depending on whether GPB was beingused for normal minute ventilation or for maximallung inflation, with the latter gulps being muchsmaller.

Fifteen patients eventually used GPB through-out daytime hours such that respiratory distressand hypercapnia developed when ceasing GPB.Thus, GPB delayed the need for daytime ventilatoruse. At the most recent evaluation, these 15 re-quired 5.3 � 1.7 mouthpiece intermittent positive-pressure ventilations per minute (1200 ml each)when not using GPB and 3.4 � 1.7 ventilations perminute when using it. All used volume-cycled ven-tilators on assist/control mode with delivered vol-umes of 850–1500 ml and a backup rate of 10–12per minute. Nine of the 32 patients did not master

GPB, and two had GPmaxSBC exceed VC by 80 and240 ml at initial mastery but have not yet returned.

Whereas 74 of 78 (95%) patients with DMD mas-tered air stacking, only 21 of 78 (26.9%) have mas-tered GPB thus far, and at least nine are likely tonever master it. Some of the remaining 42 may mas-ter it as their VCs decrease below 400 ml. Thus, for atleast 27% of DMD patients, GPmaxSBC can exceedVC, and for most of these, it can delay the need fordaytime ventilator use. For the 74 patients as a whole,and for the patients in Figures 2–4, the MIC andGPmaxSBC values significantly exceeded VC and as-sisted CPF exceeded unassisted CPF (P � 0.008).

The mean age of beginning nocturnal nonin-vasive ventilation was 19.1 � 3.3 yrs. Although allof our GPB users lost the ability to breathe unaidedby ventilator use, none have tracheostomy tubes.Six patients died during the course of the study:

FIGURE 3 Changes in vital capacity (VC) and glosso-pharyngeal breathing maximum single-breath capacity (GPmaxSBC) over time for18 patients, including eight whose GP-maxSBC did not increase with training. Ageat initial GPB mastery 24.1 � 4.3 (17.5–31.1) yrs, maximum GPmaxSBC 25.4 � 4.8(18–33.5) yrs, most recent evaluation 29.3� 5.7 (23.2–43.1) yrs of age.

FIGURE 4 Changes in vital capacity (VC) and glos-sopharyngeal breathing. Maximum sin-gle-breath capacity (GPmaxSBC) for 11patients whose GPmaxSBC increased overtime. Age at initial GPB mastery 24.4 �4.3 (17.1–31.0) yrs, at GPmaxSBC maxi-mum 27.8 � 4.3 (17.5–33.5) yrs, at mostrecent evaluation 30.2 � 5.6 (24.9–43.1)yrs of age.

FIGURE 2 Changes in vital capacity (VC), maximum insufflation capacity (MIC), and unassisted and assistedcough peak flows (CPF) over time for 47 patients with two or more data points; age initially 20.6 �4.1 (13.1–33) yrs, at maximum 21.9 � 4.5 (13.6–37.2) yrs, most recently 24.9 � 4.5 (13.7–43.1) yrs.


five from overt cardiac failure, and one, who had noassistance at home, from respiratory failure duringa respiratory tract infection.

DISCUSSIONThe VC plateau that we found for 24 patients

occurred months later than the figure reported pre-viously.7 The fact that 4 of our 24 patients receivedglucocorticoid therapy may be one of the reasonsfor this.

In a previous study, it was reported that threeDMD patients used GPB to inflate the lungs to two tothree times VC. The authors did not report any effecton cough flows or on VFBA.16 The delayed need touse daytime ventilatory support and, subsequently,the fewer assisted breaths required when using GPBthan when not using it, provided some security to thepatients because they could survive longer by usingGPB in the event of ventilator failure, and GPB savedthem some effort in having to rotate the neck andgrab the mouthpiece as much as would have beennecessary otherwise. Also, because 90% of episodes ofrespiratory failure and death for conventionally man-aged DMD patients occur as a result of ineffectivecoughing during intercurrent upper–respiratory tractinfections,17 and because cough flows correlate with(pre-) cough volumes,9 it is important for patients tobe able to autonomously increase lung volumes whenthey need to cough. Lung-inflation therapy also helpsmaintain both dynamic and static pulmonary com-pliance.18 GPB, like air stacking via a volume venti-lator, permits this. Because both GPmaxSBC and MICcan improve and plateau over a wide range of time(2–51 mos) despite declining VC, both should bemonitored regularly, and practice should be encour-aged. Pressure-cycled ventilators such as BiPAP ma-chines do not permit air stacking and should not beused for these patients.

The slight, temporary increase in VC for eightpatients was probably attributable to an improve-ment in pulmonary compliance brought about byregular lung-expansion therapy. Although we onlymeasured the effect of deeper lung volumes and ab-dominal thrusts on assisted coughing, deep lung in-sufflations and abdominal thrusts have been sepa-rately shown to increase CPF almost equally, with thegreatest increases when they are combined.9,12,13

The GPmaxSBCs and gulp volumes in DMDpatients tend to be lower than those reported forpostpolio and spinal cord–injured patients, whooften have gulp capacities over 100 ml, GPmaxS-BCs over 3000 ml, and many hours of VFBA despitehaving little or no VC.5,12,13 This is because bulbar-innervated muscles tend to be spared for the latterbut become increasingly dysfunctional in DMD.Thus, although DMD patients are usually able tospeak clearly, take food by mouth, and maintaingood vocal-fold mobility and glottic closure even

after 40 yrs of age,1 they lose the ability to close theglottis tightly enough to hold deep lung volumes.Whereas most patients with bulbar amyotrophiclateral sclerosis lose the ability to air stack, andGPB and cough become physically impossible forthem,19 we have observed losses of air-stackingability in only 2 of 74 DMD patients, despite somebeing over age 40.

When air stacking and GPB are suboptimal,lung-insufflation therapy can still be performedusing a manual resuscitator with the expiratoryvalve blocked, by delivering high volumes from avolume-cycled ventilator, or by using the CoughAs-sist machine (J. H. Emerson Company, Cambridge,MA) at insufflation pressures of 40 cm H2O or morevia an oronasal interface. Indeed, as assisted CPFand air-stacking ability decrease, our patients useCoughAssist machines for both mechanically as-sisted coughing and for maximal lung insuffla-tions. No complications have been associated withthis therapy for the �1000 patients with neuro-muscular diagnoses who have been treated in thismanner in the last 28 yrs.20

Comparing GPmaxSBC and MIC permits theevaluation of oropharyngeal muscle groups. Whenthe GPmaxSBC is greater than the MIC achieved byair stacking via a mouthpiece (as often occurs inpostpoliomyelitis patients), then the glottis is suf-ficiently intact to hold deep lung volumes, but thelips and buccal muscles are too weak to permit airdelivery past the vocal fold. Some such patients canair stack better via nasal interfaces. Likewise, whenMIC is greater than GPmaxSBC, as for these DMDpatients, the hypopharyngeal musculature is weak-ened and laryngeal mobility is impaired. Some pa-tients can only GPB or air stack with the noseplugged because the soft palate is unable to seal offthe nasopharynx.

Although we and others strongly recommendthat GPB be taught to patients with neuromusculardisorders,21 few clinicians are familiar with the tech-nique. Some patients learn it on their own. This isprobably a consequence of cerebral imprinting of thephylogenetic distribution of lung-ventilation mecha-nisms. Aspiration breathers, like mammals, ceased GPBshortly after aspiration breathing had evolved.22 In-deed, at least four of our patients who learned it ontheir own and used it spontaneously were hypercap-nic and hypoxic. It is possible that the mechanism forGPB is the same as that observed in experimentalstudies in lunged amphibians for whom buccalpumping activity increases during hypoxia and hyper-capnia.23,24 Lung packing by GPB is used by manybreath-hold divers to increase lung air volumes by upto 5.6 liters over VC to permit longer submersion, andby many competitive swimmers to increase thoracicvolumes and buoyancy.22–27 The transpulmonarypressures generated by maximal-depth GPB in swim-


mers have been reported as high as 80 cm H2O.28,29

These are similar to the pressures we use for maxi-mum-depth air stacking. Despite such high pres-sures, no barotrauma or other complications havebeen reported.

Many postpolio, spinal cord–injured, and DMDpatients can use GPB instead of a ventilator forventilatory support when awake.12,13,15,30 With de-teriorating inspiratory and bulbar-innervated mus-cle function, however, DMD patients eventually useGPB to take fewer ventilator-assisted breaths. Oncea DMD patient’s VC is inadequate to breathe with-out continuous ventilator use (VC �200 ml), bul-bar-innervated musculature is inadequate to useGPB for complete ventilator-free breathing.

GPB has been included among the techniquesthat are not well adapted to producing bronchialclearance, because “it cannot be mastered by allpatients.”31,32 However, it seems that this is rarelyeven attempted. The clinician should at least iden-tify those patients who already perform it andshould help them improve their technique by spi-rometric feedback. With training and practice,many DMD patients can exceed their VCs three- totenfold (Fig. 4)—increases comparable with thoseof spinal cord–injured and postpolio patients.5,12,13

Both air stacking and GPB can improve with prac-tice, even when VC is deteriorating. GPB provides avital advantage for those who master it. It shouldno longer be ignored for patients with DMD.

REFERENCES1. Gomez-Merino E, Bach JR: Duchenne muscular dystrophy:

prolongation of life by noninvasive respiratory muscle aids.Am J Phys Med Rehabil 2002;81:411–15

2. Bach JR: Respiratory muscle aids: patient evaluation, respi-ratory aid protocol, and outcomes, in Bach JR (ed): TheManagement of Patients with Neuromuscular Disease.Philadelphia, PA, Hanley & Belfus, 2004, pp 271–308

3. Collier CR, Dail CV, Affeldt JE: Mechanics of glossopharyn-geal breathing. J Appl Physiol 1956;8:580–4

4. Kirby NA, Barnerias MJ, Siebens AA: An evaluation of as-sisted cough in quadriparetic patients. Arch Phys Med Re-habil 1966;47:705–10

5. Bach JR, Smith WH, Michaels J, et al: Airway secretionclearance by mechanical exsufflation for post-poliomyelitisventilator assisted individuals. Arch Phys Med Rehabil1993;74:170–7

6. Sortor S, McKenzie M: Toward Independence: Assistedcough (video). Dallas, TX, BioScience Communications ofDallas Inc., 1986

7. Rideau Y, Jankowski LW, Grellet J: Respiratory function inthe muscular dystrophies. Muscle Nerve 1981;4:155–64

8. Bach J, Alba A, Pilkington LA, Lee M: Long-term rehabilitationin advanced stage of childhood onset, rapidly progressive mus-cular dystrophy. Arch Phys Med Rehabil 1981;62:328–31

9. Kang SW, Bach JR: Maximum insufflation capacity: therelationships with vital capacity and cough flows for pa-tients with neuromuscular disease. Am J Phys Med Rehabil2000;79:222–7

10. Dail CW: Glossopharyngeal breathing by paralyzed patients.Calif Med 1951;75:15–25

11. Dail CW, Affeldt JE, Collier CR: Clinical aspects of glosso-

pharyngeal breathing. Report of use by one hundred post-poliomyelitic patients. JAMA 1955;158:445–9


13. Bach JR, Alba AS: Noninvasive options for ventilatory sup-port of the traumatic high level quadriplegic. Chest 1990;98:613–19

14. Webber B, Higgens J: Glossopharyngeal Breathing: What,When and How? (video). Holbrook, England, Aslan StudiosLtd., 1999

15. Bach JR, Alba AS, Bodofsky E, et al: Glossopharyngeal breath-ing and noninvasive aids in the management of post-poliorespiratory insufficiency. Birth Defects 1987;23:99–113

16. Baydur A, Layne E, Aral H, et al: Long term non-invasiveventilation in the community for patients with musculo-skeletal disorders: 46 year experience and review. Thorax2000;55:4–11

17. Bach JR, Rajaraman R, Ballanger F, et al: Neuromuscularventilatory insufficiency: the effect of home mechanicalventilator use vs. oxygen therapy on pneumonia and hos-pitalization rates. Am J Phys Med Rehabil 1998;77:8–19

18. Lechtzin N, Shade D, Clawson L, Wiener CM: Supramaxi-mal inflation improves lung compliance in subjects withamyotrophic lateral sclerosis. Chest 2006;129:1322–9

19. Bach JR, Bianchi C, Aufiero E: Oximetry and prognosis inamyotrophic lateral sclerosis. Chest 2004;126:1502–7

20. Bach JR: The Management of Patients with NeuromuscularDisease. Philadelphia, PA, Elsevier, 2004

21. Make BJ, Hill NS, Goldberg AI, et al: Mechanical ventilationbeyond the intensive care unit: report of a consensus con-ference of the American College of Chest Physicians. Chest1998;113:289–344

22. Sheafor EA, Wood SC, Tattersall GJ: The effect of gradedhypoxia on the metabolic rate and buccal activity of alungless salamander (Desmognathus Fuscus). J Exp Biol2000;203:3785–93

23. Andersson J, Schagatay E, Gustafsson P, et al: Cardiovasculareffects of “buccal pumping” in breath-hold divers, in GennserM (ed): XXIV Annual Scientific Meeting of the EuropeanUnderwater and Baromedical Society. Stockholm, Sweden,National Defense Research Establishment, 1998, pp 103–5

24. Imhagen H, Schagatay E, Andersson J, et al. Mechanisms ofbuccal pumping (lung packing) and its pulmonary effects,in Gennser M (ed): XXIV Annual Scientific Meeting of theEuropean Underwater and Baromedical Society. Stock-holm, Sweden, National Defense Research Establishment,1998, pp 80–3

25. Brainerd EL: New perspectives on the evolution of lungventilation mechanisms in vertebrates. Exp Biol Online1999;4:11–28

26. Vann RD, Hall FG: Pressure-volume characterisics of “lungpacking.” Undersea Hyperb Med 1994;21(suppl):42

27. Lindholm P, Gennser M: Breath-hold diving—an increasingadventure sport with medical risks. Lakartidningen 2004;101:787–90

28. Seccombe LM, Rogers PG, Mai N, Wong CK, Kritharides L,Jenkins CR: Features of glossopharyngeal breathing inbreath-hold divers. J Appl Physiol 2006;101:799–801

29. Loring SH, Lindholm P, O’Donnell CR, Ferringo M:Transpulmonary pressures after both glossopharyngeal in-halation above TLC and exhalation below RV in breath-holddivers [abstract]. Proc Am Thorac Soc 2006

30. Ishikawa Y, Bach JR: Post-polio pulmonary dysfunction.Kokyu To Junkan 2003;51:1121–7

31. Soudon P, Steens M, Toussaint M: Desobstruction tracheo-bronchique chez les patients restrictifs majeurs paralyses.Respir Care 1999;3:25

32. Mazza FG, DiMarco AF, Altose MD, et al: The flow-volume loopduring glossopharyngeal breathing. Chest 1984;85:638–40


Authors:John R. Bach, MDGeorge Tewfik, BS

Affiliations:From the Department of PhysicalMedicine and Rehabilitation,University of Medicine and Dentistryof New Jersey–New Jersey MedicalSchool Newark, New Jersey.

Correspondence:All correspondence and requests forreprints should be addressed to JohnR. Bach, MD, Department of PhysicalMedicine and Rehabilitation,University Hospital B-403, 150Bergen St., Newark, NJ 07103.


DOI: 10.1097/PHM.0b013e318038d39c

Air DopingAn Expose on “Frog” Insufflation in Competitive Sports

ABSTRACT

Bach J, Tewfik G: Air Doping: An Expose on “Frog” Insufflation in CompetitiveSports. Am J Phys Med Rehabil 2007;86:301–303.

Key Words: Pulmonary, Glossopharyngeal Breathing, Insufflation, Respiratory Ther-apy, Noninvasive Ventilation, Swimming, Sports

Whereas athletes who “blood dope” add intravascular volume for a com-petitive advantage in endurance sports, other athletes attain intrapulmonaryvolumes that exceed maximum inspiratory capacity to yield equally impressivecompetitive advantages. The article by Nygren-Bonnier et al.1, entitled “Effectsof Glossopharyngeal Pistoning for Lung Insufflation on Vital Capacity inHealthy Women,” describes a little-used technique that provides importantbenefits for certain athletes as well as for patients who are too weak to breatheor cough.

Frogs do not have diaphragms. Frogs ventilate their lungs by using glos-sopharyngeal muscles to piston boluses of air past their glottises into theirchests. Because ontogeny recapitulates phylogeny, it is likely that humansperform a similar ventilation technique in utero. Although we apparently forgetit long before we are born, Nygren-Bonnier et al.1 demonstrate that we canrelearn it. They demonstrate that the great majority of a sample of 26 people inthe general population were able to master glossopharyngeal pistoning (GP) toincrease lung volumes to 15% greater than VC and to increase unassisted vitalcapacity (VC) by about 130 ml.1 This means that many people with functioningbulbar-innervated musculature can potentially benefit from GP. Like patientswith neuromuscular weakness, air bolus volumes average 60–200 ml per gulp.We, too, have observed that the majority of unaffected family members ofpatients with neuromuscular disease can learn GP along with the patients andsupplement inspiratory capacity by about 500 ml in this manner on initialattempts. The GP is most easily taught by simply imitating the instructor.

GP must always be done to provide lung volumes that exceed maximuminspiratory capacity, or else it is irrelevant and unmeasurable. Thus, patientswith little or no VC perform it to provide greater tidal volumes than they couldaccomplish autonomously to normalize alveolar ventilation. In this regard, ithas been referred to as “glossopharyngeal breathing” or “frog breathing.” Themajority of patients with little or no VC or respiratory muscle function otherthan for very functional bulbar-innervated muscles can master the techniqueand use it for ventilator-free breathing.2–5 In this way, it can provide security inthe event of sudden ventilator failure or ventilator-interface disconnection for

April 2007 Frog Breathing in Competitive Sports 301

COMMENTARY

Pulmonary

patients using noninvasive mechanical ventilation.Indeed, some patients with no measurable VC haveawakened in the middle of the night breathingglossopharyngeally before realizing that their res-pirators are no longer operating. This ability is oneof the main reasons that most patients with tra-cheostomy tubes and functional bulbar-innervatedmusculature, whether using respiratory support ornot, should be offered decanulation to noninvasivemechanical ventilation as part of their rehabilita-tion.6 Although it can be used somewhat effectivelyby some patients with capped tracheostomy tubes,for the majority of such patients, too much airleaks around the tube and out the ostomy for it tobe very useful.

Patients with reduced VCs also perform GP andother techniques such as air stacking to maximumlung volumes to raise their voices, cough moreeffectively, and expand their lungs.2–5 This hasbeen shown to result in lung volumes that areoften as much as 3 liters greater than VCs (glosso-pharyngeal maximum single-breath capacity),5,6

mild increases in non–GP-assisted VC, and pulmo-nary compliance for patients,7,8 normals,1 and ath-letes.9

Another paper that is currently in press reportsthat elite competitive swimmers use GP to increaseboth VC and chest expansion.10 Sprints of 50 m areoften done without taking a breath. Holding abreath at lung volumes that exceed maximum in-spiratory capacity results in increased buoyancythat might increase speed. In another paper, it wasshown that water submersion could be increasedfrom 309 secs with a deep breath (to maximuminspiratory capacity) to 346 secs with a deep breathsupplemented by GP.11 These athletes added 1.95liters of lung air volume to their VCs by GP.12 Itshould be noted that to increase lung volumesbeyond total lung capacity, lung and chest-wallrecoil pressures must be overcome by GP.12 Themean mouth-relaxation pressure at total lung ca-pacity supplemented by GP was 65 � 19 cm H2O;at times, it exceeded 90 cm H2O.12 These pressurelevels are comparable with those seen in patientswith neuromuscular disorders who perform airstacking to maximum insufflation capacities8 ormaximum lung insufflations with mechanical in-exsufflators three times a day to maintain pulmo-nary compliance.8 These pressures have not beenassociated with barotrauma in the normals, pa-tients, or athletes.

Is this not legal “air doping”? For GP to beeffective, the soft palate must close the nasophar-ynx off to air leakage out of the nose. Swimmers(and patients, for that matter) with incompetentsoft palates could use nose clips to seal the noseand, thereby, perform GP effectively, but thatwould mean requiring a foreign device. Individuals

whose GP is ineffective despite the use of a noseclip could fill the lungs to volumes over inspiratorycapacity by air stacking.2,8 Most simply, this wouldinvolve the use of a manual resuscitator to deeplyinflate the lungs. Although this would be quickerthan GP, racing officials would probably lookaskance at air doping in this manner.

Yet another use for GP is by breath-holdingdeep-sea divers.10 For this application, GP can beused for insufflation, and reverse GP can be usedfor exsufflation to lung volumes below residualvolume. The latter must be strictly a human en-deavor, because it is doubtful that any frog wouldever have reason to attempt such a thing. For fivedeep-sea divers, GP increased lung volumes from aVC of 6.2–9.5 liters to maximum GP lung volumesof 7.8–11.9 liters. The greater the lung volume ofair, the more oxygen is present to support longersubmersion. The increased volume of gas also bet-ter equalizes lung pressure at great depths. Deep-sea divers also master reverse GP to lung volumesbelow residual volume. The five deep-sea divers inthe study by Lindholm and Nyren10 exhaled toresidual volumes of 1.37–2.4 liters without GP andto 1.16–1.77 liters using reverse GP. Because aircannot be exhaled (to the mouth, for example)using expiratory muscles from lungs compressed atdepths to as much as 73 m,10 reverse GP for exha-lation is critical for equalizing pressures in theinner ear and sinuses at great depths. Using GP fordeep lung inflation, then reverse GP for exsuffla-tion, one individual with a VC of 5.5 liters was ableto expel 11.1 liters of air.

Thus, GP can be critical for ventilator-freebreathing ability and for coughing and speech forpatients unable to breathe and cough effectivelyunaided, for successful breath-hold deep-sea dives,and, quite possibly, for competitive swimming. Be-sides being tasty, frogs also have a lot to teach us.

REFERENCES1. Nygren-Bonnier M, Lindholm P, Markstrom A, Skedinger

M, Mattsson E, Klefbeck B: Effects of glossopharyngealpistoning on lung insufflation and vital capacity in healthywomen. Am J Phys Med Rehabil 2007;86:290–294

2. Bach JR, Bianchi C, Vidigal-Lopes M, Turi S, Felisari G:Lung inflation by glossopharyngeal breathing and “airstacking” in Duchenne muscular dystrophy. Am J Phys MedRehabil 2007;86:295–300

3. Bach JR, Alba AS, Bodofsky E, Curran FJ, Schultheiss M:Glossopharyngeal breathing and noninvasive aids in themanagement of post-polio respiratory insufficiency. BirthDefects Orig Artic Ser 1987;23:99–113

4. Bach JR, Alba AS: Noninvasive options for ventilatory sup-port of the traumatic high level quadriplegic. Chest 1990;98:613–9


6. Bach JR. Alternative methods of ventilatory support for thepatient with ventilatory failure due to spinal cord injury.J Am Paraplegia Soc 1991;14:158–74

302 Bach and Tewfik Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 4

7. Bach JR, Kang SW: Disorders of ventilation: weakness,stiffness, and mobilization. Chest 2000;117:301–3

8. Kang SW, Bach JR: Maximum insufflation capacity: therelationships with vital capacity and cough flows for pa-tients with neuromuscular disease. Am J Phys Med Rehabil2000;79:222–7

9. Nygren-Bonnier M, Gullstrand L, Klefbeck B, Lindholm P.Elite swimmers’ buoyancy: effects of glossopharyngeal in-sufflation. Med Sci Sports Exerc (in press)

10. Lindholm P, Nyren S: Studies on inspiratory and expiratoryglossopharyngeal breathing in breath-hold divers employ-ing magnetic resonance imaging and spirometry. Eur J ApplPhysiol 2005;94:646–51

11. Overgaard K, Friis S, Pedersen RB, Lykkeboe G. Influence oflung volume, glossopharyngeal inhalation and PETO2 EurJ Appl Physiol 2006;97:158–64

12. Seccombe LM, Rogers PG, Mai N, et al: Features of glosso-pharyngeal breathing in breath-hold divers. J Appl Physiol2006;101:799–801

April 2007 Frog Breathing in Competitive Sports 303

Authors:Carlos Amo, MD, PhDRobert A. Boyajian, MD, MSJohn S. Romine, MDShirley M. Otis, MD

Affiliations:From the Brain Research andTreatment Center, Division ofNeurology, Department of Medicine,Scripps Clinic, La Jolla, California.

Correspondence:All correspondence and requests forreprints should be addressed toRobert A. Boyajian, MD, MS, BrainResearch and Treatment Center,Division of Neurology MS-313,Scripps Clinic, 10666 North TorreyPines Road, La Jolla, CA 92037.

Disclosures:This work was presented in abstractand poster form at the AmericanAcademy of Neurology AnnualMeeting in San Diego, California,April 2006.


DOI: 10.1097/PHM.0b013e3180383c0d

High-ResolutionMagnetoencephalographicFunctional Mapping of the CorticalNetwork Mediating IntentionalMovement

ABSTRACT

Amo C, Boyajian RA, Romine JS, Otis SM: High-resolution magnetoencephalo-graphic functional mapping of the cortical network mediating intentional move-ment. Am J Phys Med Rehabil 2007;86:304–309.

Magnetoencephalography (MEG) is a sensitive technique that can detect andmap cortical electrophysiologic activations with high spatial (mm) and temporal(msecs) resolutions. We used 148-channel whole-head MEG to record theactivation sequence for the somatosensory and motor cortical network duringcued hand movements in a healthy 39-yr-old subject. The complex sequence andtopography of cortical activations were superimposed onto the subject’s brainmagnetic resonance images. Frontal premotor and supplementary motor andcingulate areas activated well before the primary motor area and again repetitivelyfrom 200 msecs onward with activations alternating repeatedly between frontaland parietal areas. The network’s very close functional integration of supplemen-tary motor areas suggests how brain injury that is localized to these regions, butnot to the primary motor area itself, can disrupt integrity of movement, and whypreservation of functional integrity of some areas traditionally viewed as extramo-tor may be necessary for recovery from neurologic disability.

Key Words: Magnetoencephalography, MEG, Brain Mapping, Cortical Activation,Neurophysiology

Magnetoencephalography (MEG) is a sensitive neuroimaging technique thatcombines high spatial (mm) and temporal (msecs) resolutions of functional brainactivity.1 This study seeks to use the advantages of MEG to pinpoint and functionallymap the human cortical network mediating voluntary movement. Previous MEGstudies have mapped and validated the reliability of localizing the cortical sourcesof the major latency events corresponding to the major visible peaks in theevent-related, signal-averaged magnetic waveform associated with intentionalmovement.1–9 In this study, we seek to evaluate whether magnetic dipole localiza-tions can be obtained for cortical activations occurring throughout an event-relatedperiod without limitation to the major visible latency peaks. If the technical


BRIEF REPORT

Imaging

feasibility of this concept could be established, thenexhaustive mapping of cortical activations mediatingintentional movement would be routinely available.Such capability could be used to illuminate networkcomponents and their relationships in greater detailthan has been previously reported. Such deepenedinformation regarding the functional organization ofhuman cerebral cortex could then be applied to un-derstanding healthy and neurologic disease (or in-jury) conditions.

METHODSA healthy, 39-yr-old, left-handed male subject

gave informed consent to participate in this study.The subject was instructed to respond to a sensorycue with a brisk full-hand extension and then ac-tive flexion of the wrist back to the original restingposition. A brief practice session was performed toensure optimal performance (i.e., adequate andconsistent response timing). The sensory stimuluswas a painless tactile cue consisting of a 15-psiairpuff of 30-msec duration applied at the skinsurface of the tip of the index finger of the hand tobe moved. The stimulus was applied with an inter-stimulus interval of 2 secs. Three hundred stimuli(trials) were applied to the right hand, and then300 stimuli were applied to the left hand. Head-shape digitization using anatomic and electrode(electrooculogram) fiducial points was importedinto the MEG software. MEG activity was recordedcontinuously throughout the entire 600 trials by aMagnes 2500 whole-head system with 148 channels(4D NeuroImaging Technologies, Inc., San Diego CA)inside a sound-damped and magnetically shieldedroom. Signal acquisition was conducted at a samplerate of 678.17 Hz with a band-pass filter of 0.1–100Hz. Data postprocessing for the offline analysis in-cluded digitization, filtering (0.1–40 Hz), and signalaveraging (300 trials).

A single equivalent dipole model was used(equivalent-current dipole) to calculate the spatiallocalization of the neuronal currents. The dipole-fitalgorithm selection criteria were (1) amplitude orglobal field power of each magnetic response �400fT (expressed as the root mean square of magneticflux integrated over all sensors), (2) strength of themagnetic moment had to be �400 nAm-m in am-plitude so that the current density would be con-sistent with a focal source, (3) correlation �0.90between the recorded measurements and the val-ues expected from the dipole estimate from theforward equation, (4) goodness of fit �0.90 foragreement of the observed measurements and theresulting dipole fit with the model, and (5) 95%confidence that the real dipole fell in a volume lessthan 15 cm3, on the basis of a comparison of signalstrength and the background noise as measured inan interval before the trigger.

Whole-head isocontour maps of the magneticfield strengths were evaluated visually for the pres-ence of dipoles and were also explored algorithmi-cally using the equivalent-current dipole software,with selection stringency set at the dipole-qualifi-cation criteria described above. The isocontourmaps and magnetic dipole solutions were evaluatedby scrolling millisecond by millisecond through1000 msecs (0–1000 msecs after stimulus) of theevent-related averaged waveforms. Magnetic reso-nance imaging (MRI) of the brain was performedaccording to routine clinical protocol on a GeneralElectric Signa 1.5-T system using contiguous1.0-mm T1-weighted slices. MRI was imported intothe MEG software in Digital Imaging and Commu-nications in Medicine format. Anatomic landmarkswere visualized on the MRI T1 axial plane. TheMEG dipole localizations were overlaid onto mag-netic resonance images by coregistration of MEGfiducial points and MRI anatomic landmarks usingSpatio Temporal Analysis/Review software.10 Di-poles in the hemisphere contralateral to the handmovement were counted and portrayed graphi-cally, and a schematic summary was made of thetemporal predominance of the motor- vs. somato-sensory-area dipole generators produced in contig-uous 50-msec epochs (predominance was definedas �80% of epochal dipoles). Good replicability ofthe technique has been demonstrated in a similarpneumatically cued hand-flexion study design,which showed close similarity of MEG waveformsacross three different days and close proximity ofdipolar sources computed at the peak of each com-ponent. Such MEG-derived localizations of primarysensory and motor cortex have also been verifiedusing invasive recordings during surgery.9

RESULTSSignal averaging of the 300 trials in each hand

yielded event-related waveforms corresponding to so-matosensory- and motor-evoked fields. The equiva-lent-current dipole calculations pinpointed magneticdipoles (cerebral activations) from cortical sourcesthroughout the period from the time of the centralreceipt of the stimulus to the completion of the wristmovement. All magnetic dipole source activationslocalized to a radius less than 1 cm.

Figure 1 shows the sequence of the timeframes and the changing loci of cortical activationsfrom 0 to 200 msecs after the stimulus. This se-quence of time frames is the cortical representa-tion of a physiologic event, with the limits of eachinterval (time frame) based on the latencies of thefirst and last dipoles of every group anatomicallylocated in the MRI. It can be seen from Figure 1that the spatiotemporal pattern of cortical activa-tion unfolds similarly for either right- or left-handmovement (in their contralateral hemispheres, re-

April 2007 MEG of Cortical Activity in Voluntary Movement 305

spectively). In particular, the operation of the pre-motor activations identifiable within the supple-mentary motor area (SMA) and prefrontal motor(53–63 msecs) and cingulate areas (65–108 msecs)precedes primary motor cortex activation (130–185 msecs) in a similar manner, regardless ofwhich hemisphere is engaged.

In each hemisphere, contiguous epochs of 50-msec duration from 200 to 1000 msecs, coincidentwith the actual extension–flexion wrist movement,were evaluated for predominance of motor vs. so-matosensory activations. Epochs were defined as

predominantly motor or predominantly somato-sensory when more than 80% of epochal dipoleslocalized to the motor area or the somatosensoryareas, respectively. Epochs were defined as simul-taneous motor and somatosensory when epochal di-poles localized approximately equally to the motorarea and the somatosensory areas. The schematicpresentation in Figure 2 uses these conventions toillustrate the sequence of changing predominanceof cortical activations between the motor and sen-sory areas. This pattern of reciprocating predomi-nance between the motor and somatosensory areas

FIGURE 1 Cortical dipole source representations of hand movement task during the first 200 msecs afterstimulus. The left panel shows the spatiotemporal phases of the left-hemisphere activations (corre-sponding to the right-hand task), and the right panel represents right-hemisphere activations(corresponding to the left-hand task). Five discrete sequential topographies (phases) of somatosensoryand motor response are evident in the cortical areas of each hemisphere. The limits for each interval(time frame) represent the range of latencies for each cortical phase according to the latencies of thefirst and the last dipoles of every group anatomically located in the MRI.

306 Amo et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 4

during contralateral hand movement seem to bestructured with a similar physiology (in locationsand timing), regardless of which hemisphere isbeing activated.

DISCUSSIONOur findings establish the technical feasibility

of the use of MEG to map the instantaneous tra-jectory of brain cortical processing during a cuedmovement. In the current study, this approach wasused to generate a sample operation, from a single,healthy individual, of a model cortical network.The results in this single subject unify and corrob-orate a large body of reported latency events cor-responding to select major peaks in the signal-averaged waveform across a large number ofsubjects.1–9,11 On the basis of these results, wesuggest that MEG mapping of the instantaneousoperations of the cortical networks on a millisec-ond-by-millisecond basis can now be applied tolarger populations. Methodological refinementssuch as self-triggered movements would help toisolate the pure volitional and motor planningcomponents without the influence of any externalstimuli. In principle, mapping the instantaneoustrajectory of cortical processing could be used toexplore the organization and operational status ofany desired cortical network.

The key correspondences between the modelsomatomotor network and known latencies andtheir cortical origins can be summarized in phases.In the first phase, from 16 to 51 msecs, the primarysomatosensory response forms predominantly in

the postcentral gyrus, consistent with other MEGstudies reporting 20–40 msecs.1,11 In the secondphase, between 53 and 63 msecs, cortical activationpredominates in the frontal areas for movementpreparation (SMA, prefrontal motor, etc.). This isin agreement with previous descriptions of theactivation of these areas during the preparationand execution of simple8 and sequential12 volun-tary movement. Functional MRI, a technique oflower temporal resolution, shows that the activa-tion of these areas precedes M1.13 We show that M2(premotor and supplementary motor activity) pre-cedes M1 by 70–100 msecs, in close agreementwith MEG studies that have shown that this SMAand prefrontal motor activation precede M1 by 80–120 msecs.6,9

In the third phase, from 65 to 108 msecs, thesecondary somatosensory areas, supramarginal gy-rus, posterior cingulate cortex, and insula are allactivated; others have shown between 90 and 100msecs.1,11,14 This activity has been interpreted asrelated to movement preparation through thetransmittal of sensory information to the premotorcortex.6–7,15 The posterior cingulate cortex is acti-vated by sensory stimulation16 that can be seenwith functional MRI2 and MEG.6 In the fourthphase, from 130 to 185 msecs, the primary motorcortex becomes activated along the precentral gy-rus (M1) in agreement with neuroimaging tech-niques and MEG.2,6 In the fifth phase, from 169 to198 msecs, the areas for the secondary somatosen-sory areas, supramarginal gyrus, posterior cingu-late cortex, insula, superior temporal gyrus, and,

FIGURE 2 The changing temporal predominance of motor- and somatosensory-area cortical activations foreach hemisphere during contralateral wrist movement (in contiguous epochs of 50 msecs from 200to 1000 msecs after stimulus). Epochs are labeled as predominantly motor or predominantlysomatosensory when �80% of epochal dipoles localized either to the motor or to the somatosensoryareas, respectively. Epochs are labeled as simultaneous somatosensory and motor when epochaldipoles localized approximately equally to the motor and the somatosensory areas. Within eachhemisphere, the cortical activations reciprocate between predominantly somatosensory and predom-inantly motor areas. The pattern of the temporal progression and reciprocating phenomenon seemsto unfold in the same manner regardless of which hemisphere is being activated by the contralateralhand movements.


less intensively, the SMA, are all activated. Thisactivation is described around 200 msecs1 and isinterpreted as being related to the planning andexecution of movement.6–7 In the follow-on activ-ity between 200 and 1000 msecs, corresponding tothe actual extension and flexion movements, M2and the secondary somatosensory areas, as well asother parietal areas, become activated with similarlatencies in both hemispheres, as has been re-ported from MEG.9,17

The correspondence between our findings andthese results from many studies across many sub-jects suggests that the present model network isgenerally valid. Larger studies are needed to pro-duce model refinements and validation within andacross selected populations, and to obtain norma-tive ranges for the overall trajectory and its com-ponent-activation latencies and locations. An im-portant limitation of the technique and, thus, thepresent model network, is that it identifies andincorporates only a subset of all activation events.This limitation exists because MEG is believed todetect primarily sulcal events, being insensitive toevents localized within the convexities of the cor-tices (surface of the brain) because of differingorientation of the magnetic fields generating fromthese sites.1 Therefore, contributions to networkprocesses from cortical neurons at the crowns ofthe gyri might not be registered using the MEGtechnique. This spatial gap in mapping corticalnetwork components might be complemented bythe use of other functional neuroimaging tech-niques such as functional MRI. However, no func-tional neuroimaging technique other than MEGprovides the fine-scale temporal resolution neededto precisely identify when a particular gyrus acti-vates in relation to the activation of other networkcomponents.

The present findings present a sample of amodel cortical network. This network mediatingintentional movement operates through dense re-ciprocating communications among frontal sup-plementary motor, premotor, cingulate, and so-matosensory areas. First, there is a relativelyconcentrated period of cortical processing in theprefrontal motor and SMA and cingulate gyrus,anteceding M1 activation by 70–100 msecs. Thisprocessing likely corresponds to movement plan-ning and preparation. We have recently confirmedthis through a related MEG study of imaginedmovement conducted in the same subject. Themental plan for movement was mediated predom-inantly at M2 (76–91% of frontal motor area acti-vations, left and right hemispheres, respectively).Additionally, it was also shown that the somatosen-sory stimulation alone does not elicit any M2 acti-vation.18 Second, there is the reciprocating flow ofinformation between these frontal areas and pari-

etal lobes during movement execution. This likelycorresponds to a position-checking or propriocep-tive function. We surmise, from the very closefunctional integration of these structures withinthe network, that damage to the supplementarymotor, premotor, or cingulate areas might impedeboth pre-M1 movement preparation and proprio-ceptive integrating functions during movement ex-ecution. This close functional integration may ex-plain why even small, discrete lesions localized tothese regions can produce substantial motor dis-ability, such as immediate paresis and dysarthriaafter neurosurgical resections of brain tumors lo-calized only to the SMA.19 The model’s close inte-gration of the supplementary motor, cingulate, andinsula also explains why the level of functionalintegrity of each of these areas seems to affect thedegree of impairment and recovery of motor func-tion after stroke.20

The model network presented here, based on asingle equivalent-current dipole technique for di-pole localization, provides insight that is useful forevaluating and investigating a broad range of mo-tor-associated disabilities. For example, MEG couldbe used to quickly and easily screen symptomaticindividuals to identify the level of integrity or lossof network-component functions, possibly en-abling early classification of capacity for recovery.This information is greatly needed during the ear-liest stages of brain injuries, including stroke, headinjury, or treatment of brain tumor. This principlehas been explored using MEG to monitor a unique-latency event and site, where a change in thesource strength of the first motor-evoked fieldcomponent (postmedian nerve stimulation) of themotor cortex and postcentral gyrus after stroke wasshown to correlate with the extent of recovery ofsensorimotor function.21 The present approach us-ing MEG monitors all detectable cortical activa-tions with their instantaneous trajectories acrossall gyri and throughout the full span of event-related latencies. This more comprehensive ap-proach vastly expands the capability to investigateand relate focal and global brain plasticity to clin-ical outcomes. In particular, the advantages of suchan approach could lead to the development of ob-jective quantitative indices that are highly prog-nostic regarding a patient’s individual capacity forrecovery from neurologic disability.

ACKNOWLEDGMENTSWe thank the Scripps MEG technical staff

members Patti Quint and Lacey Kurelowech fortheir expert assistance and dedicated support.

REFERENCES1. Hamalainen M, Hari R, Ilmoniemi RJ, Knuutila J, Lounasmaa

OV: Magnetoencephalography: theory, instrumentation, and

308 Amo et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 4

applications to non-invasive studies of the working humanbrain. Rev Mod Phys 1993;65:413–97

2. Ehrsson HH, Geyer S, Naito E: Imagery of voluntary move-ment of fingers, toes, and tongue activates correspondingbody-part-specific motor representations. J Neurophysiol2003;90:3304–16

3. Ikeda A, Luders HO, Burgess RC, Shibasaki H: Movement-related potentials recorded from supplementary motor areaand primary motor area. Role of supplementary motor areain voluntary movements. Brain 1992;115:1017–43

4. Rektor I, Feve A, Buser P, Bathien N, Lamarche M: Intrace-rebral recording of movement related readiness potentials: anexploration in epileptic patients. Electroencephalogr Clin Neu-rophysiol 1994;90:273–83

5. Picard N, Strick PL: Imaging the premotor areas. Curr OpinNeurobiol 2001;11:663–72

6. Huang MX, Harrington DL, Paulson KM, Weisend MP, LeeRR: Temporal dynamics of ipsilateral and contralateral mo-tor activity during voluntary finger movement. Hum BrainMapp 2004;23:26–39

7. Hoshiyama M, Kakigi R, Berg P, et al: Identification ofmotor and sensory brain activities during unilateral fingermovement: spatiotemporal source analysis of movement-associated magnetic fields. Exp Brain Res 1997;115:6–14

8. Erdler M, Beisteiner R, Mayer D, et al: Supplementarymotor area activation preceding voluntary movement isdetectable with a whole-scalp magnetoencephalographysystem. Neuroimage 2000;11:697–707

9. Castillo EM, Simos PG, Wheless JW, et al: Integratingsensory and motor mapping in a comprehensive MEGprotocol: clinical validity and replicability. Neuroimage2004;21:973–83

10. Schwartz DP, Badier JM, Bihoue P, Bouliou A: Evaluation ofa new MEG-EEG spatio-temporal localization approach us-ing a realistic source model. Brain Topogr 1999;11:279–89

11. Forss N, Silen T: Temporal organization of cerebral events:

neuromagnetic studies of the sensorimotor system. RevNeurol (Paris) 2001;157:816–21

12. Dick JP, Benecke R, Rothwell JC, Day BL, Marsden CD:Simple and complex movements in a patient with infarctionof the right supplementary motor area. Mov Disord 1986;1:255–66

13. Atlas SW, Howard RS 2nd, Maldjian J, et al: Functionalmagnetic resonance imaging of regional brain activity inpatients with intracerebral gliomas: findings and implica-tions for clinical management. Neurosurgery 1996;38:329–38

14. Forss N, Jousmaki V, Hari R: Interaction between afferentinput from fingers in human somatosensory cortex. BrainRes 1995;685:68–76

15. Petrides M, Pandya DN: Projections to the frontal cortexfrom the posterior parietal region in the rhesus monkey.J Comp Neurol 1984;228:105–16

16. Vogt BA, Finch DM, Olson CR: Functional heterogeneity incingulate cortex: the anterior executive and posterior eval-uative regions. Cereb Cortex 1992;2:435–43

17. Babiloni F, Carducci F, Cincotti F, et al: Linear inversesource estimate of combined EEG and MEG data related tovoluntary movements. Hum Brain Mapp 2001;14:197–209

18. Amo C, Criado JR, Otis SM: Magnetoencephalogram record-ing from secondary motor areas during imagined move-ments. Arg Neuropsiquiatr 2006;64:394–7

19. Krainik A, Duffau H, Capelle L, et al: Role of the healthyhemisphere in recovery after resection of the supplemen-tary motor area. Neurology 2004;62:1323–32

20. Carey LM, Abbott DF, Egan GF, Bernhardt J, Donnan GA:Motor impairment and recovery in the upper limb afterstroke: behavioral and neuroanatomical correlates. Stroke2005;36:625–9

21. Huang M, Davis LE, Aine C, et al: MEG response to mediannerve stimulation correlates with recovery of sensory andmotor function after stroke. Clin Neurophysiol 2004;115:820–33


Authors:Henry L. Lew, MD, PhDCarl P. C. Chen, MDTyng-Guey Wang, MDKelvin T. L. Chew, MD

Affiliations:From the Division of Physical Medicineand Rehabilitation, Stanford UniversitySchool of Medicine/VA Palo Alto HealthCare System, Palo Alto, California(HLL); Sports Medicine Center,Department of Orthopaedic Surgery,Alexandra Hospital, Singapore (KTLC);Department of Physical Medicine andRehabilitation, Chang Gung MemorialHospital, Linkou, Taiwan; School ofMedicine, Chang Gung University,Taiwan; King’s College, London, UnitedKingdom (CPCC); and National TaiwanUniversity Hospital, School ofMedicine, Department of PhysicalMedicine and Rehabilitation, Taiwan(T-GW).

Correspondence:All correspondence and requests forreprints should be addressed to Tyng-Guey Wang, MD, Department ofPhysical Medicine and Rehabilitation,National Taiwan University Hospital,School of Medicine, National TaiwanUniversity, Taiwan.


DOI: 10.1097/PHM.0b013e31803839ac

Introduction to MusculoskeletalDiagnostic UltrasoundExamination of the Upper Limb

ABSTRACT

Lew HL, Chen CPC, Wang T-G, Chew KTL: Introduction to musculoskeletaldiagnostic ultrasound: examination of the upper limb. Am J Phys Med Rehabil2007;86:310–321.

With recent advances in computer technology and equipment miniaturization, theclinical application of diagnostic ultrasonography (U/S) has spread across variousmedical specialties. Diagnostic U/S is attractive in terms of its noninvasiveness,lack of radiation, readiness of use, cost-effectiveness, and its ability to makedynamic examinations possible. Dynamic imaging deserves special emphasisbecause it is useful in differentiating full-thickness from partial-thickness tendontears, muscle tears, and tendon and nerve subluxations or dislocations. It is alsoa quick and easy avenue for side-to-side comparisons. When appropriately used,diagnostic U/S can be considered as an extension of one’s physical examination.However, there are limitations of U/S, which will be discussed in this reviewarticle. This is part 1 of two articles; this first part will focus on the ultrasoundexamination of the upper extremity, using selected examples relevant to muscu-loskeletal medicine. Part 2 will cover common pathologies of the lower extremity.

Key Words: Musculoskeletal, Ultrasound, Upper Limb

Ultrasonography (U/S) is an imaging modality that uses sound waves in thehigher frequency range of �20,000 Hz, which normally cannot be heard byhuman beings. Ultrasound travels as a longitudinal wave, and images aregenerated when pulses of ultrasound from the transducer produce echoes attissue or organ boundaries.1 Some of the waves are absorbed by the tissues, andthe extent to which the ultrasound is absorbed or reflected gives informationabout the structures scanned, as illustrated in Figure 1. Resolution is defined asthe smallest distance that can be discriminated in the image. Better resolutionsare attained with higher frequencies. But, in doing so, signals are attenuated,decreasing the depth of field.2 For example, a 7.5-MHz transducer wouldproduce imaging depths of up to 8 cm with an average resolution of 0.20 mm,whereas a 10.0-MHz transducer would produce imaging depths of 6 cm or less,with a sharper resolution of 0.15 mm.

Between pulse transmissions, the transducer serves as a detector of echoes,which are processed to form an anatomic image. For most musculoskeletal diag-nosis, the most useful frequency ranges for the transducer are between 7.0 and 12.0


INVITED REVIEW

Imaging

MHz. This article will review the ultrasound appear-ance of normal tissues in the upper limb and will useclinical examples to demonstrate pathologic changesas they appear on ultrasound.

TERMINOLOGY AND EQUIPMENTCONSIDERATIONS

Interpretation of ultrasound images dependson the echogenicity: the brightness of the image,depending on the degree of reflection of the ultra-sound waves. Terms used include hyperechoic,isoehoic, hypoechoic, and anechoic. The imagesare also described in terms of the plane on whichthe sonogram is viewed, which is usually longitu-dinal or transverse in relation to the structurescanned. Common terminology used in musculo-skeletal ultrasound is defined in Table 1. Correctselection and configuration of equipment is criticalfor musculoskeletal U/S. The choice of transducerused depends on the size and location of the mus-culoskeletal structure to be imaged. Generally, lin-ear transducers are used with high-frequencytransducers (7.5–20 MHz) that have higher-resolu-tion imaging but poorer tissue penetrance, makingthem ideal for small, superficial structures. Low-frequency transducers (�7.5 MHz) have poorerresolution but excellent tissue penetrance; theseare preferable for larger, deeper structures. Imagesin this review were attained from the HDI 5000(Advanced Technology Laboratories, Bothell, WA)

and Xario (SSA 660A, Toshiba Corporation, Japan)with variable-frequency transducer probes between5 and 20 MHz.

NORMAL AND PATHOLOGIC U/SAPPEARANCE OF THE TENDON

Assessment of tendon integrity is one of thebest applications of musculoskeletal U/S.3–5 Ten-dons are recognized by parallel and fine fibrillarpatterns on U/S in the longitudinal view, as shownin Figure 2. The parallel fascicles of collagen fibersproduce hyperechoic lines, whereas the interfascic-ular ground substance produce anechoic lines inbetween.6 In the transverse view, tendons appear asround or oval hyperechoic structures. Anisotropy isa characteristic feature of U/S of tendons and liga-ments, where echogenicity of the structure changesdepending on the angle of the U/S beam, as illustratedin Figure 3. The image appears hyperechoic when thebeam is perpendicular to the tendon and hypoechoicwhen the beam is oblique, which may lead to misin-terpretation.7 This characteristic is useful in identify-ing the scanned structure as either a tendon or liga-ment.

FIGURE 2 Normal biceps tendon. The ultrasound scanlongitudinal to the biceps tendon showsnormal, hyperechoic, parallel fibrillarpattern (arrows).

FIGURE 1 Ultrasound pulse transducer.

TABLE 1 Terminology in musculoskeletal ultrasound

Term Definition

Echogenicity Capacity of a structure in the path of an ultrasound beam to reflect back sound waves.Hyperechoic The structure examined in the ultrasound image shows a high reflective pattern and appears

brighter than the surrounding tissue.Isoechoic The structure demonstrates the same echogenicity as the surrounding soft tissues.Hypoechoic The structure examined in the ultrasound image shows a low reflective pattern, manifesting as

an area where the echoes are not as bright as the surrounding tissue.Anechoic The image of the structure shows no internal echoes (e.g., simple fluid).Longitudinal Scan is lengthwise and parallel to the long axis of the structure, organ, or body part.Transverse Scan is crosswise and at right angles to the long axis of the structure, organ, or body part.

April 2007 Diagnostic Ultrasound of the Upper Limb 311

The extent and mechanisms of tendon injury canbe demonstrated by U/S through passive or resisteddynamic examination.8 Tendon degeneration on ul-trasound is seen as irregularities of fibrillar appear-ance, such as thickening and fragmentation, focalhypoechoic areas, and calcifications.2,9,10 In tendonswith synovial sheath, chronic tendinosis is character-ized by widening of the tendon sheath, loss of normalfibrillar echotexture, and loss of definition of tendonmargins. In tendons without synovial sheath, thepathology is characterized by focal or diffused thick-ening of the tendon, with loss of fibrillar echotextureand patches of hypoechogenicity. Tendon ruptures,which can range from partial to complete to massive,appear as fragmented, contiguous fibrils. It is difficultto draw the distinction between tendon degenerationand intrasubstance tears in the absence of hematoma.This is especially so because the two conditions arenot mutually exclusive. Complete tears of the tendonare characterized by retraction of torn edges, withhypoechoic hematoma or granulation tissue.11 Pas-sive movement to accentuate the tendon interruptionis a useful maneuver in U/S examination of a sus-pected tendon tear.12 In tendons with synovialsheath, fluid can collect in the space between theretracted ends of the tendon.13 Partial-thickness tears

present with a combination of intact and retractedruptured portions of the tendon, often accompanied byhematoma.5

FIGURE 4 Lateral view of the right shoulder, show-ing the rotator cuff muscles.

FIGURE 5 Sequence for rotator cuff ultrasound. 1) Ul-trasound transducer placement for imagingthe biceps tendon, with the forearm restingin a supinated position on the thigh. 2)Ultrasound transducer placement for imag-ing the subscapularis, with the arm exter-nally rotated. 3) Ultrasound transducerplacement for imaging the supraspinatus,with the hand in a back pocket, palm to-ward the gluteal muscles, and the elbowdirected posteriorly. 4) Ultrasound trans-ducer placement for imaging the infraspi-natus, teres minor, and posterior glenohu-meral joint, with an arm across the chestand the hand on the opposite shoulder.

FIGURE 3 Anisotropy of normal biceps tendon. a, Transverse sonogram of hyperechoic biceps tendon (arrow)when the ultrasound beam is perpendicular to the tendon. b, Hypoechoic biceps tendon (open arrow)when the ultrasound beam is not perpendicular to the tendon.

312 Lew et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 4

ROTATOR CUFF U/SRotator cuff disease is common and is one of

the most common reasons for using U/S. Matsenand coworkers14 have suggested diagnostic U/S as aprimary imaging technique for soft-tissue injuriesof the shoulder. The main advantages are its abilityto perform dynamic examinations and to conductside-to-side comparisons on the spot. Many studiesreport excellent sensitivity and specificity of ultra-sound in diagnosing rotator cuff tears.15–20 U/Sdiagnosis of full-thickness rotator cuff tears has anoverall accuracy of up to 96%.21 Recently, Teefeyand colleagues22 found comparable accuracy of di-agnostic ultrasound and magnetic resonance im-aging for diagnosis and measurement of rotatorcuff tears. Iannotti and coworkers,23 in their eval-uation of office-based U/S by orthopedic surgeonsmeasuring full-thickness rotator cuff tears, haveyielded similar findings. However, several studieshave reported less than satisfactory results in termsof accuracy.24–26 Interstudy comparison is difficultbecause of a lack of standardization of technique,clinical experience of different operators, andchanges in equipment over time.

Understanding the anatomy of the rotator cuffis essential for successful shoulder U/S examina-tion, because the overlying bony structures createobstacles to ultrasound imaging. The rotator cuffconsists of the supraspinatus, subscapularis, in-fraspinatus, and teres minor tendons. The relationsof the rotator cuff are illustrated in Figure 4. Thelong head of the biceps runs in the interval be-tween the supraspinatus and the subscapularis ten-dons from the superior glenoid tubercle. Thesemuscles provide dynamic stability to the inherentlyunstable, extremely mobile glenohumeral joint.The subscapularis originates anteriorly from thescapula and inserts into the lesser tuberosity of thehumerus. The supraspinatus originates posteriorly

from the scapula above the scapular spine andinserts into the anterior aspect of the greater tu-berosity. The infraspinatus originates posteriorlyfrom the scapula below the scapular spine andinserts into the greater tuberosity of the humerusposteriorly to that of the supraspinatus. The teresminor arises posteriorly from the scapula and insertsposteriorly and inferiorly to that of the infraspinatusat the greater tuberosity. The three posterior muscletendons have a common insertion into the greatertuberosity and are difficult to distinguish.

Standardized Technique in Examinationof the Shoulder

The shoulder is an important, yet complicated,joint to examine. Proper positioning of the patientis important for successful U/S of the shoulder.Typically, the patient is seated upright on a revolv-ing stool. The examination should be systematic,with predetermined structures scanned step bystep. U/S of the shoulder begins with the long headof the biceps tendon, which is often used as areference landmark. The biceps tendon is examinedin the longitudinal and transverse planes with thepatient’s forearm or hand resting in a supinatedposition on the thigh. Moving from the bicepstendon medially is the subscapularis, which is bestexamined with the patient’s arm in external rota-tion. The tendon is traced from the bicipital grooveand lesser tuberosity. Imaging of the supraspinatusis obstructed by the overlying acromion. The ma-neuver for exposing the supraspinatus beneath theacromion anteriorly is to have the patient put ahand in his or her back pocket with the palmtoward the gluteal muscles while keeping the el-bow directed posteriorly. The tendon is examinedin perpendicular planes, bearing in mind that theaxis of the tendon is approximately 45° between thesagittal and coronal planes of the body. The poste-

FIGURE 6 Normal supraspinatus tendon (SST). Bur, bursal surface; Art, articular surface of the tendon; GT,greater tuberosity of the humerus (H). a, The longitudinal scan resembles a parrot’s beak. The deltoidmuscle (Del) has relatively lower echogenicity compared with the SST. b, The transverse scan showsparallel convexity.


rior glenohumeral joint, infraspinatus, and teresminor are examined by putting the patient’s armacross the chest with his or her hand on the oppo-site shoulder. The posterior cuff is examined bytracing from the bony landmark of the spine of thescapular and moving the transducer inferiorly tothe infraspinatus and then laterally to visualize theposterior glenohumeral joint.

Because the cuff tendons inserting into thegreater tuberosity are relatively indistinct fromeach other, it is difficult to distinguish them. Oneway to tell them apart is by sequential measure-ments. The supraspinatus forms approximately1.5–2 cm of width on the transverse plane, startingfrom the edge of the biceps tendon, and the in-fraspinatus forms the next 1.5 cm posteriorly. Fig-ure 5 illustrates the sequence of ultrasound exam-ination of the shoulder.

Supraspinatus Tendon PathologiesThe normal longitudinal ultrasound appear-

ance of the supraspinatus resembles a parrot’s

beak. The transverse view of the supraspinatusshows the parallel convexity of the subacromial–subdeltoid bursa above and the humeral epiphysisbelow, as illustrated in Figure 6.

Supraspinatus TendinosisThe term “tendinosis” (or “tendinopathy”) has

superseded the term “tendonitis” as studies haveshown the absence of active inflammation in theseconditions.27 Tendinosis appears as focal or diffuse,poorly demarcated hypoechoic regions accompa-nied by swelling. Confusion often occurs here be-cause partial tendon tears may appear hypoechoic.The presence of an internal fibrillar pattern and thelack of tendon atrophy differentiate tendinosisfrom partial tears. A markedly echogenic appear-ance with posterior acoustic shadowing arisingfrom the tendon substance can occur in calcifictendinopathy, as shown in Figure 7.

Full-Thickness Supraspinatus TearRotator cuff tears are characterized by the de-

gree of tear (i.e., either partial or full thickness),the amount of tendon retraction in the longitudi-nal plane, and the width of the defect in the trans-verse plane. Full-thickness tears can present withnonvisualization of the rotator cuff, where there istotal absence of the supraspinatus tendon on U/S.28

This feature can be seen in massive rotator cufftears, which are associated with a high-riding hu-meral head on radiographs. The fluid collectionbetween the deltoid and humerus may be mistakenas the supraspinatus. In the absence of the su-praspinatus tendon, compression with the ultra-sound probe will obliterate this space, as illustratedin Figure 8. In addition to compressibility, the fluidshould not be mistaken for cuff tissue, because thereis no internal fibrillar echotexture of the fluid.

The edge of the tendon stump can be taperedoff to fibrosed synovium.29 This produces a contour

FIGURE 7 Calcific tendinosis of the supraspinatus ten-don. Echogenic foci with acoustic shadow-ing (arrow).

FIGURE 8 Massive tear of the rotator cuff. Longitudinal ultrasound scan showing nonvisualization of thesupraspinatus tendon. a, Effusion in the space (open arrow) between the deltoid muscle and humeralhead. b, Compression with the ultrasound probe obliterates (arrow) this space in between.


alteration in that the normal outer convex borderof the rotator cuff is flattened or becomes concave.The most common ultrasound feature of full-thick-ness tears of the supraspinatus is the hypoechoicdefect, which appears as sharp demarcations fromthe bursal to the articular surface of the tendon.30

Shoulder effusions and bursal fluid have beenshown to correlate strongly with rotator cuff ab-normalities.31 It distributes within the glenohumeraljoint, to the tendon sheath of the long head of biceps,to the subacromiodeltoid bursa through the tear andthrough the acromioclavicular joint to produce thegeyser sign, as illustrated in Figure 9.31 The loss oftendon causes the deltoid at the bursal surface tosink into the gap to produce the deltoid herniationsign, and accompanying exaggeration of articularhyaline cartilage produces a double-cortex appear-ance as well as cortical irregularities.32 These fea-tures can be seen in Figure 10.

Partial-Thickness Supraspinatus TearA partial-thickness tear appears as a hypo-

echoic area within or at the bursal or articularaspect of the tendon, usually located at the criticalarea over the anatomic neck of the humerus.19

Differentiating between partial tears and severe lo-calized degeneration of the tendon can be difficultusing U/S, which is less sensitive in such cases thanit is for detecting full-thickness tears.21 The intra-substance tears are hypoechoic areas within thetendon substance with intact articular and bursalsurfaces. Articular-surface tear can be seen as ahypoechoic defect that continues to the articularsurface of the tendon. Cortical irregularity is acommon finding at the articular extension of atendon tear.33 In bursal-surface tears, the hypo-echoic defect is in continuity with the bursal sur-face of the tendon. Ultrasound appearance of both

FIGURE 9 a, Normal acromioclavicular joint. b, Acromioclavicular joint effusion demonstrating the geyser’s sign, asshown by arrows. Acr, acromion; Clav, clavicle.

FIGURE 10 Full-thickness tear of the supraspinatus tendon. a, Longitudinal sonogram depicting a focal hypo-echoic defect (open arrow) with deltoid muscle (Del) herniation from above. Exaggeration of cartilagereflection is seen here in the absence of an overlying tendon. Cortical irregularities are seen at thegreater tuberosity (GT). b, Transverse sonogram showing double-cortex sign (arrows), representingthe articular hyaline cartilage above and the cortex of the humeral head below.


articular- and bursal-surface partial tears of thesupraspinatus are illustrated in Figure 11.

Although most partial tears occur in the crit-ical zone of the supraspinatus tendon, some tears,commonly known as rim rent tears, involve asmall, articular surface avulsion adjacent to thegreater tuberosity.34 This type of tear appears as asmall, hypoechoic defect with a central hypere-choic line on the articular surface. The transverseview appears as a bull’s eye lesion, with centralpunctuate echo surrounded by a hypoechoic haloof fluid or edematous tendon.

Biceps TendinosisThe long head of the biceps tendon is kept in

place within the groove by the transverse humeralligament and coracohumeral ligaments. A full-thickness tear is represented by complete discon-tinuity of the fibrillar pattern of the tendon,whereas a partial-thickness or intrasubstance tearof the biceps tendon produces a hypoechoic defect.Surrounding tenosynovitis is commonly seen inbicipital tendinosis, as shown in Figure 12. Thetendon can subluxate or dislocate out of the groovewhen the integrity of the transverse humeral liga-ment is breached in association with supraspinatusor subscapularis tendon tears. Biceps tendon sub-luxation out of the groove can be demonstrated byinternal and external rotation of the shoulder; thisusually occurs medially. An empty bicipital groovecan be seen when the biceps tendon is completelydislocated or torn. Complete biceps tendon tearsare differentiated from dislocations by tracing themuscle belly, which will also be absent but onlyvisualized distally because of muscle retraction.

Tennis Elbow, Lateral EpicondylitisTennis elbow, or lateral epicondylitis, is the

most common soft-tissue injury affecting the el-bow joint. It is thought to arise from chronic re-petitive injury. The lateral epicondyle is the ori-gin of the common extensor tendons of the wrist

and hand. Tendons of the extensor carpi radialisbrevis, extensor digitorum, extensor digitiminimi, and extensor carpi ulnaris fuse to formthe common extensor tendon origin.

In lateral epicondylitis, the tendon origin ap-pears thickened and hypoechoic on ultrasound.35

There may be hypoechoic linear clefts within thetendon, representing intrasubstance tears—a com-mon occurrence in tendinopathy.10 As seen in Fig-ure 13, chronic epicondylitis is associated withtendon thickening, calcification, and cortical irreg-ularity, or spur formation of the epicondyle.10,36

De Quervain TenosynovitisDe Quervain tenosynovitis is an idiopathic

condition involving the abductor pollicis longusand extensor pollicis brevis tendons in the firstextensor compartment at the level of and proximalto the radial styloid.37 Pain is usually broughtabout by thumb movements or, specifically, by the

FIGURE 11 Longitudinal sonogram of partial-thickness tear of the supraspinatus. a, Hypoechoic defect inter-rupting the articular surface of the tendon (open arrow). b, Hypoechoic defect interrupting thebursal surface of the tendon (arrow).

FIGURE 12 Transverse sonogram of the bicipital ten-don, depicting tendinosis with effusionwithin its sheath (arrow). GT, greater tu-berosity of humerus; LT, lesser tuberosityof humerus.


Finkelstein test, in which the patient makes a fistwith his or her fingers over the thumb, with thewrist adducted. The tendon sheath appears thick-ened with hypoechoic fluid on ultrasound, asshown in Figure 14. A hypoechoic or anechoic ringsurrounds the hyperchoic tendon in peritendinealeffusion of tenosynovitis, giving the appearance ofa target sign.

NORMAL AND PATHOLOGIC U/SAPPEARANCE OF THE MUSCLE

U/S is an effective assessment tool for diagno-sis of acute muscle injury, such as muscle contu-sions, strains, tears, and hematoma, as well aschronic lesions such as fibrous scars. U/S can behelpful for predicting the expected recovery period,and it is ideal for serial assessment to documentmuscle healing and recovery.38 Muscle fibers aregrouped into fascicles and are separated by septa offibroadipose tissue. The whole muscle is enclosedin a fascial sheath. On ultrasound, muscle appearshypoechoic with hyperechoic septations, as illus-trated in Figure 15.38 The intramuscular septations

appear as hyperechoic dots combined to form areticular pattern on a hypoechoic background inthe transverse view.38 In the longitudinal view, theintermuscular septa appear markedly hyperechoic,and the intramuscular septa appear as parallel hy-perechoic striae.38 The characteristic feature ofmuscle is that its alignment varies with contrac-tion of the muscle.

Muscle strains can be classified into grade 1,which is a strain injury with no macroscopic tissuedisruption; grade 2, which is a partial-thicknesstear with associated partial loss of muscle strength;and grade 3, which is a full-thickness tear withcomplete loss of muscle strength and which may beassociated with a retraction of ruptured muscleends.39,40 Grade 1 muscle strains often appear nor-mal, but the muscle may have an increased echogenicappearance because of perifascial fluid buildup.41

Grade 2 muscle strains are represented by disrup-tion of echogenic parallel striae of the muscle, withassociated fluid collection.41 In grade 3 musclestrains, complete disruption, with retraction ofmuscle fibers, surrounded by hypoechoic hema-

FIGURE 13 Common extensor origin. a, Normal longitudinal scan of the common extensor tendon, with aparallel echogenic fibrillar pattern. b, Lateral epicondylitis. Abnormal focal swelling and hypoecho-genicity (arrow) were seen at the tendon insertion site. LE, lateral epicondyle; RH, radial head; CE,common extensor tendon.

FIGURE 14 De Quervain tenosynovitis. a, Longitudinal sonogram depicting hypoechoic fluid swelling within thetendon sheath (arrow) of the abductor pollicis longus and extensor pollicis brevis. b, Transversesonogram showing a hypoechoic ring (arrow) around the tendon, producing the appearance of atarget sign.


toma, is the characteristic feature.42 Other acutemuscle injuries include blunt injuries to muscle(contusions), in which the ultrasound appearancedepicts an ill-defined hyperechoic region in themuscle, with associated hypoechoic hematoma. Inrecurrent or chronic injuries, fibrous scar forma-tion can occur, which appears on ultrasound as ahyperechoic lesion that is unchanged with musclecontraction.

NORMAL AND PATHOLOGIC U/SAPPEARANCE OF THE BURSASubacromial Bursa

The subacromial bursa lies between the deltoidand the rotator cuff and is not easily seen in normalconditions. The opposing sides of the bursa shouldbe no more than 2 mm apart.43 The bursa mayswell in association with supraspinatus impinge-ment or tears. Supraspinatus impingement can bedemonstrated on ultrasound by pooling of fluid inthe subacromial–subdeltoid bursa with active armelevation.44 Fluid inside the subacromial bursausually collects in its caudal portion and can befound both in superficial and full-thickness tears ofthe rotator cuff. The bursa is best evaluated at thelateral aspect of the shoulder between the supraspi-natus tendon and the deltoid muscle.43 Effusionscan distend along this deltoid shelf, which is thepoint of least resistance. It produces the teardropsign, as seen in Figure 16.45 Apart from chronicrepetitive or inflammatory conditions, bursitis canresult from trauma.

NORMAL AND PATHOLOGIC U/SAPPEARANCE OF THE NERVE

Peripheral nerves have a fascicular pattern inthe longitudinal plane, as shown in Figure 17. Itdemonstrates a speckled appearance in the trans-

verse plane as the neuronal fascicles appear hypo-echoic with hyperechoic connective stroma.46

Carpal Tunnel SyndromeThe diagnosis of carpal tunnel syndrome is

usually made on the basis of clinical features and isthen confirmed by nerve conduction studies. Car-pal tunnel syndrome can also be diagnosed withU/S by demonstrating an increase in the cross-sectional area of the median nerve at the level ofthe pisiform bone, as shown in Figure 18.47 Theadvantages of U/S are that it is painless and allowsvisualization of other underlying causes, such as amass lesion. The reported cross-sectional area fordiagnosis of the condition varied mostly between 9and 11 mm2.48–50 U/S seems to be a promising toolfor the diagnosis of carpal tunnel syndrome. In onestudy comparing U/S diagnosis of carpal tunnel syn-

FIGURE 15 Normal biceps brachii muscle. a, Longitudinal sonogram showing intramuscular septations (arrows) seenas hyperechoic lines separating hypoechoic muscle bundles. b, Transverse sonogram showing the intra-muscular septations (arrows) that appear as hyperechoic dots on a hypoechoic background. Bic, bicepsbrachii muscle; Brach, brachialis muscle; H, humerus.

FIGURE 16 Longitudinal sonogram depicting a dis-tended subacromiodeltoid bursa, demon-strated by the teardrop sign (arrow). SST,supraspinatus; GT, greater tuberosity ofthe humerus.


drome with nerve conduction studies, the sensitivitywas found to be 70 vs. 98%, and specificity 63 vs.19%, respectively.50 Future research is needed to gainfurther insight into the possible additional value ofthis diagnostic modality.

OTHER APPLICATIONS OF U/SIn musculoskeletal medicine, treatment with

injections into joints, bursae, or tendon sheaths arecarried out for various pathologies. In addition todiagnosis of pathologies, ultrasound can be used tomonitor needle position during the injection pro-cedure. U/S has been shown to be an accurate, safeimaging modality for guiding musculoskeletal in-jections.51 Actual techniques for ultrasound-guidedprocedures are not within the scope of this paper.But, to mention a few applications, ultrasound canbe used for guiding glenohumeral joint injec-tions,52 subacromial injections,53 aspiration of cal-cific tendonitis,54 and elbow joint52 and carpal tun-nel injections.55

LIMITATIONS IN DIAGNOSTIC U/SThe limitations of U/S stem from operator de-

pendence for this diagnostic procedure. This diag-

nostic tool lacks uniformity because of the dynamicnature of musculoskeletal examinations. The mo-bile nature of joints (in combination with randomprobe placements), gives rise to unlimited permu-tations in image variations. This is best illustratedby ultrasound examination of the rotator cuff inthe shoulder, where clinical accuracy dependsheavily on the scanning technique. To be able tocorrectly employ the diagnostic procedure and in-terpret findings, there is a long learning curve.

CONCLUSIONMusculoskeletal ultrasound has multiple ad-

vantages as a primary diagnostic modality. It isportable and highly accessible. An important fea-ture of ultrasound is its ability for dynamic imag-ing. In addition to making side-to-side compari-sons, it allows clinicians to correlate their patients’symptoms directly with anatomic visualization.The main disadvantages are operator dependenceand the long learning curve. Nevertheless, withproper use as an adjunct diagnostic tool, it canbecome a valuable extension to one’s physical ex-amination.

FIGURE 17 Normal median nerve (arrow): longitudinal and transverse views. The nerve is relatively hypoechoiccompared with the tendons (Ten).

FIGURE 18 Carpal tunnel syndrome. a, Transverse sonogram of the median nerve (M) at the level of the pisiformbone (Pis). b, Longitudinal sonogram displaying swelling of the median nerve (M) proximal to theretinaculum.


ACKNOWLEDGMENTSWe acknowledge Dr. Kathryn Stevens from the

Department of Radiology at Stanford UniversitySchool of Medicine for her technical advice andassistance with the ultrasound images. We alsothank Ms. Sylvia Guillory for her editorial assis-tance.

REFERENCES1. Kremkau FW: Diagnostic Ultrasound: Principles and In-

struments. Philadelphia, PA, W.B. Saunders, 1998

2. Van Holsbeeck M, Introcaso J: Sonography of tendons, inBralow L (ed): Musculoskeletal Ultrasound, ed 2. St Louis,MO, Mosby, Inc., 2001, pp 77–129

3. Grassi W, Filippucci E, Farina A, Cervini C: Sonographicimaging of tendons. Arthritis Rheum 2000;43:969–76

4. Chhem RK, Kaplan PA, Dussault RG: Ultrasonography ofthe musculoskeletal system. Radiol Clin North Am 1994;32:275–89

5. Jacobson JA, van Holsbeeck MT: Musculoskeletal ultra-sonography. Orthop Clin North Am 1998;29:135–67

6. Martinoli C, Derchi LE, Pastorino C, Bertolotto M, SilvestriE: Analysis of echotexture of tendons with US. Radiology1993;186:839–43

7. Crass JR, van d, V Harkavy LA: Tendon echogenicity: ex vivostudy. Radiology 1988 ;167:499–501

8. Farin PU, Jaroma H, Harju A, Soimakallio S: Medial dis-placement of the biceps brachii tendon: evaluation withdynamic sonography during maximal external shoulder ro-tation. Radiology 1995;195:845–8

9. Fomage B, Rifkin M: Ultrasound examination of tendons.Radiol Clin North Am 1998;26 87–107

10. Connell D, Burke F, Coombes P, et al: Sonographic exam-ination of lateral epicondylitis. AJR Am J Roentgenol 2001;176:777–82

11. Kainberger FM, Engel A, Barton P, Huebsch P, Neuhold A,Salomonowitz E: Injury of the Achilles tendon: diagnosiswith sonography. AJR Am J Roentgenol 1990;155:1031–6

12. Bianchi S, Zwass A, Abdelwahab IF, Banderali A: Diagnosisof tears of the quadriceps tendon of the knee: value ofsonography. AJR Am J Roentgenol 1994;162:1137–40

13. Bianchi S, Zwass A, Abdelwahab IF, Zoccola C: Evaluation oftibialis anterior tendon rupture by ultrasonography. J ClinUltrasound 1994;22:564–6

14. Matsen FA, Arntz CT, Lippitt SB: Rotator cuff: imagingtechniques, in Rockwood CA, Matsen FA (eds): The Shoul-der. Philadelphia, PA, W.B. Saunders Co., 2006, pp 789–93

15. Hedtmann A, Fett H: Ultrasonography of the shoulder insubacromial syndromes with disorders and injuries of therotator cuff [in German]. Orthopade 1995;24:498–508

16. Middleton WD, Reinus WR, Totty WG, Melson CL, MurphyWA: Ultrasonographic evaluation of the rotator cuff andbiceps tendon. J Bone Joint Surg Am 1986;68:440–50

17. Read JW, Perko M: Shoulder ultrasound: diagnostic accu-racy for impingement syndrome, rotator cuff tear, andbiceps tendon pathology. J Shoulder Elbow Surg 1998;7:264–71

18. Roberts CS, Walker JA, Seligson D: Diagnostic capabilitiesof shoulder ultrasonography in the detection of completeand partial rotator cuff tears. Am J Orthop 2001;30:159–62

19. van Holsbeeck MT, Kolowich PA, Eyler WR, et al: US de-piction of partial-thickness tear of the rotator cuff. Radiol-ogy 1995;197:443–6

20. van Moppes FI, Veldkamp O, Roorda J: Role of shoulderultrasonography in the evaluation of the painful shoulder.Eur J Radiol 1995;19:142–6

21. Teefey SA, Hasan SA, Middleton WD, Patel M, Wright RW,

Yamaguchi K: Ultrasonography of the rotator cuff. A com-parison of ultrasonographic and arthroscopic findings inone hundred consecutive cases. J Bone Joint Surg Am2000;82:498–504

22. Teefey SA, Rubin DA, Middleton WD, Hildebolt CF, LeiboldRA, Yamaguchi K: Detection and quantification of rotatorcuff tears. Comparison of ultrasonographic, magnetic res-onance imaging, and arthroscopic findings in seventy-oneconsecutive cases. J Bone Joint Surg Am 2004;86:708–16

23. Iannotti JP, Ciccone J, Buss DD, et al: Accuracy of office-based ultrasonography of the shoulder for the diagnosis ofrotator cuff tears. J Bone Joint Surg Am 2005;87:1305–11

24. Alasaarela E, Leppilahti J, Hakala M: Ultrasound and oper-ative evaluation of arthritic shoulder joints. Ann Rheum Dis1998;57:357–60

25. Brandt TD, Cardone BW, Grant TH, Post M, Weiss CA:Rotator cuff sonography: a reassessment. Radiology 1989;173:323–7

26. Burk DL Jr, Karasick D, Kurtz AB, et al: Rotator cuff tears:prospective comparison of MR imaging with arthrography,sonography, and surgery. AJR Am J Roentgenol 1989;153:87–92

27. Khan K, Cook J: The painful nonruptured tendon: clinicalaspects. Clin Sports Med 2003;22:711–25

28. Teefey SA, Middleton WD, Bauer GS, Hildebolt CF, YamaguchiK: Sonographic differences in the appearance of acute andchronic full-thickness rotator cuff tears. J Ultrasound Med2000;19:377–8

29. Farin PU, Kaukanen E, Jaroma H, Vaatainen U, Miettinen H,Soimakallio S: Site and size of rotator-cuff tear. Findings atultrasound, double-contrast arthrography, and computedtomography arthrography with surgical correlation. InvestRadiol 1996;31:387–94

30. Wiener SN, Seitz WH Jr: Sonography of the shoulder inpatients with tears of the rotator cuff: accuracy and valuefor selecting surgical options. AJR Am J Roentgenol 1993;160:103–7

31. Hollister MS, Mack LA, Patten RM, Winter TC III, MatsenFA III, Veith RR: Association of sonographically detectedsubacromial/subdeltoid bursal effusion and intraarticularfluid with rotator cuff tear. AJR Am J Roentgenol 1995;165:605–8

32. Bouffard JA, Lee SM, Dhanju J: Ultrasonography of theshoulder. Semin Ultrasound CT MR 2000;21:164–91

33. Wohlwend JR, van HM, Craig J, et al: The associationbetween irregular greater tuberosities and rotator cufftears: a sonographic study. AJR Am J Roentgenol 1998;171:229–33

34. Tuite MJ, Turnbull JR, Orwin JF: Anterior versus posterior,and rim-rent rotator cuff tears: prevalence and MR sensi-tivity. Skeletal Radiol 1998;27:237–43

35. Miller TT, Shapiro MA, Schultz E, Kalish PE: Comparison ofsonography and MRI for diagnosing epicondylitis. J ClinUltrasound 2002;30:193–202

36. Lin J, Jacobson JA, Fessell DP, Weadock WJ, Hayes CW: Anillustrated tutorial of musculoskeletal sonography: part 2,upper extremity. AJR Am J Roentgenol 2000;175:1071–9

37. Wolfe ST: Green’s Operative Hand Surgery, ed 4. Philadel-phia, PA, Churchill Livingstone, 1999

38. Van Holsbeeck M, Introcaso J: Sonography of muscle, inBralow L (ed): Musculoskeletal Ultrasound, ed 2. St Louis,MO, Mosby, Inc., 2001, pp 23–75

39. Palmer WE, Kuong SJ, Elmadbouh HM: MR imaging ofmyotendinous strain. AJR Am J Roentgenol 1999;173:703–9

40. Boutin RD, Fritz RC, Steinbach LS: Imaging of sports-related muscle injuries. Radiol Clin North Am 2002;40:333–62

41. Lee JC, Healy J: Sonography of lower limb muscle injury.AJR Am J Roentgenol 2004;182:341–51

42. Takebayashi S, Takasawa H, Banzai Y, et al: Sonographic


findings in muscle strain injury: clinical and MR imagingcorrelation. J Ultrasound Med 1995;14:899–905

43. van HM, Strouse PJ: Sonography of the shoulder: evalua-tion of the subacromial-subdeltoid bursa. AJR Am J Roent-genol 1993;160:561–4

44. Farin PU, Jaroma H, Harju A, Soimakallio S: Shoulderimpingement syndrome: sonographic evaluation. Radiology1990;176:845–9

45. Riederer B, Chhem R, Cardinal E, Peetrons P: Shoulder:nonrotator cuff disorders, in Chhem R, Cardinal E (eds):Guidelines and Gamuts in Musculoskeletal Ultrasound.New York, NY, Wiley-Liss, 1999, pp 1–38

46. Silvestri E, Martinoli C, Derchi LE, Bertolotto M,Chiaramondia M, Rosenberg I: Echotexture of peripheralnerves: correlation between US and histologic findings andcriteria to differentiate tendons. Radiology 1995 ;197:291–6

47. Buchberger W, Judmaier W, Birbamer G, Lener M,Schmidauer C: Carpal tunnel syndrome: diagnosis withhigh-resolution sonography. AJR Am J Roentgenol 1992;159:793–8

48. Duncan I, Sullivan P, Lomas F: Sonography in the diagnosisof carpal tunnel syndrome. AJR Am J Roentgenol 1999;173:681–4

49. Lee D, van Holsbeeck MT, Janevski PK, Ganos DL, DitmarsDM, Darian VB: Diagnosis of carpal tunnel syndrome. Ul-trasound versus electromyography. Radiol Clin North Am1999;37:859–72

50. Swen WA, Jacobs JW, Bussemaker FE, de Waard JW,Bijlsma JW: Carpal tunnel sonography by the rheumatolo-gist versus nerve conduction study by the neurologist.J Rheumatol 2001;28:62–9

51. Naredo E, Cabero F, Beneyto P, Cruz A, Mondejar B, UsonJ: A randomized comparative study of short-term responseto blind injection versus sonographic-guided injection oflocal corticosteroids in patients with painful shoulder.J Rheumatol 2004;31:308–14

52. Koski JM: Ultrasound-guided injections in rheumatology.J Rheumatol 2000;27:2131–8

53. Chen MJ, Lew HL, Hsu TC, et al: Ultrasound-guided shoul-der injections in the treatment of subacromial bursitis.Am J Phys Med Rehabil 2006;85:31–5

54. Aina R, Cardinal E, Bureau NJ, Aubin B, Brassard P: Calcificshoulder tendinitis: treatment with modified US-guidedfine-needle technique. Radiology 2001;221:455–62

55. Grassi W, Farina A, Filipucci E, Cervini C: Intralesionaltherapy in carpal tunnel syndrome: a sonographic-guidedapproach. Clin Exp Rheumatol 2002;20:73–6


Synovitis of the Sternoclavicular Joint:The Role of UltrasoundSteve J. Wisniewski, MD, and Jay Smith, MDFrom the Department of Physical Medicine and Rehabilitation,Mayo Clinic, Rochester, Minnesota.

0894-9115/07/8604-0322/0American Journal of Physical Medicine & RehabilitationCopyright © 2007 by Lippincott Williams & WilkinsDOI: 10.1097/PHM.0b013e318038d264

A 55-yr-old right hand–dominant female presented toclinic with a 1-yr history of right sternoclavicular joint (SCJ)pain. The pain was focal, dull, nonradiational, and increasedwith any upper-limb activity. She reported no history oftrauma or other joint complaints. The pain started while thepatient was working in a factory; it eventually resulted in earlyretirement and a switch to a nonmanual labor job. Rheuma-tologic work-up was unremarkable. Prior treatments hadincluded numerous medications, extensive physical therapy,icing, and a non–image-guided intra-articular SCJ steroidinjection, all without any appreciable change in symptoms.

Physical examination revealed obvious swelling overthe right SCJ, where palpation produced significant ten-derness reproductive of her usual pain. All other jointswere normal.

Radiographs revealed mild degenerative changes at theright SCJ. Chest computed tomography revealed only mini-mal irregularity and narrowing of the right SCJ comparedwith the left, with moderate amounts of adjacent soft-tissueswelling. Prior bone scan was normal at the SCJ.

Because of the perception of synovitis on examination,we performed an ultrasound (US) examination of the rightSCJ with comparison views of the left SCJ (Fig. 1). The examrevealed remarkable capsular hypertrophy, a small amount ofanechoic intra-articular fluid, and a moderate amount ofhyperechoic, intra-articular tissue exhibiting increased powerDoppler flow, consistent with synovitis (Fig. 2). The visualizedbony margins were only minimally irregular.

All correspondence and requests for reprints should be addressed to SteveWisniewski, MD, Mayo Clinic Sports Medicine Center, 200 First Street SW,Rochester, MN 55905.

FIGURE 1 Ultrasound image comparing the left and rightsternoclavicular joints (SCJ). Transducer is ori-ented perpendicularly to the long axis of the SCJ.The right SCJ capsule is diffusely thickened anddisplaced, with a mixed-echogenicity SCJ effusion.MAN, manubrium.

FIGURE 2 Power Doppler evaluation of the right sterno-clavicular joint (SCJ), revealing hyperemia con-sistent with synovitis. Orientation is similar tothat of Figure 1. MAN, manubrium; CLAV, clav-icle.

FIGURE 3 Postinjection image of the right sternoclavicularjoint (SCJ). Orientation is similar to that of Figure1. After injection, the capsule is displaced by ane-choic injectate consisting of local anesthetic andcorticosteroid. MAN, manubrium; CLAV, clavicle.

VISUAL VIGNETTE

322 Wisniewski and Smith Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 4

US-guided right intra-articular SCJ aspiration and in-jection were then performed. A small amount of yellowishfluid of slightly reduced viscosity was obtained. A 1-mlmixture of methylprednisolone and 1% lidocaine werethen slowly injected under direct US guidance (Fig. 3). Thepatient tolerated the procedure well with no complica-tions.

This case illustrates numerous advantages of usingmusculoskeletal US in a physiatric practice. Despite priorunremarkable radiographs, computed tomography scan,and bone scan, the use of US with Doppler confirmed SCJsynovitis and ensured accurate needle placement for aspi-

ration and therapeutic injection. Accurate needle place-ment is particularly crucial when fluid must be obtainedfor analysis or when target structures are close to neuro-vascular structures. Finally, this procedure was easily ac-complished in the office during a single patient visit.

REFERENCES1. Ferri M, Finlay K, Popowich T, et al: Sonographic examination of the

acromioclavicular and sternoclavicular joints. J Clin Ultrasound 2005;33:345–55

2. Ernberg LA, Potter HG: Radiographic evaluation of the acromioclavic-ular and sternoclavicular joints. Clin Sports Med 2003;22:255–75

April 2007 323

AAP ANNUAL MEETING ABSTRACTS

Abstracts of Scientific Papers and Posters Presented atthe Annual Meeting of the Association of AcademicPhysiatrists

San Juan, Puerto RicoApril 10–14, 2007

The Electrode Store Best PaperAward WinnersMedical Student

VELOCITY OF MOVEMENT OR STRENGTH: WHICH ISMORE INFLUENTIAL ON THE BALANCE OF ELDERS ATRISK FOR FALLS?Douglas J. Mayson BS; Dan K. Kiely MPH, MA; Sharon I.Larose BS; Jonathan F. Bean MD

Given the expanding elderly population and the high prevalence of inju-rious falls among these individuals, optimizing balance remains an im-portant rehabilitative goal for older adults. A person’s ability to balance iscoordinated by the complex interplay of various physiologic subsystems.The musculoskeletal component of this balance system is important formaintaining both static and dynamic balance, and if diminished, may leadto an increased risk of falling. Reduced muscle strength may decrease aperson’s ability to mount an adequate response to perturbations inbalance. Also, strength has been linked to the performance of staticmeasures of balance that are predictive of falls. In addition to strength,recent studies have indicated that muscle power also plays an importantrole in the maintenance of balance and mobility. Since muscle power isthe product of the strength and velocity of movement (Power � force �velocity), a decrease in either component may lead to a diminishedcapacity to generate power.

An association between muscle power and balance would not besurprising considering the role muscle power plays in other functionalactivities. Specifically, skeletal muscle power is more predictive of per-formance in general mobility tasks than strength. It is not currentlyknown, however, which of the two components of power (i.e., strength orvelocity of movement) is most important in the maintenance of balance.In a recent article, Sayers et al. indicated that limb velocity was morepredictive of 400 meter performance than strength. Furthermore, Thelenet al. suggested that the velocity of ankle torque development plays animportant role in the ability to recover from a disturbance of balance.Taken together, these studies suggest that limb velocity may be animportant factor in performance of clinically established balance teststhat are predictive of falls.

This study evaluates clinically established balance measures amongolder adults at risk for falls to determine the relative importance of limbstrength vs. velocity of limb movement. Knowing which component of legpower is most strongly associated with balance performance will haveimportant implications for the design of rehabilitative care. We hypoth-esize that strength will have stronger associations with measures of staticbalance, while velocity will have greater associations with tests of dy-namic balance. Methods: Subjects and Setting: This study is a cross-sectional analysis of baseline data from a randomized controlled trialamong community-dwelling older adults (n � 132, mean age 75.0 yrs)with mobility limitation as defined by the Short Physical PerformanceBattery (SPPB). After an initial telephone screening, all evaluations oc-

curred at a rehabilitation research gym. Impairment Measures: Lowerlimb strength and power were measured using a pneumatic double legpress resistance machine (Keiser Sports Health Equipment Inc., Fresno,CA) as previously described. The one repetition maximum (1RM) servedas our strength measure, and was determined by progressively increasingresistance for successive repetitions until the participant could no longermove the lever arm through the full ROM. The 1RM was determined asthe highest resistance at which a full repetition was completed. Peakpower was measured as the best of 5 repetitions performed at 40% of the1RM, in which participants performed the concentric action of onerepetition as quickly as possible. We calculated each subject’s leg pressvelocity through the division of values for leg power by leg strength.Static and Dynamic Balance: All subjects underwent four balance tests,including one measure of static balance and three measures of dynamicbalance. Static balance was determined by the Unipedal Stance Test(UST). The UST is a reliable and valid test that is associated with the riskfor injurious falls. Dynamic balance measures included the Berg BalanceTest (BERG), the Dynamic Gait Index (DGI), and the Performance-Oriented Mobility Assessment (POMA). All three tests are also reported tobe reliable and valid measures among older adults and are associated withsubsequent fall risk. Data Analysis: We calculated descriptive statisticswith frequencies and proportions for categorical variables and mean andstandard deviations for continuous variables. Given the outcomes werenot all normally distributed, Spearman correlations were calculated tocompare performance on balance measures against leg strength andpower. Next, we evaluated the bivariate relationships between each of thecovariates and the mobility measures ensuring a statistically significantassociation (P � 0.05). Finally, using clinically relevant cut-offs for eachbalance measure, we created four separate multivariate logistic regressionmodels including limb velocity, limb strength, and the covariates pre-dicting good balance performance (lower fall risk). Age, BMI, gender,number of medications, number of chronic conditions, and mental statusserved as covariates in the final models. Results: The study population wascomposed of community-dwelling older adults with mobility problems.Subjects had a mean age of 75 yrs, were predominantly female (69.9%),white (82.6% white; 15.2% black, 0.8% Asian, 0.8% Native American,0.8% Other), and overweight (mean BMI 27.8 kg/cm2). The followingmean values for balance measures were determined: UST of 9.9 secs, DGI21.6 out of 24, POMA 25.8 out of 28, and the BERG 50.7 out of 56.Participants had an average leg press power of 495.6 Watts, velocity of0.90 cm/sec, and strength of 565 newtons.

After reviewing the multivariate regression models, we found thatstrength, odds ratio (OR) 1.21 (1.04–01.40), was more predictive thanvelocity OR 0.96 (0.82–01.12) of higher static balance status. In contrast,velocity was consistently more predictive of higher performance on allmeasures of dynamic balance with odds ratios of 1.26 (1.03–01.54), 1.34(1.11–01.71) and 1.84 (1.37–02.47) for the BERG, POMA and DGI re-spectively. Discussion: The major finding of our study was that of the twocomponents of muscle power, limb strength was a significant predictor ofbetter static balance performance, while limb velocity was a significantpredictor of better dynamic balance performance. Mechanistically, thisdichotomous result can be explained by looking at the dominant activitiesinvolved in these two types of balance measures.

324 Meeting Abstracts Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 4

The three dynamic balance tests, the BERG, POMA, and DGI, re-quire the subject to perform a variety of different tasks, most of whichmimic activities performed in daily life. A large number of these tasks aretimed and require the subject to perform such activities as walking, risingfrom a seated position, and transferring between chairs. As such, theBERG, POMA, and DGI may be broader measures of mobility, rather thanjust “tests of balance.” When considered as broader tests of mobility, theassociation of limb velocity with better performance is fully consistentwith the earlier report by Sayers et al. This result does not negate theutility of the BERG, POMA, and DGI as tests of dynamic balance, butinstead emphasizes the link between balance and mobility. As a result, itseems likely that training programs that aim to improve limb velocitymay positively affect both balance and mobility.

Conversely, the nature of activity required for the UST is quitedifferent from those involved in the dynamic tests. The UST is a measureof static balance that is dependant on the ability of the individual tosupport their weight on one leg. Since this activity does not involvemovement and only requires maintenance of a static position, it is notnecessarily surprising that strength plays a more dominant role than limbvelocity.

A limitation of this study is that limb strength and velocity were notmeasured within muscle groups other than the hip and knee extensors.The study design did not consider the relationship between balance andstrength or velocity at the ankle, trunk, or hip abductors, all of whichinclude muscle groups that are potentially important in the maintenanceof balance. Additionally, our study was a cross-sectional analysis evaluat-ing older adults with mobility limitations. A broader longitudinal studyincluding both healthy older adults and those with a significant fallhistory may allow us to better understand these relationships. Neverthe-less, our results are of relevance given the increased fall risk in ourstudied population on the basis of their low SPPB scores.

The results presented in this paper highlight the importance ofemphasizing enhancements in limb velocity in rehabilitation. Clinicalexperience suggests that most rehabilitative therapists will emphasizestrength training as components of balance rehabilitation. However,despite intervention studies suggesting the benefits of limb power en-hancement, use of high-velocity training among older adults is relativelyrare.

In conclusion, our study has demonstrated that higher leg pressvelocity is associated with better performance on BERG, POMA, and DGImeasures and that higher leg strength is associated with better USTperformance. These findings underscore the importance of training bothleg strength and velocity of movement when considering balance reha-bilitation among older adults at risk for falls. Further investigation ofthese relationships through future longitudinal studies is warranted.

Resident

CARPAL TUNNEL SYNDROME SYMPTOMS IN INDIVIDUALSWITH SPINAL CORD INJURY WHO USE MANUALWHEELCHAIRSJennifer Yang, MD; Michael L. Boninger, MD; Trevor A. Dyson-Hudson, MD; Robert Price, MSME; Bradley G. Impink, MS;Rachel Cowan, MS

Introduction: Due to the repetitive motion involved in wheelchair pro-pulsion and transfers, individuals with spinal cord injury (SCI) who usemanual wheelchairs are at risk for carpal tunnel syndrome (CTS). CTS isusually diagnosed based on symptoms with physical examination andnerve conduction studies providing additional information. Research onindividuals with paraplegia has directly linked the severity of mediannerve electrodiagnostic abnormalities to subject weight, wheelchairstroke propulsion patterns, wheelchair weight and wrist position. Symp-tomatic CTS is well-documented in the medical literature and was re-cently reviewed in the clinical practice guidelines for preservation ofupper limb function following SCI. Symptom surveys by Subbarao,Schroer and Pentland, and Twomey showed a 40–48% prevalence of wristpain. Separate studies by Sie and Gellman have noted a prevalence of CTSby history and physical examination ranging from 49 to 66%, with theprevalence increasing with the duration of SCI. Gellman in 1988 notedthat individuals with paraplegia had increased pressures within the carpaltunnel while the wrist was in neutral and in extension, as compared withthose without paraplegia. Studies in the general population have alsoshown that other risk factors that increase the likelihood of carpal tunnel

syndrome include female gender, increased BMI, age and symptoms inthe dominant hand.

The object of this study was to investigate the prevalence, commonsymptoms, and impact of CTS symptoms in a cohort of manual wheel-chair users with SCI. In addition, we wanted to examine the relationshipbetween symptoms and subject characteristics such as weight, gender,and years with SCI. Methods: We conducted a multicenter cross-sectionalstudy. Individuals 18 yrs old and older with traumatic spinal cord injurylevels T2 through L4 were recruited from three SCI model system centerswith large databases across the United States. To be included in the study,participants had to use a manual wheelchair for over 40 hrs/wk and nothave upper limb trauma affecting their report of symptoms. After writteninformed consent was obtained for the study, subjects were administereda questionnaire on CTS symptoms such as numbness, tingling/electricsensation and pain with wheelchair propulsion, whether they soughtmedical help for their symptoms, whether these symptoms curtailed theiractivity and the duration that they curtailed activity. Demographic datacollected included gender, age, handedness, level of injury, length of timesince injury, weight, and stature (height). BMI was calculated as weight/height squared. Demographic and clinical information were entered intoa database constructed for this study using Microsoft Access and analyzedusing SPSS. Subjects were dichotomized based on symptoms. The twogroups were then compared using the Student t test for continuousvariables the Mann-Whitney U test for ordinal variables, and the �2 testfor categorical variables. Results: Data were collected on 118 subjects, 23women and 95 men, aged 23–76 yrs (40.9 � 11.8 yrs), with length of timesince injury from 1.2 to 34.1 yrs (13.1 � 8.9 yrs). One hundred foursubjects (88.1%) were right-handed and 14 (11.9%) were left handed. Themost frequent level of injury was T12 (16.1%), followed by T10 (14.4%)and T4 (11%). Patient weight had a range of 43.1–132.8 kg (79.2 � 17.1kg), stature ranged from 1.5–1.9 m (1.7 � 0.1 m) and BMI ranged from14.9 to 40.6 (25.6 � 5.2).

Among the subjects, 57.6% had symptoms consistent with CTS.Among those who experienced symptoms, numbness was most frequentlyreported (45.8%), followed by electric sensation or tingling (39%) andpain with wheelchair propulsion (27.1%). Of the subjects with symptoms,35.2% had one symptom, 26.6% had two symptoms, and 38.2% experi-enced three symptoms. Among these subjects, 54.5% experienced symp-toms in their dominant hand (53.5% of right-handers and 13.1% ofleft-handers). Among the subjects, 41.5% had symptoms in both hands,regardless of handedness. Subjects were not found to have significantlymore symptoms in their dominant hand.

36.8% of our subjects with symptoms consulted a physician and33% were told they had CTS. Of the symptomatic subjects, 41.3% un-derwent nerve conduction studies. Of the symptomatic individuals, 25.4%curtailed their activity due to the symptoms. Subjects with a greaternumber of symptoms were more likely to curtail activity (P � 0.001). Nosignificant differences in demographic characteristics were noted be-tween persons who did and did not curtail activity.

Subjects reporting any symptoms were significantly younger (mean38.7 yrs) than subjects who reported no symptoms (mean 43.8 yrs, P �0.019). However, when comparing groups who were and were not giventhe diagnosis of CTS, a trend was found showing subjects with a CTSdiagnosis were older (mean 44.3 yrs, vs. 39.8 yrs, P � 0.089). Discussion:This study examines CTS symptoms in individuals with spinal cord injuryand their relationship to subject activity and demographics. Numbnesswas the most common complaint, consistent with what is found in thegeneral population. Most reports in the literature combine numbness andtingling under the category “paresthesias”; our study agrees with this asthese were the two most common symptoms reported by our subjects.

Unlike the general population we did not find CTS to be morecommon in the dominant hand. This may be related to the nature of oursubjects, as they most likely use both hands equally in wheelchair pro-pulsion and transfers.

Our data shows that although younger subjects are more likely toreport their symptoms, older subjects are more likely to have a diagnosisof CTS. It is possible that older subjects, presumably having had symp-toms for a longer period of time, have learned to live with their symp-toms, are less concerned about them and thus do not report thesesymptoms.

Symptoms of CTS in the wrist and hand seem to be an importantfactor limiting subject function. The magnitude of this problem can beseen in the number of subjects who consult a physician for these symp-toms, as well as those who curtail their activity due to symptoms.However there is also a population that did not seek consultation with aphysician nor curtail their activities, and these subjects may be at risk of

April 2007 Meeting Abstracts 325

further injury. This highlights the need for patient education regardingpreservation of upper limb function.

These findings provide further insight into symptoms of CTS insubjects with SCI who use manual wheelchairs. Some limitations to thisstudy must be noted. As it has been shown in the literature that femalesin the general population are more likely to have symptoms of CTS, wewould expect this in the SCI population as well. Due to the small numberof women in our study, we may not have had enough power to detect thisdifference. Although not statistically significant, it should be noted that69.6% of our female subjects reported symptom of CTS compared with54.7% of our male subjects.

We did not find a relationship between prevalence of CTS andduration of symptoms. In the absence of symptoms or if patients chooseto ignore symptoms, median nerve function may worsen with time. Bydesign this report has focused on symptoms. Future reports will be morecomprehensive, including both physical examination and electrodiagnos-tic data, both of which can assist in the diagnosis of CTS.

Fellow

NEW MODEL OF SCI-INDUCED OSTEOPOROSIS: ACUTECONTUSION SPINAL CORD INJURY IN RATS TRIGGERSRAPID BONE LOSSLeslie Morse, DO, Dou Yu, MD, Wei-Li Liao, MD, KimberlyNewton, Ricardo Battaglino, PhD, Philip Stashenko, DMD, PhD,Yang Teng, MD, PhD

Background: Limited understanding of the mechanism governing SCI-induced bone loss has led to inadequately few treatment options. As thiscondition carries significant morbidity and mortality, a need exists toadvance the general knowledge regarding its pathophysiology. Mechanis-tic studies are limited by the lack of a rodent model and difficultiesassociated with obtaining bone biopsies from patients with spinal cordinjury. In this work we present an experimental model of SCI-inducedosteoporosis in rats that is rapidly occurring, reproducible and similar innature to the bone loss experienced by individuals immediately followingcomplete spinal cord injury. Methods: Rat Model of Severe SCI: Adoles-cent male Sprague-Dawley rats (200�0225 g) were anesthetized withintraperitoneal ketamine (75 mg/kg) and xylazine (20 mg/kg). A severeT10 contusion injury was produced utilizing the New York University(NYU) impactor (10 g � 50 mm) as previously described. The controlgroup consisted of naıve, age-matched male Sprague Dawley rats. Animalswere euthanized on day 10 postinjury for subsequent analyses. Confir-mation of Severe Neurological Impairment: The well-established BBBhind limb locomotor rating scale was used to determine hindlimb func-tional deficits on day 1 and 7 postinjury. Only severely injured animals(defined as BBB score of 4 or less i.e., only very limited hindlimb reflexeswithout locomotion observed at day 7 post injury) were used for allsubsequent studies. In vivo Assessment of Bone Mineral Density: Beforesacrifice all animals were anesthetized and bone mineral density (BMD)was assessed in vivo by PIXImus Scan (Lunar PIXImus2, software version1.4X). Animals were positioned on their stomachs with the hindlimbexternally rotated. The hip and knee were flexed to 110 degrees with theankle in neutral. The forelimb was externally rotated. Two regions ofinterest were analyzed, one (16 � 17 pixel) at the distal femoral metaph-ysis and one (11 � 13 pixel) at the distal radius/ulna. Micro-ComputedTomographic Imaging (�CT): To assess bone microarchitecture, fixedfemur samples were analyzed at the Institute for Biomedical Engineering,ETH and University Zuerich, using a compact fan-beam-type tomograph(�CT 20, Scanco Medical AG, Bassersdorf, Switzerland). Samples werelocated in an airtight cylindrical sample holder filled with formalin. Thesample holders are marked with an axial alignment line to allow forconsistent positioning of the specimens. For each sample, approximately200 micro-tomographic slices with an increment of 17 �m were acquired,covering the entire width of the bone. Three-dimensional analyses wereperformed to calculate morphometric indices including total volume(TV), bone volume (BV), marrow volume (MV), bone mineral content(BMC), bone surface (BS), bone thickness (Th), and various ratios includ-ing density BV/TV, trabecular thickness [Tb.Th � 2 � BV/bone surface(BS)], trabecular number [Tb. N � (BV/TV)/Tb. Th], and trabecularseparation [Tb. Sp � (1/Tb. N) Tb. Th]. Static and Dynamic Histomor-phometry: Naıve and severely injured animals were injected with calcein,a dye which intercolates into growing bone, on day 2 and day 7 postinjury.Hindlimbs were resected and sent to the histomorphometry core facilityat Mayo clinic for dynamic (appositional growth) and static histomor-

phometry analysis. Summary of Results: We found a significant 36%relative bone loss at the distal femoral metaphysis by PIXimus scan in theinjured animals compared with naıve controls at 10 days postinjury. Wefound no significant change in the forelimb BMD (skeletal site used as asupralesional control). MicroCT analysis of the femoral metaphysisshowed a significant alteration in bone microarchitecture due to severeSCI. We detected a 48% decrease in trabecular bone mineral content withassociated decrease in trabecular thickness, number and connectivity atthe femoral metaphysis in the injured animals when compared with thenaıve controls. The cortical bone mineral content was also decreased by35%. The ratio of bone surface to bone volume was increased in bothcompartments indicating increased bone resorption. Appositional growthstudies demonstrated decreased new bone formation in the injured ani-mals with an abnormal labeling pattern suggesting a mineralizationdefect in the injured animals. Static histomorphometry studies demon-strated no difference in osteoclast or osteoblast numbers at distal femoralmetaphysis. However, there was decreased osteoid production in theinjured animals suggesting impaired ability to form new bone at that site.Conclusions: The femoral metaphysis was chosen as this is the site ofgreatest fracture frequency in individuals with spinal cord injury andtherefore has high clinical relevance. We have demonstrated that ourexperimental model of severe SCI in rats causes a rapid, severe bone lossat the femoral metaphysis. This loss is detectable 10 days postinjury. Inanimals with severe injury (BBB score less than 4), there is little vari-ability in the degree of bone loss that occurs. Furthermore, we havedemonstrated abnormal bone microarchitecture due to severe SCI withdecreased trabecular and cortical bone. These findings indicate increasedbone resorption due to osteoclast activation following SCI. We have alsodemonstrated impaired bone formation suggestive of osteoblast dysfunc-tion. Based on studies of markers of bone turnover following acute spinalcord injury, it is widely believed that bone formation is normal withabnormal bone resorption. Therefore, these findings are the first to reportboth osteoclast and osteoblast dysfunction in the development of osteo-porosis following spinal cord injury. This model has enormous potentialfor mechanistic studies that are currently lacking and greatly needed toadvance the understanding of SCI-induced bone loss. Furthermore, thismodel can be used to test therapeutic interventions aimed at preventionof osteoporosis following neurological injury.

RMSTP PRESENTATIONS

FORCE CONTROL STRATEGIES WHILE DRIVING ELECTRICPOWERED WHEELCHAIRS WITH ISOMETRIC ANDMOVEMENT-SENSING JOYSTICKSBrad E. Dicianno, MD; Donald M. Spaeth, PhD; Rory A. Cooper,PhD; Shirley G. Fitzgerald, PhD; Michael L. Boninger, MD;Karl W. Brown

Innovations to control interfaces for electric powered wheelchairs (EPWs)could benefit 220,000 current users and over 60,000 individuals whodesire mobility but cannot use a conventional motion sensing joystick(MSJ). We developed a digital isometric joystick (IJ) with sophisticatedsignal processing and two control functions. In a prior study, subject’sdriving accuracy with our IJ was comparable with using an MSJ. However,we observed subjects using excessive force on the IJ possibly because itsrigid post provides no positional feedback. Thus, this study examines thetime-series data recorded in the previous study to characterize subjects’force control strategies since fatigue and weakness are common impair-ments of individuals who need improved interfaces, and also becausestudies that compare MSJs and IJs could be biased if force control is nottaken into account. Eleven EPW users with upper limb impairmentsdrove an EPW to floor targets using an IJ with two different controlfunctions and an MSJ in a Fitts’ law paradigm. Subjects relied uponpositional feedback from the MSJ and used appropriate force. In contrast,subjects using the IJ with either control function applied significantlyhigher force than necessary (P � 0.0001 and P � 0.0058). Using higheraverage force was correlated with quicker trial times but not associatedwith accuracy. Lack of positional feedback may result in use of excessisometric force. Modifying control functions, adjusting gain, or providingadditional training or feedback might address this problem.


RELATIONSHIP BETWEEN FINGER INDIVIDUATION ANDSHAPING THE FINGERS TO OBJECT CONTOURSPreeti Raghavan, MD; John Krakauer, MD; Marco Santello,PhD; Andrew Gordon, PhD

Grasping objects is an essential component of one’s daily activities, andinvolves accurate coordination of finger movements to approximate thesize, shape, and use of an object before it is contacted. However, patientswith damage to the corticospinal tract, as a result of a subcortical stroke,show impaired shaping of the fingers to object contours. We examinedwhether the impairment in hand shaping is related to the ability to movethe fingers individually. Sixteen subjects (8 patients and their age-matched controls) reached toward and grasped objects of three differentshapes (rectangular, concave, and convex) in blocked trials. In addition,they were instructed to move each of their digits individually, whilekeeping the remaining digits stationary. Angular excursions of the thumband digit joints were recorded using an instrumented glove (Cyberglove).Discriminant analysis was used to determine the extent to which handpostures were reliably different for the 3 object shapes, and the resultswere correlated with individuation and stationarity indices, calculatedfrom the extent of movement of the instructed digit in relation to thenoninstructed digits. We found that shape discrimination was stronglydependent upon movements of the thumb, particularly the time at whichmaximum thumb rotation occurs during reach-to-grasp. Object shapediscrimination at the end of reach was related to thumb flexion, but thatat object contact was related to thumb extension. Deficits in thumbextension at the end of the reach-to-grasp movement can explain why thepatients discriminate object shape partially until reach offset, but notthereafter. In patients with stroke, individuation and stationarity indicesaccount for the differences in joint angle excursions for the respectiveobject shapes. Thus, impairments in finger individuation can explain thedeficit in shaping the fingers to object contours after stroke. Specifictraining of individuated finger movements, particularly of the thumb, canassist in the rehabilitation of hand dysfunction after stroke.

CHONDROCYTE-BASED GENE THERAPY FOR THEDEGENERATING INTERVERTEBRAL DISK IN THE RABBITYejia Zhang, MD, PhD; Frank M. Phillips, MD; Eugene J-MAThonar, PhD; Howard S. An, MD; D. Greg Anderson, MD

Introduction: Back pain associated with intervertebral disk degenerationis a common clinical problem and costs billions of dollars. Biologicaltreatments that restore the disk matrix and cellular elements are prom-ising alternatives to the surgical removal of the diseased disk. Autologouschondrocytes are readily available and are phenotypically similar to diskcells. Human bone morphogenetic protein-7 (hBMP-7) is known to stim-ulate extracellular matrix production by chondrocytes. We hypothesizethat injecting articular chondrocytes over-expressing hBMP-7 will repop-ulate the disk and sustain the release of hBMP-7. This procedure could beperformed by a physiatrist, thus providing a minimally invasive approachto disk repair. Methods: In the initial in vitro studies, disk explants andknee articular chondrocytes were isolated from young adult New ZealandWhite rabbits. Chondrocytes transduced with adenovirus expressinghBMP-7 (Ad-hBMP-7), or green fluorescnce protein (AdGFP, control), orchondrocytes (control), or medium (control) were injected into the cul-tured disk explants. The explants were cultured in DMEM/F12 media withsupplements for 1-2 mos after injection. The survival of the injectedchondrocytes was confirmed by examining chondrocytes expressing GFPwith fluorescnce microscopy. Proteoglycans accumulated by the diskexplants were assessed by a dye-binding method. Second, an in vivo studywhere allogenic rabbit articular chondrocytes transduced with Ad-hBMP-7, AdGFP, chondrocytes or culture medium, was injected into adegenerating rabbit disk. Under general anesthesia, a left abdominalincision was made, and the ventral surface of the L2/3, L3/4, L4/5, andL5/6 disks were injured with an 18-G needle. Two weeks postdisk injury,the degenerating disks were injected with chondrocytes transduced withAd-hBMP-7 or AdGFP, nontransduced chondrocytes or culture medium.At six-week postinjection, and before each surgery, a lateral x-ray wasrecorded from which the disk height index (DHI) was determined. Therabbits were subsequently killed and the spine segments harvested foranalysis. Results: In the disk organ culture system, rabbit articularchondrocytes transduced with AdGFP successfully integrated into thehost tissue and continued to express the transgene. One month after theinjection, rabbit disk explants injected with articular chondrocytes ex-pressing hBMP-7 accumulated 49.2% more proteoglycans in the nucleus

pulposus (NP) than those injected with articular chondrocytes transducedwith AdGFP (P � 0.05). The in vivo experiments have shown that theinjection of articular chondrocytes into the rabbit disk may be beneficial.Six weeks after cell transplantation, the %DHI in the disks transplantedwith chondrocytes overexpressing hBMP-7 was 25% higher than thecontrol (AdGFP) group. Conclusions: We have demonstrated, for the firsttime, that rabbit articular chondrocytes overexpressing hBMP-7 can sur-vive in a rabbit disk explant where they stimulated matrix production inthe NP. Furthermore, we have shown that the transplantation of chon-drocytes over-expressing hBMP-7 improves disk tissue proteoglycan pro-duction and disk height in the rabbit. Our encouraging findings suggestthat chondrocyte transplantation may replenish degenerating disks withvital cells and provide appropriate trophic factors; this is a promisingstrategy for the treatment of symptomatic disk degeneration.

SCIENTIFIC PAPER PRESENTATIONS

ROOM A

VITAMIN D DEFICIENCY AND OPIOID ANALGESIC USEAMONG CHRONIC PAIN PATIENTSMichael Turner, MD; Jennifer L. Kerkvliet, MS; W. MichaelHooten, MD

Introduction: Hypovitaminosis D is a prevalent vitamin deficiency amongpatients with chronic pain. Seemingly unrelated, the use of opioid anal-gesic medications for treatment of noncancer pain is increasing. Theobjective of this study was to determine the association between vitaminD deficiency and opioid analgesic use among a heterogeneous group ofchronic pain patients undergoing multidisciplinary pain rehabilitation.Methods: Serum vitamin D levels were collected at admission from 162consecutive patients admitted to the Mayo Clinic Pain RehabilitationCenter from February to August 2006. At admission, the daily opioid dosewas converted to oral morphine equivalents. Patients with vitamin Ddeficiency (� 20ng/ml) were compared with patients with levels�20ng/ml on oral morphine equivalents, demographics, pain duration,pain severity, and body mass index. Continuous variables were comparedusing two-sided t tests with a significance level of �0.05. Results: Themean age was 48 yrs (SD � 13). Seventy-nine percent (n � 128) werefemale and 97% (n � 158) were Caucasian. The mean pain duration was10.8 yrs (SD � 13.1). The mean vitamin D serum level was 29.2 ng/ml (SD� 12.0) and hypovitaminosis D was found in 24.7% (n � 40) of patients.Sixty-three percent (n � 102) of patients were using opioids uponadmission. The mean morphine equivalent dose was 95.2 mg/day. Themean morphine equivalent dose among patients with vitamin D defi-ciency was 140.2 mg/day and the mean dose among nondeficient patientswas 67.8 mg/day (P � 0.015). No significant differences in demographics,pain duration, pain severity, or body mass index were identified betweenthe two groups. Conclusions: In this consecutive series of chronic painpatients, hypovitaminosis D was associated with higher daily dosages ofopioids compared with nondeficient patients. While opioids do not di-rectly impede vitamin D metabolism, use of higher opioid dosages amongchronic pain patients may be indicative of dietary habits and impairmentsin physical functioning that increase the risk of developing hypovitamin-osis D. Further research is needed to clarify the risk factors and clinicalcorrelates of vitamin D deficiency among patients with chronic pain.

CURRENT PARTICIPATION AND INTEREST IN RESEARCHTRAINING FOR PHYSICAL MEDICINE ANDREHABILITATION (PM&R) FACULTY AND RESIDENTSScott M. Paul, MD; Frederick P. Ognibene, MD; N. Lynn H.Gerber, MD

Objective: To determine the current experience and interest in formalresearch training in PM&R faculty and residents in a major metropolitanarea, to help with planning the offering of formal didactic researcheducation. Design: World Wide Web Based Survey. Participants: Facultyand residents affiliated with 11 PM&R training programs in the New YorkCity metropolitan area. Results: Over the one month survey period 156responses (46% overall response rate) were received. Respondents in-cluded 107 residents (54% response rate) and 49 faculty physicians (35%response rate). Responses were received from persons affiliated with 9 of


11 programs. Fifty-four percent reported current involvement in PM&Rresearch, and 65% reported having some formal research training. Al-though a majority of respondents reported some training in epidemiol-ogy, ethics and regulation of human subjects, most reported no trainingin monitoring research, government regulatory issues, or preparationand funding of a research study. Sixty-five percent were definitely orprobably interested in participating in a formal course on clinical re-search. A similar percentage was equally willing to commit to attendingtwo 90-min lectures per week and traveling to a central location to receivethe course. Fewer respondents were willing to commit to five months ofinstruction. Conclusions: Residents and faculty in PM&R training pro-grams are participating in research and are interested in improving theirknowledge of the principles and practice of clinical research. Most arewilling to commit significant time and effort to obtaining that knowledge,although they may be unable to participate for an extended period of time.Given the limited resources for research training, offering a centrallylocated course, available to members of all training programs in a geo-graphic area, may be an effective way to increase research knowledge inPM&R. Plans are underway to provide such a course in the New Yorkmetropolitan area and study its impact.

CORRELATION OF CLINICAL EXPERIENCES AND CAREERPLANS OF GRADUATING RESIDENTS DURING THE 2004–2005 ACADEMIC YEARVishwa S. Raj, MD; Diana H. Rintala, PhD

The purpose of this survey was to collect data from graduating physiatryresidents during the 2004–2005 academic year and correlate clinicalexperiences during residency with future career goals. Residents wereasked to complete a survey pertaining to 11 of the primary educationalobjectives outlined by the SAE examination, as well as answer open endedquestions regarding fellowship and/or professional plans. Out of 386surveys distributed, 93 responses were returned in confidential manner.Based on preliminary results of the survey, residents planned to pursuethe following areas of specialization: interventional-based procedures(43); musculoskeletal medicine (32); general physiatry (17); sports med-icine (9); spinal cord injury medicine (6); electromyography (4); trau-matic brain injury and/or stroke (2); and industrial rehabilitation (1). Ofthe residents pursuing general physiatry, 11.8%, compared with 66.7%pursuing all other career goals, planned to pursue fellowship (exactsignificance 2 sided: �0.001 with Forward Exact Test). Of the residentspursuing interventional pain management, 95.3%, compared with 22.4%pursuing other career goals, also planned to pursue fellowship (exactsignificance 2 sided �0.001 with Forward Exact Test). Compared withresidents not pursuing interventional pain fellowship, residents pursuinginterventional pain fellowship spent less months training in traumaticbrain injury (7.45 vs. 4.10 mos; t test: 4.898; P value with equal variancesassumed �0.001), spinal cord injury (5.33 vs. 3.45 mos; t test: 2.784; Pvalue with equal variances assumed � 0.008), and medical rehabilitation(7.64 vs. 4.83 mos; t test: 2.274; P value with equal variances assumed �0.027). Compared with residents not pursuing musculoskeletal fellow-ship, residents pursuing musculoskeletal fellowship were less preparedfor physiatric therapeutics (4.28 vs. 3.88; t test: 2.176; P value with equalvariances assumed � 0.034), less confident in writing prosthetic andorthotic prescriptions (3.69 vs. 3.00; t test: 2.468; P value with equalvariances assumed � 0.017), less confident in writing physiatric thera-peutic prescriptions (4.25 vs. 3.69; t test: 2.373; P value with equalvariances not assumed � 0.027), less confident in performing interven-tional physiatric procedures (3.00 vs. 2.13 mos; t test 2.250; P value withequal variances assumed � 0.029), and less confident in performingbotulinum toxin injections and/or phenol injections (4.25 vs. 3.69; t test:2.409; P value with equal variances assumed � 0.020). These resultsindicate trends in career goals of graduating residents. In addition, whenconsidering residents pursuing interventional pain fellowship or muscu-loskeletal fellowship, certain characteristics were noted with respect tomonths of training in different subspecialities and confidence and/orpreparedness levels when dealing with specific skill sets.

ACTION OBSERVATION IMPROVES THE EFFECTS OFMOTOR TRAINING IN STROKE PATIENTSPablo Celnik, MD; Brian Webster, BS; Davis Glasser; LeonardoG. Cohen, MD

Background: Training represents the goldstandard in neurorehabilitationof motor disabilities following stroke. Similar to healthy older adults, the

ability to encode motor memories by motor training is reduced in chronicstroke patients. Observation of another individual performing a motoraction, action observation (AO), results in activation of various areas inthe motor network and could play a role in neurorehabilitation. Hypoth-esis: Action observation will enhance the beneficial effects of motortraining (MT) on motor memory formation in patients with chronicstroke. Methods: Eight chronic stroke patients (mean age 58.7�/-13.3)participated in a crossover, randomized and counterbalanced experimen-tal study testing 3 interventions on different days. The endpoint measureof the study was the ability to encode a motor memory following: MT(practice of voluntary thumb movements performed in a specific directionfor 30 min at 1Hz), MT�AO CONGRUENT (synchronous MT and obser-vation of another subject performing MT in identical direction), MT�AOINCONGRUENT (similar to MT�AO, but the movements observed are ina direction opposite to the trained movements). Motor evoked potentials(MEP) to TMS were recorded from muscles mediating movements in thetraining/observed (MEP AGONIST) and in the baseline (MEP ANTAGO-NIST) directions. Data were analyzed using ANOVARM with factors IN-TERVENTION (MT, MT�AO CONGRUENT, MT�AO INCONGRUENT)and TIME (Pre-, Postintervention), and post hoc with Forward PLSD.Results: Attention, fatigue and motor training kinematics did not differacross sessions. ANOVARM showed a significant TIME � INTERVENTIONinteraction (F2,12 � 4.2 P � 0.05) on the ability of training to encode amotor memory. Posthoc testing showed that the effects of MT�AOCONGRUENT were more prominent (from 3.4 � 2.6% to 13.4 � 5.1%,mean�SEM; P � 0.05) than those elicited by MT alone (P � 0.03) orMT�AO INCONGRUENT (P � 0.03). This effect was accompanied by anincrease in MEP AGONIST and a decrease in MEP ANTAGONIST ampli-tudes relative to baseline in the MT�AO CONGRUENT condition only (P� 0.05). Conclusions: Action observation improved training effects onmotor memory formation in chronic stroke patients. These findingssuggest that task-specific action observation may be a useful strategy toenhance the beneficial effects of motor training after stroke.

ROOM B

EXAMINATION OF SELECTED RISK FACTORS FORPNEUMONIA DURING STROKE REHABILITATIONChristina Marciniak, MD; Alexander Korutz, MS-3; Emily Lin,MS-3; Elliot Roth, MD; Linda Lovell, BS

Introduction: Nosocomial pneumonia is a frequent complication amongstroke survivors, with up to 30% of acute stroke patients developing lowerrespiratory tract infections within 3 mos following stroke. Previous stud-ies have suggested that the use of H2 receptor antagonists (H2RAs),proton pump inhibitors (PPIs), tracheotomy tubes, and feeding tubes areassociated with increased risk of pneumonia. Conversely, several studieshave suggested that ACE inhibitors may serve a protective role due totheir tussigenic effect. Although many stroke patients are placed on thesemedications or devices, studies examining the role each of these factors inthe development of pneumonia in poststroke patients undergoing inpa-tient rehabilitation are limited. This study investigates the effects ofH2RA’s, PPI’s tracheotomy tubes, feeding tubes, and ACE inhibitors in thedevelopment of pneumonia in hospitalized stroke patients. Methods: Inthis case-matched, controlled study, a database containing a prospectivecohort of patients admitted for stroke rehabilitation was queried forpoststroke patients hospitalized during a 5-yr period, and whose rehabil-itation stay was complicated by pneumonia. Subjects identified werematched with 36 controls stroke patients admitted during the sameperiod. Matching was performed on the basis of age, gender, side ofstroke, depth of stroke (cortical vs. noncortical), and NISS score. Dys-phagia severity, the presence of a tracheotomy or feeding tube, and theuse of H2RAs, PPIs, or ACE inhibitors were recorded from a subsequentchart review. Results: McNemar’s test of paired data revealed that patientson PPIs or H2Blockers had an increased risk of developing pneumoniawith an odds ratio of 3.3 [95% confidence interval (CI) 1.0–13.7]. Uni-variate analyses also revealed that significantly higher risk of pneumoniawas also found in patients with a tracheotomy tube [Odds ratio: 10; 95%CI 1.4-434], any feeding tube [Odds ratio: 5; 95% CI 1.4-27.0], and thosewith dysphagia [Odds ratio: 15; 95% CI 0.2-157 2.3631]. A conditionallogistic regression to adjust for the effects of conventional risk factors,however, did not find that these were independently associated withpneumonia risk, including decreased risk with ACE inhibitors. However,when all factors were entered into a multivariate conditional logisticregression, none of the factors proved to be statistically significant.


Conclusions: Though the individual risk factors evaluated may not inde-pendently increase the likelihood of poststroke pneumonias in acuterehabilitation, the presence of the above four risk factors is associatedwith a significantly increased risk. In our population, the use of ACEinhibitors did not decrease pneumonia incidence. Four risk factors wereassociated with increased risk of pneumonia during stroke rehabilitation(PPI or H2 Blocker use, tracheostomy, feeding tube, and the presence ofdysphagia). However, none of these factors demonstrated an independentassociation with pneumonia on multivariate analysis.

AMPUTEE BONE DENSITY EVALUATION FORCOMPETITIVE SPORTS (RUNNING) CLEARANCEMaj Derek J. Stocker, MD; Col Raul Marin; Ltc Aaron L. Stack;Maj Brandon J. Goff; Ltc Paul F. Pasquina

Introduction: Advances in body armor have lead to a higher rate ofsurviving amputees in the current conflicts in Iraq and Afghanistan.Young and active service members are aggressively recovering from theirinjuries and progressing rapidly toward pursuing increasingly vigoroussports activities, to include competitive running. As a result, many ofthese individuals are receiving specially adapted running prostheses asearly as 4–6 mos after their amputation based on the clinical recovery ofthe patient. There have been cases of residual limb bone fractures duringparticipation in the competitive running rehabilitation program attrib-uted to osteopenia. As a result, beginning in September 2005, it has beenthe policy of our service to obtain a dual-energy x-ray absorptiometry(DEXA) study of all amputees being evaluated for enrollment into oursports running prosthesis program in an attempt to assess the subsequentfracture risk. Although little is known about the nature and extent of boneloss immediately following amputation and no widely accepted criteriaare available for screening such patients for fracture risk or even whatamount of bone loss would be expected, the protocol used to performthese studies included obtaining bone density assessments of both hipsand the whole body. This retrospective study analyzed these studies todetermine the degree of bone loss in an effort to help determine thefracture risk encountered by these patients. Methods: Between September2005 and May 2006, nineteen patients with unilateral amputations re-ceived DEXA studies. Using the intact limb as a surrogate baseline, thecomparative bone density decrease was determined for the amputatedlimb at the level of the femoral neck and total hip. In addition, whole bodycomparisons were also obtained. Results: The DEXA results of 19 patientswere reviewed. One patient had retained shrapnel, limiting the measure-ments of the lower extremity and total hip assessments. The patients wereaged 29�/-1.9 yrs, predominantly male (18/19), and had undergone aunilateral amputation 9�/-1 mos before the study. Etidronate and Cele-coxib were administered to 68% (13/19) of these patients during theirrehabilitation period for heterotopic ossification prophylaxis. Significantbone loss was seen in all evaluated sites and these patients were notcleared for the sports running program. At the hip, femoral neck, andwhole body regions DEXA scans shows BMD differences of -0.1545�/-0.0241, -0.1119�/-0.0232, and -0.2876�/-0.1562 with P � 0.0005. We arenot aware of any recurrent fractures since the onset of our screeningprogram. Conclusions: Rapid bone loss of the proximal amputated limbseems to occur following traumatic amputation despite early ambulationand bisphosphonate use. Our policy of clearing amputees for aggressivesports running based on proximal amputated limb bone densities asproxies for more distal sites seems to be justified as it is presumed thatbone densities at the more distal sites are likely to be more severelyaffected. A prospective study on the natural course of amputated limbosteoporosis and the identification of distal limb bone density referencevalues is currently underway.

IMPLEMENTATION OF AN ULTRASONIC BLADDERSCANNING PROTOCOL FOR ASSESSMENT ANDMANAGEMENT OF URINARY RETENTION IS ASSOCIATEDWITH A REDUCTION IN URINARY TRACT INFECTIONS INAN ACUTE INPATIENT REHABILITATION HOSPITALKate W. Paylo, DO; Heather Cosner, RN; Alan Alfano, MD; LoriAylor, RN; Mary Bryant, MD; Jonathan Evans, MD, MPH

Background: Urinary retention (UR)is common among hospitalized pa-tients undergoing acute inpatient rehabilitation and may arise in thesetting of a variety of clinical conditions that are common in this setting,including acute stroke, brain tumor, spinal cord injury, prostatic enlarge-

ment, and drug therapy. In addition to functional dependence, UR is alsoassociated with significant morbidity, related to both renal failure andurinary tract infection (UTI). The evaluation and management of UR istherefore an important rehabilitation goal. Intermittent catheterization(IC) of the bladder has been a mainstay of evaluation, monitoring, andtreatment of UR. Unfortunately, IC may itself be a cause of UTI andbacteremia. Consequently, strategies which minimize the use of IC inevaluation and monitoring of UR are desirable in an effort to reduceinfection risk. Objective: To evaluate the effect of a hospital-wide ultra-sonic bladder scanning protocol on reducing the incidence of hospital-acquired UTI in an acute inpatient rehabilitation hospital. METHODS:Design: Intervention study (pre/post design). Intervention: Implementa-tion of a bladder scanning protocol for evaluation and management of URin all adult patients for whom assessment of postvoid residual urinevolume is clinically indicated. Setting: A 50-bed academic acute inpatientrehabilitation hospital. Main outcome measures: Rates of hospital ac-quired UTI per 1000 patient-days before and after implementation of theclinical protocol. Results: UTI rates fell significantly from 7.13–3.88 per1000 patient-days over a 2-yr period, despite an increase in severity ofclinical illness (“case-mix”) over the same time period. There was also asignificant reduction in the number of intermittent catheterizationsperformed. Discussion: Implementation of a hospital-wide bladder scan-ning protocol was associated with a clinically meaningful overall reduc-tion in the incidence of urinary tract infections. Since IC was alsoreduced, it is plausible to conclude that UTIs were prevented throughavoidance of IC, particularly for diagnostic purposes. Reduction in hos-pital acquired UTI is a desirable outcome, irrespective of the mechanismby which it was achieved. Nevertheless, there are several other potentialexplanations for the reduction in UTIs. Reported rates of hospital acquiredurinary tract infection are confounded by asymptomatic bacteriuria.Consequently, reduction in IC is expected to reduce the rate of testing forUTI, which in turn reduces the rate at which asymptomatic bacteriuria isidentified. Moreover, the implementation of this protocol may have re-sulted in other changes in physician behavior, leading to a reduction intesting for urinary tract infection even among patients without UR.

DOES METHYLPHENIDATE ENHANCE COGNITION AFTERMODERATE TO SEVERE TBI?Harvey Levin, PhD; L. Corwin Boake, PhD; Sureyya Dikmen,PhD; Nancy Temkin, PhD; Maya Troyanskaya, MD; GerardFrancisco, MD

Experimental animal models have implicated dopaminergic receptors inprefrontal cortex (PFC) in the mediation of working memory. Withestablished efficacy of methylphenidate (MPH) in the treatment of atten-tion deficit hyperactivity disorder and the widespread administration ofthis drug to treat cognitive impairment after traumatic brain injury (TBI),we completed a two center randomized, placebo-controlled double-blindclinical trial comparing the effects of MPH given in a dose of 15 mg twicea day for 28 days to placebo on mitigating working memory deficit inpatients enrolled between 3 and 12 mos after moderate to severe TBI.Prospective recruitment of patients with working memory deficit on atleast one of two screening tests (Paced Auditory Serial Addition, Self-Ordered Pointing Test) yielded 74 patients who were randomized to MPH(n � 38; mean Glasgow Coma Scale score � 6.9, SD � 3.5; mean age atstudy � 27.4 yrs, SD � 10.3; 18.4% female; mean education � 1.8 yrs, SD� 2.0; mean postinjury interval � 10.2 mos, SD � 4.3) or placebo (n �36; mean Glasgow Coma Scale score � 6.9, SD � 3.5; mean age at study �29.3 yrs, SD � 11.5; 33.3% female; mean education � 12.2 yrs, SD � 2.3;mean postinjury interval � 11.0 mos, SD � 4.6). A separate randomizationwas performed at each center. Cognitive tests were given on three occasions:pretreatment baseline, final day of treatment, and 1 mo follow-up. A visualN-back working memory task for letters, including 1-, 2-, and 3-back con-ditions, was the primary outcome measure. Secondary outcome measuresincluded the verbal selective reminding, consonant trigram distracter mem-ory, and symbol digit modality tests. There were no significant group differ-ences in patient features or pretreatment cognitive performance. Net accu-racy of working memory averaged across 1-, 2-, and 3-back conditions did notdiffer between groups F(1,60) � 0.19, P � 0.665, and there was no significantgroup difference by occasion, F(2,60) � 0.28, P � 0.757. Significant between-group differences in the posttreatment results were found on only one of thesecondary cognitive measures. On the 30-min delayed recall of the selectivereminding test there was a significant between group difference in improve-ment from baseline to 1 mo (�2 2 � 6.14, P � 0.046). MPH treated patientshad improved delayed recall from baseline (mean � 6.54, SD � 3.53) to 1 mo(mean � 7.73, SD � 3.15) (�2(1) � 9.13, P � 0.003), whereas delayed recall


by patients in the placebo group did not significantly change from baseline(mean � 7.40, SD � 3.36) to 1 mo (mean � 7.20, SD � 3.61)(�2(1) � 0.27,P � 0.603). However, total recall of words across trials of the selectivereminding test did not differ between groups. Our findings show a lack ofconsistent evidence for MPH efficacy in mitigating working memory deficitafter TBI and only an isolated effect in other cognitive domains. These resultsdiffer from a previous crossover study in which MPH increased responsespeed on reaction time measures of attention in moderate to severe TBIpatients enrolled at least 3 mos postinjury. We postulate that MPH may haveshort-term effects that diminish over long periods of treatment. In summary,the results of this randomized, placebo-controlled clinical trial offer sparsesupport for the efficacy of MPH in treating cognitive impairment in patients3–12 mos following moderate to severe TBI.

ROOM C

PREVALENCE OF ROTATOR CUFF TEARS ANDTENDONOPATHIES IN PAINFUL HEMIPLEGIC SHOULDERSepideh Haghpanah MD; Rajiv Shah MD; John Chae MD

Objective: Assess the age adjusted prevalence of rotator cuff tears andtendonopathies among stroke survivors with painful hemiplegic shoulder.Method: Shoulder MRI scans of 76 chronic stroke survivors with shoulderpain of �4 on a 0-10 numeric rating scale (Brief Pain Inventory Question12) was reviewed by a single radiologist for presence of rotator cuff tears,rotator cuff tendonopathies, subacromial bursa fluid and labral ligamen-tous complex abnormalities. Results: The mean age of the patients was 57� 11 (SD)-yrs with a mean time from stroke onset of 66 � 84-mos.Rotator cuff tears were present in 34% of patients. The Supraspinatouswas the most commonly torn (29%) followed by Infraspinatous (12%),Subscapularis (3%), and Teres Minor (1%). At least 1 tear was present in11% of patients less than 50-yrs old, 31% in age group 50–59-yrs, 42% inage group 60–69-yrs and 50% in age group older than 69-yrs. Overallprevalence of any tendonopathy was 57%. The Supraspinatus was themost commonly involved (51%) followed by Infraspinatus (21%) andSubscapularis (12%). Tendonopathy was present in 51% of patients lessthan 50-yrs old, 52% in age group 50-59-yrs, 58% in age group 60-69-yrsand 60% in age group older than 69-yrs old. Labral complex abnormalitywas found in 8% of patients, and subacromial bursa fluid in 25%. Linearregression showed that rotator cuff tears, rotator cuff tendonopathy,subacromial bursa fluid and labral ligamentous complex abnormalitywere not related to degree of hemiplegic shoulder pain. Conclusions:Rotator cuff tears and tendonopathies are common in painful hemiplegicshoulder. The prevalence of rotator cuff tears is age dependent, but theprevalence of tendonopathies is not. However, these abnormalities do notseem to be related to the severity of hemiplegic shoulder pain.

FRACTURE OF HETEROTOPIC OSSIFICATION IN A YOUNGTRAUMATIC AMPUTEE: A CASE REPORTAllison J. Franklin, DO; Donald A. Gajewski, MD

Heterotopic ossification (HO) is the formation of mature lamellar bone inthe soft tissues and periarticular areas. HO is clinically significant in only10–20% of reported cases and the reported incidence ranges from 11–76%. Heterotopic ossification is most commonly reported after electivejoint surgery, burns, and spinal cord injuries, but may also be seen intraumatic amputees. These lesions tend to be asymptomatic. Our patientwas a 22-yr-old male with a traumatic right transfemoral amputationsecondary to a blast injury. He had developed asymptomatic heteroptopicbone formation in his residual limb. Five months after his injury date hefell on his residual limb when downhill skiing. This resulted in immediateand severe pain. A hard, tender, mobile nodule was present within the HO.Radiographs demonstrated a fracture through the existing heterotopicbone. He was treated conservatively and quickly returned to his prefrac-ture activity levels. To our knowledge, this is the first report of fracturedheterotopic ossication in an amputee.

“BLIND-SIDE” TRAINING WITH FORCED-CHOICE TASKSFOR HOMONYMOUS HEMIANOPIAYonghua Tai, MD, PhD

Objective: To develop discriminative visual functions in the blind field bytraining target detection and localization with forced-choice tasks. De-sign: Case series study. Thirteen patients with homonymous hemianopia

or quadrantanopia caused by tumor, arteriovenous malformation, trau-matic brain injury, or stroke were recruited 12–144 mos beyond the acutestage. All patients’ visual fields were confirmed to be stable by serialHumphrey perimetry and SLO macular perimetry. They were alsoscreened by Sunnybrook Neglect Battery, and all had no neglect.

Each subject underwent 6 wks of computerized training of targetdetection on screen. In each trial, a transient target (200 msecs) randomlyappeared in the four quadrants at 20-degree eccentricity from centerfixation cross. The subjects were prompted to guess the target location.The home blindsight training comprised 150 trials per session, 2 sessionsdaily. The subjects responded with pressing a specific key correspondingto the target locations. In-facility training comprised 100 trials persession per week. The subjects responded with saccadic eye movementtoward the target or verbal reporting target location. Feedback wasprovided to the subjects during the training. Each subject was tested forthe above tasks. They were also tested for discriminating presence vs.absence of a target. Results: Laser scanning ophthalmoscopy confirmedthat the visual functions developed in the blind field was not a result ofscatter light detected by the normal side retina. Comparing with thepretraining performance, there were significant improvements in sac-cadic hit rate (P � 0.01), verbally expressed detection rate (P � 0.05),saccadic localization rate (P � 0.01), and verbally expressed localization(P � 0.01). Conclusions: Nonveridical vision is a true function of theblind side, rather than scatter light onto the normal retina. Patients withhomonymous hemianopia/quadrantanopia have significant improvementin verbally expressed discrimination, oculomotor accuracy performingsaccades to targets in the blind visual field after training with forced-choice tasks. Residual vision in the perimetric blind field was present incortical and subcortical homonymous hemianopia. Nonveridical visiondeveloped after training with forced-choice tasks in blind field in a periodof just weeks, even years after brain injuries.

FUNCTIONAL OUTCOME AFTER SUBTHALAMIC NUCLEUSDEEP BRAIN STIMULATION TO TREAT PARKINSONDISEASE: A LONGITUDINAL STUDYKeith M. Robinson, MD

Introduction: As many as 50% of individuals who have PD can be labeledas medically refractory. Surgical interventions including subthalamicnucleus deep brain stimulation (STN-DBS) have been viewed as effica-cious based on short-term (�1 yr) and long term (�1 yr) studies usingimpairment based outcome measures. This longitudinal study whichassesses integrative functional outcome following STN-DBS has reachedthe 18-mo postsurgery point. Methods: Thirty-two subjects have beenassessed at three points: presurgery, 6 mos postsurgery, and 18 mospostsurgery. A final assessment will occur when subjects reach the 3-yrpostsurgery point. Subjects were recruited from among PD patientsreferred regionally for STN-DBS. Criteria for inclusion included: diagno-sis of idiopathic PD; disease severity of at least Hoehn and Yahr stage 2;a history of being levodopa responsive; persistent disabling dyskinesias orrapid motor fluctuations when using dopaminergic medications; stableon medications at least 1 mo before surgery; age �21; and no comorbidconditions that increase surgical risks. Outcome measures of integrativefunctional performance included those of mobility, gait, and balance, andselected cognitive functioning (language, executive function, visuoper-ception, visual organization). Functional performance was assessed usingthe FIM (FIM), the timed “get up and go” test under single and dual taskconditions, and functional reach test forward and backward. Cognitivefunctioning was assessed using Boston Naming Task, phonemic andsemantic verbal fluency tasks, Raven’s Colored Progressive Matrices,Clock Drawing, Hayling’s Sentence Completion Test, and Oral TrailMaking Test: Alphanumeric Sequencing. Results: Two separate repeatedmeasures multivariate analysis of variance (RM-MANOVA) analyses wereperformed using the functional and cognitive measures as dependentvariables. The independent variable was time (presurgery, six monthspost, 18 mos post). Based on Wilks’ criterion, performance for the com-bined dependent functional measure variables was associated with time(F18,98 � 3.9, P � 0.000). This suggests that, averaged over subjects,functional measure scores differ as a function of time since surgery.Performance for the combined dependent cognitive measures showed noassociation between scores and time since surgery (F12,114 � 1.4, P �0.183). To examine the impact of the main effect of trends over time oneach of the functional measures, univariate analyses were conducted.Statistically significant improvements were found for performance on theSelf-Care subscale of the FIM (F1,28 � 14.3; P � 0.000), the Mobilitysubscale of the FIM (F1,28 � 8.7; P � 0.001), the Social Cognition


subscale of the FIM (F1,28 � 3.5; P � 0.038), the Communicationsubscale of the FIM (F1,28 � 8.6; P � 0.001), and the functional reachforward (F1,28 � 5.9; P � 0.005). Pairwise comparisons (using Bonfer-roni adjustments) of scores showed that performance improved morenoticeably for the first postsurgery (six month) time period, although thedifference was significant only for the functional reach forward. Conclu-sion: Early evidence indicated STN-DBS may serve to improve basicfunctional status and balance during the initial six months postsurgically,and that functional status and balance stabilized thereafter until 18 mos.Gait and cognition did not seem to be influenced either positively ornegatively by STN-DBS at 6 and 18 mos postsurgically.

RELATIONSHIP BETWEEN STRENGTH AND MUSCLE AREADIFFERS WITH OBESITY: INTERIM ANALYSISNeil A Segal, MD; James Torner PhD; Morgan Brubaker BS;Bret Goodpaster PhD; Bridget Zimmerman PhD

Objective: To assess the relationship between knee extensor strength andmuscle mass in normal weight and obese subjects, age 50–59 yrs; 2) Todetermine whether lower limb lean body mass (LBM) or lean cross-sectional area (CSA), will provide a robust and accurate means of nor-malizing strength independent of body mass. Design: Cross-sectionalobservational study. Statistical analysis included descriptive statistics andcomparison using mixed models for clustered data (two knees per sub-ject). Setting: Community-acquired population. Interventions: Not appli-cable. Main Outcome Measures: Relationship between isokinetic (IK)knee extensor strength and muscle cross-sectional area (CSA) or leanbody mass. Results: A total of 106 subjects have been enrolled to date (51male, 55 female). Two hundred eleven lower limbs were studied; 56 innormal weight, and 155 obese individuals; 62 with knee OA and 149without knee OA. Mean � SD peak IK knee extensor torque did notsignificantly differ between normal weight (108.1 � 42.2) and obese(116.7 � 46.7) subjects, nor between nonknee OA (116.0 � 43.8) andknee OA (110.4 � 49.6) subjects, whether unadjusted or adjusted for sexand OA status (P � 0.2). After scaling to adjust strength for body size,peak IK knee extensor torque did not significantly differ between groups(normal weight 110.4 � 8.2N, and obese 117.5 � 3.7N). A linear modelwith scaled peak IK knee extensor force as the response and quadricepsmuscle CSA (cm2) as the independent variable was fit for normal weightand obese individuals. In males, a significant difference was detectedcomparing the positive relationship between strength and CSA in obese(slope � 1.61 � 0.34), but not in normal weight subjects, whetherunadjusted (P � 0.0126) or adjusted for knee OA status (P � 0.0006). Infemales, there were no significant differences between slopes of theserelationships. In females, the y-intercepts significantly differed, withobese having more strength (41.9N�m) than normal weight subjects(23.4N�m) at any level of CSA. Scaled knee extensor strength correlatedbetter with quadriceps muscle CSA (r � .67; P � 0.0001) than with thighLBM (r � .24; P � 0.0006). Conclusions: Results of this interim analysissuggest that there is an altered relationship between muscle strength andbulk in obese relative to normal weight females, possibly due to aqualitative deficit in muscle function with obesity. However, a similaraltered relationship was not detected in males. Further recruitment ofnormal weight subjects will be necessary achieve adequate power to testthese relationships more fully. Quadriceps CSA is a more robust means ofnormalizing strength independent of body mass than use of LBM.

AAP Annual Meeting Poster GrandRounds

Thursday, April 12, 2007

COMPARING MOTOR RELATED CORTICAL POTENTIALSAND REACTION TIME BETWEEN HEALTHY CONTROLSAND PATIENTS WITH TRAUMATIC BRAIN INJURYHenry L. Lew, MD, PhD; Max Gray, BA; Darryl Thomander,PhD; John H. Poole, PhD

This study investigated the effectiveness with which: (i) response-lockedmotor related cortical potentials (MRCP) and reaction time (RT) mea-sures separated traumatic brain injury (TBI) patients from healthy con-

trols, and (ii) how the effectiveness changed as the ecological relevance ofthe stimulus was increased. We also examined the relationship betweenthe RT and waveform morphology. Methods: Eleven patients (mean age25.5 yrs) who had recovered from severe TBI (average 15.5 mos postin-jury) were recruited from a university affiliated hospital. Eleven age-matched volunteers served as healthy controls.Each participant performed four different discrimination tasks (two con-ventional tasks, and two other tasks in the same modalities, but designedto be more ecologically relevant). The two conventional tasks were: visualcolor (VC) and auditory tone (AT) discrimination. The ecologically rele-vant tasks were: visual facial affect (VFA) and auditory word category(AWC) discrimination. Neuroscan (El Paso, TX) hardware and softwaresystems were used for task presentation, and collection of electroenceph-alogram (EEG) and behavioral measures.

Epochs (�1000 to 50 msecs) were time-locked to the button pressfor the target trials after conventional EEG processing. Averaged wave-forms were digitally filtered (band pass 2–10 Hz) and baselined to eachparticipant’s mean 100-msec prestimulus interval. The motor peak (MP)of the preresponse negativity was measured relative to the baselineinterval at electrode C3, defined as the maximum negativity in the �200-to 0-msec interval. RT and MP data were concurrently collected. TheCoefficient of Variation (CoV) for RT was used as an index of within-participant consistency. One-tailed Wilcoxon signed ranks tests were usedto determine significant group differences for all measures, and effect sizewas used to determine effectiveness of the MRCP and RT measures indiscriminating controls from TBI individuals. Results: Response-lockedMP amplitude was significantly different (P � 0.05) between participantgroups for VC, AT, and AWC (Z � 2.67, 2.31, 1.69), but not for VFA (Z �1.25, P � 0.11). Latency of the MP did not distinguish the two groups inany condition (all Z �1.38, all P � 0.08). Mean RT was significantlydifferent (P � 0.05) between healthy participants and TBI patients in allfour tasks, VC, AT, VFA, and AWC (Z � 1.69, 1.68, 2.31, and 2.49,respectively). For CoV measures related to RT, there was no statisticaldifference between the two groups (all P � 0.08). During inspection ofindividual waveforms, we noticed an interesting pattern that divided theparticipants into two groups according to their RT measures. The fasterresponders had more closely synchronized preresponse negative peaks,while the slower responders showed a more dispersed pattern. Conclu-sion: When simple conventional stimuli were used, response-locked MPamplitude measures were more effective in discriminating TBI patientsthan RT measures. When more complex ecologically relevant stimuliwere used, the trend was reversed. Also, the observed difference inmorphology for the preresponse negativities suggests different underly-ing response execution mechanisms that deserve further research.

WHEELCHAIR USE, SOCIAL INTEGRATION, AND MOBILITYAMONG ADULTS WITH SPINA BIFIDAAnna Gaines, BS; Brad E. Dicianno, MD; Diane M. Collins, PhD

Objective: Over 70,000 Americans have Spina Bifida (SB), and many usewheelchairs to increase mobility and independence both at home and inthe community. However, individuals with SB are often sedentary andsocially isolated despite using mobility devices. Little research has beendone to characterize the use of mobility devices in adults with SB andwhether such use impacts mobility or social integration. Since the cog-nitive consequences of hydrocephalus, which is common among individ-uals with SB, and the isolating effects of depression may also impactcommunity involvement, evaluating the relationship among these factorsis vital.

The purpose of this study was to evaluate the association betweenwheelchair use, mobility status, and the extent of social integration inadults with SB. We hypothesized that, overall, social integration andcommunity mobility, as measured by the Craig Handicap AssessmentReporting Technique – Short Form (CHART-SF), will be lower for indi-viduals who use manual and power wheelchairs than for individuals whoare able to ambulate, regardless of depression scores or shunt history.Design: Eighty-one adults with SB completed surveys regarding wheel-chair use, medical history, Beck Depression Inventory (BDI), and theCHART-SF, which measures functionality in the domains of PhysicalIndependence, Mobility, Occupation, Cognitive Independence, Social In-tegration, and Economic Self-Sufficiency. Individuals were grouped asambulators, manual wheelchair (MWC) users, or power wheelchair (PWC)users. Results: The groups did not differ significantly in age, gender, race,or BDI total scores. They did differed significantly in shunt history (P �0.000), with 55.6% of ambulators, 90.9% of MWC users, and 95.2% ofPWC users having history of at least one shunt. We found associations

April 2007 2007 AAP Annual Meeting Abstracts 331

between shunt history and Physical Independence (P � 0.027) and Mo-bility (P � 0.017) CHART-SF scores but not with the other domains.Groups differed significantly in Physical Independence (P � 0.000), Mo-bility (P � 0.005), Occupation (P � 0.002), and Cognitive Independence(P � 0.000), with post hoc-analyses revealing ambulators had signifi-cantly higher scores than both MWC and PWC groups in all of thesedomains, even when shunting was used as a covariate. Mobility groups didnot differ in the CHART-SF domains of Social Integration or EconomicSelf-sufficiency. Conclusions: Ambulators were more physically indepen-dent and, as hypothesized, more mobile in daily and community activitiesthan those who used wheelchairs. However, reduced Mobility scoresacross all three groups suggests lower than average daily and communityactivity levels even for ambulators. The functional discrepancy betweenambulators and wheelchair users with respect to Physical Independenceand Mobility raises the question of whether better wheelchair provisionmay help to alleviate these significant differences.

Reduced cognitive independence and occupation scores amongwheelchair users support earlier studies that demonstrate these individ-uals often have nonverbal and organizational deficits and high rates ofunemployment. Despite other differences, mobility groups were surpris-ingly not significantly different in social integration scores, which werestrikingly low across all groups. Additional research is needed to identifyspecific barriers to social integration for all individuals with SB.

Finally, clinically significant depression, though not associated withany CHART-SF domains, was documented in a majority of participants.Research to better characterize depression among adults with SB iswarranted.

ULTRASOUND IN THE DIAGNOSIS OF MENISCAL TEARSOF THE KNEEGi-Young Park, MD, PhD; Jong-Min Kim, MD; Sung-Moon Lee,MD; Michael Lee, MD

Methods: Twenty-seven knees of twenty-two patients (16 female, 6 male;age range, 14-74 y; mean age, 50.4 y) with meniscal tears on MRI werestudied. In the 54 menisci of 27 knees, twenty-nine meniscal tears of theknee (medial 22, lateral 7) were found. Twenty eight menisci (medial 14,lateral 14) of 14 volunteers (7 females, 7 males; age range, 23-68 y; meanage 43.4 y) without tear and 25 normal menisci (medial 5, lateral 20) of22 patients formed a control group.

All patients and volunteers were prospectively examined with MRIand ultrasound. A Musculoskeletal radiologist blindly evaluated the pres-ence and location of meniscal tears on MRI and ultrasound. In MRI,meniscal tear was diagnosed by the high signal in the meniscus extendingto its articular surface, gross distortion of the normal shape, or truncationof the meniscus. In ultrasound, it was diagnosed by the appearance of adiscrete hypoechoic or anechoic cleft extending to articular surfaces or adefect in the hyperechoic meniscus. Results: Of the 22 medial meniscaltears on MRI, 20 tears (90.9%) involved in the posterior horn and 2 tear(9.1%) in the middle zone. Ultrasound showed that 18 tears were locatedin the posterior horn, 1 tear in the middle zone, and 1 tear in the anteriorhorn.

Of the 22 medial menisci diagnosed with tear on MRI, tears weredetected on ultrasound in 20 menisci (90.9%) and no tear in the other 2menisci (9.1%). Of the 19 medial menisci without tear on MRI, tears weredetected on ultrasound in 4 menisci (21.1%) and no tear in the other 15cases (78.9%). The sensitivity and specificity of ultrasound in the diag-nosis of medial meniscal tear was 90.9% (20 of 22 menisci) and 78.9% (15of 19 cases), respectively.

Of the 7 lateral menisci diagnosed with tear on MRI, tears weredetected on ultrasound in 6 menisci (85.7%) and no tear in 1 meniscus.Of the 34 lateral menisci without tear on MRI, tears were detected onultrasound in 4 cases (11.8%) and no tear in the other 30 cases (88.2%).The sensitivity and specificity of ultrasound in the diagnosis of lateralmeniscal tear was 85.7% (6 of 7 cases) and 88.2% (30 of 34 cases),respectively. The overall sensitivity, specificity, accuracy, positive predic-tive value, and negative predictive value of ultrasound for the detection ofmeniscal tears of the knee were 89.7%, 84.9%, 86.6%, 76.5%, and 93.4%.Conclusion: Ultrasound appeared to be an accurate imaging technique forthe diagnosis of meniscal tears of the knee, which strongly suggestingthat it can be a useful screening imaging technique before MRI.

Friday, April 13, 2007

RACIAL DISPARITIES IN ACCESS TO INPATIENT STROKEREHABILITATION AMONG RURAL NORTH CAROLINAAMERICAN INDIANSPatricia C. Gregory, MD; Euna Han, PhD; Kim Faurot, MPH

Objective: Prior studies have shown an urban rural disparity in access toinpatient stroke rehabilitation. This study investigated among ruralstroke patients the association between American Indian race and dis-charge to acute inpatient stroke rehabilitation. Design: A total of 17,019patients who reside in rural counties of North Carolina and were admittedto the hospital with an acute stroke (ICD codes 430-434.9 and 436) wereevaluated using the North Carolina Hospital Discharge Dataset. Out-comes evaluated included discharge to acute inpatient rehabilitation vs.(vs.) discharge to home. Factors associated with this outcome includedthe following: age (�60 yrs vs. 60� years), race (American Indian vs.white), sex, insurance status (insured vs. uninsured), days 1 day in ICUvs.�in intensive care unit (ICU) ( � 1 day in ICU), physical 3 physicaltherapy visits vs.�therapy charges �3 visits, and county level of poverty(Counties with �15% poverty). These factors were fit �15% poverty vs.into a logistic regression analysis to determine their influence on dis-charge to inpatient rehabilitation. Results: White patients were two and ½times more likely to be discharged to acute inpatient rehabilitationfacilities Crude Odds Ratio (OR) 2.51 95% Confidence Interval (95%CI)(1.11, 5.68). After controlling for age, insurance status, poverty level in thepatient’s county, physical therapy charges, and ICU charges white racewas still independently associated with discharge to acute inpatient strokerehabilitation (Or � 2.77; 95% CI � 1.16, 6.62). Conclusions: RuralAmerican Indian stroke survivors are less likely to access inpatient strokerehabilitation. Differential utilization of rehabilitation services amongAmerican Indian stroke survivors deserves further exploration to identifyways to eliminate these disparities.

GEOGRAPHIC VARIATIONS IN EPIDURAL STEROIDINJECTION USEJanna Friedly MD; Leighton Chan MD; Richard Deyo MD

Objectives: Epidural steroid injection (ESI) rates have been increasingdramatically over time with equivocal data to support their use for variouslow back pain disorders. Therefore, we sought to evaluate geographicvariations in epidural steroid injection use for low back pain within theUnited States. We also sought to determine if epidural steroid injectionrates are correlated with lumbar surgery rates or regional numbers ofphysicians performing injections. Methods: We used 2001 Medicare Phy-sician Part B claims to examine geographic variations in the use of ) codeswere epidural steroid injections. Current Procedural Technology (CPTused to identify the number of procedures performed as well as thepercentage of injections which were fluoroscopically guided. Injectionrates were compared with lumbar surgery rates. Procedure rate variationswere analyzed using state, United States. Census Bureau definitions ofregions (Northeast, South, Midwest and West) and Dartmouth Atlas ofHealth Care’s previously defined health referral regions (306 regions).Results: In 2001, there was a 5.7 fold injection rate difference between thestate with lowest rate (Hawaii at 11.5/1000 Medicare beneficiaries) vs. thestate with the highest rate (Alabama at 65.5/1000). Five of the ten stateswith the lowest injection rates are in the Northeast and six of the tenstates with the highest injection rates are in the South. Nine states hadinjection rates greater than 25% above the national average. The variationamong smaller health referral regions was even larger, with a 27-folddifference from 4.7/1000 in Owensboro, KY to 126.4/1000 in PalmSprings, CA. Seventy-one of the 306 health referral regions had injectionrates greater than 25% above the national average. Fluoroscopy usevaried dramatically between states as well as health referral regions. Only6.5% of injections performed in Vermont were fluoroscopically guided vs.88% of injections in Wyoming. Nationally, 42% of injections were per-formed using fluoroscopy. Statewide epidural steroid injection rates werepositively correlated with lumbar surgery rates (Pearson correlation co-efficient � 0.299, P � 0.033). There was also a strong correlation betweenstatewide and regional injection rates and the percentage of patients withlow back pain receiving injections (Pearson correlation coefficient �0.828, P � 0.000) as well as the percentage of patients receiving bothepidural steroid injections and lumbar surgery during the study period(Pearson correlation coefficient � 0.389, P � 0.005). There were nocorrelations between statewide epidural steroid injection rates and nor-

332 2007 AAP Annual Meeting Abstracts Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 4

malized number of primary care physicians and specialists (Pearsoncorrelation coefficient � -.193, P � 0.174) or specialists performinginjections (Pearson correlation coefficient � -.058, P � 0.688). Conclu-sions: There are substantial geographic variations in lumbosacral injec-tion rates within the United States. Southern states tend to have thehighest procedure rates, whereas Northeastern states have the lowest. Theuse of fluoroscopy was extremely variable throughout the country at thestatewide level as well as at the regional level. Injection rates are posi-tively correlated with lumbar surgery rates; states and health referralregions with higher injection rates also have higher lumbar surgery rates.In these high injection rate states, a higher percentage of patients withlow back pain are receiving epidural steroid injections. In addition, ahigher percentage of patients with low back pain are receiving bothepidural steroid injections and lumbar surgery. Thus it does not seemthat ESIs are substituting for lumbar surgeries or impacting overalllumbar surgery rates. There were no correlations between statewidenumber of physicians or specialists that perform injections and injectionrates. Therefore, these regional differences are more likely related to localpractice conventions, training styles or other factors rather than tophysician supply.

Saturday, April 14, 2007

A COMPARISON OF POST–ACUTE CARE DESTINATION FORAMPUTEES BEFORE AND AFTER IMPLEMENTATION OFTHE 75% RULELaura Lee, MD, MBA

While the 75% Rule has been notable for excluding several patientpopulations that have been treated in inpatient rehabilitation facilities(IRFs) in the past, this regulation may also affect the postacute referralpatterns of populations that are included in the 75%. Patients withamputations are included within the 75% quota, but have not usuallybeen referred to IRFs in high proportions. We hypothesized that relativelymore acute care patients would be referred to an IRF after implementa-tion of the 75% Rule.

All patients at an academic medical center who had a trans-tibial,trans-femoral, or hip disarticulation amputation during 2002 and 2005and survived to be discharged from the acute care hospital were included.The results are in table form below. Overall, a higher proportion ofpatients discharged in 2005 were discharged to an IRF than in 2002 (15%in 2002 vs. 27% in 2005). Patients who went to an IRF tended to be olderthan the ones who went home (both routine discharges with outpatientfollow-up and those who went home with home health services) butyounger that the ones who went to a subacute nursing facility. Level ofamputation did not affect discharge disposition, although more trans-tibial amputees went to an IRF and more trans-femoral amputees weredischarged to an SNF in both 2002 and 2005. More traumatic amputeestended to go home than to IRF or SNF. Lastly, comorbidities such asperipheral vascular disease (PVD), congestive heart failure, renal failure

and diabetes were examined to determine relationship to discharge dis-position, with PVD demonstrating a significant correlation (please seetable below).

THE RELATIONSHIP OF MOOD AND LIFE SATISFACTIONTO PHYSICAL FUNCTION AND DEPRESSION IN CHRONICSPINAL CORD INJURY/DISEASEMobeen Choudhri, MD; Chetan Malik, Mbbs; T. M. Srikrishnan,MD; Carl V. Granger, MD

Objective: To study the relationship of Mood and Life Satisfaction toPhysical Function and Depression in chronic spinal cord injury/disease.Study Design: Cross-sectional study of a convenience sample of 28 out-patient clinic attendees at Buffalo VAMC – SCI clinic. Persons with spinalcord injury duration of injury greater than 3 mos were selected for thestudy. Patients were classified based on the ASIA scale into paraplegicsand quadriplegics. Informed consent was obtained. Demographic datawere obtained. Physical function was measured with FIM™, BMC andLimitation, Mood as evaluated by Placid, Life satisfaction was measuredusing Satisfaction measure. BDI Fast Screen was used to assess fordepression. SPSS was used for statistical analysis. t test was used tocompare means and logistic regression was used to calculate Odds ratio.Participants: Twenty-eight chronic spinal cord injury patients (15 para-plegics, 13 quadraplegics), 27 participations were males. Mean age was 53yrs (range 20–85). Interventions: Not applicable. Results: Patients withpresence of depression had worse mood and less life satisfaction asopposed to patients with absence of depression (Placid 90 vs. 55), therelationship persistent, when adjusted other variables including paraple-gia. The adjusted Odds ratio 0.26. Quadriplegics patients also had worsemood and life satisfaction than patients with paraplegia, however therelationship did not persist when adjusted for depression. There was nodifference in the patient’s mood or life satisfaction in relation to FIM(cognitive or motor or total), BMC, and limitation scores. Paraplegicsubjects had higher motor function (BMC), and FIM (motor) scorescompared with quadriplegics. Conclusions: It seems that mood and lifesatisfaction is related the presence of depression and not BMC, FIMmotor, or degree of disability (paraplegics vs. quadriplegics).

LONG-TERM OUTCOMES AFTER TRAUMATIC BRAININJURY IN VETERANS: SUCCESSES AND CHALLENGESJohn H Poole, PhD; Marie N. Dahdah, MS; Karen Schwab, PhD;Henry L. Lew, MD PhD; Deborah L. Warden, MD; Elaine S.Date, MD

Background: To identify the rehabilitation needs of a new generation ofveterans with traumatic brain injury (TBI), research is needed on thecourse of recovery from TBI and challenges encountered by previousveterans. We recently reported on continuing problems during the firsttwo years after TBI. We now present initial findings on functional out-comes and clinical needs of veterans with TBI, five or more years postin-

No. of acute care discharges

2002 2005

n � 87 n � 75

Discharge location Routine Home Health IRF SNF Routine Home Health IRF SNFNo. of cases (%) 40 (46%) 13 (15%) 13 (15%) 21 (24%) 11 (15%) 17 (23%) 20 (27%) 27 (36%)Age, yrs 56 � 16 53 � 16 61 � 10 76 � 11 53 � 13 52 � 14 64 � 13 75 � 11% female 38% 38% 38% 48% 27% 29% 25% 41%No. of transtibial amputations 19 7 9 5 8 7 16 8No. of transfemoral amputations 21 5 4 16 2 10 4 21No. of hip disarticulations 0 1 0 0 1 0 0 0LOS, days 8 � 13 10 � 7 8 � 5 9 � 6 7 � 10 9 � 9 11 � 11 9 � 5% traumatic 10% 23% 0% 0% 27% 18% 10% 0%% PVD 80% 85% 100% 100% 45% 71% 85% 100%% ESRD 38% 23% 62% 24% 36% 12% 35% 15%% CHF 50% 46% 69% 76% 36% 18% 25% 48%%DM 70% 77% 62% 76% 36% 65% 65% 93%

April 2007 2007 AAP Annual Meeting Abstracts 333

jury. Methods: Subjects were 50 patients, currently 22–68 yrs old (medianage 39), who sustained TBI 5–15 yrs ago (median 10 yrs ago). All hadmoderate to severe TBI, and two thirds were in coma for more than 7days. Following emergence from coma, all received multidisciplinaryinpatient rehabilitation in our university-affiliated veterans medical cen-ter for an average of 4 wks. As part of an ongoing study, most subjects hadone to three follow-up evaluations during the first 2 yrs, with 40%receiving additional rehabilitation. For the present study, we conducted acomprehensive telephone interview developed by the Centers for DiseaseControl, supplemented with the Disability Rating Scale (DRS) and theCommunity Integration Questionnaire (CIQ). Results: Forty-five percentof the patients attended college classes post injury, with 25% completinga 2-yr college degree or higher. Forty percent are currently employed,with 27% working at least 30 hrs/wk. Twenty percent became married orengaged. Patient’s current disabilities range from DRS � 0 (absent) to 16(severe), and average 2.7 (partial disability). Two-thirds view themselvesas having a disability and report significant continuing problems in thefollowing areas (in descending order): Cognitive Deficits in 91% of pa-tients (mainly memory, concentration, processing speed, organization,grasp of concepts, and accident proneness); Emotional Problems in 80%

(irritability/anger, disturbed mood, suicidal ideation); Physical Symptoms in71% (balance, gait, hearing, headaches, fatigue); Social Problems in 58%(relationship with spouse, work difficulties); Independent-ADL Problems in42% (driving, money management, meal preparation); and Basic-ADL Prob-lems in 36% (mobility, dressing). Eighty percent of this veteran samplereport that they currently have health coverage. However, the majority citeunmet needs that they attribute to lack of information about availableservices, lack of transportation, and difficulty organizing services for them-selves. Conclusions: (1) A sizable minority of patients with moderate to severeTBI attain higher education, gainful employment, and marital relationships.2Even after rehabilitation and follow-up in the first two years postinjury, themajority continue to report persisting cognitive, emotional, physical andsocial problems during subsequent years.3 Due to the cognitive and emo-tional consequences of brain injury, many patients have difficulty organizingand obtaining services to address these issues.

(2) These findings highlight the importance of continuing follow-upfor patients with TBI throughout their lives. Thus, it is vital that the nextgeneration of TBI programs, now being designed, include treatment plansand a systematic schedule of follow-up appointments that are initiatedand monitored by service providers.

AAP Annual Meeting Abstractsincluding the Paper Presentationsand Poster Abstracts are availableonline as a supplement atwww.AJPMR.com

334 2007 AAP Annual Meeting Abstracts Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 4

Documents

American Journal of Physical Medicine and …lib.ajaums.ac.ir/booklist/47437.pdfAmerican Journal of Physical Medicine and Rehabilitation April 2007, Volume 86, Issue 4,pp.253-334 253