edicine and Rehabilitation

Archives of Physical M journal homepage: www.archives-pmr.org

Archives of Physical Medicine and Rehabilitation 2013;94(4 Suppl 2):S98-105

ORIGINAL ARTICLE

Acute Rehospitalizations During Inpatient Rehabilitationfor Spinal Cord Injury

Flora M. Hammond, MD,a,b Susan D. Horn, PhD,c Randall J. Smout, MS,c David Chen, MD,d

Gerben DeJong, PhD,e William Scelza, MD,f Amitabh Jha, MD,f Pamela H. Ballard, MD,e

Jessica Bloomgarden, MDg

From aCarolinas Rehabilitation, Charlotte, NC; bIndiana University, Indianapolis, IN; cInstitute for Clinical Outcomes Research, Salt Lake City,UT; dRehabilitation Institute of Chicago, Chicago, IL; eNational Rehabilitation Center, Washington, DC; fCraig Hospital, Englewood, CO; andgMount Sinai School of Medicine, New York, NY.

Abstract

Objectives: To investigate frequency of and reasons for readmission to acute care (RTAC) during inpatient rehabilitation after traumatic spinal

cord injury (SCI), and to identify factors associated with RTAC.

Design: Prospective observational cohort.

Setting: Inpatient rehabilitation.

Participants: Individuals with SCI (NZ1376) consecutively admitted for inpatient rehabilitation; 1032 randomly selected for model

development; 344 selected for model cross-validation.

Interventions: Not applicable.

Main Outcome Measures: RTAC, RTAC reasons, rehabilitation length of stay (LOS), discharge location and FIM, rehospitalization between

discharge and year 1, and 1-year outcomes: FIM, Craig Handicap Assessment and Reporting Technique, and Patient Health Questionnaire-9.

Results: Participants (nZ116; 11%) experienced RTAC with a total 143 episodesd96 patients experienced only 1 RTAC, while 14 had 2 RTACs,

5 had 3 RTACs, and 1 had 4 RTACs. The most common RTAC reasons were surgery (36%), infection (22%), noninfectious respiratory (14%), and

gastrointestinal (8%). Mean days � SD from rehabilitation admission to first RTAC was 27�30 days. Seventy-four (7%) patients had at least

1 RTAC for medical reasons and 46 (4%) for surgical reasons. Regression analyses indicated several variables were associated with RTACs:

greater admission medical severity, lower admission cognitive FIM, pressure ulcer acquired in acute care, and study site. Medical RTACs were

associated with higher body mass index, lower admission cognitive and motor FIM, payer, and study site. Predictors of surgical RTAC were longer

time from injury to rehabilitation admission and study site. After controlling for the other variables, the only outcome RTAC influenced was

longer rehabilitation LOS.

Conclusions: Approximately 11% of SCI patients experience RTAC during the course of rehabilitation for a variety of medical and surgical

reasons. RTACs are associated with longer rehabilitation LOS.

Archives of Physical Medicine and Rehabilitation 2013;94(4 Suppl 2):S98-105

ª 2013 by the American Congress of Rehabilitation Medicine

Data from the Spinal Cord Injury (SCI) Model Systems programdatabase reveals 3 interesting trends over time. First, the averageage of SCI has increased from 28.7 years in the 1973 to 1979

Supported in part by National Institute on Disability and Rehabilitation Research, Office of

Special Education Services, U.S. Department of Education (grant nos. H133A060103,

H133N060005, H133N060009, H133N060027, and H133N060014).

No commercial party having a direct financial interest in the results of the research supporting

this article has or will confer a benefit on the authors or on any organization with which the authors

are associated.

0003-9993/13/$36 - see front matter ª 2013 by the American Congress of Re

http://dx.doi.org/10.1016/j.apmr.2012.11.051

period to 40.7 years since 2005. Second, time from initial injury torehabilitation admission has decreased from an average of 24 daysin the 1973 to 1979 period to only 12 days since 2005.1 And third,the overall average length of stay (LOS) in acute rehabilitation hasdecreased from 74.1 days in 1990 to 60.8 days in 1997.2

The first of these trends stems from an aging general pop-ulation with conceivably more underlying chronic medicalconditions. The second and third trends stem from changingpolicy and practice patterns. Escalating health care costs require

habilitation Medicine

Rehospitalizations during inpatient rehabilitation for SCI S99

those with traumatic SCI to move through acute care hospitali-zation more quickly, and thus, present greater acuity duringinpatient rehabilitation. Additionally, acute inpatient rehabilitationLOS has also shortened over recent years. The shortening of bothacute and rehabilitation LOS places greater pressure on acuteinpatient rehabilitation providers to manage all aspects of medicalcare and rehabilitation in a compressed time frame.

Less clear is the frequency of readmission to acute care(RTAC) among individuals with traumatic SCI in the course ofinpatient rehabilitation, reasons for such RTACs, patient charac-teristics and care features associated with such readmissions, andassociations between RTACs and the patient’s outcomes. Under-standing these factors may help identify those at greatest risk anduncover interventions and practices that may help avert RTACs.

Studies have reported on the major causes of rehospitalizationsamong those with chronic SCI, but they have focused primarily onthose who have already completed their inpatient SCI rehabilita-tion program. The most common reasons for rehospitalizationshave consistently included disorders of the genitourinary tract (ie,urinary tract infections), respiratory complications, gastrointes-tinal (GI) disorders, and disorders of the skin and integumentarysystems (primarily complications related to pressure ulcers).3-5

The studies of specific factors contributing to RTACs duringSCI rehabilitation care have not been reported in the literature.

In 1999, Chen et al6 reported on the most common medicalcomplications experienced by persons with new SCI during acuterehabilitation who were treated within the SCI Model Systemsprogram. However, the study was unable to differentiate whetherthe medical condition developed during the acute care or acuterehabilitation phase of care after injury. Moreover, the data relatedto acute care complications had a limited look-back period,because the main data source, the SCI Model Systems database,started to collect such data only in the previous 2 years. Chen6 didnot explore patient characteristics or health-related factors thatmay have increased the risks for developing these complications,nor did they delve into the setting of care in which these medicalconditions were treated (acute care vs rehabilitation) or the impactthat they had on the rehabilitation course and outcome.

The full impact of RTAC episodes remains unknown. Anec-dotally, we know that an RTAC can disrupt rehabilitation progress,create emotional stress on the patient and family, increase healthcare costs, and undermine patient outcomes. Concurrently, payersand quality monitoring organizations look to readmissions withinthe first 30 days after acute care discharge as a quality-of-careindicator denoting substandard care. In a recent literature reviewof acute care hospital admissions (for any diagnosis, not just SCI),it was estimated that 9% to 48% of acute care rehospitalizationsare related to substandard care during acute care hospitalization.7

Of note, this study examined readmissions primarily after acute

List of abbreviations:

AIS ASIA Impairment Scale

ASIA American Spinal Injury Association

BMI body mass index

CHART Craig Handicap Assessment and Reporting Technique

CSI Comprehensive Severity Index

GI gastrointestinal

HL Hosmer-Lemeshow

LOS length of stay

RTAC readmission to acute care

SCI spinal cord injury

www.archives-pmr.org

care to community discharge as opposed to acute care to inpatientrehabilitation specifically. Adequate preparation for acute caredischarge to inpatient rehabilitation may differ from the needs foracute care discharge to the community. The medical preparationand needs of the SCI population in transition from acute care toacute inpatient rehabilitation and during inpatient rehabilitationremain unclear. Many RTACs may be planned events, forexample, readmissions for a planned surgery and not related tomedical complications that develop or worsen during the acutephase of rehabilitation.

The aim of this article is to describe the medical complicationsthat precipitate an RTAC during inpatient rehabilitation for SCI byassessing the incidence of and reasons for RTACs and the factorsassociated with RTACs. A secondary aim is to evaluate the rela-tion between RTACs and outcome at rehabilitation discharge and1 year postinjury.

Methods

The SCIRehab Project is a multicenter, 5-year investigationrecording and analyzing the details of the SCI inpatient rehabili-tation process for 1032 patients. The introductory article in thissupplement of SCIRehab project articles describes the studydesign, including use of practice-based evidence researchmethods, inclusion criteria, data sources, and the analysis plan.8,9

Briefly, the SCIRehab sites included 6 inpatient rehabilitationfacilities: Craig Hospital, Englewood, CO; Shepherd Center,Atlanta, GA; Rehabilitation Institute of Chicago, Chicago, IL;Carolinas Rehabilitation, Charlotte, NC; Mount Sinai MedicalCenter, New York, NY; and National Rehabilitation Hospital,Washington, DC. Institutional review board approval was obtainedat each center. Participants were �12 years of age, gave (or theirparent/guardian gave) informed consent, and were admitted to thefacility’s SCI unit for initial rehabilitation after traumatic SCI.

Patient, injury, and outcome variables

Trained data abstractors collected patient and injury data frompatient medical records. The International Standards of Neuro-logical Classification of SCI10 were used to describe the motor leveland completeness of injury. Patients with American Spinal InjuryAssociation (ASIA) Impairment Scale (AIS) grade D were groupedtogether regardless of injury level. Patients with AIS grades of A, B,and C were grouped together and separated by motor level todetermine the remaining 3 injury groups: high tetraplegia (C1-4),low tetraplegia (C5-8), and paraplegia (T1 and below). These injurycategories were selected because they were each large enough foranalysis and created groupings thought to have relatively homog-enous functional ability within groups and clear differencesbetween groups. Height and weight were collected to calculatebody mass index (BMI).

The Comprehensive Severity Index (CSI), the study’s principalseverity measure, was used to score the extent of deviation fromnormal physiological status for each medical complication andcomorbidity present during the first 3 days after rehabilitationadmission.11 Higher CSI scores denote increased medical severity.We also assessed several medical conditions that occurred duringacute care prior to the initial rehabilitation admission and theirpossible association with RTAC: deep venous thrombosis, leukocy-tosis, pneumonia, pressure ulcers, sepsis, and urinary tract infection.

The FIM was used to describe a patient’s independence inmotor and cognitive abilities at rehabilitation admission and

S100 F.M. Hammond et al

discharge and 1 year postinjury. All FIM data were Rasch-adjusted, as described in the first article in this supplement.8

Participation in mobility, occupation, and social interaction at1 year postinjury was assessed by various subscales of the CraigHandicap Assessment and Reporting Technique (CHART).12 Thepresence of depression at 1 year was measured by the PatientHealth Questionnaire-9,13 although it is important to note thatpositive responses to the Patient Health Questionnaire-9 itemsmay be present because of SCI or other reasons and may notalways represent depression per se. The LOS in rehabilitationexcluded days during which the patient was returned to acute care.

Definition of RTAC

All interruptions of inpatient rehabilitation requiring readmissionto an acute care hospital were considered to represent an RTAC. Foreach such event, the dates of interruption and reason(s) wereabstracted from the medical chart. In many instances, multiplereasons were listed along with the presenting signs, symptoms, anddiagnoses recorded by the clinical team. In such cases, we reviewedthe data and selected the 1 primary reason that best represented thecause for the RTAC episode. Hence, no RTAC episode is countedmore than once, and only 1 reason is represented even whenmultiple causes may have prompted the RTAC. Because the unit ofanalysis is the patient, a patient with 1 or multiple RTACs is countedonly once in the RTAC group. The RTAC reasons were grouped intoeither surgery or 1 of the following medical categories: infection(eg, pneumonia, aspiration pneumonia, urinary tract infection,sepsis, skin infection), respiratory (eg, pneumothorax, respiratoryfailure, pleural effusion, tracheal stenosis, anaphylaxis), GI, venousthromboembolism, cardiac (eg, congestive heart failure, chest pain,arrhythmia), mental status change, neurologic, anemia, orthostatichypotension, pain, and unknown.

Data analysis

The sample (NZ1376) was divided by random selection intoa primary analysis dataset with 75% of the cases (nZ1032) anda validation set with the other 25% (nZ344). Stratification wasused to ensure equal representation by level/completeness of injury,treatment center, and availability of follow-up interview data.Statistical tests showed no significant differences between thedevelopment and validation subgroups on any dependent or inde-pendent variable used in the regression models. All analyses wereinitially performed using the development dataset. For ordinaryleast-squares regressions, the adjusted R2 reduces the unadjusted R2

to take into account the number of predictors in the model. Theadjusted R2 value indicates the strength of the model or the amountof variation explained in the outcome by the independent variables,and values range from 0 (no prediction) to 1 (perfect prediction);values that are closer to 1 indicate better models. The models werevalidated by calculating the R2 for the 25% dataset using only thestatistically significant predictors from the original model anddetermining the reduction in R2 (relative shrinkage) that occurred.A relative shrinkage of <0.1 is interpreted as strong validation ofthe original model, 0.1 to 0.2 is considered a moderate validation,and >0.2 is considered a poor validation.

For logistic regression, discrimination was assessed by usingthe area under the receiver operator characteristic curve (c) toevaluate how well the model distinguished patients who did notachieve a specified outcome from patients who did achieve the

specified outcome. Values of c range from 0.5 to 1; values that arecloser to 1 indicate better discrimination. The maximum rescaledR2, also known as the Nagelkerke pseudo R2, was also reported asa measure of the strength of the model and is scaled the same asthe R2 value (0e1). The models were validated by comparing theHosmer-Lemeshow (HL) goodness of fit test P value for theoriginal model with the same parameter for the validation model.Models validated well if the HL P was >.10 for both, whichindicates no lack of fit in either model.

RTAC during inpatient rehabilitation was assessed as toprobable cause, associated factors, and relation with functionaloutcome at the time of rehabilitation discharge and year 1 follow-up. Descriptive analyses examined the occurrence of RTACs bydemographic and premorbid characteristics, injury and medicalvariables, and outcome at rehabilitation discharge and at 1 yearpostinjury. Demographic and premorbid characteristics included:age at the time of injury, sex, highest level of education, maritalstatus at injury, employment status at time of injury, and BMI (atthe time of rehabilitation admission). Injury and medical variableswere: categories of individuals with SCI (paraplegia/tetraplegia,complete/incomplete, ASIA motor score), cause of injury, payer,CSI at the time of rehabilitation admission, Rasch-transformedadmission cognitive and motor FIM subscores, days from traumauntil rehabilitation admission, deep venous thrombosis duringacute care, and presence of pressure ulcer during acute care.Outcome at rehabilitation discharge was measured by FIM motorand cognitive subscores, residence type, and rehabilitation LOS.Longer-term outcomes measured at the 1-year injury anniversarywere major depression, cognitive and motor FIM, CHARTdimensions of physical independence, mobility, occupation, andsocial interaction, participation in work or school, and rehospi-talization occurrence during year 1 after rehabilitation discharge.Bivariate analyses were conducted to determine association ofRTAC with these variables.

Prediction of RTACLogistic regression analyses were used to predict the occurrenceof RTAC using only the variables available at the time ofadmission (ie, demographic, premorbid, injury, severity, func-tional, acute care, and rehabilitation admission data).

RTAC as predictor of outcomesOrdinary least-squares regression models were used to determinethe influence of RTACs during acute inpatient rehabilitation onrehabilitation LOS, Rasch-transformed discharge FIM scores, and1-year postinjury outcomes.

Results

The demographic and injury characteristics are summarized for thesample as a whole and for each injury group in the first article in thissupplement.8 The sample was 81% men, 71% white, 38% married,83% not obese (83% had a BMI of <30), and 66% employed at thetime of injury. Average age of the sample was 38 years. The mostcommon cause of injury was vehicular collisions (49%), followedby falls or flying objects (25%), and sports and violent etiologies(11% each). Mean rehabilitation LOS � SD was 56�37 days(range, 2e267; median, 45). The mean Rasch-transformed motorFIM subscore at admission � SD was 18�13, and the meanRasch-transformed cognitive score � SD was 74�18. A mean timefrom injury to rehabilitation admission � SD was 31�28 days.

www.archives-pmr.org

Rehospitalizations during inpatient rehabilitation for SCI S101

Table 1 describes the characteristics of participants who didand did not have an RTAC during rehabilitation. Among the 1032study participants, 116 (11%) had �1 RTAC episodes for a total of143 episodes. Ninety-six patients experienced only 1 RTAC, while14 patients had 2 RTACs, 5 had 3 RTACs, and 1 had 4 RTACs. The

Table 1 Patient and injury characteristics by RTAC occurrence

Characteristic

Age at injury (y)

Sex, % men

Marital status at injury, % married

Race/ethnicity

White

Black

Hispanic

Other

Employment status before injury

Working

Student

Retired

Unemployed/other

BMI at admission

<30

30e40

>40

Payer

Medicare

Medicaid

Private insurance/payer

Worker’s compensation

Highest education achieved

<12y

High school

College

Other/unknown

Cause of injury

Vehicular

Fall

Sports

Violence

Medical/surgical/other

ASIA group

C1-4 AIS grades AeC

C5-8 AIS grades AeC

Paraplegia AIS grades AeC

AIS all grade Ds

Admission CSI score

FIM motor at rehabilitation admissiondRasch-transformed

FIM cognitive at rehabilitation admissiondRasch-transformed

Days from trauma to rehabilitation admission

Rehabilitation LOS in days with interruptions removed

Complications during acute care, % yes

Deep venous thrombosis

Leukocytosis

Pneumonia

Pressure ulcers

Sepsis

Urinary tract infection

NOTE. Values are mean � SD or % or as otherwise indicated.

* Statistically significant difference (P<.05) in bivariate testing by t test

www.archives-pmr.org

primary reasons for RTAC are summarized in table 2; the mostcommon reasons are surgery (36%), followed by infection (22%),respiratory (14%), and GI (8%). Mean days from rehabilitationadmission to first RTAC � SD was 27�30 days. Six participantshad an episode classified as medical and a different RTAC

RTACs (nZ116) No RTAC (nZ916) P*

42.7�17.6 37.0�16.5 <.001*

84 81 .530

46 37 .042*

.240

73 70

16 22

5 3

5 5

.045*

68 66

8 16

12 7

12 11

.115

75 82

22 15

3 3

.061

12 7

15 19

59 64

15 10

.444

17 20

53 51

28 25

2 4

.676

53 49

28 25

9 11

9 12

3 4

.023*

40 27

19 20

32 4

10 16

26.5�20.2 19.3�17.5 <.001*

14.3�12.9 18.2�12.5 .002*

69.0�19.3 74.2�17.9 .004*

35.8�43.0 30.4�25.2 .182

74.0�43.8 53.3�34.9 <.001*

22 14 .025*

6 8 .706

32 36 .411

28 20 .068

3 7 .225

22 27 .264

or chi-square test.

Table 2 Primary reasons for RTAC occurrences

RTAC Reason Frequency (%)

Surgery/procedure 51 (36)

Infection 31 (22)

Noninfectious respiratory 20 (14)

GI 12 (8)

Venous thromboembolism 7 (5)

Cardiac 5 (4)

Mental status change 5 (4)

Neurologic 5 (4)

Anemia 2 (1)

Orthostatic hypotension 1 (<1)

Pain 1 (<1)

Unknown 3 (2)

Total 143*

* Some patients had >1 RTAC. Primary reason for all RTACs is listed.

S102 F.M. Hammond et al

classified as surgical, and thus, were included in both the surgicaland medical groups for analysis.

Medical RTAC

Among the sample of 1032, 74 (7%) patients had at least 1 RTACfor medical reasons. Mean days from rehabilitation admission tofirst RTAC � SD was 25�34 days in the medical group. Infection,respiratory, and GI are broad, catch all categories that requiremore specific explanation of the causes within these categories.The causes of the 31 infection RTACs were: pneumonia (nZ7;23%), abscess (nZ6; 19%), sepsis (nZ6; 19%), urosepsis (nZ5;16%), wound (nZ4; 13%), aspiration pneumonia (nZ1; 3%),isolation (nZ1; 3%), and clostridium difficile (nZ1; 3%). The 20respiratory causes included: respiratory distress/failure (nZ12;60%), pneumothorax (nZ3; 15%), ventilator wean (nZ2; 10%),tracheal stenosis (nZ1; 5%), pleural effusion (nZ1; 5%), andanaphylactic reaction (nZ1; 5%). Of the 12 GI RTACs, 6 (50%)were because of GI bleeding, in addition to the 2 RTACs becauseof anemia. Four (33%) of the GI RTACs occurred because ofobstruction, 1 (8%) because of gastroparesis, and 1 (8%) becauseof other. Neurologic reasons included 3 seizures and 2 instances ofspinal cord compressions (although not recorded as undergoingsurgical decompression).

Surgical RTAC

At least 1 RTAC because of a surgical reason occurred in46 patients (4% of the total study sample) with a total of 51surgical RTACs. Mean days from rehabilitation admission to firstRTAC for surgical causes � SD was 29�21 days. The mostcommon reason was spine stabilization, which accounted for22 (43%) of the surgical RTAC occurrences. Of the surgicalRTACs, 14 (27%) were skin reasons, specifically, myocutaneousflaps (nZ10; 20%), nonspine wound debridement (nZ2; 4%),and spine wound washout (nZ2; 4%). Nonspine fracture repairprompted 6 RTACs (12%). Most surgical RTACs were recordedas unscheduled, while 31% were noted as scheduled (14 partic-ipants had 16 RTAC occurrences for preplanned surgeries).Preplanned surgeries were comprised of spine stabilization(nZ9; 56%), fracture repair (nZ4; 25%), tracheostomy (nZ1;6%), and colostomy (nZ1; 6%).

Prediction of RTAC

Table 3 presents logistic regression models for the prediction ofRTAC overall and RTAC based on medical and surgical reasonsseparately. RTAC overall was predicted by higher admission CSI,lower admission cognitive FIM, pressure ulcer occurrence duringacute care, and study site. A c statistic of .72 indicates the strengthof the prediction is quite good. This model also validated well withan HL PZ0.7. The HL P value of the model using the validationdataset was 0.3. RTAC for medical reasons was predicted by higherBMI, lower admission cognitive and motor FIM subscores,worker’s compensation as payer source, and study site. A c statisticof .769 indicates the strength of the prediction is quite good. Thismodel also validated well with an HL PZ0.2. The HL P value ofthe model using the validation dataset was .102. Predictors ofsurgical RTAC included longer time from injury to rehabilitationadmission and study site. ASIA groups were not predictive ofRTAC. A c statistic of .706 indicates the strength of the prediction isquite good. This model also validated well with an HL PZ0.6. TheHL P value of the model using the validation dataset was 0.5.

RTAC as predictor of outcomes

Using bivariate analyses, we found statistically significant associ-ations between RTAC occurrence and rehabilitation outcomes atdischarge and at the 1-year anniversary for the following outcomes:lower discharge motor FIM, discharge to a nonhome residence,postdischarge rehospitalization in the 1-year follow-up period, lackof participation in work or school at 1 year, and rehabilitation LOS.However, when using multivariate regression modeling andallowing other patient variables (as listed in table 1) in addition toRTAC, the only outcome in which RTAC occurrence was signifi-cant was rehabilitation LOS (table 4). This model’s predictionstrength was also quite good with an adjusted R2Z.594 and can beconsidered well validated with relative shrinkage of .035.

Discussion

In our study, 11% of study participants undergoing acute rehabili-tation were transferred to acute care either for a medical compli-cation or elective/unexpected surgerydultimately resulting in aninterruption of their course of rehabilitation. Surgery was the mostfrequently reported primary reason for RTAC. RTACs for medicalreasons mostly occurred because of infections, respiratory, and GIissues, which are consistent with medical conditions commonlyobserved in this patient population. The study findings point to skincomplications as a major reason for RTACs with 4 wound infec-tions, 10 myocutaneous flaps, 2 nonspine wound debridement, and2 spine incision washouts. This highlights the potential relevance ofearly efforts toward skin protection, monitoring, infectionprophylaxis, and attention to surgical attendance as a means ofpossibly preventing RTAC. The findings only partially represent theprevalence of skin-related issues during SCI rehabilitation, becausethe study only looks at those skin issues that prompted RTAC.

RTAC occurrence, without regard to reason, was predicted byhigher CSI at rehabilitation admission, lower FIM cognitive sub-score, pressure ulcer during acute care, and study site. However,the CSI was not predictive of RTAC for medical reasons. The CSIis an indicator of medical acuity/complexity, which is intuitivelyassociated with risk of complications, which may lead to RTAC.From a prevention and treatment perspective, medical complexity

www.archives-pmr.org

Table 3 Prediction of RTAC occurrence

Outcome All RTACs Medical RTACs Surgical RTACs

No. of observations used* 1031 (yesZ115, noZ916) 1031 (yesZ74, noZ957) 1031 (yesZ45, noZ986)

c statistic .720 .769 .706

Maximum rescaled R2 .122 .175 .086

Parameter

Estimate

Odds Ratio

Estimate P

Parameter

Estimate

Odds Ratio

Estimate P

Parameter

Estimate

Odds Ratio

Estimate P

Admission CSI score .01 1.01 .029y 0.02 1.02 .051

Admission FIM motor

scoredRasch-transformed

�0.02 0.98 .031y

Admission FIM cognitive

scoredRasch-transformed

�.02 0.98 .007y �0.02 0.98 .018y

Pressure ulcer formed during

acute hospital stay

.54 1.72 .032y

BMI �30 0.59 1.80 .041y

Age at injury 0.02 1.02 .055

No. of days from trauma to

rehabilitation admission

0.01 1.01 .016y

Location <.001y <.001y .010

Site 1 �.97 0.71 <.001y �0.90 1.39 .002y �0.55 0.45 .190

Site 2 �.68 0.94 .065 �0.59 1.89 .140 �0.51 0.47 .425

Site 3 .78 4.08 .002y 0.66 6.64 .022y 1.31 2.86 <.001y

Site 4 .73 3.88 .021y 1.07 9.99 .002y �0.28 0.58 .657

Site 5 .76 3.99 .001y 0.99 9.21 <.001y �0.23 0.62 .642

Site 6 (reference) .00 NA NA 0.00 NA NA 0.00 NA NA

Primary payerdworker’s

compensation

0.94 2.56 .009y

Abbreviation: NA, not applicable.

* A total of 1031 observations was used instead of 1032 because of the lack of CSI data for 1 study participant.y Statistically significant predictor (P<.05); variables not listed did not have statistically significant association with RTACs (overall, surgical, or

medical).

Rehospitalizations during inpatient rehabilitation for SCI S103

can be tracked early in the acute hospital and rehabilitation courseand may serve as a marker for greater risk of subsequentcomplications requiring RTAC. Similarly, a pressure ulcer duringacute care and a lower FIM cognitive subscore can serve aspotential markers that can help foster both vigilance and strategiesto prevent medical complications. The association with the FIMcognitive subscore may reflect the presence of a coexisting trau-matic brain injury. It is not clear why FIM cognitive subscorespredict RTAC. It is expected that the individuals in this study withconcurrent traumatic brain injury who were cared for in an SCIrehabilitation unit had mild or moderate traumatic brain injury.Interestingly, the CSI at rehabilitation admission was significantlyassociated with RTAC occurrence during rehabilitation, while thediagnosis of specific medical problems during acute care (deepvenous thrombosis, leukocytosis, pneumonia, pressure ulcers,sepsis, and urinary tract infection) was not significantly associatedwith RTAC. This may be because these issues represent theoccurrence and not the severity of the medical problem. Thisfinding may also represent the impact of the early diagnosis andtreatment initiation before rehabilitation admission.

Medical RTAC

The medical reasons for RTAC in our study mirror those found instudies of rehospitalization after completion of SCI rehabilita-tion,3-5,14 suggesting robustness of these findings as well as the

www.archives-pmr.org

difficulty in preventing these complications. Infection, respiratory,and GI complications are common causes of rehospitalization,both during inpatient rehabilitation and again after completion ofrehabilitation. Infection was the most common cause of rehospi-talization, as observed in these other studies.

Pulmonary compromise was a common RTAC trigger with 30rehospitalizations being due to some type of pulmonary cause:7 pneumonia, 1 aspiration pneumonia, 20 respiratory causes, and2 surgical RTACs for tracheotomy/tracheoplasty. Pulmonaryproblems have been previously found to be a major cause ofhospitalization, as well as mortality.15 Who is at greatest risk forthese pulmonary complications and the role of vigilance,screening, and prophylactic measures in preventing RTAC was notstudied here.

GI complications have been found to be a frequent cause ofrehospitalization after rehabilitation discharge in some studies,16

but not others.3 After SCI, the GI system can exhibit relativelycommon conditions, such as gastroparesis, paralytic ileus andconstipation, and serious conditions, such as perforated ulcer,17

obstruction, appendicitis, gastritis, colitis, cholecystitis, pancrea-titis, and the need for a colostomy. In the present study, the mostcommon GI cause of RTAC was GI bleeding. The use andeffectiveness of prophylaxis with H2 antagonists and proton pumpinhibitors, which often are used to prevent these complications,were not studied here. Clinicians should remain vigilant for thesepotential GI complications.

Table 4 Prediction of rehabilitation LOS

Outcome Rehabilitation LOS

No. of observations used 1025*

R2 0.601

Adjusted R2 0.594

Parameter

Estimate P

Primary payer <.001y

Medicare 0.64 .829

Medicaid 3.97 .045y

Worker’s compensation 11.93 <.001y

Private insurance/payer (reference) 0.00 NA

Highest education achieved .010y

Other/unknown �6.92 .110

High school 2.44 .200

College 5.46 .014y

<12y (reference) 0.00 NA

ASIA group <.001y

C1-4 AIS grades AeC 15.73 <.001y

C5-8 AIS grades AeC 22.66 <.001y

Paraplegia AIS grades AeC 12.24 <.001y

All AIS grade Ds (reference) 0.00 NA

Admission CSI score 0.28 <.001y

FIM motor scoredRasch-transformed �0.65 <.001y

FIM cognitive scoredRasch-transformed �0.10 .032y

Ventilator use at rehabilitation admission 13.21 <.001y

RTAC during rehabilitation 8.29 .001y

Rehabilitation location <.001y

Site 1 36.35 <.001y

Site 2 1.78 .521

Site 3 18.35 <.001y

Site 4 3.85 .274

Site 5 24.42 <.001y

Site 6 (reference) 0.00 NA

Abbreviation: NA, not applicable.

* A total of 1025 observations was used instead of 1032 because of

a lack of CSI data for 1 study participant and lack of ventilator use at

rehabilitation admission on 6 patients.y Statistically significant predictor (P<.05).

S104 F.M. Hammond et al

RTAC for medical reasons was predicted by higher BMI,lower FIM motor and cognitive subscores at rehabilitationadmission, worker’s compensation as main payer source, andstudy site. Patient factors, such as BMI and functional status, maypoint to the potential benefit of targeted programs for moreintensive monitoring and surveillance for complications andpotentially different or more intense prophylaxis. System factors,such as study site and insurance coverage, likely reflect differ-ences in the capabilities of rehabilitation facilities to managemedical problems and the effect of case management programs tomove injured patients efficiently through their acute care and intorehabilitation.

Surgical RTAC

Four percent of study participants had at least 1 RTAC for surgicalreasons, making surgery the most common single cause for RTAC.The majority of procedures involved the spine or skin wounds.

Most of the surgical RTACs were unscheduled, although it is notclear that the categorization of scheduled versus unscheduled ismeaningful. It is unclear if the surgical needs were a continuationof a problem that first arose during acute hospitalization oremerged for the first time during inpatient rehabilitation. It also isunclear if the surgical procedures could have been anticipatedprior to rehabilitation admission. It is presumed that the surgerywas important enough to interrupt the rehabilitation course.

Longer time from injury to rehabilitation admission and studysite predicted surgical RTAC. This may represent a greatermedical complexity leading to longer acute care LOS. There isoften a difficult balancing act that occurs during the acute reha-bilitation admission screening process: should rehabilitationadmission be delayed to allow for improved stability or comple-tion of planned procedures, or would such a delay allow for theaccumulation of more complications that need to be identified andresolved once the patient has been admitted to rehabilitation?Perhaps earlier or more aggressive use of rehabilitation services inthe acute care setting or earlier surgical intervention wouldattenuate the potential deleterious effect of RTAC on acute reha-bilitation outcomes (rehabilitation LOS). In particular, extendedacute care hospitalization may increase the risk for pressureulcers, which may later require surgical RTAC.

RTAC as predictor of outcomes

This is the first study, to our knowledge, to look at RTAC as itrelates to outcomes. RTAC occurrence predicted longer rehabili-tation LOS, even controlling for other variables, such as CSI andpatient factors. This likely represents the impact of medicalconditions on achievement of rehabilitation milestones andmedical stability for discharge readiness.

Effect of site

The 6 study sites were selected to participate based on theirwillingness, geographic diversity, and expertise in treatment ofpatients with SCI. They offer variation in setting and care deliverypatterns, specifically integration with or immediate proximity toacute care hospital and patient clinical and demographic charac-teristics. Rehabilitation centers vary in their breadth of medicalexpertise and ability to evaluate and manage medical complexityand instability. This may also be reflected in the center effectsobserved here. It is expected that rehabilitation centers may varyin their threshold to treat medical complications and the need totransfer, as exemplified by the study findings. The decision to keepa sick patient in the rehabilitation setting versus transfer to acutecare is complex: partly expertise and capacity, partly allocation ofresources, and partly payer driven. All of these considerations mayhave center or regional variations.

Future clinical and research directions

Research is needed to understand if RTACs are preventable, whichones are preventable, and how. The study findings may informfuture research to identify potential surveillance and preventionstrategies aimed at reducing RTAC occurrence and shorteningLOS. Such strategies could then be subjected to a rigorous qualityimprovement or research protocol to discern which efforts aremost effective and efficient in reducing RTACs. These strategiesmight include improved preadmission screening criteria, earlier

www.archives-pmr.org

Rehospitalizations during inpatient rehabilitation for SCI S105

medical management and rehabilitation, better admission timing,and better timing of planned surgeries.

Study limitations

The SCIRehab sites are highly specialized centers for SCI reha-bilitation, and thus, findings may not be generalizable to allrehabilitation facilities that provide care for patients with SCI.Often, there are several medical issues contributing to the need toRTAC. For this study, the authors reviewed the multiple reasonsand selected the main reason prompting transfer. We often found itdifficult to isolate the exact reason, and thus, in many cases, weused the main sign or symptom of instability as the main reasonwhen other factors may have precipitated the RTAC (eg, acutemental status change). Thus, the reasons for RTAC may notrepresent all medical problems that contributed to the need totransfer to another level of care. The reason for RTAC was basedon initial clinician documentation in the medical record, and thus,is subject to clinical opinion. It is often difficult to determine theexact reason for medical instability, and often there are multipleissues that are cooccurring. In order to avoid counting an episodemore than once, clinical judgment of the researchers had to beapplied to the selection of the primary reason. During datacollection, the surgical RTACs were recorded as scheduled versusunscheduled, but it is not clear if this categorization was suffi-ciently defined to be meaningful in the present study. Scheduledcould mean scheduled in advance of the rehabilitation stay orcould mean scheduled the day before the surgical RTAC. Deter-mining whether or not a surgery was scheduled through chartreview is subject to inaccuracies. The acute care hospital medicalrecords, either for the stay prior to the first rehabilitation admis-sion or for the RTACs, were not available (other than summariesincluded in the rehabilitation record), which would have contrib-uted to a more complete assessment of the reasons for RTACs andtheir avoidability.

Conclusions

Approximately 11% of patients with traumatic SCI experience anRTAC during the course of their initial rehabilitation stay, mainlyfor reasons associated with the overall severity of their medicalcondition and functional status. Site of care may interact withreason to cause RTAC. RTACs occur during inpatient rehabilita-tion for a variety of medical complications and need for surgicalintervention. RTACs are associated with longer LOSs inrehabilitation.

Keywords

Comorbidity; Patient readmission; Rehabilitation; Spinal cordinjuries

www.archives-pmr.org

Corresponding author

Flora M. Hammond, MD, 4141 Shore Dr, Indianapolis, IN 46254.E-mail address: flora.hammond@rhin.com.

References

1. Facts and figures at a glance. Birmingham: National Spinal Cord

Injury Statistical Center; 2011.

2. Eastwood EA, Hagglund KJ, Ragnarsson KT, Gordon WA, Marino RJ.

Medical rehabilitation length of stay and outcomes for persons with

traumatic spinal cord injurye1990-1997. Arch Phys Med Rehabil

1999;80:1457-63.

3. Cardenas DD, Hoffman JM, Kirshblum S, McKinley W. Etiology and

incidence of rehospitalization after traumatic spinal cord injury:

a multicenter analysis. Arch Phys Med Rehabil 2004;85:1757-63.

4. McKinley WO, Jackson AB, Cardenas DD, DeVivo MJ. Long-term

medical complications after traumatic spinal cord injury: a regional

model systems analysis. Arch Phys Med Rehabil 1999;80:1402-10.

5. Davidoff G, Schultz JS, Lieb T, et al. Rehospitalization after initial

rehabilitation for acute spinal cord injury: incidence and risk factors.

Arch Phys Med Rehabil 1990;71:121-4.

6. Chen D, Apple DF, Hudson LM, Bode RK. Medical complications

during acute rehabilitation following spinal cord injuryecurrent expe-

rience of themodel systems.ArchPhysMedRehabil 1999;80:1397-401.

7. Benbassat J, Taragin M. Hospital readmissions as a measure of quality

of health care: advantages and limitations. Arch Intern Med 2000;160:

1074-81.

8. Whiteneck GG, Gassaway J. SCIRehab uses practice-based evidence

methodology to associate patient and treatment characteristics with

outcomes. Arch Phys Med Rehabil 2013;94(4 Suppl 2):S67-74.

9. Whiteneck G, Gassaway J, Dijkers MP, Heinemann AW, Kreider SE.

Relationship of patient characteristics and rehabilitation services to

outcomes following spinal cord injury: the SCIRehab project. J Spinal

Cord Med 2012;35:484-502.

10. Clinical assessment after acute cervical spinal cord injury. Neurosur-

gery 2002;50(3 Suppl):S21-9.

11. Ryser DK, Egger MJ, Horn SD, Handrahan D, Gandhi P, Bigler ED.

Measuring medical complexity during inpatient rehabilitation

following traumatic brain injury. Arch Phys Med Rehabil 2005;86:

1108-17.

12. Krontke K, Spitzer RL. The PHQ-9: a new depression diagnostic and

severity measure. Psychiatric Ann 2002;32:1-7.

13. Whiteneck GG, Charlifue SW, Gerhart KA, Overhosler JD,

Richardson GN. Quantifying handicap: a new measure of long-term

rehabilitation outcomes. Arch Phys Med Rehabil 1992;73:519-26.

14. DeJong G, Wenqiang T, Hsieh C-H, et al. Rehospitalization in the first

year of traumatic spinal cord injury after discharge from medical

rehabilitation. Arch Phys Med Rehabil 2013;94(4 Suppl 2):S87-97.

15. DeVivo MJ, Krause JS, Lammertse DP. Recent trends in mortality and

causes of death among persons with spinal cord injury. Arch Phys Med

Rehabil 1999;80:1411-9.

16. Jaglal SB, Munce SE, Guilcher SJ, et al. Health system factors asso-

ciated with rehospitalizations after traumatic spinal cord injury:

a population based study. Spinal Cord 2009;47:604-9.

17. Mowrey K. The challenge of assessing and diagnosing acute abdomen

in tetraplegics: a case study. J Neurosci Nurs 2007;39:5-8.