ORIGINAL STUDIES

Anemia Increases Risk for Falls inHospitalized Older Adults: AnEvaluation of Falls in 362 Hospitalized,Ambulatory, Long-Term Care, andCommunity Patients

T.S. Dharmarajan, MD, FACP, AGSF, Sai Avula, MD, and Edward P. Norkus, PhD, FACN

Objective: The objective of this study was to determineif a relationship exists between the presence of ane-mia and the occurrence of falls during hospitalizationin ambulatory older adults from long-term care andcommunity settings. All individuals were hospitalizedfor acute conditions not related to a fall.

Participants: Three hundred sixty-two hospitalized,ambulatory older (59–104 years) adults.

Measurements: Laboratory values (hemoglobin [Hb],hematocrit [Hct]), routine laboratory tests, pertinentmedical history, and demographics.

Results: Ambulatory hospitalized patients who fell werecompared to controls (no falls during hospitalization) ofsimilar age (P � .283) and gender distribution (P � .554).Patients who fell had significantly lower Hb (P � .00005),lower Hct (P � .00005) and were more likely to be ane-

DOI: 10.1016/j.jamda.2006.12.015

SUPPLEMENT

regression model examined the effect of Hb level andanemia on falls and included the covariates of age, gen-der, place of residence, and race. The model described a22% decreased risk of falls for every 1.0 g/dL increase inHb (P � .0005) and an overall 1.9-fold increased risk offalls in anemic patients (P � .001). Patient age, gender,and place of residence were not related to the risk offalls during hospitalization.

Conclusions: These findings suggest a potentially impor-tant link between anemia and the risk of falls duringhospitalization in ambulatory older patients. Furtherstudies are needed to determine if the risk of falls can bemodified by correction of anemia and to determine theapplicability of these findings to older adults in differentsettings. (J Am Med Dir Assoc 2007; 8: E.9–E.15)

Keywords: Anemia; hemoglobin; fall risk; older

mic (56% vs 38%, P � .001) than controls. A logistic adults; falls

The World Health Organization (WHO) defines anemia as ahemoglobin (Hb) �12 g/dL in females and �13 g/dL in males1;

Department of Medicine (T.S.D.), Division of Geriatrics (T.S.D., S.A.), and De-partment of Medical Research (E.P.N.), Our Lady of Mercy Medical Center,Bronx, NY; University Hospital of New York Medical College, Valhalla, NY.

This study did not receive external funding. This study was internally fundedas part of the Geriatric Medicine Fellowship program.

This work was presented, in part, at the Annual Meeting of the American Geri-atrics Society Annual Meeting in 2005. Dharmarajan TS, Avula S, Jayakrishnan L,Jospeh P, Talamati J, Karbasian M. Norkus EP. Mild anemia increases the risk offalls in hospitalized older adults. J Am Geriatr Soc 2005;53:S94.

Address correspondence to T.S. Dharmarajan, MD, FACP, AGSF, 31 PheasantRun, Scarsdale, NY 10583. E-mail: dharmarajants@yahoo.com

Copyright ©2006 American Medical Directors Association

using this definition, anemia appears to be common in olderadults in the United States.2,3 The third National Health andNutrition Survey (NHANES III) identified anemia in 5% to8% of older adults between 60 and 69 years (combined malesand females, all race/ethnic groups), with an increase to 12% to15% in adults older than 69 years.3 NHANES III also foundthat African Americans have the highest rates of anemia be-tween 60 and 69 years (15% to 20%), increasing to more than25% at ages older than 69 years.3 A recent comparison of datafrom NHANES III and NHANES 1999–2000 further suggeststhat, among older women (all race/ethnic groups), the preva-lence of anemia is increasing.4 Others have suggested evenhigher rates of anemia among older males and females, partic-ularly among those older than 85 years.5,6 However, anemia is

often under-recognized and consequently under-treated.5,7 Ane-

Dharmarajan et al. E.9

mia in older adults may not be given the consideration itdeserves because the risks associated with untreated anemia arepresently poorly defined and because some clinicians may be-lieve that lower levels of Hb and hematocrit (Hct) are accept-able or normal in the older population.

Recent reports suggest a link between anemia and both adecline in physical performance and an increase in disabilityin older adults.8,9 Even older adults with borderline anemia(within 1.0 g/dL above the WHO limits) appear to exhibitgreater physical decline than individuals with higher Hblevels.9 Low Hb levels also have been identified as a risk factorfor falls in older women10; we recently reported a link be-tween anemia and a history of falls occurring prior to hospi-talization in older nursing home and community residentswho subsequently suffered a hip fracture.11

In this study, we examined the relationship between ane-mia (and/or Hb level) and the risk of falls occurring duringhospitalization in a sample of ambulatory older adults fromthe community and from nursing homes.

METHODS

Data were collected from 362 ambulatory older adults (59–104 years) hospitalized for acute care at our university medicalcenter from June 2001 through December 2004. Fall patientswere identified by examination of the Daily Incident Reportsand were prospectively compared with a group of controlpatients of similar age and gender who did not sustain a fallduring hospitalization. The vast majority of fall patients andcontrols were acute-care patients from a 70-bed, dedicatedgeriatrics unit within the medical center. During the studyperiod (2001 to 2004), the mean age of patients in thegeriatric unit was 82 � 12 (SD) years (range of 60 to 106years). For this study, a fall was defined as the sudden, unin-tentional loss of posture causing an individual to inadver-tently rest at a lower level, without use of overwhelmingexternal force. Patient demographics, medical history includ-ing comorbidity, history of prior falls, routine and additionallaboratory values (including Hb, Hct, serum iron, total iron-binding capacity [TIBC], ferritin, serum B12, and serum fo-late) were collected (when available) from medical records.Anemia was defined using WHO criteria (Hb �12.0 g/dL forwomen and Hb �13.0 g/dL for men). Approval was obtainedfrom our institutional review board prior to study initiationand the researchers conducting this study were not involvedin the management of any patients included in this report. Allstudy patients were identified during hospitalization and theirmedical records were accessed after hospital discharge.

Patient comorbidity was classified into disease classes (car-diac, diabetes, electrolyte imbalance, ethanol abuse, hyper-tension, incontinence, musculoskeletal, neurological, pulmo-nary, renal). The major disease categories were classified asfollows: cardiac � coronary artery disease (CAD), congestiveheart failure (CHF), arrhythmia; diabetes; hypertension; in-continence; musculoskeletal disease � arthritis, myopathy,bone disease other than fracture; neurological � dementia,delirium, cerebrovascular disease, Parkinson’s disease, neurop-

athy; pulmonary disease � chronic obstructive pulmonary

E.10 Dharmarajan et al.

disease (COPD), pneumonia; and renal insufficiency or failure(chronic kidney disease [CKD] � creatinine � 1.5).

Statistical analyses, STATA 8.2 (Stata Corp, College Sta-tion, TX), provided descriptive summaries expressed as actualnumbers, percentages, or mean values (� SD). Student t testsand 1-way analysis of variance (ANOVA) were used to detectsignificant differences, respectively, between the mean valuesof 2 or 3 independent, continuous variables. Chi-square anal-yses or Fisher’s exact test (for small cells) were used to com-pare categorical variables. To identify the significance of po-tential relationships between the presence of anemia and falls,logistic regression analyses were performed. Additional mul-tiple regression analyses examined the relationship betweenHb and Hct levels and falls. A model was constructed todetermine if additional independent relationships existed be-tween falls and the covariates of age and Hb level (g/dL) ascontinuous variables and gender, place of residence (nursinghome vs community), race (white, African American, His-panic, Asian), anemia (present vs absent) as categorical orindicator variables. P values less than .05, in 2-tailed testing,were considered statistically significant.

RESULTS

Patient Characteristics

The study included 362 patients, hospitalized over a 3-yearperiod, with a mean age of 76.9 � 9.9 years (range, 59–104years). Twenty-seven percent were 60 to 69 years old, 31%were 70 to 79 years old, 30% were 80 to 89 years old, and 12%were older than 89. The sample included 166 males and 196females; 210 patients lived in the community and 152 werenursing home residents.

Table 1 compares study variables by patient gender. Femalepatients were significantly older than male patients (P � .002),while the gender distribution between community and nursinghome patients was similar (P � .257). As expected, mean Hblevels were significantly but not dramatically higher in malesthan females (12.7 vs 12.3 g/dL, P � .036). Anemia, as definedby WHO, occurred more often in males than females (54% vs42%, P � .032) but the length of hospital stay (P � .100) andserum albumin levels (P � .750) were similar between genders.

Hospitalized patients from nursing homes were slightly

Table 1. Patient Characteristics by Gender

Variable Males(n � 166)

Females(n � 196)

P Value

Age, y 75.2 � 9.5 78.4 � 10.0 .002*Residence, n in NH/

community75/91 77/119 .257†

Hemoglobin, g/dL 12.7 � 1.9 12.3 � 2.0 .036*Anemia, % 54 42 .032†Length of stay, d 12.4 � 7.9 9.9 � 8.7 .100*Serum albumin, g/dL 3.1 � 0.7 3.1 � 0.7 .750*

NH, nursing homes.Results expressed as means � SD, percentages, or actual numbers.

Statistical comparisons used * Student t tests or † chi-square analysis.

older (78.2 � 9.6 years vs 76.0 � 10.0 years, P � .036) than

JAMDA – March 2007

ers. St

patients from the community (Table 2). No significant differ-ences in gender distribution (P � .257), mean Hb level (P �.534), proportion of patients with anemia (P � .368), lengthof hospital stay (P � .4547), and serum albumin levels (P �.1411) were observed between patients from nursing homesand the community.

Table 3 compares study variables by patient race. Hispanic,African American, white, and Asian patients did not differsignificantly with respect to age (P � .397), place of residence(P � .115), or length of hospital stay (P � .826). However, inthis sample, Hispanic patients had significantly higher meanhemoglobin levels (P � .018) and significantly less anemia (P� .001) than African American, white, and Asian patients.

Figure 1 presents the frequency of comorbid conditions inthis sample of hospitalized older adults by the entire sample,by fall patients alone and by control patients alone. For theentire sample, hypertension (68%), cardiac disease (51%),anemia (47%), diabetes (41%), and neurological diseases(38%) occurred most often, followed by renal disease (26%),incontinence (15%), and musculoskeletal disease (12%). Pul-monary disease, ethanol abuse, and electrolyte imbalanceoccurred in fewer than 10% of patients. Anemia occurredsignificantly more often in fall patients compared with con-trols (P � .001); otherwise, fall patients and controls hadsimilar frequencies for comorbidity (P � .05).

Additional comparisons, not included in Figure 1, foundthat renal disease and ethanol abuse occurred more often inmales (32% vs 21%, P � .017, and 10% vs 2%, P � .004,

Table 2. Patient Characteristics by Place of Residence

Variable NursingHome(n � 152)

Community(n � 210)

P Value

Age, y 78.2 � 9.6 76.0 � 10.0 .035*Gender, F/M ratio % 51/49 57/43 .257†Hemoglobin, g/dL 12.5 � 1.9 12.4 � 2.1 .534*Anemia, % 45 50 .368*Length of stay, d 11.7 � 13.7 10.6 � 13.8 .455*Serum albumin,

g/dL3.0 � 0.7 3.1 � 0.7 .141*

Results expressed as means � SD or percentages. Statistical com-parisons were made using * Student t tests or † chi-square analysis.

Table 3. Patient Characteristics by Race

Variable Hispanic(30.9%)

Africa(31.3

Age, y 74.9 � 9.6 75.9 �Residence, % NH/community 33/67 37/63Hemoglobin, g/dL 12.9 � 2.1 11.9 �Anemia, % 33 54Length of stay, d 9.4 � 9.6 9.8 �Serum albumin, g/dL 2.9 � 0.7 3.3 �

NH, nursing homes.Results expressed as means � SD, percentages, or actual numb

or † chi-square analysis.

SUPPLEMENT

respectively). All other comorbid disorders occurred withsimilar frequency between males and females (P � .05).Patients from nursing homes had more neurological disease(60% vs 23%, P � .001) and incontinence (22% vs 10%, P �.003) but less diabetes (33% vs 47%, P � .009) than com-munity-dwelling patients. The frequencies of all other comor-bidities were similar in patients from nursing homes and thecommunity (P � .05). A comparison of comorbidity by race,in this sample, observed that whites had less hypertension(55%) than Hispanics (72%), African Americans (75%), andAsians (67%), respectively (P � .031). Whites also had lessdiabetes (27%) than Hispanics (58%), African Americans(42%), and Asians (67%), respectively (P � .001). Hispanicshad less anemia (31%) than African Americans (54%),whites (52%), and Asians (100%), respectively (P � .001).The frequencies of all other comorbidities did not differ byrace (P � .05).

Falls During Hospitalization and the Relationshipto Anemia

Of the 362 patients studied, 198 (54.7%) experienced a fallduring hospitalization (Table 4). Patients who fell and non-fall controls had similar ages (P � .283) and gender distribu-tion (P � .554). However, patients with falls manifested

0 20 40 60 80

Percent Occurence

electrolyte imbalance

alcohol abuse

pulmonary

musculoskeletal

incontinence

renal

neurological

diabetes

anemia

cardiac

hypertension

Fig. 1. Frequency of comorbidity in 362 ambulatory older adults.The major comorbidity classes are graphed for all patients (blackbars), fall patients only (dark grey bars), and control patients only(light grey bars).

erican White(35.5%)

Asian(2.3%)

P Value

77.2 � 9.3 78.5 � 14.6 .397*49/51 33/67 .115†12.5 � 1.7 11.5 � 1.3 .018*52 100 .001†10.3 � 8.5 12.0 � 5.5 .826*3.1 � 0.7 2.7 � 0.4 .022*

atistical comparisons used * 1-way analysis of variance (ANOVA)

n Am%)

9.5

2.0

7.50.7

Dharmarajan et al. E.11

significantly lower mean Hb levels (12.0 vs 13.0 g/dL, P �.00005), mean Hct levels (36.6 vs 39.2%, P � .00005), ahigher prevalence of anemia (56% vs 38%, P � .001), andlonger length of hospital stay (14.2 vs 7.3 days, P � .00005)compared with patients who did not fall. In addition, His-panic patients appeared less likely to fall (39%) than AfricanAmerican (60%), white (59%), and Asian patients (83%),respectively (P � .007). Figure 2 shows the percentage of fallspatients in relation to their serum Hb status. Forty-nine per-cent of patients with hemoglobin levels between 12.0 and12.9 g/dL sustained a fall and the percentage of patientsexperiencing a fall increased as the hemoglobin levels de-clined below this range (P � .001). Eighty-two percent of allfalls occurred in patients with anemia or borderline anemic(� 1.0 g/dL � the WHO limits).

Results of our multivariate logistic regression model aresummarized in Table 5. The model adjusted for covariates ofage, gender, race, and place of residence and found thatanemic patients had a significantly higher risk of falls thannonanemic patients (odds ratio [OR] � 1.86; 95% confidenceinterval [CI]: 1.16–2.82; P � .008). When Hb was substitutedfor anemia in the regression, it was identified as a moresignificant predictor of falls during hospitalization (OR �0.78; 95% CI: 0.69–0.88; P � .0005) indicating that the riskof falls occurring during hospitalization decreased by 22% forevery 1 g/dL increase in Hb above 6.8 g/dL. The model alsoidentified that the risk of falls was significantly less amongHispanic patients (OR � 0.45; CI: 0.25–0.79; P � .005)compared to the other three race groups (white, AfricanAmerican, and Asian) and was independent of the risk fromlow Hb status or presence of anemia. In this model, age,gender, and place of residence were not independent predic-tors of falls during hospitalization.

DISCUSSION

Meeting the health care needs of an aging population withincreasingly evident comorbid conditions will be a major

Table 4. Comparisons Between Fall Patients and Controls

Variable FallPatients(n � 198)

Controls(n � 164)

P Value

Age, y 77.3 � 9.3 76.3 � 10.5 .283*Gender, F/M ratio % 56/44 48/52 .554†Hemoglobin, g/dL 12.0 � 2.1 13.0 � 1.6 �.00005*Hematocrit, % 36.6 � 6.4 39.2 � 5.2 �.00005*Anemia, % 56 38 .001†Length of hospital

stay, d14.2 � 16.9 7.3 � 7.0 �.00005*

Race, %Hispanic 39 61 —African American 60 40 —White 59 41 —Asian 83 17 .007‡

Results expressed as means � SD or percentages. Statistical com-parisons were made using * Student t tests, † chi-square analysis,or ‡ Fisher’s exact test.

public health concern throughout this century. We now re-

E.12 Dharmarajan et al.

alize that falls and consequent injuries in the geriatric popu-lation contribute substantially to social problems and nega-tively impact quality of life and survival.12 Falls occur as aresult of multiple factors, requiring a multifactorial approachto prevention.13–16 Individuals fall more often as they growolder,14 with reports of prevalence rates as high as 30% inadults older than 64 years of age15,17 and approaching rates of40% in adults older than 80 years.15 Up to 60% of ambulatorynursing home residents have experienced a fall16 and 50% offalls may be recurrent.18 Many falls result in mild or no injury,while some result in moderate to severe injury19 or even causeinjury-related deaths.12 Falls are responsible for more than90% of hip fracture cases.11,20 With 6.3 million hip fracturesprojected by the year 2050 worldwide,21 the impact of falls onmedical care costs to treat hip fracture alone will be substan-tial.22 Clearly, measures that reduce the risk of falls will havetremendous potential benefit to the health status and qualityof life of older adults and would contribute to a reduction inoverall medical care costs. Our finding that patients who fallin the hospital tend to have prolonged hospitalization (pa-tients who fell averaged twice the length of hospital stay inour study, 14 vs 7 days), simply highlights the increasedmedical care costs that are associated with falls.

Aside from the aging process, medications (especiallypolypharmacy), arthritis, impaired vision, gait and balancedisorders, decline in cognitive function, and other comorbidi-ties are recognized as risk factors for falls in older adults.13

Some factors may be easy to modify (eg, scrutiny and revisionof medication regimen, correction of refractive errors, cata-racts, and so forth). Other factors such as environmentalfactors may be modified to a limited extent.14 Still others (eg,permanent neurological deficits) cannot be modified.10,18,23

With the focus on prevention and risk reduction, currentthinking suggests that the goals should be to identify any andall risk factors that contribute to falls and to target efforts atthose conditions that can be corrected, either partially orcompletely. Efforts to improve functional status and physicalability, including balance control and muscle strength

0

20

40

60

80

100

Pe

rce

nt

Wh

o F

ell

, %

<8.08.0-8.9

9.0-9.9

10.0-10.9

11.0-11.9

12.0-12.9

13.0-13.9

>14.9

Hemoglobin Level, g/dL

Fig. 2. Percentage of falls during hospitalization by hemoglobin

level.

JAMDA – March 2007

through exercise and fitness programs in older adults, appearto show promise in fall prevention in a few,12,24 but not all,25

studies. In this regard, anemia has largely escaped attention inthe prevention of falls and fall-related injuries.

Anemia, arising from nutritional deficiencies or fromchronic disease such as renal insufficiency, cancer, or rheu-matoid arthritis, is common in older people.3,26 Anemia lead-ing to weakness, fatigue, limitation in activity,27 and anincreased risk of cardiovascular events is well documented inthe nephrology literature.28,29 In addition, the impact of fa-tigue from anemia on exercise compliance, mobility, andcognitive function in older adults deserves further study.9

Given the recognized association between anemia and phys-ical decline in older persons,8,9 corrective measures such asexercise and improved balance training, aimed at preventingfalls and injuries, may have diminished effectiveness in thepresence of anemia simply because patients may be too fa-tigued to derive optimal benefit.

With specific regard to risk factors for falls in older subjects,anemia has received surprisingly little attention. Herndon andcolleagues23 observed that self-reported anemia was associatedwith a 50% higher risk of fall injury in community-dwellingolder adults. More recently, Lawlor et al10 found that womenreporting at least one fall per year had significantly, althoughnot clinically important, lower mean Hb concentrations (13.4vs 13.5 g/dL) compared with women who did not report a fall.We recently reported anemia to be a significant risk factor forfalls in a group of community and nursing home patients whosustained falls and consequent hip fractures prior to hospital-ization for hip fracture.11 Those findings provided the impetusfor us to further explore the link between anemia and falls ina different setting, namely that of hospitalized ambulatory olderadults.

The present report demonstrates a strong relationship be-tween falls during hospitalization in ambulatory older patientsand the presence of anemia or low Hb levels, and is consistentwith our earlier findings that dealt with falls prior to hospital-ization in older adults. Although we recognize that additionalfactors probably contribute to the risk of falls, we observed a22% decreased risk for falls for every 1.0 g/dL increase in Hb(P � .001) and a 1.9-fold increased risk for falls in thepresence of anemia (P � .008). These findings support andextend the few previous studies,10,11,23 in that it demonstratesthat, in ambulatory older persons, the risk of falls appears

Table 5. Logistic Regression Model: Risk of Falls Predicted by Study

Independent Variables Odds Ratio

Age 1.01Gender 0.92Residence 0.93Hemoglobin, g/dL 0.78Anemia* 1.86Hispanic race† 0.45Other races† 0.92–2.91

* When anemia was included, Hb was dropped from the model.† Categorical variables (0/1) were created for each race.

clearly correlated with reduced hemoglobin levels.

SUPPLEMENT

An earlier report has suggested that white women are morelikely to fall than men or African American women.30 We didnot observe any effect of gender on falls risk. However, we didobserve a link between falls and race. In our regression model(Table 5), Hispanic patients were found to have a 55%decreased risk of falls during hospitalization compared withAfrican American, white, and Asian patients (P � .005); thisobservation was independent of the increased risk of falls dueto decline in hemoglobin levels or presence of anemia. Thereason(s) for this decrease is unclear. We also did not examineseveral other suggested risk factors for falls such as visual orhearing impairment, gait and balance disorders, depression,and environmental factors.12,18,30–33 As all falls occurred inan acute care hospital geriatric unit that was well-designedwith features suited to elder needs, we believe that the envi-ronmental impact on falls would have been minimal andwould have affected all subjects equally.

The body of research on falls in older subjects suggests thatmany factors, acting together, influence the occurrence offalls; in other words, falls are multifactorial.13,16,18 Some ofthese factors cannot be corrected. We tend to agree withviews expressed in a recent editorial that suggests a focus ononly those factors that are correctable to lower the risk offalls.34 At present, anemia (particularly mild anemia) appearsto be under-recognized and has escaped attention as a poten-tial correctable risk factor for falls. A recent study in thegeriatric wards of a rehabilitative hospital did not consideranemia in the suggested clinical features and diagnoses thatare part of a risk assessment.35 Neither the guidelines pub-lished by the American Geriatrics Society Panel in 200136 northe comparative study of the use of 4 fall risk assessment toolsused in acute medical wards published in 2005, mentionedanemia as a risk factor with predictive value for falls.37 How-ever, these reports did stress the need for more researchtoward interventions to help prevent falls.

We recognize that our study had limitations. First, all of ourpatients resided in the Bronx, NY, and so our study lackedgeographic diversity. Second, we matched controls, post hoc,to fall patients by age and gender (Table 4). Our controls werechosen from the same acute geriatric care setting in ourhospital so that they would mimic the fall patients to theextent possible with regard to comorbidity (Figure 1). How-ever, we did not attempt to examine or control for other riskfactors for falls such as visual and hearing impairment, gait or

es

95% Confidence Interval P Value

0.98–1.03 .6800.59–1.45 .7260.59–1.46 .7450.69–0.88 �.00051.16–2.82 .0080.25–0.79 .0050.32–25.9 all �.05

Variabl

balance disorders, or medication use between our fall patients

Dharmarajan et al. E.13

and controls. Third, our study was observational not inter-ventional. Future studies should be designed to examine therelationship between falls and anemia in different settings andmay be enriched from interventional approaches. We believethat efforts should focus on the impact that anemia correctionmay have on fall prevention in older subjects.

Anemia is easily detected and its etiology can be identifiedin the majority of cases.5 The American Medical DirectorsAssociation Foundation has recently recognized the need todevelop practice guidelines for anemia and to evaluate therelationship between anemia and falls. In the long-term caresetting, an interdisciplinary approach is recommended toevaluate the interaction of medical conditions (includinganemia where appropriate), medications and the environ-ment.38 Using the WHO definition for anemia, we observedthat the risk of falls increased approximately 2-fold in thepresence of anemia and that this risk was independent offactors that included age, gender, place of residence, and race.The extent to which even mild anemia may be associatedwith falls warrants further investigation, as borderline anemiahas been recognized to be associated with physical decline inelderly persons.8,9 In our study, 82% of all falls occurred inpatients with anemia or borderline anemia (� 1.0 g/dL � theWHO limits).

Recognition of anemia is easily achieved using routinelaboratory tests. Its evaluation and correction should becomepart of future interventional studies that aim to improvefunctional status and risk for falls in older adults. One third ofall cases of anemia in older subjects result from nutritionalinadequacy, one third results from chronic disease includingchronic kidney disease, and the remaining cases of anemia areof undetermined etiology.3,5 Data are beginning to emerge onthe relationship between nutritional anemia and falls.39 Thus,it would seem possible that more than 50% of cases of anemiain older adults are correctable through nutrient replacementand/or by stimulating erythropoiesis. The impact of maintain-ing Hb at or near normal levels on functional capacity, organfunction, risk for falls, and health care economics needs to bestudied in varied population samples and in different settingsincluding long-term care.38 A recent editorial comment onmore than 60 randomized controlled trials that focused on fallprevention states that “we need to target interventions atpeople most likely to benefit from them” and further suggeststhat interventions should be effective, and that providers setachievable targets.34 Anemia certainly appears to be a riskfactor that should be targeted for correction with likely ben-eficial results.

REFERENCES1. World Health Organization. Nutritional Anemias. Technical Report

Series No. 405. Geneva, Switzerland: WHO, 1968.2. Beard JL, Ashraf MH, Smiciklas-Wright H. Iron nutrition in the elderly.

In: Watson RR, ed. Handbook of Nutrition in the Aged, 2nd ed. BocaRaton, FL: CRC Press, Inc, 1994, pp. 393–414.

3. Hollowell JG, Van Assendelft OW, Gunter EW, et al. Hematologicaland iron-related analyses—Reference data for persons aged 1 year andover; United States, 1988–1994. National Center for Health Statistics.

Vital Health Stat 2005;11(247).

E.14 Dharmarajan et al.

4. Looker AC, Cogswell ME, Gunter EW. Iron deficiency—United States,1999–2000. MMWR 2002;51:897–899.

5. Guralnik JM, Eisenstaedt RS, Ferrucci L, et al. Prevalence of anemia inpersons 65 years and older in the United States: Evidence for a high rateof unexplained anemia. Blood 2004;104:2263–2268.

6. Balducci L. Epidemiology of anemia in the elderly: Information ondiagnostic evaluation. J Am Geriatr Soc 2003;51:S2–S9.

7. Schrier SL. It’s tough to be old and anemic. Blood 2004;104:2214–2215.8. Chaves PHM, Ashar B, Guralnik JM, et al. Looking at the relationship

between hemoglobin concentration and prevalent mobility difficulty inolder women. Should the criteria currently used to define anemia in olderpeople be reevaluated? J Am Geriatr Soc 2002;50:1257–1264.

9. Penninx BW, Guralnik JM, Onder G, et al. Anemia and decline inphysical performance among older persons, Am J Med 2003;115:104–110.

10. Lawlor DA, Patel R, Ebrahim S. Association between falls in elderlywomen and chronic diseases and drug use: Cross sectional study. BMJ2003;327:712–717.

11. Dharmarajan TS, Norkus EP. Mild anemia and the risk of falls in olderadults from nursing homes and the community. J Am Med Dir Assoc2004;5:395–400.

12. Bloem BR, Steijns JA, Smits-Engelsman BC. An update on falls. CurrOpin Neurol 2003;16:15–26.

13. King MB, Tinetti ME. A multifactorial approach to reducing injuriousfalls. Clin Geriatr Med 1996;12:745–759.

14. Chang JT, Morton SC, Rubestein LZ, et al. Interventions for the pre-vention of falls in older adults: Systematic review and meta-analysis ofrandomized clinical trials. BMJ 2004;328:680–683.

15. Prudham D, Evans JG. Factors associated with falls in the elderly: Acommunity study. Age Ageing 1981;10:141–146.

16. Tinetti ME. Factors associated with serious injury during falls byambulatory nursing home residents. J Am Geriatr Soc 1987;35:644 –648.

17. Stalenhoef PA, Diederiks JP, Knottnerus JA, et al. A risk model for theprediction of recurrent falls in community-dwelling elderly: A prospec-tive cohort study. J Clin Epidemiol 2002;55:1088–1094.

18. Tinetti ME, Speechley M, Ginter SF. Risk factors for falls amongelderly persons living in the community. N Engl J Med 1988;319:1701–1707.

19. Alexander BH, Rivara FP, Wolf ME. The cost and frequency of hospi-talization for fall-related injuries in older adults. Am J Public Health1992;82:1020–1023.

20. Marks R, Allegrante JP, Ronald MC, et al. Hip fractures among theelderly: Causes, consequences and control. Ageing Res Rev 2003;2:57–93.

21. Cooper C, Campion G, Melton LJI. Hip fractures in the elderly: Aworld-wide projection. Osteoporos Int 1992;2:285–289.

22. Englander F, Hodson TJ, Terregrossa RA. Economic dimensions of slipand fall injuries. J Forensic Sci 1996;41:733–746.

23. Herndon JG, Helmick CG, Sattin RW, et al. Chronic medical condi-tions and risk of fall injury events at home in older adults. J Am GeriatrSoc 1997;45:739–743.

24. Robertson MC, Campbell AJ, Gardner MM, et al. Preventing injuries inolder people by preventing falls: A meta-analysis of individual-level data.J Am Geriatr Soc 2002;50:905–911.

25. Moreland J, Richardson J, Chan DH, et al. Evidence-based guidelines forthe secondary prevention of falls in older adults, Gerontology 2003;49:93–116.

26. Balducci L. Epidemiology of anemia in the elderly: Information ondiagnostic evaluation. J Am Geriatr Soc 2003;51(Suppl):S2–S9.

27. Aapro MS, Cella D, Zagari M. Age, anemia, and fatigue. Semin Oncol2002;29:55–59.

28. Eckardt K-U. Cardiovascular consequences of renal anaemia and eryth-ropoietin therapy. Nephrol Dial Transplant 1999;14:1317–1323.

29. Levin A. Clinical epidemiology of cardiovascular disease in chronickidney disease prior to dialysis. Semin Dial 2003;16:101–105.

30. de Rekeneire N, Visser M, Peila R, et al. Is a fall just a fall: Correlates offalling in healthy older persons. The Health, Aging and Body Compo-

sition Study. J Am Geriatr Soc 2003;51:841–846.

JAMDA – March 2007

31. Cesari M, Landi F, Torre S, et al. Prevalence and risk factors for falls inan older community-dwelling population. J Gerontol A Biol Sci Med Sci2002;57:M722–M726.

32. Ryynanen OP. Health, functional capacity, health behaviour, psycho-social factors and falling in old age. Public Health 1994;108:99–110.

33. Smith RG. Fall-contributing adverse effects of the most frequently pre-scribed drugs. J Am Podiatr Med Assoc 2003;93:42–50.

34. Gillespie L. Preventing falls in elderly people [editorial]. BMJ 2004;328:653–654.

35. Vassallo M, Vignaraja R, Sharma JC, et al. The effect of changingpractice on fall prevention in a rehabilitative hospital: The hospital

injury prevention study. J Am Geriatr Soc 2004;52:335–339.

SUPPLEMENT

36. Kenny RA, Rubenstein LZ, Martin FC, et al. Guideline for theprevention of falls in older persons. J Am Geriatr Soc 2001;49:664 –672.

37. Vassallo M, Stockdale R, Sharma JC, et al. A comparative study of theuse of four fall risk assessment tools on acute medical wards. J Am GeriatrSoc 2005;53:1034–1038.

38. Feinsod AM, Capezutti EA, Felix V. Reducing falls risk in long-term careresidents through interdisciplinary approach. Annals of Long Term Care2005;13:24–33.

39. Dharmarajan TS, Avula S, Jospeh P, et al. Falls in ambulatory commu-nity older adults are linked to anemia and nutritional status [abstract]. J

Nutr Health Aging 2005;9:136.

Dharmarajan et al. E.15