One leg balance is an important predictor of injurious falls in older persons

One-Leg Balance Is an Important Predictor of Injurious Falls in Older Persons OBJECTIVE: To test the hypothesis that one-leg balance is a significant predictor of falls and injurious fa

One-Leg Balance Is an Important Predictor of Injurious Falls

in Older Persons

OBJECTIVE: To test the hypothesis that one-leg balance is a

significant predictor of falls and injurious falls

DESIGN: Analysis of data from a longitudinal cohort study

SUBJECTS: Healthy, community-living volunteers older

than age 60 enrolled in the Albuquerque Falls Study and

followed for 3 years ( N = 316; mean age 73 years)

MAIN OUTCOME MEASURES: Falls and injurious falls

detected via reports every other month

INDEPENDENT VARIABLES: Baseline measures of demo-

graphics, history, physical examination, Iowa Self Assess-

ment Inventory, balance and gait assesessment, and one-leg

balance (ability to stand unassisted for 5 seconds on one leg)

RESULTS: At baseline, 84.5% of subjects could perform

one-leg balance (Impairment was associated with older age

and gait abnormalities.) Over the 3-year follow-up, 71%

experienced a fall and 22% an injurious fall The only inde-

pendent significant predictor of all falls using logistic regres-

sion was age greater than 73 However, impaired one-leg

balance was the only significant independent predictor of

injurious falls (relative risk: 2.13; 95% CI: 1.04, 4.34; P =

.03)

CONCLUSION: One-leg balance appears to be a significant

and easy-to-administer predictor of injurious falls, but not of

all falls In our study, it was the strongest individual predic-

tor However, no single factor seems to be accurate enough to

be relied on as a sole predictor of fall risk or fall injury risk

because so many diverse factors are involved in falling J Am

Geriatr SOC 45:735-738, 1997

everal risk factors for falls have been documented repeat-

S edly The most important of these risk factors include leg

weakness, impaired gait, and balance dysfunction.'-" How-

~ _ _ _ _ _ - - - ~

From the 'New Mexico Aging Process Study, Clinical Nutrition Program, De-

partment of Pathology, University of New Mexico, School of Medicine, Albu-

querque, New Mexico; tService dc Gerontoltrgie Clinique, Departement de Me-

decine Interne, C.H.U Toulousc-Purpan, Toulouse, France; *Department of

Biochemistry, University of New Mexico, School of Medicine, Albuquerque,

New Mexico; and SGeriatric Research Education and Clinical Center (GRECC)

Sepulveda VA Medical Center (1 l E ) , Sepulveda, California

Supported by Grants AG 02049, AG10149,HL36158, and GCR DRR, 5 M01-

00997-13-13,14 from the United States Public Health Service

Address correspondence to Laurence 2 Rubenstein, MD, MPH, Director,

GRECC, Sepulvcda VA Medical Center ( 1 1E) Sepulveda, CA 91.343

ever, these factors are complex and often require special expertise or explicit protocols to measure ac~urateIy.~,' There is need for simple, accurate, reproducible indicators of

fall risk Based on observational experience with a number of

balance maneuvers and discussion with other investigators,

we hypothesized that one-leg balance (defined as the ability to

stand on one leg unsupported for 5 seconds) might be a

simple predictive factor for injurious falls and frailty To test this hypothesis, we studied the relationship between one-leg balance and falls among subjects in the Albuquerque Falls Study, a prospective study of falls in healthy older persons living in the community,6 which is part of a larger longitudi- nal study on aging: the New-Mexico Aging Process Study

METHODS

Subjects

Entrance to the Albuquerque Falls Study (AFS)(n = 482)

was limited to volunteers older than 60 years of age with no known serious medical conditions All participants were

white, more than 40% had attended college, and most of

them lived at home or in residences for older persons

Baseline Examination

All subjects underwent an initial examination consisting

of an interview, a physical examination, and an assessment of

gait and balance performance These tests are described in detail below

The Iowa Self-Assessment Inventory (ISAI)' was admin-

istered to all subjects The ISAI is a 56 item self-administered

questionnaire reported to be a reliable and valid instrument

to assess sense of well being The subjects underwent a medical history questionnaire and a general physical exam including blood pressure, Mini-mental State Exam (MMSE),' and assessment of drug prescription and consumption Balance and gait performance was assessed using the instrument developed by Tinetti.' In addition, we tested one-leg balance, i.e., asking the subject to stand unassisted on one leg for 5 seconds, again using a method reported by Tinetti We asked the subject being tested to choose a leg to stand on (whichever he felt more comfortable with), flex the opposite knee allowing the foot to clear the floor, and balance

on one leg as long as possible; the nurse recorded whether the subject was able to balance for 5 seconds These balance and

gait tests satisfied the need for an instrument that ( 1 ) required

n o equipment and little experience to master so that it could

be used in a typical clinical setting without intensive training;

- - _ _ _ _ - _ _ - - _ _ _ _

JAGS 45:735-738,1997

0 1997 by the American Geriatrics Society

73fi VELLAS ET AL _ _ _ _ ~

-(2) was reliable yet sensitive to significant changes; and (3)

most importantly, reflected position changes and gait maneu-

vers used during normal daily activities

The geriatric assessment was repeated annually to assess

the evolution of gait, balance, activity and health status

Repeat measures allow investigation of possible associations

of the above changes with falls and better documentation of

post-fall sequellae

Assessment of Falls

The subjects were instructed to report a fall event by

calling the study coordinator We used the definition of fall

defined by the World Health Organization (WHO) and the

Kellogg International Work Group on the Prevention of Falls

by the Elderly: “A fall is an event which results in a person

coming to rest inadvertently on the ground or other lower

level and other than as a consequence of a violent blow; loss

of consciousness or sudden onset of paralysis” In order to

capture unreported falls, all participants were sent a stamped

post-card every other month inquiring specifically about

whether or not a fall had occurred If no response was

received after 15 days, a second card was sent Each subject

was interviewed as soon as possible after each reported fall in

order to review the circumstances and determine whether the

incident met the study definition of a fall If it did meet the

definition, a W H O Falls Questionnaire was completed by the

study coordinator during a telephone interview We were

also able to assess aspects of the severity of falls Injurious

falls were defined as falls for which medical attention was

sought Follow-up data were collected on all study subjects

for 3 years

Statistical Analyses

Analysis was conducted using the Statistical Analysis

System for personal computers Tests of statistical signifi-

cance were done using chi-square tests when comparing

percents and either a t test (for normally distributed data) or

a Wilcoxon rank sum test (for non-normally distributed data)

when comparing means A significance level of 05 was used

for all tests

The present report gives results for the 36-month study

period All falls and fall-related injuries reported during this

period constituted the outcome variables Predictor variables

are those assessments (including one-leg standing balance)

made at the subjects’ annual visit before the study period

RESULTS

The mean age (SD) of the participants at entry into the

study was 72.7 (k6.1) years, and 59% percent were women

As reported previously, the cognitive status of this population

was generally excellent, and no subject was severely de-

mented The community mobility was generally good, and

few participants had difficulty with transportation or doing

their own shopping as determined by specific questions in the

ISAI questionnaire.6

In 1990,483 subjects were enrolled in the Albuquerque

Falls Study Six of these subjects were dropped because they

did not have a one-leg balance measured in 1990 Another

113 subjects dropped out of the study in the subsequent 3

years, leaving 364 remaining subjects In order to use a

dichotomous outcome for falls and injurious falls, it was

necessary for subjects to have nearly complete follow-up over

the 3 years we were following them for occurence of falls and

JUNE 1997-VOL 45, NO 6 JAGS

~~

injuries We decided to exclude subjects who did not respond

to four or more postcards Because postcards were mailed every other month, this meant that subjects could miss up to

6 months of follow-up (of a total of 36 months) and still be included in the study This final exclusion resulted in 48 subjects being dropped and gave a final sample size of 316 The subjects who dropped out or failed to respond were, on average, about 2 years older that those who completed the 3-year follow-up, but they did not differ in terms of sex, gait abnormality, Mini-Mental State Exam score, or Body Mass Index They were, however, more likely to fail the balance test at baseline (38% vs 26% of those who completed follow- UP)

In 1990, 267 of the 316 subjects (84.5%) were able to

stand on one leg unsupported for 5 seconds The remaining

49 subjects, who were unable to balance this way for 5 seconds, were hypothesized to be more likely to fall or to have

an injurious fall during the subsequent 3 years

Table 1 shows baseline data by one-leg balance in 1990

As a result of many non-normal distributions, continuous variables were dichotomized as shown in the table Subjects who were unable to balance on one leg were significantly older than subjects who had normal balance They also had significantly more gait abnormalities than the comparison group There was also a difference of borderline significance

( P = .08) between the two groups with regard to MMSE score in which subjects with abnormal one-leg balance had lower scores for cognitive functions than did the comparison group Women were slightly more likely to have abnormal balance, but this was not significant There was little differ- ence between the groups with respect to number of medica- tions, Body Mass Index, and any of the self-reported mea- sures

Table 1 Baseline Data from 1190 by One-Leg Standing Balance

in 1990

One Leg Balance in

1990

Age

573

> 73

Women Men

0 1+

0

1+

30

Sex

Gait abnormalities

Number of medications

MMSE score

<30 Body Mass Index

<24.6 224.6

150 56.2 1 1 22.4

117 43.8 38 77.6 .OOO

156 58.4 33 67.3

1 1 1 41.6 16 32.7 242

243 91.0 31 63.3

24 9.0 18 36.7 OOO

58 21.8 9 18.4

208 78.2 40 81.6

150 56.4 21 42.9

116 43.6 28 57.1

129 48.5 26 53.1

137 51.5 23 46.9

JAGS JUNE 1997-VOl 45, NO 6

Over the next 3 years (1991-1993 inclusive) 225 of the

316 subjects (71.2%) experienced a fall; 70 of these subjects

had a fall that was classified as injurious Table 2 shows the

results of logistic regression using all falls as the dependent

variable Variables included in the model were those vari-

ables that were statistically significant in Table 1 (age and gait

abnormalities), variables of borderline significance (MMSE

score and self-reported mobility), along with sex and one-leg

balance Variables were coded 0/1 as defined in Table 1 The

results show one-leg balance was not significantly associated

with the risk of falling when the other variables are included

in the model The only statistically significant predictor of

whether a subject would fall was age, with subjects 73 years

and older having a risk 1.80 times greater than younger

subjects

Table 3 repeats the analysis of Table 2 using injurious

falls as the dependent variable instead of all falls In this case,

one-leg balance is a statistically significant predictor of inju-

rious falls Subjects who were unable to balance on one leg in

1990 had 2.1 times the risk of incurring an injurious fall in

the subsequent 3 years as the comparison group, controlling

for age, sex, and other important variables in the logistic

regression model Age was also significantly associated with

the risk of incurring an injurious fall The sensitivity of

abnormal one-leg balance at baseline to predict an injurious

faller during follow up was 36%, specificity was 76%, and

positive predictive value was 31% These relatively low fig-

ures are reflective of the difficulty in accurately predicting

injurious falls, reported by many others, and are consistent

with the multifactorial and largely random nature of injuri-

ous falls We found no association between number of med-

ications taken a t baseline and either all falls or injurious falls,

consistent with another recent study of predictors of falls and

fall injuries

DISCUSSION

One-leg standing balance appears to be a promising

predictive marker for injurious falls In our study population,

falls occured frequently, and we found that more than 7 0 %

of these healthy older persons had a t least one fall during the

36-month period Inasmuch as it is clearly not possible to

prevent every fall in every older person, a more practical and,

probably more cost-effective, approach, would be to aim

prevention measures toward the smaller proportion of falls

that result in injury In our population, one-leg standing

balance was a reasonably good single clinical marker to

predict those who have greater risk of injurious falls and who

are possibly more likely to benefit from an intervention trial

If this finding can be confirmed in other populations, such a

simple marker of risk for injurious falls would have impor-

ONE-LEG BALANCE PKEDICTS INJUKIOUS F A I L S

tant applications The test is very easy to perform in clinical practice and can be part of the routine geriatric assessment Moreover it would give us the opportunity to focus our fall prevention interventions on the less than 25%) of the total population who are at greater risk of injury

Why one-leg balance was a more important predictor of injurious falls in our study than some of the other factors reported (e.g., global gait and balance measures, muscle weakness, female sex, functional impairment) is an important question and may relate to the study population: our study population was more homogeneous and healthier than groups used in earlier studies It may also be that one-leg balance can be used as a simple proxy measure for the more complicated measures used by others The biological associ- ation between abnormal balance and fall injuries may relate

to an increased tendency of persons with balance problems to

fall to the side, and falls to the side are more likely to result in injury than falls in other directions."

Our study has some limitations First, the Albuquerque

Falls Study population is comprised of volunteers who are likely not representative of the entire US older population, and, thus, generalization must be made with caution after confirmation elsewhere Second, we only measured single stance time at baseline and did not update it during the 3-year follow-up It is certainly possible that many persons experi- enced some change in their balance during the follow-up period that might have changed, even strengthened, the rela- tionship between single stance and occurence of falls o r

fall-related injuries Third, o u r determination of single stance was dichotomized a t 5 seconds rather than used as a contin-

uous variable, which would have provided considerably more data and possibly improved the predictive relationship Un- fortunately, we did not record the actual stance time at baseline, but we recommend that this be done in subsequent investigations of this relationship Finally, even though the one-leg balance test was able to predict a subgroup of older persons at significantly increased risk of fall injury, it was not sensitive enough by itself to predict the majority of such events We feel this to be a reflection of the complex and interacting nature of falls, injuries, and their etiologies and risk factors It is extremely unlikely that any single factor will prove to be an adequate sole predictor to be used without other clinical information Thus, its ultimate purpose as an identifier of persons at increased risk of fall injury who might

be targeted for fall prevention activities (and thus potentially reduce the cost of offering these activities to a larger group) must be weighed against the real risk of missing large num- bers of persons at risk of fall injury who have normal one-leg balance We must certainly not conclude from this study that one-leg balance is an adequate predictor, in itself, to be used

Table 2 Results of Logistic Regression for All Falls

Relative Risk

~~

MMSE -0.0724 0.2622 1.06 (0.64, 1.80) 78

JUNE 1997-VOL 45, NO 6 JAGS

.~

738 VELLAS ET AL

Table 3 Results of Logistic Regression Analysis for Injurious Falls

Relative Risk

The relative risk for an iniurious fall given the inability to stand on one leg for 5 seconds is about twice that for a person able to perform the task-a clinically and statistically significant risk

to exclude subjects from fall prevention activities; rather it

seems to be a promising marker of a high-risk group to

include in such activities

We have found in recent studies that one-leg balance can

be improved by some low-intensity physical exercises.” It

will be important to see if this kind of exercise can be of help

in decreasing the incidence of injurious falls in this particu-

larly high risk population This strategy should certainly not

preclude the use of other fall-prevention interventions (e.g.,

environmental modification, medication adjustment, gait

training) targeted to this and other high risk subgroups

Invidualization of fall-prevention intervention strategies will

likely remain an important concept

REFERENCES

1 O’Loughlin JL, Robitaille Y, Boivin JF, Suissa S Incidence of and risk fac-

tors for falls and injurious falls among the community-dwelling elderly Am J

Epidemiol 1993;137:342-3.54

2 Rubenstein LZ, Josephson KR Falls in the elderly: Causes and preventive a p

proaches In: Vellas B, Toupet M, Rubenstein LZ, Albarede JL, eds Falls, Bal-

ance, and Gait Disorders in the Elderly Amsterdam: Elsevier, 1992, pp 21-38

3 Nevitt MC, Cummings SR, Hudes ES Risk factors for injurious falls: A pro-

spective study J Gerontol 1991;46:M164-170

duce risk of falling among elderly people living in the community N Engl

J Med 1994;331:821-827,

5 Robbins AS, Rubenstein LZ, Josephson KR et al Predictors of falls among elderly people: Results of two population-based studies Arch Intern Med

6 Vellas BJ, Baumgartner RN, Romero LJ et al Incidence and consequences of falls in free-living healthy elderly persons Facts and Research in Gerontol-

7 Morris WW, Buckwalter KC, Cleary TA et al Refinement of the IOWA Self-

8 Folstein MF, Folstein S, McHuth PR Mini-Mental State: A practical method

4 Tinetti ME, Baker DI, McAvay G et al A multifactorial intervention to re-

1989;149:1628-1633

ogy 1993;7:131-142

Assessment Inventory Gerontologist 1990;30:243-248

for grading cognitive state of patients for the clinician J Psychiatr Res

9 Tinetti ME Performance-oriented assessment of mobility problems in the

10 Tinetti ME, Doucette JT, Claus EB, Marottoli R Risk factors for serious

1975;12:189-198

elderly J Am Geriatr SOC 1986;34:119-126

injury during falls by older persons in the community J Am Geriatr Soc

11 Nevitt MC, Cummings SR, and the SOF Research Group Type of fall and risk of hip and wrist fractures J Am Geriatr SOC 1993;41:1226-1234

12 Rubenstein LZ, Josephson KR, Robbins AS et al Effects of an exercise inter- vention for elderly men with impairments Gerontologist 1995;35:165 1995;43:1212-1221