Abstract
Background Screening should have simple and easy-to-administer methods that identify impairments associated with future fall risk, but there is a lack of literature supporting validation for their use.
Objective The aim of this study was to evaluate the independent contribution of balance assessment on future fall risk, using 5 methods to quantify balance impairment, for the outcomes “any fall” and “any injurious fall” in community-dwelling older adults who are higher functioning.
Design This was a prospective cohort study.
Methods A sample of 210 community-dwelling older adults (70% male, 30% female; mean age=79.9 years, SD=4.7) received a comprehensive geriatric assessment at baseline, which included the Berg Balance Scale to measure balance. Information on daily falls was collected for 12 months by each participant's monthly submission of a falls log calendar.
Results Seventy-eight people (43%) fell, of whom 54 (30%) sustained an injurious fall and 32 (18%) had recurrent falls (≥2 falls). Different balance measurement methods identified different numbers of people as impaired. Adjusted relative risk (RR) estimates for an increased risk of any fall were 1.58 (95% confidence interval [CI]=1.06, 2.35) for self-report of balance problems, 1.58 (95% CI=1.03, 2.41) for one-leg stance, and 1.46 (95% CI=1.02, 2.09) for limits of stability. An adjusted RR estimate for an increased risk of an injurious fall of 1.95 (95% CI=1.15, 3.31) was found for self-report of balance problems.
Limitations The study was a secondary analysis of data.
Conclusions Not all methods of evaluating balance impairment are associated with falls. The number of people identified as having balance impairment varies with the measurement tool; therefore, the measurement tools are not interchangeable or equivalent in defining an at-risk population. The thresholds established in this study indicate individuals who should receive further comprehensive fall assessment and treatment to prevent falls.
Although impairment in balance is acknowledged as a major predictor of falls, systematic reviews have been limited in recommending specific clinical assessment scales of balance.1–4 There are many risk factor studies in which the results from univariate analysis have driven the regression modeling procedures, but few prospective epidemiologic studies have focused a priori on balance impairment as the exposure of interest. Additional limitations in the literature evaluating balance impairment as the exposure of interest include the failure to take account for multicollinearity among covariates and the inclusion of multiple measures of balance in the same regression model.5,6 Measures of balance may be highly related to other variables such as lower-extremity strength (force-generating capacity), and a thorough evaluation of the independent contribution of balance requires evaluating these other relationships to accurately determine the true magnitude of effect.
Postural stability is a complex process that involves the rapid, automatic integration of information from the vestibular, somatosensory, visual, and musculoskeletal systems, in the presence of cognition, which includes attention and reaction time.7,8 The measurement tools used to evaluate balance in the clinical setting are a means of quantifying the working capacity of the sum of the components that enable postural stability. The goal of screening is to identify individuals who are at an elevated risk for falling and who should receive further assessment. Screening should have simple and easy-to-administer measurement tools that identify impairments associated with future fall risk.
We hypothesized that the clinical assessment of balance can identify future fall risk even after the adjustment for other factors related to an increased fall risk. Measurement scales that involve a difficult balance task (eg, using a reduced base of support and movement) should be the most successful, as they reduce ceiling effects. The primary objective of the study was to evaluate the independent contribution of the clinical assessment of balance on future fall risk, using 5 methods to quantify balance impairment, for the outcome of “any fall” in community-dwelling older adults who are higher functioning. A secondary objective was to perform the same analysis for the outcome of “any injurious fall.”
Method
Participants
This study was a secondary analysis of data from phase II of the Project to Prevent Falls in Veterans (PPFV). The PPFV, funded jointly by Veterans Affairs Canada and Health Canada, was approved by the University of Western Ontario Research Ethics Board for Health Sciences Research Involving Human Subjects. The sampling and data collection procedures for phase I of the PPFV have been described in detail elsewhere.9
Briefly, in phase I of the PPFV, a questionnaire was mailed to 3 simple random samples of addresses of 1,000 Canadian veterans of WWII and the Korean War living in southwest Ontario. Canadian veterans of WWII and the Korean War, as determined by Veterans Affairs Canada, and their caregivers who were living independently in the community and who were able to understand and provide responses to the questionnaire were eligible.
Participants from 2 regions in phase I of the PPFV became eligible for participation in phase II, a risk factor modification trial. Participants from phase I who consented to be re-contacted and who self-reported at least one modifiable risk factor for falling were stratified into 5 groups by the number of modifiable risk factors from the questionnaire (lower-extremity weakness; 4 or more prescription medications; and balance, foot, and vision problems). Randomization was performed within each stratum to 1 of 2 groups: the specialized geriatric services (SGS) group or the family physician (FP) group. The SGS group received a comprehensive geriatric assessment performed by a geriatrician or physical therapist and was provided with individual recommendations for fall risk factor reduction. The intervention was the assessment and did not involve the provision of any follow-up treatment. A letter summarizing the risk factors identified on the mailed questionnaire was sent to the participants in the FP group, and a similar letter was sent to each participant's family physician. Any treatment was left at the discretion of the family physician. People with no reported modifiable risk factors for falling on the questionnaire formed an open arm in the study. This group received the same comprehensive geriatric assessment as the SGS group and was given educational materials on fall prevention and healthy living.
There were no statistically or clinically significant differences among the study arms in the proportion falling, the number of falls, or the time to first fall. Because there were no detectable differences on any outcome among the study groups, the opportunity existed to use the data collected to evaluate the association between balance and fall risk.
In phase II of the PPFV, 210 people received the comprehensive geriatric assessment. Four people who were less than 60 years of age were excluded from the analysis, ensuring all people in the sample were older than 60 years of age, and 2 people had incomplete balance measurement data. Twenty-two people did not have complete follow-up information on fall outcome data, leaving a sample of 182 participants for analysis in this study. Reasons for missing follow-up data were: no fall calendars submitted (n=15), dropped out because not interested in participating anymore (n=6), and deceased (n=1).
Assessment
The baseline comprehensive fall assessment utilized the basic version of the interRAI Community Health Assessment (CHA), a subset of the Minimum Data Set for Home Care (MDS-HC) version 2.0.10,11 Reliability and validity of all items of the MDS-HC in community settings have been reported.12,13 A study supplement included the Berg Balance Scale (BBS), a valid and reliable scale for the assessment of balance in older adults.14–16 The BBS consists of 14 balance tasks scored on a scale of 0 to 4, where 0 indicates the inability to perform a task and 4 indicates that the task is performed independently. The maximum possible score of 56 indicates no identified balance difficulties.
The individual items of the BBS represent assessment of different aspects of balance, including maintaining a fixed position, dynamic balance, and movement over a fixed base. No single measurement tool can evaluate all aspects of or mastery of balance control, but it may be possible that a single tool can optimally identify people at an elevated risk for falling in a screening format. Individual items of the BBS, which have the potential to be independent measurement tools to predict falls in a screening situation, and additional measures of postural stability from the mailed questionnaire and interRAI assessment were available within the data set for analysis. The present analysis, with a priori specified associations for exposure and confounding, evaluated whether simple and easy methods of assessing postural stability were independently associated with the outcome of any fall. The following clinical assessment methods were used to evaluate balance.
One-leg stance with eyes open.
Each participant was asked to stand on his or her preferred leg. The ability to assume the position independently and maintain it for longer than 10 seconds was recorded. The need for assistance to assume the starting position or inability to maintain the position for 10 seconds was considered impaired ability. The validity and reliability of this test have been established.17–22
Tandem stance with eyes open.
The heel of one foot was placed in front of and touching the toes of the other foot. The ability to independently assume this position and maintain it for 30 seconds was recorded. The need of assistance to assume the starting position or inability to remain in balance for 30 seconds was considered impaired ability. The validity and reliability of this test have been established.18,22–25
Observational gait assessment.
The participant was asked by the assessor to ambulate at a comfortable speed, and the presence of any of the following features during an observational gait assessment was used to denote an unsteady or impaired gait: steppage; leg-length discrepancy; waddling, antalgic, ataxic, spastic, or frontal-lobe gait; or vestibular or parkinsonian movements.
Limits of stability in standing.
The participant was asked to lift an arm to 90 degrees with fingers stretched out in front and then, without changing foot position, reach as far forward as possible. The maximum distance, measured as the distance forward that the finger reached while the individual was in the most forward lean position, was recorded. A distance of less than 25.4 cm (10 in) was denoted as impairment. The validity and reliability of this test have been established.22,24,26–29
Self-report of balance impairment.
On a self-administered questionnaire, the following question was asked, “Sometimes people get dizzy or light-headed, and lose their balance. Other people report a loss of balance in their legs. Do you ever feel that you are losing your balance, other than when you feel dizzy or light-headed? By that, we mean do you feel the problem is in your legs rather than your head?”30 A “yes” response to this question denoted impairment.
A new variable was created as the summed number of balance tests with impairment from the individual tests of one-leg stance, tandem stance, limits of stability, and unsteady gait. Each test represents an evaluation of a different aspect and challenge to postural stability; the association of the composite may be superior to each test alone.
Data Collection
Prospective information on daily falls was collected for 12 months by each participant's monthly submission of a falls log calendar. A fall was defined as the person coming to rest unintentionally on the floor or ground. An injurious fall was defined as a fall resulting in an injury that required the person to see a physician. Participants who indicated falling in the previous month were interviewed by telephone to obtain detailed information about the specifics of the fall. Data collection for the baseline comprehensive geriatric assessment was started in May 2002, and collection of 1-year follow-up information on prospective falls was completed in January 2004.
Data Analysis
Descriptive statistics for study participant characteristics using data from the questionnaire and comprehensive geriatric assessment, including demographics and measures of functional ability, were calculated.
First, a univariate analysis between each measure of balance and the dependent variable (any fall, treated as a dichotomous variable) was performed to generate unadjusted relative risk (RR) estimates and 95% confidence intervals (CIs). Adjustment for confounding, based on clinical significance and previously identified fall risk factors, was accomplished by controlling for age (≥80 years), sex, number of prescription medications (≥4), history of falls in the previous 12 months, and the treatment group in the PPFV. Analysis of multicollinearity among the independent variables was performed using the phi coefficient (Φ) to test for correlation among categorical variables. For pairs of variables that were highly correlated (Φ≥.30),31 only one variable was entered into the modeling procedure, and then the effect of placing correlated variables in the same model was evaluated. The same analyses were repeated for the outcome of any injurious fall, also treated as a dichotomous (yes/no) variable.
The presence of multiplicative effect measure modification was evaluated for each model, prior to the calculation of adjusted effect measure estimates, using logistic regression. Regression diagnostics were performed on the full adjusted analyses using the Hosmer-Lemeshow test for goodness of fit. As falls were a common outcome, the adjusted odds ratios will overestimate the RR; thus, a modified Poisson regression model was applied to the final logistic models to directly obtain adjusted relative risk values.32
The presence of biologic interaction, as described by Rothman and Greenland,33 among balance exposure, history of falls, and number of prescription medications and their association with the outcomes of any fall and any injurious fall was evaluated. The calculation of the biologic interaction necessitated the creation of a composite variable with 4 levels for balance exposure/history of falls and balance exposure/number of prescription medications: (ab, unexposed to both factors; Ab, exposed to non–balance factor only; aB, exposed to balance factor only; and AB, exposed to both non–balance factor and balance factor).34 The RR was calculated for each category using a modified Poisson regression adjusting for age, sex, and treatment group in the PPFV. A biologic interaction effect was defined as the departure from additivity of the absolute effects, and the excess RR caused by the interaction (RERI) of the 2 terms was calculated using the equation notation from Hosmer and Lemeshow34:
An RERI of zero means no interaction, but an RERI of 0.5 would indicate that, because of interaction between the 2 variables, the RR is 0.5 greater than expected, based on the addition of the 2 risks.34 The percentage of falls among those participants with both exposures that was attributable to the interaction (AP(AB)%) was calculated as follows34:
All analyses were performed using SAS, version 8.2.*
Results
Baseline characteristics of the sample are presented in Table 1. In the complete sample, the average age was 79.9 years (SD=4.7) and 127 participants (70%) were male. Over the 1-year follow-up, 78 people (43%) fell, of whom 54 (30%) sustained an injurious fall and 32 (18%) had recurrent falls (≥2 falls).
Baseline Characteristics Presented for the Whole Sample (n=182) and Then Stratified by Fall Status at the End of the Study
Different balance measurement tools identified different numbers of people as impaired (Tab. 2). The assessment of both one-leg stance and tandem stance, which produce the narrowest base of support, identified the largest number of participants as having deficits (63% and 43%, respectively). The distribution of deficits per score level of the sum of balance impairments variables were: (1) 85% had impairment in one-leg stance; (2) 88% had impairment in one-leg stance and tandem stance; and (3) 64% had combined impairment in one-leg stance, tandem stance, and limits of stability, and another 28% had combined impairment in one-leg stance, tandem stance, and unsteady gait. Fallers had a higher prevalence of all measured impairments compared with nonfallers. A trend of decreasing total BBS score was demonstrated, with an increase in number of impaired test results.
Frequency of Impairment in Balance by Clinical Measurement Scale Assessed at Baseline of the Study for the Complete Sample and Stratified by Fall Status at the End of the Study, and Mean Total Berg Balance Scale Scores for People Identified as Impaired and Nonimpaired in Each Category
In unadjusted analyses, each method of assessing balance demonstrated that impairment was associated with an increased risk for any fall, with only tandem stance not statistically significant. Multiple measures of postural stability were not included in the same regression model due to the variables being highly correlated. The adjusted RR estimates demonstrated a statistically significant association with increased risk for any fall for self-report of balance problems (RR=1.58; 95% CI=1.06, 2.35), one-leg stance (RR=1.58; 95% CI=1.03, 2.41), and limits of stability (RR=1.46; 95% CI=1.02, 2.09) (Tab. 3). For the outcome of any injurious fall, although all RR estimates were associated with an increased risk, only the RR estimate for self-report of balance problems (RR=1.95; 95% CI=1.15, 3.31) was statistically significant. The number of tests positive for an impairment (one-leg stance, tandem stance, limits of stability, and unsteady gait) demonstrated a statistically significant association per unit increase in the score (from 0 to 4) for both outcomes, such that for each test with impairment, the risk for any fall and for any injurious fall increased by 20% and 23%, respectively.
Unadjusted and Adjusted Relative Risk Estimates (With 95% Confidence Intervals) for Balance Impairment as Measured by 5 Different Methods on the Outcomes of “Any Fall” and “Any Injurious Fall” Calculated From a Modified Poisson Regression
The presence of multiplicative effect measure modification was evaluated, and the estimated effect of interaction was not statistically significant for either outcome of any fall or any injurious fall. The presence of biological interaction, or additive interaction, was evaluated between the balance exposures that were significant in adjusted analyses and the risk factors of history of falling in the previous 12 months and the number of prescription medications (≥4). Due to the interaction, the RR was greater than expected, based on the addition of the 2 factors alone for the outcome of any fall: self-report of balance problems and history of falls, self-report of balance problems and number of prescription medications, one-leg stance and number of prescription medications, and limits of stability and number of prescription medications (Tab. 4). For the outcome of any injurious fall, the RR was greater than expected for the 2-factor combinations of self-report of balance problems and history of falls and self-report of balance problems and number of prescription medications.
Values of Excess Relative Risk and Percentage of Falls Among Study Participants With Both Exposures That Was Attributable to Their Interaction
A sensitivity analysis to evaluate the robustness of the measures of association using the best- and worst-case scenarios was performed by assuming that all individuals with missing fall outcome data either had or did not have the outcome of falling. The magnitude of association in the adjusted analyses did not change by greater than 10%, except for unsteady gait, which changed by 13% assuming all participants with missing data did not fall. The measures that were statistically significant in the final adjusted analyses remained statistically significant in the scenario that people with missing fall outcome data did not fall for both fall outcomes.
Discussion
The clinical assessment of balance is used as a means for measuring the integrity of the postural stability system, which involves the integration of information from somatosensory, musculoskeletal, visual, and vestibular systems and cognition. This study demonstrated the clinical assessment of balance impairment using simple and easy-to-administer measures encompassing self-report of balance problems, one-leg stance, and limits of stability, which were shown to be independent predictors of any fall over the subsequent year. None of the clinical assessment scales alone was associated with an injurious fall, although the self-report of balance problems was associated with an increased risk. When multiple clinical assessment scales were considered together, the risk for any fall or any injurious fall increased as the number of tests indicating impairment increased.
As systematic reviews have been limited in their ability to recommend specific measurement tools, this study presents important information that can be transferred directly to clinical practice to facilitate the identification of community-dwelling older adults at an elevated risk for falling. As this population of older adults was high functioning with very few functional limitations, the test maneuvers that identified future fall risk were challenging tasks that would decrease a ceiling effect to facilitate exposing deficits. Additionally, the time requirement to administer the complete BBS may not be possible for a screening situation, but the identified task items could be performed in this situation. Administering multiple clinical assessment scales also may be seen as a limitation for implementation in a screening situation, but it does provide a more comprehensive evaluation of different aspects of postural stability, and the number of tests with impairment is associated with both fall outcomes.
Importantly, not all of the methods of evaluating balance impairment are independently associated with future falls. The use of an observational gait assessment and tandem stance were not statistically associated with an increased risk for any fall and should be discouraged from use in a screening situation to identify individuals who are at risk for falls. If the intention of assessment was to identify risk for an injurious fall, none of the clinical tests was statistically significantly associated with this outcome. This study also demonstrated that the number of people identified as having balance impairment varied with the measurement tool used to clinically assess balance. The measurement tools, therefore, are not interchangeable or equivalent in defining an at-risk population. This information has implications for clinical practice, as well as for the comparison of results across studies.
Some prospective studies have shown the assessment of one-leg stance not to be an independent predictor of future falls.35–40 There are some methodological issues with the regression modeling procedures that may have affected the findings. The lack of an a priori objective to evaluate one-leg stance, a variation in the threshold used to denote impairment, and lack of evaluation of multicollinearity among covariates may have influenced this measurement tool being found not significant in adjusted analyses. Conversely, Vellas et al21 did find the assessment of one-leg stance time to be an independent predictor of injurious falls. The present study adds further information to support the assessment of one-leg stance time as independently associated with falls in a sample of individuals who are high functioning. Bergland et al40 found one-leg stance was not associated with injurious falls in a sample of older women.
The self-report of balance problems is on par with clinical measures of balance as a predictor of future fall risk. The self-report of balance problems was statistically significantly associated with history of falls, both for any fall and for any recurrent falls, and may not add additional information than that obtained from a report of a fall history. That said, the self-report of balance problems appears to be an important piece of information that should be sought during a self-report history for its association with any fall and any injurious fall. A positive report of problems should be enough to warrant further comprehensive evaluation and may be such an important factor, as it encompasses problems experienced over a prolonged period of time and in many situations that cannot be captured in a single test.41 Covinsky et al42 also found that the self-report of either dizziness or balance problems was associated with increased fall risk. They found an odds ratio of 1.83, which may overestimate the RR, and the inclusion of symptoms of dizziness does not separate problems of balance from potential cardiac-mediated symptoms. The findings from our study reinforce the importance, both for its inclusion and its diagnostic weight, of the self-report of balance problems in screening community-dwelling older adults.
Testing the limits of stability in this study was equivalent to the Functional Reach Test (FRT). The present findings are consistent with previous work that showed the FRT is associated with future fall risk.27 The FRT, evaluated as a continuous variable of the maximum reaching distance (in centimeters), has not been found to be associated with falls in adjusted analyses.37,40,43 Duncan et al27 found a distance of less than 25.4 cm as the critical threshold that denotes the onset of compromised ability, a finding corroborated in the present study.
The regression modeling maintained power to find an association, but potential reasons for the lack of an association include: validity of the measurement tool; the threshold value; chance; balance impairment being on the causal pathway for falls, with a resultant attenuation of association; and the inability of the measurement tool to detect a response to a challenge of the postural control systems. The analyses in this study used an a priori causal model where balance was treated as the sum result of multiple systems; control over confounders deliberately did not include variables such as vision impairment or lower-extremity muscle weakness, as their effects are subsumed within the measure of balance. Power was acceptable for the outcome of any injurious fall, but sensitivity analyses did not change the magnitude and statistical significance of the findings. We have confidence in the results found for this outcome.
The evaluation of biologic interaction was a novel contribution to the falls literature, establishing that the joint effects of some variables exceed that expected from the sum of the individual components. Information on biologic interaction has important implications for the prevention and treatment of falls. If the joint additive effect of 2 factors is in excess of the sum of the individual effects, then reduction in either factor would reduce the risk of the other factor producing falls. The strength of the interaction was not consistent across the measures of balance impairment. The self-report of balance problems and the limits of stability when each measure was combined with the measure of polypharmacy produced the strongest evidence of effect measure modification. This finding suggests that a review of medications should be a standard procedure to limit fall risk, particularly in people with identified balance impairments.
The present study's participants represented a relatively high-functioning group of older adults with few functional limitations, although fall events were still a common occurrence over the follow-up period. Risk for any fall in this sample (43%) was higher than the population average of 30% to 35%. Risk for any fall may have been higher in this group of older adults because they participated at their usual level of functioning, not curtailed or modified activity, which may have exposed them to more situations that could result in a fall. In our study, we used a general definition of an injurious fall, including potentially minor and major injuries, which resulted in a fall risk of 30%. Comparison with other studies on fall injuries is difficult due to the many ways that injuries are reported in the literature. Bergland and Wyller,40 however, found that 24% of falls resulted in an injury, which is comparable to values obtained in our study, but is surprising as their sample comprised only older women.
A potential limitation of the present study is that prospective falls data were derived from an intervention study. The participants' exposure to falls prevention information may have produced a decrease in the number of falls, attenuating the association between balance impairments and new falls and producing conservative estimates. The use of data from 2 arms of an intervention study is not a design flaw, as there was no detected treatment effect from the intervention, which provides reasonable grounds to combine data for prognostic research.42 The treatment group in the PPFV was included as a covariate in the adjusted analyses to fully account for any differences that the intervention group may have introduced. Sensitivity analysis including and excluding the PPFV treatment group variable did not change the statistical significance of associations or the magnitude of association in any important way.
Another potential limitation is the lack of generalizability of the results to the target population of all older adults at risk for falling. First, the sample may not be representative of the general population of community-dwelling older adults because they were volunteers. The men were from a representative sample of male veterans, but they may have been different from their contemporaries who were not veterans due to possible unique health consequences from active military service. This study used the gold standard for collecting fall occurrence information, but there is still a possibility of underreporting of fall events, in particular injurious falls. The self-report of injurious falls was not validated with medical records or documentation of a clinic visit for medical attention, and thus the RR estimates may be conservative. As all individuals volunteered to participate in the study, we felt there would not be a reporting bias by sex, and we are unaware of any research that would suggest male veterans would be differentially inclined to underreport or overreport falls compared with the average male older adult. Participants were aware the study was supported by Veterans Affairs Canada, and this partnership may have further ensured accurate reporting of events.
The strengths of the study include the prospective design and the use of a reliable method to collect falls data in a large sample of community-dwelling older adults.44 The contact with study participants when falls were identified on calendars enhanced information collection, and the contact with participants who did not submit calendars aided in maximizing follow-up of study participants. Finally, all participants received the same standardized, comprehensive geriatric assessment at baseline.
Conclusions
The self-report of balance problems, the one-leg stance test, and the limits of stability test are associated with increased risk for any fall in community-dwelling older adults who are high functioning. The self-report of balance problems also is strongly associated with an individual sustaining an injurious fall. Balance impairment in the presence of polypharmacy increases fall risk beyond simple addition of the individual risks producing biologic interaction. A finding of impairment using the thresholds established in this study for these simple and easy-to-administer screening measures indicates an individual who should be referred for further comprehensive fall risk assessment and interventions to prevent falls.
Footnotes
-
Dr Muir provided concept/idea/research design and data analysis. All authors provided writing. Dr Speechley provided data collection and fund procurement.
-
A poster presentation of this research was given at the XIXth International Congress of the International Association of Gerontology and Geriatrics; July 5–9, 2009; Paris, France.
-
This study was a secondary analysis of data from phase II of the Project to Prevent Falls in Veterans (PPFV). The PPFV, funded jointly by Veterans Affairs Canada and Health Canada, was approved by the University of Western Ontario Research Ethics Board for Health Sciences Research Involving Human Subjects.
-
↵* SAS Institute Inc, PO Box 8000, Cary, NC 27513.
- Received May 19, 2009.
- Accepted September 15, 2009.
- © 2010 American Physical Therapy Association
References
Issue highlights
