Falls continue to be a major concern among older persons and clinicians who care for them. One in 3 adults aged 65 years or older falls,1 with 20% to 30% of these individuals suffering moderate to severe injuries.2 Even if individuals are uninjured by a fall, between 3% and 85% of them develop a fear of falling or loss in balance confidence.3 Still others may not have fallen themselves but lack confidence.4 Fortinsky and associates5 labeled these individuals as timid in that they had low confidence but were not otherwise at risk for falls (Table 1). Scheffer and associates3 reported that the main risk factors for developing a loss of balance confidence are at least 1 fall, being female, and being older. With a loss in balance confidence, individuals often limit their activity leading to a decline in physical and mental performances, an increased risk of falling, and progressive loss of health-related quality of life. Furthermore, some individuals may have high balance confidence despite multiple falls and poor balance performance. In other words, they should be afraid of falling but are not. Fortinsky et al5 termed these people overconfident. If clinicians do not routinely assess fear of falling or balance confidence, terms often used interchangeably,6 they might not be aware of this paradox.5 Two commonly used scales that measure an individual's self-efficacy regarding balance are the Activities-Specific Balance Confidence (ABC) Scale7 and the Falls Efficacy Scale (FES).8
Balance ability contributes to functional mobility because an individual controls his or her upright posture over a moving base of support.9 The Dynamic Gait Index (DGI), Functional Gait Assessment (FGA), and Timed Up and Go (TUG) are commonly used to assess functional mobility and dynamic balance ability in older participants. In addition, the DGI, FGA, and TUG have often been used to determine fall risk. Well-documented and accepted cutoff scores for fall risk have been established for the DGI and the FGA.10–12 Clinicians commonly use 13.5 seconds as the cutoff score for the TUG,13 inappropriately applied to the normal-paced version. Several other cutoff scores have been suggested for the normal-paced TUG, but none of these scores have been validated.1,12,14
A fourth test of functional mobility, the L test, is not widely known or used but may provide a better assessment of fall risk than the TUG, especially for high-functioning older adults in the community. The L test has been studied in young participants with amputations15 and frail older adults.16 No published performance data are available for community-dwelling older adults. In addition, the association between L test performance and fall risk is not known, nor has a cutoff score been determined. With the exception of the L test, we know that functional mobility performance declines with age but the contribution of balance confidence to functional mobility is unknown. In addition, the association of balance confidence and fall risk is not widely recognized.
Therefore, the purpose of this study was 3-fold. First, we wanted to determine the contribution of age and balance confidence on selected functional mobility outcome measures in a sample of older adults living in the community. Because these functional mobility outcome measures are often used to determine fall risk, a second purpose was to determine the association between balance confidence and fall risk. Although both the FGA and DGI have well-established cutoff scores indicating fall risk, the normal-paced TUG and L test do not. Therefore, a third purpose was to propose cutoff scores with adequate diagnostic accuracy for these 2 outcome measures.
This study was a prospective cohort design with 3 age groups. Beginning at about 60 years of age,17,18 balance confidence and functional mobility decline with age. Therefore, we consecutively recruited participants in 3 cohorts (60-69, 70-79, and 80 years and over). Prospective participants lived in urban, suburban, and rural north Texas. In addition, they were community dwelling and apparently healthy without any short-term-medical conditions. Participants were eliminated if they could not follow simple commands or ambulate without an assistive device. Each participant (N = 118) provided written informed consent in compliance with institutional review board requirements at the Texas Woman's University. Because some cases included missing items, data from only 105 participants were used for analyses.
Testers were 6 second- and third-year physical therapy students. They were taught, practiced, and tested on all measures as part of their entry-level curriculum and had completed at least one 6-week clinical experience. In addition, the authors reviewed all standardized procedures with the testers and supervised data collection.
Falls and Balance Confidence
Participants completed a demographic questionnaire that confirmed inclusion and exclusion criteria and included questions regarding fall history. A fall was defined as an episode of unintentionally coming to rest on the ground or lower surface that was not the result of dizziness, fainting, sustaining a violent blow, loss of consciousness, or other overwhelming external factor.19
To divide participants into high and low balance confidence groups, the ABC Scale was chosen over the FES for 2 reasons. The FES has noted ceiling effects in community-dwelling samples,20 and the ABC Scale includes items performed outside the home appropriate for community-dwelling populations.4 The ABC Scale was administered but not scored so that testers were blinded to the ABC score. For analysis, participants were divided into high and low balance confidence groups. High balance confidence was defined as a score of 80% or greater on the ABC Scale,4 whereas low balance confidence was as a score of less than 80%.
Physical Performance Assessment
The outcome measures were administered in random order and scored after all tests were completed. The outcome measures included the DGI, FGA, TUG, and L test. Both the DGI and FGA assess balance during gait activities. The DGI was developed specifically for use with older adults.10 The 8-item test assesses the effects of varying demands, such as walking at different speeds or walking with head turns, on a person's balance. It has been studied with older adults and people with vestibular disorders. Individuals who score less than 19 on the DGI are at increased risk for falls.10,11 The DGI cutoff score served as the fall reference standard for determining diagnostic accuracy. The FGA is a 10-item assessment that is based on the DGI.21 The authors developed the measure to identify walking limitations not detected by the DGI in younger participants with vestibular problems. This test may be appropriate for assessing balance during high-level gait activities in all age groups and other diagnoses. Individuals living in the community, who score 22 or less on the FGA are at increased risk for falls.12
Both the TUG and L test assess functional mobility including transitional movements. The TUG specifically measures the time in seconds it takes for individuals to stand from a chair, walk 3 m at their normal pace, turn, and return to sitting.22 Past research has indicated that people who can perform this task in less than 20 seconds are able to function independently. Authors have suggested an 11-second12 or 13-second14 cutoff score for the normal-paced version; however, these cutoff scores have not been validated or widely used. Other authors have suggested 13.5 seconds as indicating a risk for falls; however, this cutoff is based on the fast version of the TUG.13 Many clinicians apply the 13.5-second cutoff to the normal-paced TUG. The tool has been used in various settings and is often used with participants, who have long-term conditions. In some populations, such as very fit older adults or younger people with a unilateral problem, the TUG is too easy and does not provide useful information for the clinician.
The L test was developed by clinicians working with individuals following a lower extremity amputation to challenge the younger, more fit participants during gait activities.15 Test procedures are the same as for the TUG with the addition of 2 tasks, a right and left turn. To perform the test, a person stands up from a chair, walks 3 m, turns either right or left, walks 7 m, turns 180°, and walks the same path to return to sitting. The score is the time in seconds that it takes a person to complete the sequence. This test has been studied only with people who had lower extremity amputations15 and hospitalized frail older adults.16 It is not known how people in the community perform on this test.
All data were analyzed with SPSS 17.0 software (SPSS Inc, Chicago, Illinois). Descriptive analyses were conducted to describe the participant characteristics and performance. On the basis of the work by Fortinsky and associates,5 participants were classified as timid, congruent, or overconfident with respect to their balance confidence and fall risk as determined by physical performance (Table 1). The Pearson product moment coefficient of correlation determined the relationship among the mobility tests and age and ABC score. Three linear regression models were computed for each outcome measure. In model 1, the univariate association between each outcome measure and age was obtained. In model 2, the univariate association between each outcome measure and balance confidence was examined. In model 3, multivariate analysis revealed the contribution of both predictor variables to outcome measure performance. Multivariate analysis of covariance controlling for age tested for differences in functional mobility between participants with high and low confidence. Finally, we explored the possibility of a normal-paced TUG and L test cutoff score indicating fall risk as determined by the DGI cutoff (<19). Specifically, we identified the receiver operating characteristic (ROC) curve point that balanced specificity and sensitivity. We also estimated the area under the curve to measure the model's fit (C statistic). Higher C statistic values indicate a better model fit with values ranging from 0.5 to 1.0.
Of the 118 participants enrolled in the study (2006-2009), 13 did not complete all tests; therefore, the data analysis was performed on the remaining 105 individuals (mean age 75.7 [9.8] years). On the basis of 2010 census data, our sample was similar to older Americans in the US population with respect to sex, marital status, and living situation.23 The percentage of participants who engaged in regular exercise was comparable with that reported in the Gallop-Healthways 2013 poll for adults aged 18 years and older.24 Likewise, the percentage of our sample, who took 5 or more medications, is similar to that reported by Qato et al.25 Thirty-seven percent of our sample expressed a fear of falling compared with 45% in community-dwelling older adults studied by Oh-Park et al.26 Table 2 provides sample characteristics, and Table 3 provides a summary of performance on the outcome measures of participants in each decade. No adverse events were associated with testing. Performance across all tests declined with age. The correlation analysis revealed moderate to strong relationships among the variables.27 As expected, the strongest relationships were between measures, such as the DGI and FGA or the TUG and L test (Table 4).
The best single predictor of functional mobility performance across outcome measures was age (Table 5). The univariate analyses (model 1) tested the question, does age alone affect older adults' functional mobility as measured by each outcome measure better than chance? The F value for all models was significantly different from zero. Therefore, as participant age increased, functional mobility decreased. Because balance confidence had similar but slightly weaker correlations, we performed a second set of univariate analyses, model 2, which tested the question, does balance confidence alone affect older adults' functional mobility as measured by each outcome measure better than chance? The F value for all models was significantly different from zero. Therefore, as participant balance confidence increased, functional mobility also increased. The R 2 explains the extent of the association between the variables of interest and the outcome measures. The resulting R 2s for both age and balance confidence were similar across outcome measures (0.293-0.450).
The multivariate analyses (model 3) tested the question, does the combination of age and balance confidence affect older adults' functional mobility as measured by each outcome measure better than chance? The F value for all models was significantly different from zero. Age was still the best predictor of functional mobility. However, when both variables were considered together, the percent change in R 2 increased by 16% to 28%.
Participants with low balance confidence were older than those with high balance confidence. Therefore, we conducted a multivariate analysis of covariance with age as the covariate to test for differences in functional mobility between participants with high and low balance confidence. Participants with low balance confidence still performed worse on all outcome measures compared with those with high balance confidence (Table 6).
Table 7 depicts the association between balance confidence and fall risk. Depending on the outcome measure, anywhere from 16% to 30% of participants were incongruent, meaning that their balance confidence and physical performance-based fall risk did not match.
Contingency tables were constructed to compare the diagnostic accuracy of the normal-paced TUG and the L test. Figure 1 is a flow diagram illustrating how the contingency table results were determined for the normal-paced TUG as an example. We sought optimum balance between test sensitivity and specificity. Table 8 is the final contingency table with 12 seconds or more as the fall risk cutoff score for the normal-paced TUG, and 25.5 seconds or more for the L test. The ROC analysis supported these fall risk cutoff scores (Figure 2). The area under the curve (C statistic) was 0.88 and 0.89, respectively, suggesting a high moderate accuracy for both models.28 On the basis of the proposed TUG cutoff, we found that 43.8% of our participants were at risk for falls (95% confidence interval [CI], 34.3%-53.3%). For the proposed L test cutoff, 41.9% of our participants were at risk for falls (95% CI, 32.5%-51.3%).
The current health care environment requires outcome measures to compare individuals' data to expected age-related performance for the purpose of setting goals, monitoring progress with intervention, and predicting other variables such as fall risk. Anecdotally, many clinicians focus only on a participant's mobility performance related to age norms and do not consider balance confidence when making clinical decisions. We confirmed that functional mobility does decline with age (Table 4), supporting the need to regularly assess physical performance. However, our findings also indicate that balance confidence contributes to functional mobility performance and the 2 constructs do not match (are incongruent) about 25% of the time. For example, Table 7 shows that 17 participants with low balance confidence on average performed better than the published cutoff score for the DGI. Alternatively, 8 participants with high balance confidence on average performed worse than expected. As suggested by Fortinsky and associates,5 this mismatch between balance confidence and physical performance may contribute to fall risk.
One cannot discount the psychological aspects of fall risk because psychological trauma may cause individuals to limit their activity in the same way as physical trauma.8 Conversely, some individuals are overconfident, do not limit their activities, and therefore take unnecessary risks. Fortinsky and associates5 explored the alignment between balance confidence and fall risk with a special focus on those whose confidence seems incongruent with their risk. The authors classified participants into the following 3 categories: timid (least confident and low risk for falls), congruent (aligned confidence and fall risk), and overconfident (most confident and at moderate to high fall risk). Fortinsky et al did not use any physical performance measures but relied solely on fall history to determine fall risk. We expanded on their study by aligning balance confidence and fall risk with physical performance measures in addition to expressed fear of falling and fall history requiring medical attention. Our findings suggest that one can explain more of the variance in functional mobility when both age and psychological measures are used. Data in Tables 3 and 6 show that participants with low balance confidence consistently performed worse than those with high balance confidence even when age was controlled. Given our findings, to assess fall risk, one must look beyond physical performance and take into account age and balance confidence (Table 5). However, age and balance confidence alone do not explain the majority of the variance in functional mobility performance. Within the framework of a biopsychosocial model, other factors such as strength, range of motion, pain, and motivation that we did not measure may explain the remaining variance in functional mobility. Future studies should address other possible contributing factors.
Nevertheless, many clinicians seek the one best outcome measure for a particular purpose. In writing about diagnostic tests, Fritz and Wainner29 caution readers that diagnostic tests cannot be deemed good or bad; their value is predicated on the unique presentation of different individuals. This concept applies to both clinical and prevention practices. Our population of interest is community-dwelling older adults. Determining who is at risk for falls is a high public health priority, given that unintentional injury is the ninth leading cause of death among older Americans,30 and just over half of these deaths can be attributed to a fall. In our prospective cohort design, we were blind to the participants' fall history. Because we were unable to follow our participants for a long term to document actual falls, we chose the DGI as the reference standard given its well-established study-based threshold for fall risk in community-dwelling older adults.10,11 Although some study-based cutoff scores have been suggested for the TUG, the commonly used 13.5-second cutoff is inappropriately applied to the normal-paced TUG protocol and may be too slow for the fast-paced TUG as suggested by Jernigan et al.31 Other authors have suggested 13 seconds14 and 11 seconds,12 but these values are based on less than ideal methodologies and clinical judgment. The L test has very limited clinical or prevention usefulness to date because it does not have age-related reference values or a study-based cutoff score. Therefore, our final purpose was to propose cutoff scores that could be validated in future longitudinal studies using the number of falls as the reference standard. Our proposed cutoff of 12 seconds or more for the normal-paced TUG and 25.5 seconds or more for the L test to indicate fall risk among community-dwelling older adults is based on the optimal design for examining a diagnostic test.29 The area under the curve (C statistic) for the ROC curves indicates that our models and proposed cutoff values are accurate. The fall risk positive and negative predictive values for both the TUG and L test are similar. Nothing in our findings suggests that either test should be abandoned in favor of the other. Rather, Deathe and Miller's15 rationale for development of the L test should guide clinicians in test selection. The L test is more appropriate for older adults who are more physically fit, where the need for a more demanding test that includes both left and right turns is necessary.16 Now, with an available cutoff score and age-related performance values, the L test is clinically useful. Clinicians with similar higher-functioning populations would be wise to use the L test in place of the TUG.
The reference standard for determining cutoff scores (DGI < 19) for the normal-paced TUG and the L test may be a limitation of our study. The number of actual falls over a long period of time (6-12 months) is the actual gold standard and would be a better reference standard for determining cutoff scores for these tests. However, following our participants for a longer period of time was not practical. Therefore, we used the DGI cutoff score that has been widely used and validated in community-dwelling older adults.10,11 We did ask the participants about past falls requiring medical attention, but the numbers of individuals who reported this type of fall were few. To compare outcome measures with a reference standard, all participants must have data for the outcome measure of interest and the reference standard.29 If we had used the number of falls requiring medical attention as the reference, our sample size would have been reduced to 22 instead of 105. Although not ideal, the DGI was the best choice to ensure an adequate sample size. Future studies should validate the proposed cutoff scores using a prospective cohort design with actual falls as the reference standard. Furthermore, the proposed cutoff scores for identifying persons at risk for falls are valid only in community-dwelling older adults. These cutoff scores may not be appropriate for determining fall risk in patient populations or frail elders living in institutional settings. A second limitation of this study is that we did not measure the reliability of the testers. This limitation was mitigated by standardized procedures and supervision by the authors. Blinding of testers controlled for any bias.
Older adults will benefit if clinicians assess balance confidence and functional mobility. In addition, functional mobility and confidence may be incongruent, further supporting the need to formally assess balance confidence.
The authors thank the graduate student research team and the older adults who participated in data collection for their time and effort in completing this study. Both the authors were equally responsible for the study concept, study design, analysis and interpretation of data, and article preparation.
1. Hausdorff JM, Rios DA, Edelberg HK Gait variability and fall risk in community living older adults: 1-year prospective study. Arch Phys Med Rehabil. 2001;82(8):1050–1056.
2. Alexander BH, Rivara FP, Wolf ME The cost and frequency of hospitalization for fall-related injuries in older adults. Am J Public Health. 1992;82(7):1020–1023.
3. Scheffer AC, Schuurmans MJ, van Dijk N, et al. Fear of falling: measurement strategy, prevalence, risk factors and consequences among older persons. Age Ageing. 2008;37(1):19–24.
4. Myers AM, Powell LE, Maki BE, et al. Psychological indicators of balance confidence: relationship to actual and perceived abilities. J Gerontol A Biol Sci Med Sci. 1996;51(1):M37–M43.
5. Fortinsky RH, Panzer V, Wakefield D, et al. Alignment between balance confidence and fall risk in later life: has over-confidence been overlooked? Health Risk Soc. 2009;11(4):341–352.
6. Rand D, Miller WC, Yiu J, et al. Interventions for addressing low balance confidence in older adults: a systematic review and meta-analysis. Age Ageing. 2011;40(3):297–306.
7. Powell LE, Myers AM The Activities-specific Balance Confidence (ABC) Scale. J Gerontol A Biol Sci Med Sci. 1995;50A(1):M28–M34.
8. Tinetti ME, Richman D, Powell L Falls efficacy as a measure of fear of falling. J Gerontol. 1990;45(6):P239–P243.
9. Duncan P ed. Balance: Proceedings of the APTA Forum
. Alexandria, VA: American Physical Therapy Association; 1990.
10. Shumway-Cook A, Baldwin M, Polissar NL, et al. Predicting the probability for falls in community-dwelling older adults. Phys Ther. 1997;77(8):812–819.
11. Whitney SL, Hudak MT, Marchetti GF The dynamic gait index relates to self-reported fall history in individuals with vestibular dysfunction. J Vestib Res. 2000;10(2):99–105.
12. Wrisley DM, Kumar NA Functional gait assessment: concurrent, discriminative, and predictive validity in community dwelling older adults. Phys Ther. 2010;90(5):761–773.
13. Shumway-Cook A, Brauer S, Woollacott M Predicting the probability for falls in community-dwelling older adults using the Timed Up and Go Test. Phys Ther. 2000;80(9):896–903.
14. Medley A, Thompson M Usefulness of variations of the Timed Up and Go in apparently healthy individuals. Phys Occup Ther Geriatr. 2005;23(4):1–23.
15. Deathe AB, Miller WC The L test of functional mobility: measurement properties of a modified version of the Timed “Up & Go” Test designed for people with lower-limb amputations. Phys Ther. 2005;85(7):626–635.
16. Nguyen VC, Miller WC, Asano M, et al. Measurement properties of the L test for gait in hospitalized elderly. Am J Phys Med Rehabil. 2007;86(6):463–468.
17. Bohannon RW, Larkin PA, Cook AC, et al. Decrease in timed balance test scores with aging. Phy Ther. 1984;64(7):1067–1070.
18. Vereeck L, Wuyts FL, Truijen S, et al. The effect of early customized vestibular rehabilitation on balance after acoustic neuroma resection. Clin Rehabil. 2008;22(8):698–713.
19. Tinetti ME, Baker DI, Gottschalk M, et al. Systematic home-based physical and functional therapy for older persons after hip fracture. Arch Phys Med Rehabil. 1997;78(11):1237–1247.
20. Hatch J, Gill-Body KM, Portney LG Determinants of balance confidence in community-dwelling elderly people. Phys Ther. 2003;83(12):1072–1079.
21. Wrisley DM, Marchetti GF, Kuharsky DK, et al. Reliability, internal consistency, and validity of data obtained with the Functional Gait Assessment. Phys Ther. 2004;84(10):906–918.
22. Podsiadlo D, Richardson S The Timed “Up & Go”: a test of basic functional mobility for frail elderly persons. J Am Geriatr Soc. 1991;39(2):142–148.
23. Werner CA The Older Population, 2010. Washington, DC: US Department of Commerce, Economics and Statistics Administration, US Census Bureau; 2011.
25. Qato DM, Alexander GC, Conti RM, et al. Use of prescription and over-the-counter medications and dietary supplements among older adults in the United States. JAMA. 2008;300(24):2867–2878.
26. Oh-Park M, Xue X, Holtzer R, et al. Transient versus persistent fear of falling in community-dwelling adults: incidence and risk factors. J Am Geriatr Soc. 2011;59(7):1225–1231.
27. Portney LG, Watkins MP Foundations of Clinical Research Applications to Practice. 3rd ed. Upper Saddle River, NJ: Pearson-Prentice Hall; 2009.
28. Akobeng AK Understanding diagnostic tests 3: receiver operating characteristic curves. Acta Paediatr. 2007;96(5):644–647.
29. Fritz JM, Wainner RS Examining diagnostic tests: an evidence-based perspective. Phys Ther. 2001;81(9):1546–1564.
31. Jernigan SD, Pohl PS, Mahnken JD, Kluding PM Diagnostic accuracy of fall risk assessment tools in people with diabetic peripheral neuropathy. Phys Ther. 2012;92(11):1461–1470.