FALLS ARE COMMON, WITH UP TO 45% of communitydwelling

Transcription

1 1636 ORIGINAL ARTICLE A Pilot Study to Explore the Predictive Validity of 4 Measures of Falls Risk in Frail Elderly Patients Janet I. Thomas, MSc, Judith V. Lane, MSc ABSTRACT. Thomas JI, Lane JV. A pilot study to explore the predictive validity of 4 measures of falls risk in frail elderly patients. Arch Phys Med Rehabil 2005;86: Objectives: To test the hypothesis that scores on 4 falls risk measures will differ significantly in patients reporting recurrent falls compared with those who do not; and to explore the validity of each measure to predict such falls status. Design: A convenience sample was tested to establish the sensitivity and specificity of the Functional Reach Test, Timed Up & Go test, one-leg stance test (OLST), and balance subsection of the Performance Oriented Mobility Assessment (B- POMA). A 12-month retrospective falls history was used to identify recurrent fallers. Setting: A day hospital for the elderly. Participants: Convenience sample of 30 day hospital patients. The inclusion criteria were: ability to rise from a chair and walk 6m; no severe cognitive impairment or blindness; age 65 years or older. Interventions: Not applicable. Main Outcome Measures: Scores on the 4 tests and retrospective falls histories. Results: Scores on the B-POMA and OLST showed significant differences between fallers and nonfallers (P.05). An OLST time of 1.02 seconds or less (odds ratio [OR] 15.2; 95% confidence interval [CI], ) and B-POMA score of 11 or less (OR 18.5; 95% CI, ) were predictive of day hospital patients having a history of recurrent falls. Conclusions: OLST and B-POMA both have potential as screening tools for risk of falls, but this observation requires confirmation in a prospective study. Key Words: Accidental falls; Aged; Predictive value of tests; Rehabilitation by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation FALLS ARE COMMON, WITH UP TO 45% of communitydwelling elderly reporting at least 1 fall a year, 1 resulting in a number of health and social repercussions. Within the National Service Framework for Older People, 2 identification of those who are at risk of falling is a key issue. During initial evaluation of patients, validated tools ought to be used to ascertain the person s likelihood of falling due to physical From the Physiotherapy Department, Whitefield Day Hospital, Queen Margaret Hospital, Dunfermline, Fife (Thomas); and Department of Physiotherapy, Queen Margaret University College, Edinburgh (Lane), UK. No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the authors(s) or upon any organization with which the author(s) is/are associated. Correspondence to Janet Thomas, Physiotherapy Dept, Whitefield Day Hospital, Queen Margaret Hospital, Dunfermline, Fife KY12 0SU, UK, janet.thomas@faht.scot.nhs.uk. Reprints are not available from the author /05/ $30.00/0 doi: /j.apmr factors 3,4 and many such tools have been identified. 5 However, studies 6 have revealed that in 63% of case notes there was no record of a standardized test or measure being used. It is unclear why this is the case, but guidance on which test is most appropriate for use in a specific setting has been lacking, as well as clinical time in which to carry out the tests, which can be lengthy. No clear picture has emerged from studies reporting comparative assessments of falls risk measures Some studies have reported that no test could differentiate fallers 9,10 while others concluded that tests such as the Berg Balance Scale (BBS), 7,11 the Fast Evaluation of Mobility Balance and Fear (FEMBAF), 8 and the Elderly Mobility Scale 12 were effective in discriminating fallers from nonfallers. Many of the tests already mentioned, such as the BBS and the FEMBAF, take up to 20 minutes to complete, 13 making them less likely to become part of regular clinical assessment or a screening tool. Four validated falls risk measures were identified that take less than 5 minutes to complete: Functional Reach Test (FRT), 14 one-leg stance test (OLST), 15 Timed Up & Go (TUG) test, 16 and the balance section of the Performance Oriented Mobility Assessment (B-POMA). 17 Functional reach has good intrarater (intraclass correlation coefficient [ICC].92) 18 and interrater reliability (r.97). 19 Reported sensitivity is 62% 9 and specificity is 92%, 20 with suggested cutoff values of 22 to 25.4cm. 9,20 The OLST also has excellent interrater reliability (ICC.99). 21 A sensitivity of 95% and specificity of 58% with a cutoff time of 30s were identified with older outpatients. 22 The TUG developers report good inter- and intrarater reliability (ICC.99). 16 The sensitivity is 87% 23 and specificity, 100%. 9 Suggested cutoff points for identifying fallers have varied from 13.5 seconds 23 to 23.7 seconds. 11 The B-POMA has adequate interrater reliability (ICC.75). 24 The sensitivity of the B-POMA is excellent at 95% but the test has a very poor specificity (16%). 25 It is difficult to identify an accurate cutoff score due to the wide range of versions in use. 26 Despite these encouraging results, most studies have reported results from fit community-dwelling adults, with little research comparing the effectiveness of these measures in the frail elderly. Only one focused on patients from a specific health care setting: the day hospital. 12 Day hospitals are outpatient health care facilities in which multidisciplinary assessment, treatment, and rehabilitation are available on attendance for a full or part day for elderly people in the community. 27 Day hospitals have been suggested as a suitable setting in which assessment and rehabilitation of fallers can take place 27 and the percentage of patients with fall-related problems is approximately 29% of referrals. 28 The first aim of the study was to establish whether there were significant differences among the B-POMA, TUG, FRT, and OLST between frail elderly fallers and nonfallers. The secondary aim was to establish optimal cutoff points for predicting falls status in frail elderly patients using the B-POMA, TUG, FRT, and OLST when compared with a retrospective falls history.

2 PILOT STUDY PREDICTIVE VALIDITY FALLS RISK, Thomas 1637 Table 1: Subject Demographics Characteristics Fallers Nonfallers Mean age SD (y) Mean BMI SD (kg/m 2 )* % living in own home % females % using any walking aid % with general frailty or mobility problems % with a stroke % with orthopedic problems % with Parkinson s disease % with other musculoskeletal problems Abbreviations: BMI, body mass index; SD, standard deviation. *Significant at P.05. METHODS Participant Selection A convenience sample of 30 subjects was recruited from day hospital patients between December 2002 and February The inclusion criteria were: ability to rise from a chair and walk 6m; no severe cognitive impairment or blindness; and age 65 years or older. Subjects had been referred for a variety of reasons, and none had commenced rehabilitation prior to entry to the study. Ethics approval was obtained from the local research ethics committee and written informed consent was obtained from all participants. Data Gathering Measurements were taken in the order stated for the TUG, FRT, OLST, and B-POMA by a single investigator (JTT). A pilot study was undertaken to determine intrarater reliability for all 4 tests. A set protocol and instructions for each test, taken from the source articles, were used to minimize variation in test performance. During each test, subjects had 1 trial attempt to familiarize themselves with the procedure, followed by a 1 minute rest, before repeating the test for data gathering. A 2-minute rest was enforced before continuing with the next measure. All tests were carried out in usual footwear, with their usual walking aid during the TUG and B-POMA. The same equipment was used for all tests. The researcher was unaware of patients falls status during data gathering. A 12-month retrospective falls history was subsequently obtained by giving a description of a fall based on the Kellogg working party definition 29 and asking if the subject had fallen in the past 12 months. This self-report was later verified by inspection of the patients day hospital medical and nursing case notes, including general practitioner referral letters where available. Subjects were defined as fallers if they had 2 or more falls in the past year, because a single fall could have been due to overwhelming extrinsic factors. Information on age, sex, walking aid used, weight, height, diagnostic category, and place of residence was also collected. Data Analysis Intrarater reliability. The Shrout and Fleiss ICC 3,1 was used. 30 Descriptive statistics. Means and 95% confidence intervals (CIs) were calculated for normally distributed data (established with the Shapiro-Wilks test); medians and interquartile ranges (IQRs) were calculated for nonnormally distributed and ordinal data. Frequency counts were established for nominal data. Univariate analyses. Data were grouped as fallers and nonfallers. Frequencies were compared using the Fisher exact test. 31 Variables were compared using an independent t test (for normally distributed data) and the Mann-Whitney U test (for nonnormally distributed and ordinal data). Effect sizes and powers were calculated according to Cohen. 32 The level was set at.05. Multivariate analysis. Receiver operating characteristic (ROC) curves were plotted to determine maximum sensitivity and specificity 33,34 and cutoff points were defined. The cutoff points (and any other significant covariates) were then used in logistic regression analysis to establish the risk of being a faller or nonfaller. RESULTS Intrarater Reliability ICCs were in the range of fair to excellent 35 (TUG test.75; FRT.87; OLST.69; B-POMA.84). Demographics Thirty subjects consented to take part in the study, of whom 18 were fallers and 12 were nonfallers. Demographics are described in table 1. There was a significant difference in the body mass index (BMI) between the 2 groups (t 3.011, P.005, effect size [ES] 1.07). We found no significant difference in the proportion of fallers using a walking aid (Fisher exact test.066) or in sex of fallers (Fisher exact test.136). There were no other significant differences between the groups. Univariate Analyses A Shapiro-Wilks test indicated that the TUG (P.010) and OLST (P.010) were not normally distributed. Normal distributions, however, were found for the FRT (P.083) and B- POMA (P.295). Descriptive statistics are provided in table 2. Mann-Whitney U tests showed no significant difference between fallers and nonfallers for the TUG (U 67.0, P.087, power.34); however, significant differences existed for the OLST (U 39.0, P.003, ES.43). An independent t test showed a significant difference in the B-POMA (t 3.245, P.003, ES 1.05), but not in the FRT between fallers and nonfallers (t 2.024, P.053, power.46). Sensitivity and Specificity Sensitivity and specificity were calculated for each outcome measure for a variety of scores. The cutoff points that provided optimum sensitivity and specificity (as defined by ROC curve analysis) are shown in table 3. Logistic Regression Analysis Logistic regression analysis was used to determine odds ratios (ORs) of being a faller from the cutoff points identified by ROC analysis (table 4). Additional variables of BMI and any walking aid were also evaluated in the model. The only Table 2: Descriptive Statistics Test Fallers Nonfallers TUG (s) (33.88) (22.63) OLST (s)* 0.43 (1.57) 2.71 (2.59) FRT (cm) B-POMA* NOTE. Values are median (IQR) or mean SD. *Significant at P.05.

3 1638 PILOT STUDY PREDICTIVE VALIDITY FALLS RISK, Thomas Test Table 3: Sensitivity and Specificity Cutoff Points Cutoff Point Sensitivity (%) (95% CI) Specificity (%) (95% CI) TUG 32.6s 75 (.47.91) 67 (.44.84) OLST 1.02s 67 (.39.86) 89 (.67.97) FRT 18.5cm 75 (.46.91) 67 (.44.84) B-POMA (.55.95) 72 (.49.87) other significant variable was BMI. In all cases, the model was significantly improved by the addition of the predictor variables to the constant (for models TUG, OLST, and B-POMA, P.001; for FRT, P.05). DISCUSSION The B-POMA and OLST differentiated between fallers and nonfallers and predicted fallers in a day hospital setting, whereas TUG and FRT did not differentiate between fallers and nonfallers, although a TUG time above 32.6 seconds was associated with a significant OR of being a faller. Several versions of the B-POMA are in use, and the version we used was taken from Tinetti et al. 17 The results showed good sensitivity and specificity, although neither was as high as those reported by Chiu et al, 11 who used a longer version of the B-POMA with falls clinic patients. Additional research is required to determine if use of the extended scale would further increase sensitivity and specificity in day hospital patients. Use of logistic regression indicated that the B-POMA could be a significant predictor of falls. No other research has attempted to establish the OR of falling given a positive B-POMA score. Although this study gives indicative results, further research with a larger sample and using a prospective design is needed. The B-POMA had the best sensitivity of the 4 tests, and is therefore best at identifying the true positives. 36 The B- POMA incorporates many functional maneuvers and although there are inherent problems in summing activities in an ordinal scale, it allows many factors to have an impact on the total score. The B-POMA tests both predictive and reactive balance mechanisms 37 as well as the impact of reducing visual input. Maintaining balance is an integration of many body systems, 38 so the inclusive approach of the B-POMA is a strength of the measure. Because falls have been linked to lower-leg weakness 1 the inclusion of a sit-to-stand maneuver is important and, indeed, this was the item for which the majority of subjects had some difficulty. In the current analysis, the summed B-POMA score was considered as interval data and analyzed with parametric statistics as appropriate for a normal distribution. Although this approach to the analysis of summed ordinal data has been a topic of debate for many years, 31 we believe that the benefits of parametric analysis outweigh the negatives and that a nonparametric approach would not have altered the conclusions. The median FRT was lower than other studies 18 and there was no statistically significant difference between fallers and nonfallers, a finding that replicates the results from other studies. 19,39-41 Although Spilg et al 12 found that FRT was predictive of recurrent falls in day hospital patients, their values for the FRT appear to be much lower than those reported in the current study. The weight of evidence supports our findings that FRT is not useful in discriminating between subjects who may be at risk of falling and, regardless of its widespread use, the FRT cannot be recommended as a falls risk measure in day hospitals. Because the FRT is influenced more by trunk flexibility than displacement of the center of pressure, 42 it may not be a true measure of balance. The TUG times were far slower than the normative values established in community-dwelling adults. 43,44 Despite there being no significant difference between the times of fallers and nonfallers on the TUG, the test was still able to predict fallers. This predictive ability of the TUG could be expected, because components such as sit-to-stand, gait, and turning require many aspects of postural control. Additionally, previous studies have found that TUG can discriminate between fallers and nonfallers 11,23,45 in community-dwelling elderly adults and those attending falls clinics. TUG times were slowest among those subjects using walking aids, especially walking frames. Despite the current study not finding that walking aid use was linked to falls, it has had some impact on TUG times. TUG was the only measure tested that includes a gait component, which is functionally important because many falls occur during ambulation. Previous studies investigating OLST 22,46 have reported mean times far longer than that in the present study. Although the OLST was previously thought to be most useful in fit elderly adults, the current research extends that applicability to a much frailer population. The high specificity of the test allows for correct identification of true negatives. It is therefore best at identifying persons who are nonfallers. 36 The OLST only tests 1 aspect of balance that of maintaining equilibrium on a reduced base of support. As such, its excellent results in identifying elderly fallers are unexpected. However, the greater task constraints and the need to cope with greatly increased postural sway may challenge a particular aspect of balance that is needed in saving a person during a potential fall. Further research into the responses of elderly subjects whilse performing OLST may offer greater insights into why it is such an effective measure. The investigators intrarater reliability for the OLST was low, due in part to the short times achieved, leaving little margin for error in starting and ceasing timing. Even a small inaccuracy in timing may lead to a misclassification in falls status. The results must, therefore, be interpreted with caution and use of the OLST may be improved by taking the mean of several attempts. As a further recommendation, one should recognize that an inability to stand on 1 leg should be a trigger for further investigation, thus removing the need for timing altogether. The supposition that day hospital patients are frailer than elderly persons dwelling in the wider community was borne out by the results from the present study. Our subjects mean ages were greater than previous studies, which reported ages up to 10 years younger. 43,47 Use of walking aids was also prevalent, compared with between 12.4% and 49% of subjects in prior studies. 48,49 The present study included subjects with a variety of diagnoses. Other researchers excluded these subjects to increase control of their study, 14,21,23,39,43 but their inclusion in the present study strengthens its external validity. The proportion of subjects defined as fallers was 60%. This proportion is in line with other retrospective studies of a frail population, 50,51 but greater than the incidence of fallers in a communitydwelling population. 1 This change in the population demo- Table 4: ORs of Being a Faller for Each Outcome Variable Test OR 95% CI P TUG OLST FRT B-POMA

4 PILOT STUDY PREDICTIVE VALIDITY FALLS RISK, Thomas 1639 graphics highlights the importance of this research, because one cannot assume that previous results are applicable to day hospital patients. Although the present study is limited by use of a retrospective falls history, patient-reported falls status was verified by medical records and the study provides a basis on which further prospective work on predicting falls in frail elderly home dwellers can be based. The study would also have been strengthened by randomizing the order of the tests. The tests were done in a pragmatic order because of the facility s physical layout and in an attempt to minimize fatigue in this frail group of subjects by limiting the walking between test areas. CONCLUSIONS The B-POMA, OLST, and TUG can be recommended as measures to identify patients at risk of falling because of physical factors in a day hospital setting, although this finding must be confirmed in a larger scale prospective study. The OLST and TUG may be more useful as a quick screening tool or objective marker. The B-POMA examines several different aspects of balance and so may be seen as the measure of choice during a full physical assessment. After rehabilitation, the measures can be used to reassess whether a person s falls risk attributable to modifiable physical factors has reduced. Acknowledgment: We thank Alasdair MacSween, PhD, for his helpful comments in the preparation of this manuscript. References 1. Guideline for the prevention of falls in older persons. American Geriatrics Society, British Geriatrics Society, and American Academy of Orthopaedic Surgeons Panel on Falls Prevention. J Am Geriatr Soc 2001;49: Department of Health (UK). National Service framework for older people: executive summary. London: Department of Health; Gillespie LD, Gillespie WJ, Robertson MC, Lamb SE, Cumming RG, Rowe BH. Interventions for preventing falls in the elderly. Cochrane Database Syst Rev 2003;(4):CD Lamb SE. Effectiveness of falls prevention and rehabilitation strategies in older people: implications for physiotherapy. London: Chartered Society of Physiotherapy; Chartered Society of Physiotherapy, College of Occupational Therapists. Audit pack: guideline for the collaborative, rehabilitative management of elderly people who have fallen. London: Chartered Society of Physiotherapy; Chartered Society of Physiotherapy. Falls audit short report. London: Chartered Society of Physiotherapy; Available at: 7 fec/fallsauditsr.rtf. Accessed January 8, Shumway-Cook A, Baldwin M, Polissar NL, Gruber W. Predicting the probability for falls in community-dwelling older adults. Phys Ther 1997;77: Di Fabio RP, Seay R. Use of the fast evaluation of mobility, balance and fear in elderly community dwellers: validity and reliability. Phys Ther 1997;77: O Brien K, Pickles B, Culham E. Clinical measures of balance in community-dwelling female fallers and non-fallers. Physiother Can 1998;50: Boulgarides LK, McGinty SM, Willett JA, Barnes CW. Use of clinical and impairment-based tests to predict falls by communitydwelling older adults. Phys Ther 2003;83: Chiu AY, Au-Yeung SS, Lo SK. A comparison of four functional tests in discrimination of fallers from non fallers in older people. Disabil Rehabil 2003;25: Spilg EG, Martin BJ, Mitchell SL, Aitchison TC. Falls risk following discharge from a geriatric day hospital. Clin Rehabil 2003;17: Perell KL, Nelson A, Goldman RL, Luther SL, Prieto-Lewis N, Rubenstein LZ. Fall risk assessment measures: an analytical review. J Gerontol A Biol Sci Med Sci 2001;56:M Duncan PW, Weiner DK, Chandler J, Studenski S. Functional reach: a new clinical measure of balance. J Gerontol A Biol Sci Med Sci 1990;45:M Vellas BJ, Wayne SJ, Romero L, Baumgartner RN, Rubenstein LZ, Garry PJ. One-leg balance is an important predictor of injurious falls in older persons. J Am Geriatr Soc 1997;45: Podsiadlo D, Richardson S. The timed Up and Go : a test of basic functional mobility for frail elderly persons. J Am Geriatr Soc 1991;39: Tinetti ME, Williams F, Mayewski R. Fall risk index for elderly patients based on number of chronic disabilities. Am J Med 1986;80: Rockwood K, Awalt E, Carver D, McKnight C. Feasibility and measurement properties of the functional reach and timed up and go tests in the Canadian study of health and aging. J Gerontol A Biol Sci Med Sci 2000;55:M Franzen H, Hunter H, Landreth C, Beling J, Greenberg M, Canfield J. Comparison of functional reach in fallers and non fallers in an independent retirement community. Phys Occup Ther Geriatrics 1998;15: Behrman AL, Light KE, Flynn SM, Thigpen MT. Is the functional reach test useful for identifying falls risk among individuals with Parkinson s disease? Arch Phys Med Rehabil 2002;83: Franchignoni F, Tesio L, Martino MT, Ricupero C. Reliability of four simple, quantitative tests of balance and mobility in healthy elderly females. Aging Clin Exp Res 1998;10: Hurvitz EA, Richardson JK, Werner RA, Ruhl AM, Dixon MR. Unipedal stance testing as an indicator of fall risk among older outpatients. Arch Phys Med Rehabil 2000;81: Shumway-Cook A, Brauer S, Woollacott M. Predicting the probability for falls in community-dwelling older adults using the Timed Up & Go Test. Phys Ther 2000;80: McGinty SM, Master LD, Till DB. Inter-tester reliability using the Tinetti gait and balance assessment scale. Iss Aging 1999;22: Topper AK, Maki BE, Holliday PJ. Are activity-based assessments of balance and gait in the elderly predictive of risk of falling and/or type of fall. J Am Geriatr Soc 1993;41: VanSwearingen JM, Brach JS. Making geriatric assessment work: selecting useful measures. Phys Ther 2001;81: British Geriatrics Society. Geriatric day hospitals and ambulatory care Compendium Doc D5. Available at: Accessed January 8, McIntyre A. Elderly fallers: a baseline audit of admissions to a day hospital for elderly people. Br J Occup Ther 1999;62: Lord SR, Sherrington C, Menz HB. Falls in older people: risk factors and strategies for prevention. Cambridge: Cambridge Univ Pr; Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull 1979;28: Munro BH. Statistical methods for health care research. 4th ed. Philadelphia: Lippincott Williams & Wilkins; Cohen J. Statistical power analysis for the behavioral sciences. Hillsdale: Lawrence Erlbaum Associates; Greenhalgh T. How to read a paper. 2nd ed. London: BMJ Publishing Group; Deeks JJ. Systematic reviews of evaluations of diagnostic and screening tests. BMJ 2001;323: Blesh TE. Measurement in physical education. 2nd ed. New York: Ronald Pr; Black ER, Panzer RJ, Mayewski RJ, Griner PF. Characteristics of diagnostic tests and principles for their use in quantitative decision making. In: Black ER, Bordley DR, Tape TG, Panzer RJ, editors.

5 1640 PILOT STUDY PREDICTIVE VALIDITY FALLS RISK, Thomas Diagnostic strategies for common medical problems. Philadelphia: American College of Physicians & American Society of Internal Medicine; Huxham FE, Goldie PA, Patla AE. Theoretical considerations in balance assessment. Aust J Physiother 2001;47: Ragnarsdóttir M. The concept of balance. Physiotherapy 1996:82: Cho CY, Kamen G. Detecting balance deficits in frequent fallers using clinical and quantitative evaluation tools. J Am Geriatr Soc 1998;46: Wernick-Robinson M, Krebs DE, Giorgetti MM. Functional reach: does it really measure dynamic balance? Arch Phys Med Rehabil 1999;80: Wallmann HW. Comparison of elderly nonfallers and fallers in performance measures of functional reach, sensory organisation and limits of stability. J Gerontol A Biol Sci Med Sci 2001;56: M Jonsson E, Henriksson M, Hirschfeld H. Does the functional reach test reflect stability limits in elderly people? J Rehabil Med 2003; 35: Thompson M, Medley A. Performance of community dwelling elderly on the timed up and go test. Phys Occup Ther Geriatrics 1995;13: Siggeirsdóttir K, Jónsson BY, Jónsson H, Iwarsson S. The timed Up & Go is dependent on chair type. Clin Rehabil 2002;16: Okumiya K, Matsubayashi K, Nakamura T, Fujisawa M, Osaki Y, Doi Y. The timed Up & Go test is a useful predictor of falls in community-dwelling older people. J Am Geriatr Soc 1998;46: Gustafson AS, Noaksson L, Kronhed AA, Möller M, Möller C. Changes in balance performance in physically active elderly people aged Scand J Rehabil Med 2000;32: Cwikel JG, Fried AV, Biderman A, Galinsky D. Validation of a fall risk screening test, the Elderly Fall Screening Test (EFST), for community-dwelling elderly. Disabil Rehabil 1998;20: Studenski S, Duncan PW, Chandler J, et al. Predicting falls: the role of mobility and non physical factors. J Am Geriatr Soc 1994;42: Newton RA. Balance screening of an inner city older adult population. Arch Phys Med Rehabil 1997;78: Piotrowski A, Cole J. Clinical measures of balance and functional assessment in elderly persons. Aust J Physiother 1994;40: VanSwearingen JM, Paschal KA, Bonino P, Yang JF. The modified Gait Abnormality Rating Scale for recognizing the risk of recurrent falls in community dwelling elderly adults. Phys Ther 1996;76: