Falls among the older adults are a major health concern and may be associated with significant health care costs. Falls are the leading cause of injury deaths, nonfatal injuries, and hospitalizations for trauma among the older adults. In 2000, direct medical expenses for fatal falls totaled $179 million, and $19 billion for nonfatal injuries that may reach up to $54 billion by 2020.1,2 Defining falls and balance, predicting who is at risk for falls, and improving balance in the older adults are some of the leading topics in geriatric physical therapy today. Balance has been defined in terms of specific pathologies by several authors including participants with neurological diseases, orthopedic deficits, and vestibular disorders.3,4 Falls, for the purposes of this paper, are defined as a precipitous unplanned descent to the floor resulting in either injury or no injury. Many instruments have been developed to assess balance and predict falls in the older adults. One of the most reliable5–10 and valid10–17 outcomes measures tested in a variety of settings with differing populations and diagnoses developed today is the Berg Balance Scale (BBS).
THE BERG BALANCE SCALE
The BBS, created in 1989, contains 14 static and dynamic activities related to everyday living.18 The BBS assesses balance and risk for falls through direct observation of the participant's performance by trained health care professionals in a variety of settings. The BBS tasks progress in challenges: from sitting to standing, standing with narrow base of support, and finally to tandem and single-leg stance. Scoring is on a 5-point ordinal scale with 0 indicating an inability to complete the task and 4 as independent with completing the task. The maximum score of 56 indicates good balance. The scale takes approximately 10 to 20 minutes to complete requiring minimal equipment (chair, stopwatch, ruler, and step) and minimal space. The BBS can be obtained from the Internet Stroke Center web site19 or video demonstrations can be viewed online.20
Evidence of Validity
The BBS has been well established as a valid measure for predicting falls in a variety of patient populations. It has been validated as a predictor of length of stay and discharge destination in stroke rehabilitation,21,22 and as a predictor of length of stay and outcomes in acute inpatient rehabilitation when used in conjunction with the Functional Independence Measure.23
Validity using the BBS with participants with Parkinson disease (PD) was established by several authors and correlated with the Unified Parkinson Disease Rating Scale (r = −0.50, P < .001,14 and r = −0.58, P < .005).13 In participants with multiple sclerosis, concurrent validity was examined with respect to the Dynamic Gait Index (r = 0.78), Timed Up and Go Test (r = −0.62), Deambulation Index (r = −0.74), Activities-specific Balance Confidence (r = 0.48), and the Dizziness Handicap Inventory (r = −0.32, P < .001).12
Construct validity was examined in community-dwelling elderly Taiwanese in relation to the Timed Up and Go Test (r = −0.53, P = .01).10 Concurrent validity was studied with correlations to the Postural Assessment Scale for Stroke Patients (r = 0.93-0.95), and the Fugl-Meyer balance subscale (r = 0.90-0.92, P < .0001).16
Evidence of Reliability
Moderate to high reliability has been established for the BBS within a variety of settings and diagnoses. Relative intrarater reliability for the BBS was high (Interclass Correlation Coefficient [ICC] = 0.97) when used with participants in residential care facilities, while absolute reliability indicated a large change of 8 points was needed to detect change in function (95% confidence interval).8 Moderate to good scores were evaluated for personal care home residents for interrater reliability (ICC = 0.88) and test-retest reliability (ICC = 0.77).24 Good interrater reliability (ICC = 0.87) was seen for the community-dwelling elderly Taiwanese.10
In adults with learning disabilities, interrater reliability was established (κ = 0.74-1.00) but test-retest reliability varied (κ = 0.37-1.00).25 Interrater and intrarater reliability were deemed excellent in acute stroke and elderly populations in a study by Berg et al.6 The interrater values for stroke were ICC = 0.98 and ICC = 0.92 for the older adults. Intrarater reliability values were ICC = 0.99 for stroke and ICC = 0.91 for older adults.6 A second study with stroke patients established a high interrater reliability (ICC = 0.95).16 Finally, 1 study including participants with parkinsonism revealed high test-retest reliability for the BBS (ICC = 0.90).26
Kornetti et al27 performed a Rasch Analysis on the original rating scale performance and described the 45 cutoff score in functional terms. The probabilities of each rating (0-4) for each item on the BBS was examined. The Rasch Analysis was used to compare the ability of a person to the item difficulty on the BBS.27 They identified items that participants scoring 45 or higher should be able to successfully complete. Four items were specifically identified as being important for participants to achieve a score at or near the 45 cutoff. These included alternating foot, looking behind, and standing on 1 leg. The fourth item, tandem stance, was found to be the most difficult item on the scale. Kornetti et al27 concluded that scale item scoring revisions on the BBS would help to further distinguish and connect a participant's functional level to the scale results and clinical assessments. Additionally, they provided a functional definition of what a person should be able to accomplish to score a 45 by putting a person's ability and item difficulty on the same linear continuum.
Wang et al10 performed an item analysis of the BBS with 268 community-dwelling elderly Taiwanese residents (age 73.84 ± 5.2). The mean (SD) score was 53.3 (4.1). The item response profile showed that more than 90% of the participants attained the highest score on all but 3 items (reach forward, tandem stance, and 1-leg stance). They concluded the 2 most difficult items, tandem stance and 1-leg stance, were useful to discriminate participants with history of falls from those without history of falls.
The objective of this systematic review was to examine the sensitivity, specificity, and recommended cutoff scores of the BBS to predict falls in the older adults. Systematic reviews have been conducted to assess the usefulness of the BBS in stroke rehabilitation28 and in combination with other measures for fall risk assessment.29 However, no systematic review has been conducted examining the ability of the BBS alone to predict falls in the older adults with and without pathological conditions. Therefore, a systematic review was conducted on the available literature pertaining to the ability of the BBS to predict falls in the older adults with and without pathology. Emphasis was placed on the specific cutoff scores that are most effective in predicting falls in the older adults, concentrating on the sensitivity and specificity of the BBS to predict falls.
A search of the English-language rehabilitation literature published between 1985 to March 2009 was performed using electronic databases (OVID, CINAHL, MEDLINE, and PubMed). Articles were searched using the following key terms: falls, elderly, balance scale, BBS, stroke, cerebrovascular accident, CVA, multiple sclerosis, Parkinson disease, neurological, psychometrics testing, validity, reliability, specificity, and predicting falls. The reference sections in all found articles were examined to identify other relevant articles. Finally, authors of the selected publications were searched to obtain other relevant articles not initially found.
Article Selection and Review
Articles selected were reviewed by 3 authors each for meeting the objectives and the inclusion criteria of psychometrics identifying sensitivity and specificity in relation to a cutoff score associated with the BBS to predict falls with a study sample's mean age of 65 ± 10 years. Included articles were rated using the Physiotherapy Evidence Database (PEDro) scale.30 The PEDro scale is widely used in rehabilitation literature to quantify a study design and is scored 0 (no) or 1 (yes) for a maximum score of 10.31
Decision rules were established for any discrepancy in PEDro scores as follows: (1) the criterion would be defined and explained to ensure the rater understands, (2) a specific reference in the article would be cited for clarity, and whether the raters continued to not agree, (3) the criterion would be scored by a majority vote from all authors. Only articles with a PEDro score of 5 or more out of 10 were included in this review. Finally, levels of evidence were identified for each included article.32
The initial search of databases identified 244 publications related to the BBS. The number of publications reviewed was reduced to 90 when search strategies for keywords or phrases were included. After determining duplication of titles, non-English titles, and articles with modification of the BBS, 58 publications were identified as relating to the objective of this review. Of these, 48 articles were excluded for age (n = 15), not pertaining to fall prediction psychometrics or cutoff score (n = 29), or an analytical or systematic review (n = 4). The 10 remaining articles7,11,33–40 were rated using the PEDro scale. One study33 scored less than 5 out of 10 resulting in 9 publications for this review.7,11,34–40 The PEDro score and level of evidence for each article included is provided in Table 1.
A description of the included articles are listed in Table 2, detailing the study design, age and gender, primary diagnosis, sample size, and inclusion/exclusion criteria used for each study. Several of the studies included were case-control studies,35–38,40 meaning the authors were looking at whether participants have the disorder in question (ie, fall risk). The remaining articles were prospective cohort studies,7,11,34,39 following participants over time to determine whether the BBS could predict falls. The majority of the studies reported more female participants than male, and all focused on older adults with a mean age greater than 68.1 years. Sample sizes ranged from 44 to 187 participants with the 3 primary diagnoses noted as community-dwelling older adults (without pathology),7,11,35,37,38 stroke,34,39 and PD.36,38
The psychometric properties of the BBS obtained for each study are provided in Table 3, including the recommended cutoff score, sensitivity and specificity, and other relevant information. Sensitivity and specificity were addressed in all 9 articles.7,11,34–40 Sensitivity for the BBS ranged from 25%11 to 95.5%35 and specificity ranged from 20.8%38 to 100%.34 These sensitivities and specificities are calculated on the basis of recommended BBS cutoff scores, ranging from 3335 to 54.11,36 Eight of the articles made recommendations about the appropriate cutoff score.11,34–40
Five studies investigated the BBS in relation to the elderly population. Four of these studies addressed community-dwelling older adults,7,11,35,40 while 1 addressed nursing home and community-dwelling older adults.37 These studies reported a range in sensitivity between 53%7 and 88.2%,35 specificity between 53%11 and 96%,7 and cutoff scores between 4637 and 54.11 Bogle-Thorbahn and Newton7 found that older adults fell more frequently when BBS scores were closer to the recommended cutoff of 45 and fall frequency was not as high in participants with scores further away from the cutoff. They concluded that participants scoring much lower than the 45 point cutoff were the most impaired and had adopted strategies to minimize the risk of falling, including the use of an assistive device and/or companion during mobility. In addition, they concluded that participants with the greatest physical impairments did not actually have the highest risk of fall. The results of this study indicate the BBS was not sensitive (53%) to identify participants with history of falls, but was specific (92%-96%) to identify those without history of falls. A ceiling effect was reported, resulting in the BBS being insensitive to differences among participants with very high levels of balance.7
Chiu et al35 identified sensitivity and specificity values at 3 different cutoff scores. A cutoff of 47 had an 88.2% sensitivity and 76.5% specificity. At a cutoff of 38 and 33, the sensitivity increased to 95.5% and 94.1%, respectively, with specificity increasing to 95.5% and 90.9%, respectively. They also used logistic regression to identify 2 items that were significant to discriminate participants with history of single fall from those without history of falls: picking up an object from the floor and standing on 1 leg. These items had an odds ratio (OR) = 0.27 and OR = 0.21, respectively. Chiu et al35 determined that picking up an object from the floor (OR = 38.0) and placing alternate feet on a stool (OR = 46.0) were significant for discriminating participants with history of single fall from those with history of multiple falls.
Muir et al11 compared different fall groups: any fall, multiple falls, and injurious falls. This study compared the previously identified cutoff of 4529,41 to their recommendation of 53 to 54 as the cutoff score. Refer to Table 3 for detailed information on sensitivity, specificity, and cutoff scores for the different groups. The authors recommended a higher cutoff score of 53 to identify participants with history of multiple falls and 54 to identify any elderly person at risk for a fall for optimal fall prediction sensitivity.11
Shumway-Cook et al40 identified that a BBS score of 49 or higher has a sensitivity of 77% and specificity of 86%. By plotting the predictive probability for falls as a function of the BBS, they demonstrated the BBS has a nonlinear and inverse relationship to fall risk. They determined for BBS scores between 56—and 54, each point lower on the BBS increased fall risk by 3% to 4%. Fall risk increased by 6% to 8% for each score lower between 54 and 46. Fall risk was nearly 100% for scores less than 36 and further decline in the BBS score did little to change the participant's fall risk. The authors concluded that a single-point change on the BBS can lead to different predicted probabilities of falling, and they suggested that this model can be used to quantify the relative fall risk, detect clinically relevant changes in fall risk, and measure the changes from various interventions used.
In contrast to the other studies including only community-dwelling older adults, the study by Lajoie and Gallagher37 included community-dwelling and nursing home older adults. They identified 46 as the recommended cutoff point for the BBS in their sample, yielding 82.5% sensitivity and 93% specificity. They determined that 50 points on the BBS was associated with a 10% risk for fall, whereas a score of 38 was associated with a 90% risk of fall. These authors also determined that the 1-leg stance item was the most significant predictor on the BBS to predict falls.
Two studies examined individuals with a history of stroke.34,39 Ashburn et al34 recommended a cutoff at 48.5, with sensitivity and specificity in the elderly stroke population at 85% and 49%, respectively. Mackintosh et al39 recommended a similar score of 49, with sensitivity and specificity in this population at 92% and 65%, respectively. Mackintosh et al39 also showed that BBS scores less than 49 could independently predict 2 or more falls in the 6 months after discharge from stroke rehabilitation. This predictive value was strengthened when combining history of fall in the hospital/rehabilitation with a BBS score less than 49, producing sensitivity and specificity at 83% and 91%, respectively.
These same studies34,39 addressed both positive predictive values (PPV) and negative predictive values (NPV). PPV is the estimate of the likelihood that a participant with a positive test actually has the condition of interest, in this case falls. NPV is the proportion of participants who are correctly identified as not having the condition of interest, for example, a fall. Ashburn et al34 determined PPV and NPV as 55% and 83%, respectively, at the 48.5 cutoff. Mackintosh et al39 identified PPV and NPV at 42% and 97%, respectively, with a cutoff of 49.
The last diagnostic group identified in this systematic review examines participants with PD.36,38 The authors of these studies recommended different cutoff scores for this population. Dibble and Lange36 recommended using a cutoff of 54, while Landers et al38 recommended using 43.5 as the cutoff. The 54-point cutoff yielded sensitivity of 79% and specificity of 74%,36 whereas the 43.5 point cutoff yielded 68% sensitivity and 95.8% specificity.38 Both studies applied various cutoff points to their data to determine sensitivity and specificity at individual points to determine what was thought to be the best predictive cutoff point.
Landers et al38 disagreed with the 54 point cutoff recommended by Dibble and Lange36 because the higher score may have placed too many participants without history of falls into the category of participants with history of falls. Landers et al38 found that a cutoff of 43.5 had a 94.4% posttest probability to classify participants with history of falls. In comparison, the cutoff of 54 proposed by Dibble and Lange36 had a posttest probability of 56.8% to accurately classify participants with history of falls.
One objective of this systematic review was to determine the sensitivity and specificity of the BBS to predict falls in the older adults. Sensitivity for the BBS would indicate the ability of the test to identify a balance deficit in participants who have decreased balance abilities.42 This review identified a wide range of sensitivity values depending on the study and the cutoff value. Bogle-Thorbahn and Newton7 attributed a low BBS sensitivity to the relationship between physical impairments (those who may develop compensatory strategies or use an external device) and risk for falls (those who have not developed compensatory strategies). This underscores the findings that participants who scored closer to the cutoff score of 45 fell more than those with a lower score. Dibble and Lange36 found a low sensitivity at a cutoff of 46 for persons with PD and noted that sensitivity was increased at a higher cutoff score of 54. Finally, Muir et al11 attributed low sensitivity to an inadequate cutoff score and thus raised the cutoff for community-dwelling older adults in their study to 54. Even with the higher cutoff score the sensitivity was still low. They based their findings on the multi-factorial nature of falls, concluding the use of a single tool to predict falls was inaccurate. They recommended using the BBS as a part of a more comprehensive falls assessment.
Specificity is the probability that a person who does not have the condition will have a negative test, or does not identify a balance deficit in those with good balance.42 As with sensitivity, specificity significantly varied on the basis of recommended cutoff scores. Ashburn et al34 reported the lowest of the 9 articles. They explained that the wide range of factors related to falls and the variety of settings used in their study may have contributed to a lower specificity. Landers et al38 modified the recommended cutoff score to focus on increasing specificity for fall prediction in participants with PD and suggested their intent was to improve specificity to reduce the chance of incorrectly identifying a participants with history of falls as a participants without history of falls. When Muir et al11 focused on improving the sensitivity of the BBS by raising the optimal single cutoff score, specificity decreased. Overall, all publications except Ashburn et al34 revealed moderate to high specificity, even with the variances in cutoff scores.
A second objective of this study was to identify the best cutoff score to predict falls in the older adults with and without pathological conditions. Our review was unable to identify a predictive cutoff score for use in the practice setting. One reason may be the diagnostic groups included in our review were too broad to draw conclusions for the entire elderly population. However, even within the same diagnostic groups (community-dwelling and PD) the recommended cutoff scores varied. The 2 studies34,39 regarding recommended cutoff scores for stroke were in near agreement, however, the findings were not statistically significant. In addition, the low number of accepted articles into our study may have limited the information available for cutoff score recommendations. Therefore, the BBS needs to be studied in various diagnostic groups to identify the best predictive score for each particular diagnosis, or whether a narrow range of cutoff scores can be used across healthy older adults and adults with a particular diagnosis.
BBS cutoff score recommendations appear to lie on a continuum for predicting falls in the older adults. As previously stated, Shumway-Cook et al40 concluded fall probability has an inverse, nonlinear relationship to the BBS score. As the BBS score decreased, the risk for falling increased at different rates, depending on where, along the scale, the BBS score falls. A 1-point change in the BBS score led to a 6% to 8% increase in fall risk. Conversely, Bogle-Thorbahn and Newton7 found that participants fell more frequently when scores were closer to the cutoff point of 45 than to lower scores. They suggest, this was related to compensation strategies used by those with decreased balance to minimize falls. Since no score indicates that a person will not fall, it might be better to determine when the risk of fall is low enough for a clinician to determine client safety and independence.
Some authors have suggested that individuals who score 40 or less use mobility aides and recommend balance interventions to reduce the risk for falling.7,11,40 This recommendation may relate in part to the findings by Shumway-Cook et al,40 in which older adults with BBS scores at 40 were at nearly 100% risk for falling, and Lajoie and Gallagher,37 who found a 90% risk for scores 38 or less. Additionally, Shumway-Cook et al40 found that scores near the originally recommended cutoff of 45 revealed an increased risk of falls when compared with lower scores. The use of a mobility aide for clients who score less than 45 may be an adequate compensation for safety and may explain a decrease in falls compared with those with a BBS near 45.7 These studies seem to imply that a score of 40 may be a useful cutoff point to indicate a client may need a mobility device; however, additional research is warranted to confirm these findings
We recognize that our inclusion and exclusion parameters may have potentially biased our review process. Omitting articles on the basis of age may have excluded studies just outside the set age or excluded certain pathologies (ie, multiple sclerosis and brain injury) that tend to occur in younger people. Also, the PEDro rating inclusion criteria may have excluded some articles that have good results but do not fit the scale properly, therefore resulting in a lower rating. Another limitation may be the inclusion of English-language publications only; therefore, valuable data could have been overlooked.
Recommendations for Future Research
To determine fall risk, particularly in individuals with comorbidities, modifications of BBS items based on the Rasch analysis27 may help differentiate participants with history of falls from participants without history of falls more accurately. Such modifications may allow for increased accuracy to determine a person's true functional ability as related to the scale and clinical assessments. Such an item analysis of the BBS may help distinguish items that are most pertinent to fall risk,10,35,37 and better classify individuals who have history of single, multiple, and recurrent falls.
The BBS alone is not able to definitively predict fall risk and no cutoff score was identified in this review as the optimal score for fall risk prediction. Rather, a range of scores to identify people who have an increased risk for falls and those who require mobility devices should be used. The BBS is only one test that a clinician can use to help identify and measure changes to elderly clients' fall risk as a part of a total balance evaluation. We recommend that the BBS be used in conjunction with other tests or measures as a total balance assessment. It should be combined with unique patient factors to quantify an older adult's chances of falls and to help guide a clinician's recommendation for client safety and interventions.
2. Stevens JA, Corso PS, Finkelstein EA, Miller TR, The costs of fatal and nonfatal falls among older adults. Inj Prev. 2006;12:290–295.
3. Pollock AS, Durward BR, Rowe PJ, Paul JP, What is balance? Clin Rehabil. 2000;14:402–406.
4. Berg KO, Kairy D, Balance interventions to prevent falls. Generations. 2003;26:75–78.
5. Berg KO, Wood-Dauphinee SL, Williams JI, Gayton D, Measuring balance in the elderly: preliminary development of an instrument. Physiother Can. 1989;41:304–311.
6. Berg KO, Williams JI, Wood-Dauphinee SL, The balance scale: reliability assessment with elderly residents and patients with an acute stroke. Scand J Rehabi Med. 1995;27:27–36.
7. Bogle Thorbahn LD, Newton RA, Use of the Berg balance test to predict falls in elderly persons. Phys Ther. 1996;76:576–785.
8. Conradsson M, Lundun-Olsson L, Lindelof N, et al., Berg balance scale: intrarater test-retest reliability among older people dependent in activities of daily living and living in residential care facilities. Phys Ther. 2007;87:1155–1163.
9. Liston RA, Brouwer BJ, Reliability and validity of measures obtained from stroke patients using the Balance Master. Arch Phys Med Rehabil. 1996;77:425–430.
10. Wang CY, Hsieh CL, Olson SL, Wang CH, Sheu CF, Liang CC, Psychometric properties of the Berg balance scale in a community-dwelling elderly resident population in Taiwan. J Formos Med Assoc. 2006;105:992–1000.
11. Muir SW, Berg K, Chesworth B, Speechley M, Use of the Berg balance scale for predicting multiple falls in community-dwelling elderly people: a prospective study. Phys Ther. 2008;88:449–459.
12. Cattaneo D, Regola A, Meotti M, Validity of six balance disorders scales in persons with multiple sclerosis. Disabil Rehabil. 2006;28:789–795.
13. Qutubuddin AA, Pegg PO, Cifu DX, Brown R, McNamee S, Carne W, Validating the Berg balance scale for patients with Parkinson's disease: a key to rehabilitation evaluation. Arch Phys Med Rehabil. 2005;86:789–792.
14. Brusse KJ, Zimdars S, Zalewski KR, Steffen TM, Testing functional performance in people with Parkinson disease. Phys Ther. 2005;85:134–141.
15. Whitney S, Wrisley D, Furman J, Concurrent validity of the Berg balance scale and the dynamic gait index in people with vestibular dysfunction. Physiother Res Int. 2003;8:178–186.
16. Mao HF, Hsueh IP, Tang PF, Sheu CF, Hsieh CL, Analysis and comparison of the psychometric properties of three balance measures for stroke patients. Stroke. 2002;33:1022–1027.
17. Riddle DL, Stratford PW, Interpreting validity indexes for diagnostic tests: an illustration using the Berg balance test. Phys Ther. 1999;79:939–948.
18. Berg KO, Williams JI, Wood-Dauphinee S, Maki B, Measuring balance in the elderly: validation of an instrument. Can J Public Health. 1992;83:S7–S11.
21. Wee JY, Bagg SD, Palepu A, The Berg balance scale as a predictor of length of stay and discharge destination in an acute stroke rehabilitation setting. Arch Phys Med Rehabil. 1999;80:448–452.
22. Wee JY, Wong H, Palepu A, Validation of the Berg balance scale as a predictor of length of stay and discharge destination in stroke rehabilitation. Arch Phys Med Rehabil. 2003;84:731–735.
23. Juneja G, Czyrny JJ, Linn RT, Admission balance and outcomes of patients admitted for acute inpatient rehabilitation. Am J Phys Med Rehabil. 1998;77:388–393.
24. Holbein-Jenny MA, Billek-Sawhney B, Beckman E, Smith T, Balance in personal care home residents: a comparison of the Berg balance scale, the multi-directional reach test, and the activities-specific balance confidence scale. J Geriatr Phys Ther. 2005;28:48–53.
25. Sackley C, Richardson P, McDonnell K, Ratib S, Dewey M, Hill H, The reliability of balance, mobility, and self care measures in a population of adults with a learning disability known to a physiotherapy service. Clin Rehabil. 2005;19:216–223.
26. Steffen T, Seney M, Test-retest reliability and minimal detectable change on balance and ambulation tests, the 36-item short-form health survey, and the unified Parkinson disease rating scale in people with parkinsonism. Phys Ther. 2008;88:733–746.
27. Kornetti DL, Fritz SL, Chiu YP, Light KE, Velozo CA, Rating scale analysis of the Berg balance scale. Arch Phys Med Rehabil. 2004;85:1128–1135.
28. Blum L, Korner-Bitensky N, Usefulness of the Berg balance scale in stroke rehabilitation: a systematic review. Phys Ther. 2008;88:559–566.
29. Perell KL, Nelson A, Goldman RL, Luther SL, Prieto-Lewis N, Rubenstein LZ, Fall assessment measures: an analytic review. J Gerontol Biol Sci Med Sci. 2001;56:M761–M766.
31. Bhogal SK, Teasell RW, Foley NC, Speechly MR, The PEDro scale provides a more comprehensive measure of methodological quality than the Jadad scale in stroke rehabilitation literature. J Clin Epidemiol. 2005:58;668–673.
33. Anderson AG, Kamwendo K, Seiger A, Appelros P, How to identify potential fallers in a stroke unit: validity indexes of four test methods. J Rehabil Med. 2006;38:186–191.
34. Ashburn A, Hyndman D, Pickering R, Yardley L, Harris S, Predicting people with stroke at risk of falls. Age Ageing. 2008;37:270–276.
35. Chiu AY, Au-Yeung SS, LO SK, Comparison of four functional tests in discriminating fallers from non-fallers in older people. Disabil Rehabil. 2003;25(1):45–50.
36. Dibble LE, Lange M, Predicting falls in individuals with Parkinson disease: a reconsideration of clinical balance measures. J Neurol Phys Ther. 2006;30:60–67.
37. Lajoie Y, Galagher SP, Predicting falls within the elderly community: comparison of postural sway, reaction time, the Berg balance scale and the activities-specific balance confidence (ABC) scale for comparing fallers and non-fallers. Arch Gerontol Geriatr. 2004;28:11–26.
38. Landers MR, Backlund A, Davenport J, Fortune J, Schuerman S, Altenburger P, Postural instability in idiopathic Parkinson's disease: discriminating fallers from nonfallers based on standardized clinical measures. J Neurol Phys Ther. 2008;32:56–61.
39. Mackintosh SF, Hill KD, Dodd KJ, Goldie PA, Culham EG, Balance score and history of falls in hospital predict recurrent falls in the 6 months following stroke rehabilitation. Arch Phys Med Rehabil. 2006;87:1583–1589.
40. Shumway-Cook A, Baldwin M, Polissar N, Gruber W, Predicting the probability for falls in community-dwelling older adults. Phys Ther. 1997;77:812–819.
41. Scott V, Votova K, Scanlan A, Close J, Multifactorial and functional mobility assessment tool for fall risk among older adults in community, home-support, long term and acute care settings. Age Ageing. 2007;36:130–139.
42. Portney LG, Watkins MP, Foundations of Clinical Research: Applications to Practice. 3rd ed. Upper Saddle River, NJ: Prentice Hall; 2009.