Alzayer, Lamia PT, MS; Beninato, Marianne PT, DPT, PhD; Portney, Leslie G. PT, DPT, PhD, FAPTA
Falls are a major health risk factor after a stroke. The incidence of falls is as high as 73% in the first six months after discharge home from hospitalization.1 Several studies have reported an incidence of falls between 23% and 50% in community-dwelling adults with chronic stroke.2–4 This rate is much higher than for older adults without stroke (11%-30%).5 It is reported that 28% of people with chronic stroke experience an injury as a result of a fall.6 Falls bring added risk because this population has more than seven times the risk of experiencing fractures as a result of a fall.7 More common complications are hip fracture, soft tissue injuries, fear of falling, and hospitalization,1,8,9 which may lead to further disability and possible death.10 Therefore, the development and use of falls screening tools as part of an effective fall prevention programs for this population is imperative.
Scores from objective clinical measures of balance have been frequently associated with falls in older adults and in people with stroke. Perhaps the most commonly used clinical tool to assess balance in these populations is the Berg Balance Scale (BBS).11 In the elderly, several studies of fall risk using the BBS as a dichotomous scale based on defined cutoff scores have demonstrated mixed results.11–15 Recently in a prospective study of fall risk, Muir et al,16 using dichotomous scoring based on the standard cutoff of 45 and other higher scores derived from receiver operating characteristic (ROC) curves, found the BBS to be only moderately effective for identifying older adults who fell, with sensitivities ranging from 0.25 to 0.69. Study of the association between the total BBS score and falls in community-dwelling people with chronic stroke has yielded equivocal results. Belgen et al4 identified a cutoff score of 52 (sensitivity [Sn; true positive rate] = 91%; specificity [Sp; true negative rate] = 42%) for classifying community-dwelling individuals with stroke who had a history of multiple falls during the past six months compared with those with one or no falls. Similarly, Mackintosh et al17 found that a BBS score of <49 predicts recurrent falls in people with stroke after hospitalization (Sn = 92% and Sp = 65%). Ashburn et al18 also found the BBS to be predictive of multiple falls posthospitalization (Cutoff = 48.5, Sn = 85%, and Sp = 49%). Hyndman and Ashburn2 found significantly lower BBS scores (P < 0.05) in individuals with multiple falls compared with those with no falls in a sample of community-dwelling people with stroke. In contrast to the these studies, Harris et al,6 in a prospective study of people with chronic stroke living in the community, found no significant difference in BBS scores between those who fell multiple times and those who did not. These equivocal results demonstrate that the association of BBS scores to falls in community-dwelling people with chronic stroke still warrants further investigation.
One of the limitations of the interpretation of the total BBS score may lie in the scoring of the individual BBS items. In a Rasch analysis of the BBS in the elderly, Kornetti et al19 performed a rating scale analysis and found that the score associated with success on an individual item varies among the 14 BBS items. “Success” in this context refers to performing the defining task for a particular BBS item based on the operational definitions described by the developers of the BBS.20 For example, for the item tandem stance (B13), success can only be reached with a score of 4 because that is the only score that requires true tandem stance. For other items such as placing alternate foot on the step of stool (B12), success can be reached with a score of 3 and faster performance earns a score of 4. For the item standing on one leg (B14), a score of 2 indicates success and higher scores are earned by maintaining the posture for a longer period of time.19 This scoring inconsistency, that is, the lack of a consistent score across items that indicates success may result in differential weighting of individual BBS items in the total score and may explain some of the variability seen in total BBS scores used as fall screening measures.
To date, there has been limited analysis of individual BBS items relative to fall risk in older adults. Kornetti et al19 found that a total BBS score of 45 was associated with an increased probability of succeeding on items alternating foot on stool (B12), stand on one leg (B14), turn and look behind (B10), and stand with feet together (B7). Chiu et al,15 through logistic regression analysis, found picking up an object from the floor (B9; odds ratio [OR] = 0.27, confidence interval [95% CI] = 0.08-0.88) and stand on one leg (B14; OR = 0.21, CI = 0.06-0.72) contributed most to the discriminative ability of the BBS in identifying older adults with a history of single falls. They also found that the items picking up an object from the floor (B9; OR = 0.38, CI = 2.3-658) and alternate foot on stool (B12; OR = 46, CI = 9.8-2121) are useful for discriminating older adults with multiple falls.15 Similar item analysis relative to fall risk has not been performed on the BBS in people with stroke. In addition, although these individual items have been associated with fall risk, the relevance of individual scores on those items remains unclear. In other words, there may be a critical pivot score on individual items that is most closely related to risk of falls and that, if identified, would allow for simpler, dichotomized scoring of items. This approach would circumvent some of the scoring issues raised previously about the BBS by Kornetti et al19 Additionally, if these optimal pivot scores could be identified, the clinical usefulness of the BBS as a fall screening tool may be improved.
In the clinical setting, where time is at a premium and patient fatigue is a concern, the 15-20 minutes required to complete the BBS may limit its usefulness. This practical concern, in addition to the scoring system variability issues19 and variable accuracy in detecting fall risk, may limit the utility of the BBS as a screening tool for falls. Identifying a cluster of BBS items and understanding the association of these items to fall history may lead to a more accurate and efficient way for assessing fall risk. The purpose of this study was to explore a dichotomized scoring system of individual BBS items and to determine the Sn and Sp of each in classifying individuals with a history of multiple falls in a sample of community-dwelling people with chronic stroke. Our second purpose was to determine whether individual items or a combination of BBS items would have greater overall accuracy than the total BBS in classifying community-dwelling people with stroke with a history of multiple falls.
This study was a secondary data analysis performed on a subset of subjects from a sample that has been described previously.4 The original inclusion criteria were as follows: onset of stroke >one month before study, able to follow three-step commands, able to walk 10 m independently with or without an assistive device, living at home or in an assisted living facility, and able to provide informed consent. For purposes of this study, a subset of subjects with chronic stroke (>12 months) were selected (Fig. 1) to limit the influence of spontaneous recovery during the past six months. The exclusion criteria were as follows: presence of neurological disorder other than stroke, presence of major musculoskeletal problem or lower extremity fracture, and having undergone a surgical procedure in the past six months. The study was approved by the Human Subjects Research Institutional Review Board of Spaulding Rehabilitation Hospital.
The study design was a cross-sectional case series. An interview was conducted to collect demographic data including age, sex, stroke history, time since the stroke, side of weakness, use of assistive device, and walking ability. Walking ability was self-reported as indoors only, outdoors fewer than two blocks (community limited), and outdoors more than two blocks (community unlimited). Subjects were asked about the number of falls in the past six months. For purposes of this study, 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” (modified after Tinetti et al21).
Our primary variable was the BBS.20 The BBS consists of 14 items, each rated on a five-point ordinal scale ranging from 0 (cannot perform the task) to 4 (independence). Total scores can range from 0 to 56.20 The BBS is reported to have good internal consistency (Cronbach α = 0.92-0.98).22–24 Intrarater reliability is excellent in older adults (intraclass correlation coefficient [ICC] = 0.91)25 and in people with stroke (ICC = 0.99).26 Test-retest reliability has been established at 0.98.27 Interrater reliability is also excellent in older adults (ICC = 0.9226) and in people with stroke (ICC = 0.9826 and 0.9523). The original developers of the measure identified a score <45 as an indication of impaired mobility in older adults.11 The measurement characteristics of the BBS in people with stroke was recently reviewed by Blum and Korner-Bitensky.28
Participants were categorized as having a history of multiple falls or having a history of one or no falls in the past six months. We chose to examine those people with a history of multiple falls because it has been suggested that a single fall may be a random event, and those with multiple falls are more likely to be true fallers.29 Several researchers in the study of falls in people with stroke have followed this approach.1,4,6,18
Descriptive statistics were generated for demographic variables. Group comparisons were made based on falling category using t test (age and time since stroke), χ2 (stroke side, sex, and use of assistive device), or Mann-Whitney U test (BBS). Data analysis was conducted using SPSS 16.0 (SPSS, Inc., Chicago, IL) statistical software with a significance level of P ≤ 0.05.
BBS Item Analysis
Analysis of individual BBS items was performed to identify the individual BBS items with the highest Sn and Sp relative to classifying participants by fall history (history of multiple falls vs one or no falls). We dichotomized subjects based on scoring of each BBS item using three different pivot points on the scoring scale: between the score of 1 and 2, 2 and 3, and 3 and 4. Therefore, for example, with the pivot point between 1 and 2, scores of 0 and 1 were considered positive tests for falls, whereas scores of 2 or better were considered negative tests for falls. Likewise, when using the pivot point of 2 and 3, scores of 0, 1, or 2 were considered positive and scores of 3 and 4 considered negative. Similarly with the pivot point between 3 and 4, scores 0 through 3 were considered positive and only a score of 4 was considered negative. The BBS item standing on one leg (B14) was performed and scored twice based on standing on the more affected limb (hemiparetic) and on the less affected limb.
The Sn and Sp for classifying people with a history of multiple falls were calculated for each of the 14 BBS items for each of the three scoring dichotomies. Based on these results, we created various combinations of items with an Sn >60% to determine whether more than one item together would improve the Sn and Sp over any single item. When items were combined, a test result was considered “positive” (ie, associated with falls) when the subject scored less than the pivot point on each BBS item.
Positive (Sn/1 − Sp) and negative (1 − Sn/Sp) likelihood ratios (+LR and −LR) were calculated. The +LR is a measure of the likelihood of having a history of multiple falls with a positive test result (score below the dichotomized pivot point), and the −LR is the likelihood of identifying those with a history of falls with a negative test results (score better than the pivot point).
We also performed an ROC curve analysis on the BBS total score using falling status (multiple falls vs one or no falls) as the dichotomous outcome to determine the usefulness of the total BBS in classifying people according to fall history. The best total score, considering the balance between Sn and Sp, was identified as the point nearest the upper left hand corner of the ROC curve. The area under the curve (AUC) is a measure of the overall ability of the BBS test to distinguish between the two groups.30 We then performed a similar ROC curve analysis using a BBS subscore composed of the sum of all BBS items with an Sn >60%.
For Sn, Sp, and LR, 95% CIs were calculated based on a method described by Simel et al.31
The study included a total of 44 participants. The characteristics of the study population are given in Table 1. With regards to six-month fall history, 25 (57%) participants had no falls, nine (20%) had one fall, and 10 (23%) had a history of multiple falls. Ambulation status was self-reported as indoors only by three participants (7%), community limited by 12 participants (27%), and community unlimited by 29 participants (66%). Participants with multiple falls more often used an assistive device for ambulation compared with those with one or no falls (n = 34; 77%).
The results from individual BBS item analysis based on dichotomous scoring are shown in Tables 2–4. The items that yielded Sn >60% at the scoring pivot point between 3 and 4 were turning 360 degrees (B11), alternating foot on stool (B12), tandem stance (B13), and standing on one leg (B14; Table 4). At the scoring pivot point between 2 and 3 (Table 3), only tandem stance (B13) yielded Sn >60%. This item was not analyzed further, however, because the definition for a score of 3 was not clearly delineated for the item.20 For example, to achieve a score of 3, the subject is required to put one foot in front of the other but not in tandem. This definition allows participants to assume a variety of positions and still earn a score of 3. Similarly, the score of 2 is defined as the person needing help to step. This also does not specify a tandem position, and the subject may step in any direction. To achieve a score of 4, the subject is required to place feet in full tandem stance (on foot directly in front of the other), agreeing with the original operational definition of the item,20 and so we limited our further analysis to the 3-4 dichotomy for this item.
The +LRs and −LRs for the four items B11, B12, B13, and B14 are reported in Table 5. The Sn for B14 when calculated with the participant standing on the more involved limb yielded an Sn = 0.90 and Sp = 0.18. When calculated for performance on the less involved limb, the Sn was the same but the Sp was higher (0.50). We therefore included only the results from the less involved limb for B14 in further analyses.
Because BBS item 14 has the best Sn, we paired items B11, B12, or B13 with B14 to see whether pairing any of the items with item B14 improved the Sn over B14 alone. Pairing items did not improve the Sn but resulted in slight improvements in Sp (Table 4).
The ROC curve analysis of the total BBS score (Fig. 2) revealed that the optimal single cutoff value is 52 with an Sn = 0.90 (95% CI = 0.71-1.09) and Sp = 0.41 (95% CI = 0.24-0.58). The AUC was 0.67 (95% CI = 0.48-0.87). The ROC curve analysis for the subscore based on the raw scores of the four items with a Sn >60% (B11 + B12 + B13 + B14) is shown in Figure 3. Based on a maximum total score of 16, the optimum cutoff score was 11.5 with a Sn = 0.80 (95% CI = 0.55-1.05) and Sp = 0.53 (95% CI = 0.36-0.70) and an AUC of 0.67 (95% CI = 0.48-0.86).
With ongoing changes in healthcare, there is increasing pressure to more efficiently and effectively evaluate patients relative to fall risk. This study analyzed individual items of the BBS in people with chronic stroke in association with fall history using a dichotomized scoring system to eliminate some of the ambiguities of the scoring system. To our knowledge, this was the first study to analyze the BBS in this way. Our exploratory findings suggest that certain individual BBS items may be as accurate as the total BBS in identifying subjects with a history of multiple falls.
Relative to our study demographics, our 23% multiple fall rate compares with multiple fall rates of 11%,5 15%,32 22%,4 and 35%2 of other studies of community dwelling people with stroke. The current sample was relatively high functioning, as indicated by the high median BBS score of 49.5 and the high level of mobility with more than half of the participants able to ambulate in the community independently without using any assistive device.
Although there are no similar analyses of individual BBS items relative to fall risk, the four items found most closely associated with falls are also the four most difficult items according to Berg et al.33 Similarly, Kornetti et al19 reported B12, B13, and B14 as the most difficult items. Mao et al23 demonstrated a ceiling effect in the BBS at six months poststroke, so considering the high functioning level of the participants in this sample, it is not surprising that the four items identified here were the most challenging BBS items.
In older adults, Kornetti et al19 found turning 360 degrees (B11) and tandem stance (B13) to be closely associated with a total score of 45. In contrast to our findings, they also found that the items turn to look behind (B10) and stand unsupported with eyes closed (B6) were closely associated with a total score of 45.19 Our results agree with those of Chiu et al15 who found that standing on one leg (B14) was one of two items that best discriminated people with single falls, and placing alternate foot on stool (B12) was one of two items that best discriminated people with multiple falls in older adults. In contrast, Chiu et al15 found that picking an object up from the floor (B9) also discriminated older adults with multiple falls. The BBS items related to fall history in our sample were movements that required increased speed of movement (B11 and B12) or changes in base of support (B13 and B14), motor control requirements that are related to common impairments and postural control constraints in people with hemiparesis.34–37
The current findings indicate that a dichotomized scoring system using the pivot point between scores of 3 and 4 yielded the highest Sn and Sp compared with other pivot points. This cut point represents the most challenging performance on each item. This finding may be due, in part, to the high functioning level of our sample. In addition, on the four items with an Sn >60%, the scoring guidelines are unambiguous with the difference between scoring a 3 versus 4 being a matter of either performing the task faster (B11 and B12) or holding the posture longer (B13 and B14). The efficacy of these items at this scoring pivot point in accurately classifying participants according to fall history may be due in part to these clear scoring definitions. Interestingly, a minimal score of 4 matches the original operational definition of success19,20 only on B13 (tandem stance). For the other three most sensitive items (B11, B12, and B14), a score of 4 represents performance beyond basic successful performance of the item according to the operational definition of success.19,20
We used a dichotomous scoring approach to classify participants based on fall history for both the individual BBS item analysis and for the BBS total score and subscore ROC analyses. This approach has its shortcomings because fall risk more likely exists along a gradient rather than as a dichotomy of present/absent. Muir et al16 demonstrated an approach to analyzing fall risk along a gradient by using categories of total BBS scores and identifying their fall prediction value. This approach merits further examination in people with stroke.
Our results that B14 (standing on one leg) was the item with the single highest Sn in identifying those with a history of multiple falls is in agreement with those others who have found single limb stance to be a good predictor of falls in the older adults. Vellas et al,38 in a sample of community ambulatory older adults, reported that the ability to stand on one leg unassisted for less than five seconds was a significant predictor of injurious falls (Sn = 36%, Sp = 76%, positive predictive value of 31%). Likewise, Hurvitz et al39 found that unilateral stance time <30 seconds was associated with a history of falls (Sn = 91% and Sp = 75%). Together with the current findings, these findings suggest that single limb stance as a single predictor of falls may be clinically useful in older adults and in people with stroke.
On the basis that B14 yielded the highest single Sn, we combined B14 with B11, B12, or B13 to explore whether the Sn and Sp could be improved. Our approach was based on the approach used in the development of clinical prediction rules.40 Combining items did not improve Sn, but our conclusions on this approach are tentative because we were hampered by our small sample size. Nevertheless, similar combination of items has not been reported previously and may warrant further investigation. A larger sample size would allow for other combinations of items than was possible with our sample.
Our findings suggest that administration of the whole BBS may not be necessary and using specific items of the BBS may prove more efficient than administering the total BBS without sacrificing accuracy. Our ROC analysis of the four items with an Sn >60% yielded approximately the same results as the total BBS in identifying participants with multiple falls. This suggests that the discriminative power of the BBS to classify individuals based on fall history is largely derived from these four items. We interpret these findings with caution, however, because our sample was high functioning and the items associated with fall history would likely vary with samples at a lower level of function. Our results highlight the importance of matching examination tools with the functioning level of the sample. Future research may explore testing various subsets of BBS items matched to a group’s functional level to more accurately identify fall risk. A simple definition of functional level may include, for example, ambulation status as nonambulatory and ambulatory with device or ambulatory without device. Alternatively, for an ambulatory cohort, one might consider functional category according to gait speed. These concepts need further exploration, and their relationship to specific groups of BBS items need to need determined. McGinnis et al41 recently identified perceived usefulness as an important influence on whether a clinician uses a particular balance assessment tool. A more precise matching of sets of BBS items to functional level, thereby increasing efficiency and accuracy, may increase the perceived usefulness of the BBS on the part of clinicians and may lead to more widespread use of the BBS as a standardized outcome measure.
There are a number of limitations to this study. The 95% CI for some Sn and Sp exceeded 1.00. This bias nearly always leads to overestimation of the predictive ability of the test and is likely the result of our small sample size.42 Similarly, some of the 95% CIs for Sn and Sp, the +LRs and −LRs contained the null value (0.50 for Sn and SP, 1.0 for LR). We suggest that the current results be interpreted with caution and that a larger sample is needed to confirm these current exploratory results. In addition, the conclusions are limited to ambulatory community-dwelling people with chronic stroke and therefore cannot be generalized to all people with stroke, particularly those at lower functional levels. Data on fall frequency were collected retrospectively and were dependent on the subject’s recall of the past six months, which may have introduced some error. Finally, the predictive value of the current findings needs to be confirmed with a prospective study because associations with fall history may not equate to predicting fall risk.
Individual item analysis of the BBS in community-dwelling ambulatory people with chronic stroke indicated that administering the total BBS may not be necessary. Using selected BBS items may be as useful and accurate in classifying people based on fall history as using the total BBS score, which would improve efficiency. A dichotomous approach to scoring may eliminate some of scoring variability and ambiguity and seems to hold promise as a system that could be used to measure fall risk, but these exploratory results need to be investigated prospectively.
1. Forster A, Young J. Incidence and consequences of falls due to stroke: a systematic inquiry. Br Med J. 1995;311:83–86.
2. Hyndman D, Ashburn A. People with stroke living in the community: attention deficits, balance, ADL ability and falls. Disabil Rehabil. 2003;25:817–822.
3. Lamb SE, Ferrucci L, Volapto S, et al. Risk factors for falling in home-dwelling older women with stroke: the Women’s Health and Aging Study [see comment]. Stroke. 2003;34:494–501.
4. Belgen B, Beninato M, Sullivan PE, et al. The association of balance capacity and falls self-efficacy with history of falling in community-dwelling people with chronic stroke. Arch Phys Med Rehabil. 2006;87:554–561.
5. Jorgensen L, Engstad T, Jacobsen BK. Higher incidence of falls in long-term stroke survivors than in population controls: depressive symptoms predict falls after stroke. Stroke. 2002;33:542–547.
6. Harris JE, Eng JJ, Marigold DS, et al. Relationship of balance and mobility to fall incidence in people with chronic stroke. Phys Ther. 2005;85:150–158.
7. Kanis J, Oden A, Johnell O. Acute and long-term increase in fracture risk after hospitalization for stroke. Stroke. 2001;32:702–706.
8. Davenport RJ, Dennis MS, Wellwood I, et al. Complications after acute stroke. Stroke. 1996;27:415–420.
9. Hyndman D, Ashburn A, Stack E. Fall events among people with stroke living in the community: Circumstances of falls and characteristics of fallers. Arch Phys Med Rehabil. 2002;83:165–170.
10. Ramnemark A, Nilsson M, Borssen B, et al. Stroke, a major and increasing risk factor for femoral neck fracture. Stroke. 2000;31:1572–1577.
11. Berg KO, Wood-Dauphinee SL, Williams JI, et al. Measuring balance in the elderly: Validation of an instrument. Can J Public Health. 1992;83(Suppl 2):S7–S11.
12. Bogle Thorbahn LD, Newton RA. Use of the Berg balance test to predict falls in elderly persons [see comment]. Phys Ther. 1996;76:576–583.
13. Shumway-Cook A, Baldwin M, Polissar NL, et al. Predicting the probability for falls in community-dwelling older adults. Phys Ther. 1997;77:812–819.
14. Riddle DL, Stratford PW. Interpreting validity indexes for diagnostic tests: An illustration using the Berg balance test. Phys Ther. 1999;79:939–948.
15. Chiu AY, Au-Yeung SS, Lo SK. A comparison of four functional tests in discriminating fallers from non-fallers in older people. Disabil Rehabil. 2003;25:45–50.
16. Muir SW, Berg K, Chesworth B, et al. Use of the Berg balance scale for predicting multiple falls in community-dwelling elderly people: A prospective study. Phys Ther. 2008;88:449–459.
17. Mackintosh SF, Hill KD, Dodd KJ, et al. Balance score and a history of falls in hospital predict recurrent falls in the 6 months following stroke rehabilitation. Arch Phys Med Rehabil. 2006;87:1583–1589.
18. Ashburn A, Hyndman D, Pickering R, et al. Predicting people with stroke at risk of falls. Age Ageing. 2008;37:270–276.
19. Kornetti DL, Fritz SL, Chiu YP, et al. Rating scale analysis of the Berg balance scale. Arch Phys Med Rehabil. 2004;85:1128–1135.
20. Berg KO, Wood-Dauphinee S, Williams JI, et al. Measuring balance in the elderly: Preliminary develpment of an intrument. Physiother Can. 1989;41:304–311.
21. Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med. 1988;319:1701–1707.
22. Berg K, Wood-Dauphinee S, Williams JI. The balance scale: Reliability assessment with elderly residents and patients with an acute stroke. Scand J Rehabil Med. 1995;27:27–36.
23. Mao HF, Hsueh IP, Tang PF, et al. Analysis and comparison of the psychometric properties of three balance measures for stroke patients. Stroke. 2002;33:1022–1027.
24. Chou CY, Chien CW, Hsueh IP, et al. Developing a short form of the Berg balance scale for people with stroke. Phys Ther. 2006;86:195–204.
25. Stevenson TJ, Garland SJ. Standing balance during internally produced perturbations in subjects with hemiplegia: Validation of the balance scale. Arch Phys Med Rehabil. 1996;77:656–662.
26. Botner EM, Miller WC, Eng JJ. Measurement properties of the activities-specific balance confidence scale among individuals with stroke. Disabil Rehabil. 2005;27:156–163.
27. 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.
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. Overstall PW. Falls after strokes [see comment]. Br Med J. 1995;311:74–75.
30. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology. 1982;143:29–36.
31. Simel DL, Samsa GP, Matchar DB. Likelihood ratios with confidence: Sample size estimation for diagnostic test studies. J Clin Epidemiol. 1991;44:763–770.
32. Mackintosh SF, Goldie P, Hill K. Falls incidence and factors associated with falling in older, community-dwelling, chronic stroke survivors (>1 year after stroke) and matched controls. Aging Clin Exp Res. 2005;17:74–81.
33. Berg KO, Maki BE, Williams JI, et al. Clinical and laboratory measures of postural balance in an elderly population. Arch Phys Med Rehabil. 1992;73:1073–1080.
34. Shumway-Cook A, Wollacott MH. Motor Control: Translating Research into Clinical Practice. 3rd ed. Baltimore: Lippincott Williams & Wilkins; 2007.
35. Laufer Y. The effect of walking aids on balance and weight-bearing patterns of patients with hemiparesis in various stance positions. Phys Ther. 2003;83:112–122.
36. Laufer Y, Dickstein R, Resnik S, et al. Weight-bearing shifts of hemiparetic and healthy adults upon stepping on stairs of various heights. Clin Rehabil. 2000;14:125–129.
37. Turnbull GI, Charteris J, Wall JC. Deficiencies in standing weight shifts by ambulant hemiplegic subjects. Arch Phys Med Rehabil. 1996;77:356–362.
38. Vellas BJ, Wayne SJ, Romero L, et al. One-leg balance is an important predictor of injurious falls in older persons. J Am Geriatr Soc. 1997;45:735–738.
39. Hurvitz EA, Richardson JK, Werner RA, et al. Unipedal stance testing as an indicator of fall risk among older outpatients. Arch Phys Med Rehabil. 2000;81:587–591.
40. Childs JD, Cleland JA. Development and application of clinical prediction rules to improve decision making in physical therapist practice. Phys Ther. 2006;86:122–131.
41. McGinnis PQ, Hack LM, Nixon-Cave K, et al. Factors that influence the clinical decision making of physical therapists in choosing a balance assessment approach. Phys Ther. 2009;89:233–247.
42. Deeks JJ, Altman DG. Sensitivity and specificity and their confidence intervals cannot exceed 100%. BMJ. 1999;318:193–194.