The Association of Balance Capacity and Falls Self-Efficacy With History of Falling in Community-Dwelling People With Chronic Stroke

Article Outline

• Abstract
• Methods
  • Participants
  • Procedure
  • Measures
  • Statistical Analyses
• Results
• Discussion
• Conclusions
• Acknowledgment
• References
• Copyright

Abstract 

Belgen B, Beninato M, Sullivan PE, Narielwalla K. The association of balance capacity and falls self-efficacy with history of falling in community-dwelling people with chronic stroke.

Objectives

To describe the frequency of falls; to relate capacity-based and self-efficacy measures to fall history; and to determine to what extent capacity-based and self-efficacy measures are explained by subject characteristics and stroke impairments.

Design

Cross-sectional.

Setting

Community.

Participants

Convenience sample of 50 people with chronic stroke.

Interventions

Not applicable.

Main Outcome Measures

Fall history, Falls Efficacy Scale–Swedish Version, fear of falling, and the mood subscore of the Stroke Impact Scale. Balance, strength, and functional mobility were measured using the Berg Balance Scale, timed sit to stand, and Timed Up & Go, respectively.

Results

Falls were reported by 40% (n=20) of subjects; 22% (n=11) reported multiple falls. Subjects with fall history had more fear of falling (relative risk [RR], 2.4; 95% confidence interval [CI], 1.1−4.9), had less falls-related self-efficacy (P=.04), and more depressive symptoms (P=.02) than nonfallers. Subjects with multiple fall history had poorer balance (P=.02), more fear of falling (RR=5.6; 95% CI, 1.3−23), and used a greater number of medications (P=.04) than non- and 1-time fallers. Strength partially explained balance, mobility, and falls-related self-efficacy.

Conclusions

Balance and falls-related self-efficacy are associated with fall history and should be addressed in people with chronic stroke.

Key Words:  Accidental falls , Balance , Cerebrovascular accident , Rehabilitation , Self efficacy

STROKE IS A MAJOR HEALTH problem worldwide with incidence per 100,000 varying from 101 to 285 in men and 47 to 198 in women.¹ Stroke is the leading cause of serious, long-term disability in the United States, with over 1 million adults reporting difficulties in function as a result of stroke.² Recent statistics in the United States also indicate that the number of people living in the community with chronic stroke rose to 2.4 million by the early 1990s.² In addition to the primary deficits associated with stroke, there is a high rate of secondary complications, including falls.

A high risk of falling after stroke has been reported during hospital stay³, ⁴, ⁵, ⁶, ⁷ and after discharge.⁸, ⁹, ¹⁰, ¹¹ Incidence of falls in inpatient settings have ranged from 14% to as high as 64.5%.³, ⁴, ⁵, ⁶, ⁷ Several reports have documented the incidence, risk factors, and consequences of falls for community-dwelling people with stroke. The incidence of 1-time falls varies from 23% to 73%, with multiple fall rates ranging from 12% to 47%.⁸, ⁹, ¹⁰, ¹² Some of the variability in the reported fall rates may be related to whether subjects were studied shortly after onset of stroke⁸, ¹⁰ as compared with studies where mean time poststroke was greater than 1 year.⁹, ¹¹, ¹³ Other factors may be whether the study design was prospective⁸, ¹⁰, ¹¹, ¹³ or cross-sectional⁹, ¹² and whether the time period of falls history was 4,¹³ 6,⁸, ¹⁰, ¹² or 12⁹, ¹¹ months. Despite these variations, fall rates in people with stroke are higher than in the general elderly, where the fall rate has been reported as 32%,¹⁴ 34%,¹⁵ increasing to 45% for those over 80 to 89 years of age.¹⁵

The consequences of falling include hip fractures, soft tissue injuries,³, ⁴, ⁸, ⁹ fear of falling,⁹ hospitalization, increased immobility, and greater disability.⁸ The potentially disabling consequences of falls, high fall rates in community-dwelling stroke survivors, and number of people living in the community with the effects of stroke all indicate the need for better understanding of contributing factors to balance and falls in community-dwelling people with chronic stroke.

Identifying fall risk factors is an important first step in maintaining optimal function and preventing falls. Risk factors related to falls, as well as incidences and consequences of falls in the general elderly population, have been well studied.¹⁴, ¹⁵, ¹⁶, ¹⁷, ¹⁸ However, it is unclear whether, and to what extent, these factors implicated in the well elderly can be generalized to community-dwelling persons with stroke. The higher incidence of falls in the stroke population may be accounted for by the addition of stroke-specific impairments. These impairments include deficits in motor function,⁹, ¹⁰ sensation,¹⁰ vision,¹⁰ and cognitive factors, including depression⁹, ¹⁹ and attention impairments.²⁰ Some studies have examined the role of stroke specific characteristics and impairments that may be associated with fall history⁸, ⁹, ¹⁰, ¹³, ²⁰ while others have utilized generic measures such as the Berg Balance Scale (BBS)¹², ²⁰ and gait speed¹² or self-report of function, strength, and balance.¹¹ The association between stroke-specific measures and falls history is unclear and varies somewhat on the measure that was used. For example, Yates et al,¹⁰ utilizing the Fugl-Meyer Assessment (FMA) of sensorimotor impairment to measure motor function, found a relationship between falls history and motor function, but Jorgensen et al,¹³ using the Scandinavian Stroke Scale, did not. Diagnosis- and impairment-specific measures may be more associated with fall history than generic tools in people with chronic stroke, but further study is necessary to elucidate these relationships.

The BBS²¹, ²², ²³, ²⁴ and the Timed Up & Go (TUG) test²⁵ are commonly used clinical tools that were developed as measures of balance and functional mobility, respectively, in the elderly. Subsequently, several investigators have demonstrated the clinical usefulness and accuracy of the BBS²⁶, ²⁷, ²⁸ and the TUG¹⁷, ²⁸ to differentiate elderly with a history of falls from those who had not fallen. In the existing literature, cutoff scores used to identify fall risk vary from 38 to 49 on the BBS²², ²⁶, ²⁷, ²⁸ and 13.5¹⁷ or 20.1²⁸ seconds for the TUG. These various cutoff points may reflect different patient subgroups being tested and raise the question that a single cutoff score may not be appropriate across all groups of elders, much less all diagnostic categories. While some subjects with stroke were included in studies of the BBS,²⁷, ²⁸ and the TUG²⁸ and the BBS has been validated as a measure of balance for people with stroke,²³ the ability of the TUG and BBS to differentiate based on fall history specifically in persons with stroke is not well studied.

The results are thus far equivocal with regards to the relationship of BBS scores and falls history in people with stroke. Hyndman and Ashburn²⁰ found lower BBS scores associated with history of multiple falls when compared with subjects with no fall history in a sample of independent ambulating community-dwelling people with chronic stroke. In contrast, Harris et al¹² found no difference in BBS scores between subjects with either 1-time or multiple fall history when compared with those with no fall history. No studies have reported the ability of the TUG to differentiate subjects with stroke based on fall history. Further investigation of the usefulness of these tools in differentiating based on fall history in community-dwelling people with chronic stroke is warranted.

Fear of falling²⁹, ³⁰ and falls-related self-efficacy¹⁸, ³¹, ³² have been studied in the elderly and have been related to fall status,²⁹, ³⁰, ³¹, ³² activity restriction,³¹ depression,³¹, ³³ and balance capacity as measured on the BBS.¹⁸ Fear of falling has also been shown to exist in community-dwelling people with stroke among both people who have fallen and those who have not.⁹ No study, however, has reported the relationships among falls-related self-efficacy, capacity-based measures of balance, and fall history in this population. Better understanding of the relationship between these physical and psychologic factors will enable more focused intervention directed toward the goal of improving balance and reducing falls.

The aims of this study were to determine the frequency and circumstances of falls of community-dwelling persons with chronic stroke. We also aimed to determine how well the Falls Efficacy Scale–Swedish Version (FES-S), TUG, BBS, and the stroke-specific measures of the FMA and Stroke Impact Scale (SIS) mood and emotion subscore as well as other demographic characteristics could distinguish between groups of subjects based on their history of falling. Finally, we aimed to measure the extent to which the variance of the BBS, TUG, and FES-S could be explained by subject characteristics and impairments.

Back to Article Outline

Methods 

Participants 

We recruited a convenience sample of 50 community-dwelling people with chronic stroke from community stroke support groups or through posted advertisements. Volunteers underwent a telephone screening to determine eligibility and were scheduled for a single study session. People were included if they had a stroke onset more than 1 month prior, were able to follow 3-step commands, lived at home or in an assisted living facility, were able to walk 10m with no physical assistance with or without any assistive device, and were able to provide informed consent. Exclusion criteria were other neurologic diagnoses, lower-extremity fracture, or other surgical procedures during the last 6 months. All participants gave written informed consent prior to participation in the study. The study was approved by the Spaulding Rehabilitation Hospital Institutional Review Board for Human Subject Research.

Procedure 

The study design was cross-sectional. The 4 investigators performed separate portions of the examination. A standardized questionnaire was administered and physical impairments and functional performance were assessed. The 3 investigators who performed the impairment and functional measures were blinded to the information on falls history. Time to complete a single study session ranged from 60 to 90 minutes.

Measures 

Demographic characteristics and medical history recorded by questionnaire included age, height, weight, alcohol use, comorbidities, number of medications, use of assistive device, living environment, date of stroke onset, and side of stroke. We obtained side of and time since stroke from subjects’ self-report, but specific details regarding whether the stroke was hemorrhagic or ischemic were not obtained. Subjects were asked for a list of any other known medical conditions and the number of prescription medications they were currently taking. They were asked if they drink alcohol (yes, no) and how many drinks they have per week on average. With regard to living situation, subjects were asked if they live alone, or with their spouse, other family, friends, or paid help. They were also asked if they lived in their own residence, assisted living, senior housing, or in a family member’s residence. They were also asked if they used an assistive device (yes, no) and if so, which one (straight cane, quad cane, hemiwalker, or walker). Subjects were also asked if they primarily walked at home indoors only, outdoors but less than 2 blocks (community limited) or outdoors and more than 2 blocks (community unlimited).

Falls history included number of falls during last 6 months or number of falls since the stroke onset, if stroke was less than 6 months prior. 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 al¹⁴). Subjects reported circumstances of the falls, including the location and associated activities. Subjects were also asked if they knew someone who had fallen (yes, no). Fear of falling was assessed by asking subjects if they were afraid of falling (yes, no) and also ranking fear of falling on a 10-point ordinal scale with 1 representing “no fear” and 10 representing “the worst fear you can imagine.” Falls-related self-efficacy was measured using the FES-S, which has been shown to be a reliable tool for use with stroke survivors with a test-retest interclass correlation coefficient (ICC) of .97.³⁴ Subjects were asked to rate their level of self-efficacy in regard to performing common daily living activities without falling on a scale ranging from 1 (extreme confidence) to 10 (no confidence at all). Activities such as getting in and out of bed, getting dressed, or walking around the house are included. Hellstrom and Lindmark³⁴ added 3 items to the original 10 items on the FES³¹ to make the questions more specific to the functional difficulties seen in stroke patients and to decrease the possibility of ceiling effect. The total score of the FES-S on 13 questions ranges from 13 to 130 with low scores representing a greater level of self-efficacy. Only the total score was used in the analysis.

We measured composite functional lower-extremity muscle strength by the timed sit-to-stand (STS) test. This test was introduced by Csuka and McCarty³⁵ as a composite measure of lower-extremity strength. Csuka and McCarty determined the average coefficient of variation (repeatability) across 3 trials to be 6.8%±3.4% (range, 0%−17.3%) in a sample of 12 subjects.³⁵ The test has been modified for use in stroke patients by reducing the number of repetitions from 10 to 5.³⁶ Lord et al³⁷ have established ICC of .87 for intratester reliability of this version of the test in elderly subjects. Subjects begin in the seated position in a standard height chair (seat height, 45cm) with their back against the chair and feet on the ground. The time to rise from sitting to full standing and back again without the use of the upper limbs 5 times was measured with a stopwatch. Timing was stopped when the subject’s buttocks touched the chair seat at the end of the last repetition.

We measured lower-extremity motor control with the lower-extremity portion of the FMA, which is a standardized,³⁸ reliable,³⁹, ⁴⁰ and valid³⁸ measure. This test includes the assessment of reflexes, synergistic patterns, speed, tremor, and dysmetria. Sanford et al⁴⁰ reported high interrater reliability (ICC=.92) for this subscore of the FMA. All tasks are scored on a 3-point ordinal scale (range, 0−2) with the total lower-extremity score ranging from 0 to 34.

We tested balance performance using the BBS, a 14-item tool used to assess the ability to maintain positions of varying difficulty and perform specific balance-related movements. The scoring method is a 5-point ordinal scale (range, 0−4) for each task, with the total score ranging from 0 to 56. Interrater (ICC=.98) and intrarater (ICC=.99) reliability of the BBS has been established with a sample of subjects with stroke.²³

We used the TUG²⁵ to assess functional mobility where “functional mobility” refers to balance and gait maneuvers used in daily activities. The TUG has been shown to have good interrater (ICC=.99) and intrarater (ICC=.99) reliability.²⁵ It was administered according to standard protocol.²⁵ The subject sat in a chair with arms, with their back against the chair and feet on the floor. The time it took for the subject to rise, walk 3m at their usual pace, turn around, walk back to the chair and sit down was timed using a stopwatch. Ambulatory assistive devices were used if needed.

The SIS, a self-report, stroke-specific outcome measure, examines several aspects of health developed by Duncan et al.⁴¹ We used the mood and emotions subscore of this tool that is comprised of 9 questions such as: In the past week, how often did you … a) Feel sad? b) Feel that there is nobody you are close to? c) Feel that you are a burden to others? Responses are rated on a 5-point ordinal scale (5, none of the time; 1, all of the time) with a possible total ranging from 9 to 45. The test-retest reliability of mood and emotion domain of the SIS has been established (ICC=.57).⁴¹

Statistical Analyses 

Descriptive statistics were generated for all variables. Subjects were categorized based on their falling history as having a history of falls compared with those who had not fallen, and also as having a history of multiple falls compared with those who had fallen once or not at all. These 2 falling classifications were used to compare our results to others. Some authors⁸, ⁹, ¹¹, ¹² consider those who have fallen more than once as having a significant falls history and most susceptible to falls, because a single fall may have happened by chance as a random event.

To determine the association between variables recorded dichotomously and falling status, relative risk (RR) analysis was performed according to methods described by Portney and Watkins.⁴² The variables analyzed in this way included age (≥60y of age, <60y of age), time since stroke (≥12mo, <12mo), sex, side of stroke, use of assistive ambulation device (yes, no), alcohol use (yes, no), and afraid of falling (yes, no). The Mann-Whitney U test was used to compare the means between groups of continuous and ordinal variables (BBS, TUG, FMA, STS, FES-S, SIS mood and emotion, number of medications, age, time since stroke) because the data were not normally distributed.

Based on the results of the Mann Whitney U tests, variables that were found to be significantly different between groups were entered into logistic models to generate receiver operating characteristic (ROC) curves⁴³, ⁴⁴ to determine how well the variables differentiated subjects who had fallen from those who had not and subjects with a history of multiple falls from subjects who had fallen once or not at all. The area under the ROC curve is an indication of the overall ability of the variable measure plotted on the curve to detect the presence or absence outcome condition,⁴⁵, ⁴⁶ in this case fall category. Sensitivity and specificity indicate the true positive and true negative rate, respectively. The graph is plotted with sensitivity on the y axis over 1 minus specificity (false positive rate) on the x axis. The cutoff score on the measure (eg, BBS, FES-S) that best distinguished between the 2 categories of falls history was the point on the curve closest to the upper left-hand corner. This point corresponded to the score with the best overall accuracy, considering sensitivity and specificity,⁴⁵, ⁴⁷ in identifying subjects according to category of falls history.

The factors that explained the variance in BBS, TUG, and the FES-S were analyzed by multiple stepwise regression. Subject characteristics that were entered into the regression analysis are shown with an asterisk in table 1. In addition to the variables indicated on table 1, the FMA and STS were entered into equations for all 3 dependent variables. For analysis of FES-S, the BBS and TUG scores were also included. For the analysis of the variance in the BBS and TUG, fear of falling (yes, no) was entered as an additional variable. Variables were entered into the equation at the .05 level and removed at the .10 level.

Table 1. Subject Demographics (N=50)

Characteristics	n (%)	Mean ± SD (Range)
Age (y)	NA	59.9±11.9 (35–87)
Age >60y⁎	20(40)	NA
Time since stroke (mo)	NA	62.2±62.1(3–312)
Time since stroke ≥12mo⁎	42(84)	NA
Sex⁎
Female	19(38)	NA
Male	31(62)	NA
Ambulation status
Indoors only	3(6)	NA
Community limited	16(32)	NA
Community unlimited	31(62)	NA
Stroke side⁎
Left	18(36)	NA
Right	32(64)	NA
Assistive device type
Straight cane	16(32)	NA
Quad cane	4(8)	NA
Walker	1(2)	NA
None	27(54)	NA
Reported comorbidities
Diabetes	11(22)	NA
Cancer	5(10)	NA
Heart disease	16(32)	NA
Osteoporosis	4(8)	NA
Osteoarthritis	14(28)	NA
Vertigo/dizziness	9(18)	NA
Joint replacement	4(8)	NA
Rheumatoid arthritis	2(4)	NA
Fracture	13(26)	NA
Visual impairment	27(54)	NA
Fainting/syncope	2(4)	NA
Alcohol use (yes or no)⁎	21(42)	NA
No. of drinks per week	NA	1.4±2.7(0–10)
No. of medications	NA	5.1±3.1(0–13)
Seven or more medications⁎	15(30)	NA
Know somebody who had a fall⁎	28(56)	NA

Abbreviations: NA, not applicable; SD, standard deviation.

Statistical analyses were performed using SPSS software package, version 11.5.^a An α level of .05 determined significance unless stated otherwise.

Back to Article Outline

Results 

Subject demographics and medical history are shown in table 1. Average age of the subjects was 59.9±11.9 years (range, 35−87y), with 20 subjects less than 60 years and 10 subjects less than 50 years. Mean time since stroke was 62.2±62.1 months (range, 3−312mo), with 38 (76%) subjects more than 12 months since stroke onset. All subjects were independent in ambulation. Community ambulation was reported by 47 (94%) subjects; 62% reported unlimited community ambulation and 32% ambulation in a limited fashion. No assistive device was used by 27 (54%) subjects. In the 6 months prior to the study session, 20 (40%) subjects had fallen; 9 (18%) fell only once and 11 (22%) fell more than once. The mean number of falls was 0.84±1.13 (range, 0−4; 95% confidence interval [CI], 0.52−1.15). Of the subjects who fell, 17 (85%) fell at home, 8 (40%) fell outside, and 2 (10%) fell at a workplace. Eighteen subjects (90%) who had fallen reported that they were “walking” when they fell, 5 (25%) reported “dressing,” 7 (35%) reported “misstepping,” 6 (30%) reported “foot getting stuck,” and 6 (30%) reported “imbalance” as reasons for falling. Trips, slipping on ice, and pain following turning of the knee were other reasons.

Scores of standardized measures are shown in table 2 and results from RR analysis are in table 3. One subject was not able to perform the STS without the use of his upper limbs, so no score was recorded. Persons who reported falling once had lower levels of falls-related self-efficacy as measured by FES-S (P=.04) and had lower scores on SIS mood subscore (P=.02). RR analysis showed that subjects who reported falling were 2.4 times more likely to be afraid of falling (95% CI, 1.1−4.9). However, no differences were found in demographic, impairment, or functional measures between those who fell and those who did not. Subjects who reported multiple falls had significantly lower BBS scores (P=.02) and took a greater number of medications (P=.04) than subjects who fell once or not at all.

Table 2. Subject Characteristics by Falling Status (N=50)

NOTE. Values are mean ± SD and range or as otherwise indicated.

Table 3. RR Analysis

Variable	Fall History Versus No Falls RR (95% CI)	Multiple Fall History Versus No or 1 Fall RR (95% CI)
Age (≥60y or <60y)	0.64(.029–1.39)	1.40(0.50–4.11)
Time since stroke (≥12mo or <12mo)	0.57(0.16–2.00)	1.37(1.13–1.67)
Sex	1.17(0.57–2.40)	0.40(0.16–1.23)
Stroke side	1.18(0.59–2.35)	1.10(0.38–3.22)
Assistive device use (yes, no)	0.69(0.35–1.38)	1.16(1.12–2.54)
Alcohol use (yes, no)	0.48(0.20–1.11)	0.33(0.07–1.37)
Afraid of falling (yes, no)	2.36(1.14–4.90)	5.58(1.35–23.05)

Subjects who fell multiple times were also 5.6 times more likely to be afraid of falling (95% CI, 1.3−23) and 6.0 times more likely to use an assistive device (95% CI, 1.4−25).

The ROC curve analysis revealed that the FES-S best differentiated subjects who had fallen from those who had not at a threshold score of 17.5 (sensitivity, .90; specificity, .53) and an area under the curve (AUC) of .71. The BBS best differentiated those reporting multiple falls from subjects with 1 or no falls at a threshold score of 52 (sensitivity, .91; specificity, .42) with an AUC of .72. Refer to table 4 for ROC results and figure 1 and figure 2 for the ROC curves.

Table 4. ROC Curves With Falling Status as the Outcome State

Variable	Nonfallers vs 1- and Multiple Time Fallers AUC (95% CI)	Non- and 1-Time Fallers vs Multiple Fallers AUC (95% CI)
BBS	.61 (.44–.77)	.72(.54–.90)
TUG	.46(.29–.64)	.59(.39–.81)
FES-S	.71(.56–.85)	.66(.51–.82)

• 
  • View Large Image
  • Download to PowerPoint

• Fig 1. 

  ROC curve for the FES-S to discriminate between fallers and nonfallers.

• 
  • View Large Image
  • Download to PowerPoint

• Fig 2. 

  ROC curve for the BBS to discriminate between multiple fallers and non- and 1-time fallers. ⁎Cutoff score of 45 reported by Berg et al²¹ and Bogle Thorbahn and Newton²⁷. †Cutoff score of 49 reported by Shumway-Cook et al.²⁶ ‡Cutoff of 38 reported by Chiu et al.²⁸

Multiple stepwise regression analysis revealed that a total of 52% of the variance in the performance of the BBS was explained by the composite strength of lower extremities, measured by STS and age (table 5). The 3 factors that entered the stepwise regression to explain the FES-S were “knowing somebody who had fallen,” use of alcohol, and the STS (table 6). Together they explained a total of 28% of the variance in FES-S score. The STS explained 64% of the variance of the TUG with no additional variables entering into the equation (table 7).

Table 5. Stepwise Regression Analysis Results for the BBS

NOTE. Age entered as dichotomous variable >60 or ≤60 years.

Table 6. Stepwise Regression Analysis Results for FES-S

Table 7. Stepwise Regression Analysis Results for the TUG Test

Back to Article Outline

Discussion 

This study determined the relationship of falls-related self-efficacy, balance capacity, and functional mobility to fall history in a sample of community-dwelling people with chronic stroke. We also identified strength as a variable that explained a portion of the variance in all 3 of these constructs. These findings identify potentially modifiable features of the fall risk profile of the population of people with chronic stroke and may influence intervention strategies aimed at preventing falls.

The overall fall rate of our sample of community-dwelling stroke survivors was 40%. This rate was lower than the 48%,¹¹ 50%,⁹, ¹² 51%,¹⁰ and 73%⁸ reported by other studies of people with chronic stroke but higher than rates of 32%¹⁴ and 34%¹⁵ reported for the elderly. Our sample was recruited from community-based stroke support groups, which may have had an effect on the fall rate we observed. This sampling may account, in part, for the younger age (mean, 59.9y) and the high mobility level with 62% of subjects ambulating more than 2 blocks outdoors and over half reported ambulating without an assistive device. Our mean BBS score above 45 also indicates that subjects were high functioning.²² This level of functional independence may have contributed to the relatively low incidence of falling in our sample compared with other studies of community-dwelling people with stroke. Variations in subjects’ time poststroke and fall recording period utilized in different studies may also account for some of these differences. Nevertheless, a 40% fall rate is still of concern in this high functioning sample and successful falls prevention intervention may reduce the many unfortunate outcomes of falls.

Consistent with many studies of elderly¹⁴, ¹⁸, ²¹ and community-dwelling people with stroke,⁹, ¹¹, ¹³, ³⁶ the primary reason for falls reported by subjects was imbalance and the common activities being performed were walking followed by dressing.

This highlights the importance of patient education on safe walking and dressing. Dressing may consist of single-leg standing activities and reaching while in sitting that may be beyond the person’s lower-extremity and trunk balance capability. Lamb et al¹¹ suggest that in the assessment of stroke patients, questions relating to the frequency of balance problems during complex tasks such as dressing could be useful in identifying people at risk for falls.

In agreement with Hyndman and Ashburn²⁰ subjects with a history of multiple falls had lower BBS scores compared with subjects with 1 or no falls. In contrast, Harris et al¹² did not find a difference in BBS scores between their group of subjects with multiple falls compared with subjects who had not fallen. Harris¹² suggested that the inclusion of primary wheelchair users may have confounded their results. We excluded subjects who could not ambulate at least 10m without physical assistance. Additionally, we only had 1 subject who used a wheelchair as the primary mode of mobility so we could not establish any relationships between wheelchair use and balance. Harris et al¹² also excluded subjects with less than 1 year since stroke, whereas Hyndman and Ashburn²⁰ and the current study included subjects as early as 3 months poststroke. People with long-term chronic stroke may learn to adapt to their balance limitations and avoid compromising situations that would put them at risk for falls. Yates et al¹⁰ suggested that subjects with a greater number of stroke impairments and resulting limited mobility are at a lower risk of falling than those with fewer impairments who are more active. Therefore, community-dwelling persons most at risk as determined by performance-based measures may be taking fewer chances with their mobility and therefore, falling less often than predicted. This may account for the differential between the fall rate and the presumed fall risk in this population.

Considering fall risk in the above context may also explain why, although the BBS best differentiated subjects with multiple falls history, it was not strongly associated with falls history as indicated by the area under the ROC curve (.71). Additionally, the BBS cutoff score of 52 that had the best sensitivity in differentiating subjects with multiple falls was higher than other cutoff scores of 45,²², ²⁷ 38,²⁸ and 49²⁶ reported by others. These findings suggest the use of caution when applying previously published cutoff scores to select samples of any given population and that even when subjects with stroke were included in a sample,²², ²⁸ values generated from that sample may not be generalized to a stroke-specific population.

Methodologic variations in determining cutoff scores may also contribute to these differences. Berg et al²² established a cutoff score of 45 on the BBS on the basis of clinical experience and then used RR analysis to determine risk of falls with a score of less than 45. Bogle Thorbahn and Newton²⁷ used the recommendation of Berg when using 45 as the cutoff in their chi square analysis for fall risk. Shumway-Cook et al²⁶ used logistic regression models to generate cutoff scores that maximized both sensitivity and specificity and had a predicted probability of .5 or higher of classifying subjects with a history of falls and those who had not fallen. Chiu et al²⁸ derived the cutoff scores from ROC curves relative to the best overall sensitivity and specificity. Another factor that may account for the differences among these studies is the criteria for subject inclusion. Bogle Thorbahn and Newton²⁷ included subjects with both single and multiple fall history whereas other investigators²², ²⁶, ²⁸ considered only subjects with a history of multiple falls.

This was the first study to report the ability of the TUG to differentiate subjects with a history of falls in a sample of people with chronic stroke. We did not find any significant differences in TUG scores between groups. Our ROC curve analyses supported these findings with areas under the curves near .50 indicating that the ability of the TUG scores to differentiate between groups was near chance. This is in contrast to findings in studies of the elderly¹⁷, ²⁸ that found the TUG to be a good predictor of falls. These current findings suggest, as with the BBS, that tools shown to be effective in identifying fall risk in one population need to be specifically tested to determine if the same measurement applications transfer to other patient groups.

In contrast with the finding of Hyndman⁹ and Harris¹² and colleagues, we found that subjects with a history of multiple falls were 6 times more likely to use an assistive device. In a follow-up analysis, we found that subjects in the present study who used an assistive device were significantly older (P=.009). The impaired balance as a result of aging process and/or as a result of stroke in these subjects may require the use of the assistive device. Harris et al¹² found that those subjects who used a wheeled walker or wheelchair were less likely to fall. In the present study, only 1 subject used a walker, so a similar relationship could not be estimated.

These associations between assistive devices for ambulation and falls suggest that additional training with the appropriate assistive device may be indicated.

Our finding of a lower score on the mood and emotions domain of the SIS in subjects with a history of falls is supported by findings of several other researchers,⁸, ¹³, ¹⁹ who also reported an association between falling and depression after stroke. Hyndman et al⁹ found higher anxiety and depression scores in community-dwelling stroke survivors with a history of multiple falls. Depression may be both a physiologic and psychologic consequence of stroke and have a psychologic association with falling.⁸ Lower levels of falls-related self-efficacy have also been associated with depression in the elderly³¹, ³³ Jorgensen et al¹³ found a relationship between impaired lower-extremity mobility and a depression score. A causal relationship between falling and depression, however, is still not clear. Jorgensen¹³ explained that falls may lead people with stroke to limit their mobility, which in turn increases disability and results in depression. Alternatively, the authors also suggested that depression might cause impaired attention, which may lead to an increased risk for falling. The exact relationship between falls and depression requires further investigation.

Consistent with studies that have been reported in the general elderly¹⁶, ¹⁸, ³⁰ and stroke populations⁹, ²⁹, ³⁴ subjects with a history of falls expressed fear of falling and impaired fall-related self-efficacy in our sample. Similar to Friedman et al,²⁹ where people with stroke who had a falls history were more likely to develop fear of falling, the present study found subjects who had fallen to have lower levels of falls-related self-efficacy and to be more likely to be afraid of falling as compared with those who had not fallen. Subjects with a history of multiple falls were more likely to be afraid than those who had fallen once or not at all. Regardless of fall history, Friedman²⁹ found that people with fear of falling at baseline were 1.79 times more likely to fall and people with a history of falls were 5.4 times more likely than those with no falls history to develop fear of falling. Others have shown that fear of falling leads to restriction of activities and avoidance of risky behavior⁸, ¹⁶, ³⁰ raising the importance of assessing this construct. Hatch et al¹⁸ found an association between decreased balance confidence and falls in community-dwelling elders. We found falls-related self-efficacy as measured by the FES-S to best discriminate subjects who had fallen from those who had not with an area under the ROC curve of .71. While the cross-sectional design of this study does not allow an estimation of the causal relationships, the current findings agree with previous studies and suggest that falls-related self-efficacy and fear of falling may have predictive value and should be a standard part of fall risk assessment.

The variance of the FES-S was partially explained by 3 factors. The only capacity-based measure to explain the FES-S was the STS at 8%. The majority (72%) of the variance of falls-related self-efficacy was unexplained by the self-report and capacity-based variables included in this study. Interestingly, neither the TUG nor the BBS explained any part of falls-related self-efficacy. This may be because self-efficacy as measured on the FES-S relates to performance in a task-specific manner and the tasks used in the BBS and the TUG may not well reflect those in the FES-S. These findings differ from those of Hatch et al¹⁸ who found 57% of the balance confidence in elderly was explained by the BBS. The current analysis may have been affected by the use of the FES-S compared with the use of the Activities-Specific Balance Confidence (ABC) scale³² by Hatch.¹⁸ The ABC scale assesses more difficult activities and may be a better choice for use in this higher functioning population to avoid a ceiling effect.

We found no relationship between motor impairment as measured by the FMA and fall status. Yates et al¹⁰ studied community-dwelling stroke survivors in the first 6 months after stroke onset. Using the FMA lower-extremity score, the authors found that people with motor impairment (FMA score ≤28) were 2.2 times more likely to fall. The mean FMA lower-limb score for their population was 18.7, as compared with our mean of 23.8, suggesting that FMA may be a better tool for detecting people at fall risk in acute stage of stroke.

The number of medications taken was related to fall status in the case of the multiple fallers. Medications,¹⁶ specifically sedative medications,¹⁵ have been shown to be a risk factor for falls in elderly. However, in multivariate analyses, neither Lamb¹¹ nor Jorgensen¹³ and colleagues found a relationship between use of specific types of medications and risk for falling in subjects with chronic stroke. Our analysis was limited in that we analyzed only the number, not the type, of medications used. Our finding of the association of number of medications to risk for a single fall is notable considering the negative finding of others¹¹, ¹³ who have analyzed the relationship by type of medication. Future studies should investigate the effects of specific type of medications on falling in this population.

Composite muscle strength measured by STS was found to explain a significant portion of the variance of the BBS, TUG, and FES-S. The BBS includes many tasks that require muscle strength. It is reasonable then, that muscle strength explains some of the variance in the BBS. Because the action required for the STS is a component of the TUG it was expected that the STS would explain a large percentage of the variance in TUG. Weiss et al³⁶ demonstrated an improvement in STS after strength training in subjects with stroke suggesting that strength is a key element of this task. They recommend strength training to improve strength, motor performance, and balance in this population.³⁶ As a composite measure of lower-extremity strength, the STS may have limitations, including that both lower extremities contribute to the movement. Balance reactions may require strength of either or both lower extremities in order to prevent a fall. By design, the STS does not differentiate the contribution of the individual limbs to the accomplishment of the task. While the STS was not identified as an individual risk factor for falls in this study, the task of sit to stand was related to 3 of the variables significantly related to falls history, namely, balance capacity, functional mobility, and falls-related self-efficacy. These findings suggest that strength is an important contributing component of these constructs and should be addressed in people with long-term stroke. A large portion of the variance of the BBS and the TUG remained unexplained and may be accounted for by range of motion, individual muscle weakness, reaction time, and/or other cognitive impairments not included in this study.

This study does have some limitations. Because this was a cross-sectional study, predictive relationships cannot be assumed. Additionally, the functional state at the time of the fall may not have been the same as at the time the subject was assessed. Further studies to address the causal relationships between stroke-specific physical impairments, fear of falling, and falls are needed. As would be the case in most retrospective studies, false recall may have produced inaccuracies in reporting of the fall history retrospectively over 6 months. Whether the actual number of falls would be greater or less than 40% we cannot estimate. Also, because our sample consisted of fairly high functioning community-dwelling people with chronic stroke the current findings cannot be generalized to all people with stroke, particularly those at lower functional levels.

Back to Article Outline

Conclusions 

Falls at a rate of 40% are still occurring even in this high functioning population with stroke, warranting continued falls prevention intervention and education in this population. Subjects with a history of falls have fear of falling and decreased falls-related self-efficacy, suggesting that these constructs should be included in falls risk assessment and intervention. Subjects with a history of multiple falls have poorer balance capacity necessitating balance training to be performed even in these independently ambulating community-dwelling people.

Lower-extremity strength was found to be a determinant of balance capacity (BBS), of functional mobility (TUG), and of falls-related self-efficacy (FES-S). Continued strength training, as a part of rehabilitation, may be useful in people with chronic stroke to improve balance, mobility, and falls-related self-efficacy.

Supplier

Back to Article Outline

Acknowledgment 

We acknowledge Poonam Pardasaney, PT, MS, for her assistance with data analysis.

Back to Article Outline

References 

1. Thorvaldsen P , Asplund K , Kuulasmaa K , Rajakangas AM , Schroll M . Stroke incidence, case fatality, and mortality in the WHO MONICA Project . Stroke . 1995;26:361–367
   • View In Article
   • MEDLINE
   • CrossRef
2. American Heart Association . Heart disease and stroke statistics–2003 update . Dallas: AHA; 2002;
   • View In Article
3. Davenport RJ , Dennis MS , Wellwood I , Warlow CP . Complications after acute stroke . Stroke . 1996;27:415–420
   • View In Article
   • MEDLINE
   • CrossRef
4. Nyberg L , Gustafson Y . Patient falls in stroke rehabilitation. A challenge to rehabilitation strategies . Stroke . 1995;26:838–842
   • View In Article
   • MEDLINE
   • CrossRef
5. Byres V , Arrington ME , Finstuen K . Predictive risk factors associated with stroke patient falls in acute care settings . J Neurosci Nurs . 1990;22:147–154
   • View In Article
   • MEDLINE
   • CrossRef
6. Tutuarima JA , De Haan RJ , Limburg M . Number of nursing staff and falls (a case-control study on falls by stroke patients in acute-care settings) . J Adv Nurs . 1993;18:1101–1105
   • View In Article
   • MEDLINE
7. Teasell R , McRae M , Foley N , Bhardwaj A . The incidence and consequences of falls in stroke patients during inpatient rehabilitation (factors associated with high risk) . Arch Phys Med Rehabil . 2002;83:329–333
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (81 KB)
   • CrossRef
8. Forster A , Young J . Incidence and consequences of falls due to stroke (a systematic inquiry) . BMJ . 1995;311:83–86
   • View In Article
   • MEDLINE
   • CrossRef
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
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (69 KB)
   • CrossRef
10. Yates JS , Lai SM , Duncan PW , Studenski S . Falls in community-dwelling stroke survivors (an accumulated impairments model) . J Rehabil Res Dev . 2002;385–394
    • View In Article
11. Lamb SE , Ferrucci L , Volapto S , Fried LP , Guralnik JM . Risk factors for falling in home-dwelling older women with stroke . Stroke . 2003;34:494–501
    • View In Article
    • CrossRef
12. Harris JE , Eng JJ , Marigold DS , Tokuno CD , Louis CL . Relationship of balance and mobility to fall incidence in people with chronic stroke . Phys Ther . 2005;85:150–158
    • View In Article
    • MEDLINE
13. Jorgensen L , Engstad T , Jacobsen BK . Higher incidence of falls in long-term stroke survivors than in population controls . Stroke . 2002;33:542–547
    • View In Article
    • CrossRef
14. Tinetti ME , Speechley M , Ginter SF . Risk factors for falls among elderly persons living in the community . N Eng J Med . 1988;319:1701–1707
    • View In Article
15. Campbell AJ , Reinken J , Allan BC , Martinez GS . Falls in old age (a study of frequencies and related clinical factors) . Age Aging . 1981;10:264–270
    • View In Article
16. Tinetti ME , Williams TF , Mayewski R . Fall risk index for elderly patients based on number of chronic disabilities . Am J Med . 1986;80:429–434
    • View In Article
    • Abstract
    • Full-Text PDF (732 KB)
    • CrossRef
17. 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;80:896–903
    • View In Article
    • MEDLINE
18. Hatch J , Gill-Body KM , Portney LG . Determinants of balance confidence in community-dwelling elderly people . Phys Ther . 2003;83:1072–1079
    • View In Article
    • MEDLINE
19. Ugur C , Gucuyener D , Uzuner N , Ozkan S , Ozdemir G . Characteristics of falling in patients with stroke . J Neurol Neurosurg Psychiatry . 2000;69:649–651
    • View In Article
    • MEDLINE
    • CrossRef
20. Hyndman D , Ashburn A . People with stroke living in the community (attention deficits, balance, ADL ability and falls) . Disabil Rehabil . 2003;25:817–822
    • View In Article
    • MEDLINE
    • CrossRef
21. Berg K , Wood-Dauphinee SL , Williams JI , Gayton D . Measuring balance in the elderly (preliminary development of an instrument) . Physiother Can . 1989;41:304–311
    • View In Article
    • CrossRef
22. Berg KO , Wood-Dauphinee SL , Williams JI , Maki B . Measuring balance in the elderly (validation of an instrument) . Can J Public Health . 1992;83:7–11
    • View In Article
23. Berg KO , Wood-Dauphinee SL , Williams JI . The balance scale (reliability assessment with elderly residents and patients with an acute stroke) . Scand J Rehabil Med . 1995;27:27–36
    • View In Article
    • MEDLINE
24. Berg KO , Maki BE , Williams JI , Holliday PJ , Wood-Dauphinee SL . Clinical and laboratory measures of postural balance in an elderly population . Arch Phys Med Rehabil . 1992;73:1073–1080
    • View In Article
    • MEDLINE
25. Podsiadlo D , Richardson S . The timed “Up & Go” (a test of basic functional mobility for frail elderly) . J Am Geriatr Soc . 1991;39:142–148
    • View In Article
    • MEDLINE
26. Shumway-Cook A , Baldwin M , Polissar NL , Gruber W . Predicting the probability of falls in community-dwelling older adults . Phys Ther . 1997;77:812–819
    • View In Article
    • MEDLINE
27. Bogle Thorbahn LD , Newton RA . Use of the Berg Balance Test to predict falls in elderly persons . Phys Ther . 1996;76:576–582
    • View In Article
    • MEDLINE
28. 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
    • View In Article
    • MEDLINE
    • CrossRef
29. Friedman SM , Munoz B , West SK , Rubin GS , Fried LP . Falls and fear of falling: which comes first? A longitudinal prediction model suggests strategies for primary and secondary prevention . J Am Geriatr Soc . 2002;50:1329–1335
    • View In Article
    • MEDLINE
    • CrossRef
30. Maki BE , Holliday PJ , Topper AK . Fear of falling and postural performance in the elderly . J Gerontol . 1991;46:123–131
    • View In Article
31. Tinetti ME , Richman D , Powell L . Falls efficacy as a measure of fear of falling . J Gerontol . 1990;45:239–243
    • View In Article
32. Powell LE , Myers AM . The Activities-specific Balance Confidence (ABC) scale . J Gerontol A Biol Sci Med Sci . 1995;50:28–34
    • View In Article
33. Kressig RW , Wolf SL , Sattin RW , et al.   Association of demographic, functional and behavioral characteristics with activity-related fear of falling among older adults transitioning to frailty . J Am Geriatr Soc . 2001;49:1456–1462
    • View In Article
    • MEDLINE
    • CrossRef
34. Hellstrom K , Lindmark B . Fear of falling in patients with stroke (a reliability study) . Clin Rehabil . 1999;13:509–517
    • View In Article
    • MEDLINE
    • CrossRef
35. Csuka M , McCarty DJ . Simple method for measurement of lower extremity muscle strength . Am J Med . 1985;78:77–81
    • View In Article
    • Full-Text PDF (1011 KB)
    • CrossRef
36. Weiss A , Suzuki T , Bean J , Fielding RA . High intensity strength training improves strength and functional performance one year after stroke . Am J Phys Med Rehabil . 2000;79:369–376
    • View In Article
    • MEDLINE
    • CrossRef
37. Lord SR , Murray SM , Chapman K , Munro B , Tiedemann A . Sit to stand performance depends on sensation, speed, balance and psychological status in addition to strength in older people . J Am Geriatr Soc . 2002;57:M539–M543
    • View In Article
38. Fugl-Meyer AR , Jaasko L , Leyman I , Olsson S , Steglind S . The post-stroke hemiplegic patient. A method for evaluation of physical performance . Scand J Rehabil Med . 1975;7:13–31
    • View In Article
    • MEDLINE
39. Duncan PW , Propst M , Nelson SG . Reliability of the Fugl-Meyer assessment of sensorimotor recovery following cerebrovascular accident . Phys Ther . 1983;63:1606–1610
    • View In Article
    • MEDLINE
40. Sanford J , Moreland J , Swanson LR , Stratford PW , Gowland C . Reliability of the Fugl-Meyer assessment for testing motor performance in patients following stroke . Phys Ther . 1993;73:447–454
    • View In Article
    • MEDLINE
41. Duncan P , Wallace D , Lai SM , Johnson D , Embretson S , Laster LJ . The stroke impact scale version 2.0 (Evaluation of reliability, validity and sensitivity to change) . Stroke . 1999;30:2131–2140
    • View In Article
    • MEDLINE
    • CrossRef
42. Portney LG , Watkins MP . In: Foundations of clinical research (applications to practice) . 2nd ed.. Upper Saddle River: Prentice Hall; 2000;p. 333–334
    • View In Article
43. Metz CE . Basic principles of ROC analysis . Semin Nucl Med . 1978;8:283–298
    • View In Article
    • Abstract
    • Full-Text PDF (1846 KB)
    • CrossRef
44. Hanley JA , McNeil BJ . The meaning and use of the area under a receiving operator characteristic (ROC) curve . Radiology . 1982;143:29–36
    • View In Article
    • MEDLINE
45. Stratford PW , Binkley FM , Riddle DL . Health status measures (strategies and analytic methods for assessing change scores) . Phys Ther . 1996;76:1109–1123
    • View In Article
    • MEDLINE
46. Deyo RA , Centor RM . Assessing the responsiveness of functional scales to clinical change (an analogy to diagnostic test performance) . J Chronic Dis . 1986;39:897–906
    • View In Article
    • MEDLINE
    • CrossRef
47. Stratford PW , Binkley JM , Riddle DL , Guyatt GH . Sensitivity to change of the Roland-Morris Back Pain Questionnaire (part 1) . Phys Ther . 1998;78:1186–1196
    • View In Article
    • MEDLINE

• a SPSS Inc, 233 S Wacker Dr, 11th Fl, Chicago, IL 60606.