Clinical Analysis of Risk Factors for Falls in Home-Living Stroke Patients Using Functional Evaluation Tools

Methods 

Subjects were 112 stroke outpatients of our affiliated hospital, Hidakakai Hidaka Hospital. All patients had hemiparesis and were home living, and could walk independently with or without supporting devices such as a cane, orthosis of the lower extremities, and walker. All underwent a rehabilitation program involving walking, stretching, and muscle strengthening once to 3 times a week run by physical therapists and occupational therapists under the Japanese long-term care insurance system. After initial evaluation they were observed for 12 months, from January 1 to December 31, 2003. Among the 11 patients who dropped out, 3 died and 8 were readmitted due to recurrent stroke or other diseases not associated with falls. The remaining 101 stroke patients were included in the study. The study was approved by the local ethics committee at the hospital and written informed consent was obtained from all subjects and/or their families.

We evaluated the following baseline characteristics for all patients: sex, age, duration of disease, stroke type (ischemic stroke, hemorrhagic stroke, subarachnoid hemorrhage), affected side of the body (left, right), frequency of rehabilitation per week, and use of sedatives (yes, no) such as antiepileptics, psychotropic medicine, and sleeping medicine considered relevant with respect to increased fall risk. In addition, the Mini-Mental State Examination (MMSE), which has a maximum score of 30, was administered to assess cognitive function.22

We used the SIAS and FIM to evaluate various aspects of impairment and disability related to falls in stroke patients. The SIAS, developed for stroke outcome research in Japan,23 has a total of 22 items, with each item scored from 0 (severely impaired) to 5 (normal) for motor function, or 0 to 3 (normal) for muscle tone, sensory function, range of motion (ROM), pain, trunk strength, higher cortical function, and unaffected side function. The SIAS assesses various aspects of impairment in hemiplegic patients and shows interrater reliability, predictive validity, sensitivity, and scale quality.24 The FIM, which is used most often worldwide for evaluating the function of stroke patients,25 has motor and cognitive scores and is subdivided into 18 subcategories, each scored on a scale of 1 (total assistance) to 7 (complete independence). Total FIM scores range from 18 to 126. It has been found to show good reliability.26

We defined falls as any unexpected touch of the floor by any part of a patient’s body except for the soles of the feet. The information of falls was obtained from the patients or their family through interview with staff at each periodic rehabilitation. Patients were divided into 3 groups according to the frequency of falls: nonfallers, occasional fallers (patients experiencing only 1 fall during the study period), and repeat fallers (patients experiencing 2 or more falls during the study period). Table 1 summarizes the clinical characteristics of the stroke patients.

Table 1. Characteristics of Stroke Patients and Comparisons Among Nonfallers and Fallers

	Characteristics	Total (N=101)	Nonfallers (n=56)	Fallers (n=45)
				Total	Occasional (n=25)	Repeat (n=20)
	Sex (men/women)	62/39	35/21	27/18	15/10	12/8
	Age (y)	67.2±10.0	66.9±9.2	67.5±11.1	67.4±10.7	67.6±11.9
	Duration of disease (y)	6.1±4.6	5.7±4.4	6.6±4.7	6.8±4.6	6.5±5.0
	Stroke type
	Ischemic	48	27	21	14	7
	Hemorrhagic	49	27	22	10	12
	Subarachnoid hemorrhage	4	2	2	1	1
	Affected side of body (right/left)	51/50	32/24	19/26	15/10	4/16⁎
	Frequency of rehabilitation (/wk)	1.8±0.6	1.8±0.5	1.8±0.6	1.6±0.6	2.0±0.5
	Use of sedatives	29	18	11	6	5
	MMSE score	24.5±4.7	24.4±4.5	24.7±4.9	24.9±5.7	24.4±3.9
	Total SIAS score	46.1±11.1	46.7±11.6	45.5±10.4	45.1±9.2	46.0±11.9
	Total FIM score	107.4±13.1	109.0±9.3	105.5±16.5	110.2±9.9	99.6±21.1
	Motor FIM score	75.7±10.3	77.0±8.5	74.1±12.2	77.8±7.9	69.5±15.0
	Cognitive FIM score	31.9±4.7	32.2±3.8	31.4±5.7	32.4±4.1	30.1±7.1

NOTE. Values are n or mean ± standard deviation (SD).

⁎ Nonfallers versus repeat fallers (P<.05).

Results 

Total fallers (occasional and repeat fallers) comprised 45 (44.6%) of the 101 patients. Repeat fallers totaled 20 (19.8%) patients. The frequency of falls in repeat fallers ranged between 2 and 10 times (mean, 2.1) during the study period. Four patients sustained fractures; 1 had a humeral fracture, 1 had a hip fracture, and 2 had a lumbar compression fracture. Most fallers fell in their home during ADLs.

No significant difference in variables between nonfallers and total fallers was recognized except for the memory score on the FIM.

Bivariate analysis revealed the following factors to differ significantly between nonfallers and repeat fallers: self-care for toileting, bladder management, transfer from bed to chair, transfer to toilet and transfer to tub or shower, as well as memory score on the FIM, foot tap test, gripping power score, and ROM of the lower extremities score of SIAS, duration of disease, and affected side (all P<.10) (Table 2, Table 3). As shown in table 4, some variables were highly correlated and as such they could not be entered simultaneously in stepwise logistic regression analysis. The logistic model was refined until it included only 4 predictors: memory score on the FIM, ROM of the lower extremities on the SIAS, duration of disease, and affected side (table 5). The logistic model using the 4 predictors showed a high predictive value (table 6). Sensitivity of fall prediction was 81.3% and specificity was 88.1% (Hosmer-Lemeshow goodness of fit test, P=.692).

Table 2. Comparison of Nonfallers and Repeat Fallers (FIM)

	Subscale and Items	Nonfallers	Repeat Fallers	P
	Motor
	Self-care
	Eating	6.6±0.8	6.7±0.7	.586
	Grooming	6.0±1.3	5.3±1.9	.320
	Bathing	4.3±2.2	3.6±2.5	.145
	Dressing upper body	5.5±1.6	4.6±2.4	.697
	Dressing lower body	5.5±1.6	4.6±2.3	.382
	Toileting	6.4±0.8	5.5±1.9	.086⁎
	Sphincter control
	Bladder	6.9±0.3	6.4±1.2	.037⁎
	Bowel	6.8±0.4	6.6±1.0	.977
	Transfers
	Bed/chair	6.7±0.5	6.3±1.1	.087⁎
	Toilet	6.4±0.5	6.0±0.9	.066⁎
	Tub/shower	5.1±1.6	4.3±2.0	.073⁎
	Locomotion
	Walk/chair	6.0±0.8	5.8±1.1	.496
	Stairs	4.6±1.9	4.1±1.8	.270
	Cognitive communication
	Comprehension	6.5±1.1	6.4±1.3	.347
	Expression	6.3±1.2	6.2±1.6	.734
	Social cognition
	Social interaction	6.5±1.1	6.0±1.8	.139
	Problem-solving	6.1±1.1	5.6±1.8	.311
	Memory	6.8±0.4	6.0±1.6	.015⁎

⁎ P<0.1.

Table 3. Comparison of Nonfallers and Repeat Fallers (SIAS)

	Subscale and Items	Nonfallers	Repeat Fallers	P
	Motor function
	Knee-mouth	2.3±1.5	2.2±1.5	.878
	Finger function	1.5±1.5	1.9±2.0	.597
	Hip flexion	3.1±1.2	3.1±1.1	.779
	Knee extension	2.8±1.2	2.8±1.2	.990
	Foot tap	2.0±1.7	1.4±1.6	.066⁎
	Muscle tone
	DTR: UE	1.8±0.7	2.0±0.7	.229
	LE	1.7±0.7	1.8±0.7	.429
	Tone: UE	1.6±0.8	1.7±0.9	.402
	LE	1.8±0.9	1.9±0.9	.694
	Sensory function
	Touch: UE	1.9±0.9	1.5±1.0	.124
	LE	2.0±1.0	1.5±1.1	.135
	Position: UE	2.0±1.1	2.0±1.1	.914
	LE	2.0±1.1	1.8±1.1	.541
	ROM
	Shoulder abduction	1.8±0.7	2.0±0.9	.172
	Ankle dorsiflexion	1.6±0.8	2.1±0.8	.010⁎
	Pain	2.5±0.6	2.3±1.0	.803
	Trunk
	Verticality	2.7±0.5	2.7±0.6	.801
	Abdominal MMT	2.2±0.7	2.2±1.0	.897
	Higher cortical function
	Visuospatial	2.6±0.7	2.7±0.6	.567
	Speech	2.6±0.7	2.8±0.5	.144
	Unaffected side function
	Grip strength	2.2±0.6	1.9±0.7	.058⁎
	Quadriceps MMT	2.2±0.8	2.1±0.8	.257

Abbreviations: DTR, deep tendon reflex; LE, lower extremities; MMT, manual muscle test; UE, upper extremities.

⁎ P<0.1.

Table 4. Spearman Correlation Coefficients Among Variables

		Toileting	Bladder Control	Bed/Chair	Toilet	Tub/Shower	Memory	Foot Tap	GP	ROM (LE)	Duration	Side
	Toileting	1.000	0.277⁎	0.542‡	0.602‡	0.640‡	0.140	0.262⁎	0.182	0.042	0.293⁎	−0.110
	Bladder control		1.000	0.250⁎	0.265⁎	0.221	0.293⁎	0.187	0.094	−0.002	0.053	−0.136
	Bed/chair			1.000	0.526‡	0.506‡	0.313‡	0.230	0.219	0.074	0.090	−0.079
	Toilet				1.000	0.519‡	0.145	0.264⁎	0.277⁎	0.059	0.024	−0.119
	Tub/shower					1.000	0.105	0.480‡	0.133	−0.004	0.072	−0.135
	Memory						1.000	−0.019	−0.026	0.227	−0.068	0.039
	Foot tap							1.000	0.307	−0.137	−0.056	−0.171
	GP								1.000	−0.204	−0.086	−0.037
	ROM (LE)									1.000	−0.043	0.022
	Duration										1.000	−0.102
	Side											1.000

Abbreviations: Bed/Chair, transfer of bed, chair score in FIM; Bladder control, bladder control score in FIM; Duration, duration of disease; Foot tap, foot tap test in SIAS; GP, gripping power score in SIAS; Memory, memory score in FIM; ROM (LE), ROM of lower extremities in SIAS; Side, affected side of body; Toilet, transfer of toilet score in FIM; Toileting, self-care of toileting score in FIM; Tub/shower, transfer of tub or shower score in FIM.

⁎ P<.05; ‡ P<.01.

Table 5. Stepwise Logistic Models for Predicting Falls

	Predictors	OR	95% CI	P
	FIM
	Memory	0.252	0.093–0.679	.006
	SIAS
	ROM of lower extremities	4.278	1.637–11.179	.003
	Duration of disease	1.262	1.074–1.483	.005
	Affected side of body (right/left)	0.076	0.013–0.444	.004

Table 6. Predictive Value of the Logistic Model

	Predicted Outcome	Observed Fall (case)	Observed No Fall (case)	Predictive Value (%)
	Predicted fall	13	3	81.3
	No predicted fall	7	52	88.1
	Overall			86.7

Discussion 

This is the first study to report the ability of the SIAS and FIM to predict falls in home-living stroke patients. The fall rate in this study (44.6%) was lower than in previous studies (50%−73%)7, 19, 20 aimed at home-living patients. Participants in our study continued rehabilitation during the study period, which may have reduced the risk of falls.

Comparing nonfallers and total fallers, we found no significant difference except for memory score on the FIM. Cognitive scores on the FIM reflect communicative and social cognitive function in daily life, and cognitive deficits may lead to reduced attention and thereby to an increased risk of falling.27, 28

An important area in the research of falls is the study of repeat fallers. One fall can be a chance occurrence, but repeat falls can lead to an increased risk of injury. Therefore, we also compared the differences between nonfallers and repeat fallers, a classification of falls that has been used in several studies.19, 20, 21

Aside from the FIM and SIAS scores, affected side and duration of disease were the only other variables related to fall risk in the present study. Rapport et al29 when reporting the predictors of falls among stroke patients with an affected right hemisphere, described the causes of falls in affected left-hemisphere stroke patients to be inattention, perceptual deficits, and hemispatial neglect. Although the visuospatial function of the SIAS score was not associated with fall risk in the present study, we often find a difference in hemispatial neglect between static and dynamic situations. Several studies have shown, however, no significant relationship between the risk of fall and the affected side.14, 15

The bivariate analysis showed that 9 category scores of the SIAS and FIM were related to fall risk: self-care for toileting, bladder management, transfer from bed to chair, transfer to toilet and transfer to tub or shower, as well as memory score on the FIM, foot tap test, gripping power score and ROM of the lower extremities score of SIAS, duration of disease, and affected side. It seems reasonable to assume that memory disturbance leads to repeat falls due to a lack of awareness of the risk. Several studies have shown the relationship between falls and cognitive deficit.8, 30, 31 However, the MMSE score showed no relationship to fall risk in the present study. We suggest that because the cognitive score on the FIM was evaluated through observation of the patient’s daily activities, it slightly differed from the MMSE score evaluated through examination.

The low score on self-care for toileting and bladder management on the FIM increased fall risk in this study. A possible explanation for this may be that urinary incontinence is also a plausible risk factor for falls because it is often coincident with a general decline in physical and mental functioning.32, 33 These findings indicate we must pay attention to patients who have such disorders to prevent falls.

It is interesting to note that unlike other functions, limitation in the ROM of the ankle reduced fall risk in this study. This finding suggests that immobilization of the distal portion of the affected lower extremity stabilized motion and posture.

Analysis of the total score of the FIM and scores of the motor FIM subscale and cognitive FIM subscale revealed no significant relations to falls through bivariate analysis or multivariate analysis. Whitson and Pieper34 reported the nonlinear relationship between total FIM score and the fracture risk. In their study, patients with intermediate functional impairment had higher fracture risk than patients with mild or severe functional impairment. They thought that severely impaired patients had limited mobility and little opportunity to fracture. In our study, all the participants were living in the community and could walk independently, so their total score of the FIM was high (mean, 107.4; range, 56−124), but we found no significant relationship to falls. However, we found significant relations to falls for several individual variables including the memory score and the bladder management score, indicating that change in an individual item of the FIM and SIAS multi-item tests has little effect on the total scores. The total scores of these tests are measures of overall function, and it is therefore difficult to pinpoint which function affects fall risk. We analyzed individual items on these tests in this study in order to determine specifically what kind of impairment and disability increases the risk of falls.

Several studies have attempted to predict falls in stroke patients using balance scores among other factors.12, 21, 35, 36 In these studies, balance problem, arm function, ADL ability, and depressive symptoms were reported as risk factors. Our study shows that functional evaluation tests such as the SIAS and FIM are sufficiently sensitive to detect functional deficits related to falls. These evaluation tools are generally used at initial assessment and the scores can therefore enable close attention to be paid to patients with the aforementioned risk factors during rehabilitation or daily activities. The findings of the present study are also useful for designing interventions to reduce fall frequency among home-living stroke patients.

Conclusions 

Falls occurred at a rate of over 40% in home-living stroke patients. The study revealed the usefulness of clinical evaluation using the SIAS and FIM to predict repeat falls in the high-risk group of home-living stroke patients.