Home > RNJ > 2008 > May/June > An Analysis of Falls Occurring in Patients with Stroke on an Acute Rehabilitation Unit

An Analysis of Falls Occurring in Patients with Stroke on an Acute Rehabilitation Unit
Meheroz H. Rabadi, MD MRCPI Freny M. Rabadi, BS (Hons) Margaret Peterson, PhD

Falls are a common occurrence in stroke patients admitted to a rehabilitation unit. This study evaluates the effect of current fall risk screening and prevention strategies on the number of falls in stroke patients admitted to an acute rehabilitation facility. All stroke patients admitted to a designated acute rehabilitation unit were considered at risk for falls by virtue of their diagnosis. These patients were studied retrospectively during a 24-month period. Wheelchair lap belts and bedrails were provided to all patients admitted to the unit. Further measures consisting of bed and chair alarms, enclosed beds, and placement in rooms close to the nursing station were implemented with high–fall risk patients. One hundred seventeen (15.5%) of the 754 patients in this study fell. The fall index rate was 8.2 falls, based on patient care days during the study period. No injury was observed in 143 of 159 fall cases (90%). In 13 cases (8%) there were only minor injuries, and 3 falls (2%) resulted in serious injuries. Current preventive strategies decrease the number of falls and the severity of fall-related injuries. Patients who do fall are cognitively impaired on admission and have lower ambulation speed.

Falls are a common occurrence in stroke patients admitted to a rehabilitation unit. The reported fall frequency rate is 14% in the acute hospital (Tutuarima, de Haan, & Limburg, 1993) and 25%–39% in the acute rehabilitation setting (Dromerick & Reding, 1994; Nyberg & Gustafson, 1995). Falls are more common in rehabilitation because of its setting, the patient population’s age, level of staffing, and length of stay (Nakai, Akeda, & Kawabata, 2006). The increased risk for falls in an acute rehabilitation setting results from the active promotion of patient mobility and independence.

Numerous risk factors for falls have been identified in stroke patients. These include age, right hemisphere lesion and impulsivity (Rapport, Webster, & Fleming, 1993), cognitive impairment, severity of neurological impairments, poor balance (Cheng et al., 1998), urinary incontinence (Nyberg & Gustafson, 1997), and sedating and psychotropic drugs (Leipzig, Cumming, & Tinetti, 1999).

Falls lead to loss of confidence, reluctance to ambulate, and varying degrees of injuries, from minor lacerations and abrasions to severe injuries such as hip fractures. Falls in stroke patients are a leading cause of hip fractures (Ramnemark, Nilsson, Borssen, & Gustafson, 2000; Ramnemark, Nyberg, Borssen, Olsson, & Gustafson, 1998), which usually occur on the affected side because of weakness and disuse osteoporosis (Mulley & Espley, 1979). Therefore, the authors of this study decided to examine whether fall prevention strategies used with stroke patients admitted to an acute rehabilitation unit could decrease the number of falls and facilitate patient participation in rehabilitation programs without lowering their level of activity. In addition, this study attempted to predict which risk factors were responsible for falls in all patients who had undergone initial fall risk screening and preventive measures.

The goal of this study was to define the fall incidence rate in patients who routinely undergo fall prevention measures on an acute stroke rehabilitation unit because this has not been previously defined.



All stroke patients consecutively admitted to a designated stroke rehabilitation unit during a 24-month period were studied retrospectively. The inclusion criteria were first ischemic or hemorrhagic stroke, admission within 4 weeks of the acute event, and neuroimaging (computed tomography [CT] or magnetic resonance imaging [MRI]) showing the lesion corresponding to the clinical signs and symptoms. The exclusion criteria were hemorrhage within a brain tumor and sudden onset of clinical signs and symptoms due to brain lesion other than vascular cause. The data collected included age, sex, lesion type as defined by the International Classification of Disease Version 9 (e.g., thrombotic, embolic, carotid occlusion, and acute ill-defined stroke), lesion location determined by CT or MRI (e.g., right, left, or bilateral hemisphere lesions; cerebellar lesions; brainstem lesions), stroke severity based on the neurological examination given at admission, rehabilitation hospital admission and discharge total Functional Independence Measure (FIM™) scores, and length of stay. Approval from an institutional review board was not obtained, and Health Information Portability and Accountability Act regulations do not apply because this was an observational study performed as part of a routine performance improvement audit of fall risk screening and preventive strategies mandated by the Joint Commission.

Stroke severity was assessed by a neurologist at admission to the rehabilitation facility and was defined by neurological impairments (in cognition, visual field defects, motor weakness, and sensory impairment). Stroke severity was graded using an ordinal neurological impairment scale of 1 (motor impairment only), 2 (motor plus hemianopic vision or motor plus somatic sensory impairment), or 3 (motor plus hemianopic vision plus somatic sensory impairment) (Reding & Potes, 1988). Cognitive state was examined using the Mini-Mental State Examination (MMSE). The MMSE is a reliable (O’Connor et al., 1989) and valid (Tombaugh & McIntyre, 1992) screening tool used to measure cognitive function and help quantify the severity of cognitive impairment (Folstein, Folstein, & McHugh, 1975). It has a total score of 30, with a cutoff score of 25, above which the person is considered cognitively intact. The test evaluates orientation, memory, attention and calculation, language, and visuospatial domains. The main limitations of the MMSE are that it is influenced by age, education, language, and cultural background. Motor weakness was evaluated using the Motricity Index, a weighted score derived from Medical Research Council grades. The patient is examined while lying in bed, and three arm movements (pinch grip, elbow flexion, and shoulder abduction) of the unaffected side are compared with those of the affected side. Similarly, three leg movements (ankle dorsiflexion, knee extension, and hip flexion) of the unaffected side are compared with those of the affected side (Sunderland, Tinson, Bradley, & Langton-Hewer, 1989). The MI is scored for the arm and the leg separately. The total score for the arm and the leg ranges from 1 (no movement) to 100 (normal power for age). This is a valid and reliable scale for evaluating motor weakness of the upper motor neuron type. The scale is sensitive to change associated with stroke recovery (Cameron & Bohannon, 2000; Collin & Wade, 1990). Homonymous visual field deficits (hemianopsia or visual neglect) were assessed at the bedside by confrontation testing. Sensory impairment and proprioceptive loss were evaluated using the Limb Placement Task (Yarnell & Boutell-Friedman, 1987). A mean error in the most affected quadrant of more than 6 in. indicates a significant somatic sensory deficit. Patients unable to comprehend this task after repeated gestural clues were scored as abnormal. This assessment scale was reliable in 90% of cases (Yarnell & Boutell-Friedman). Controlled comparisons indicate that the Limb Placement Task is a more objective than subjective assessment of primary sensations (i.e., touch, temperature, pinprick, vibration sense) (Reding & Potes).

The Fugl-Meyer (F-M) balance subscore (Fugl-Meyer, Jääskö, Leyman, Olsson, & Steglind, 1975) and Berg Balance Scale scores for balance (Berg, Wood-Dauphinee, & Gayton, 1989; Berg, Wood-Dauphinee, & Williams, 1995; Berg, Wood-Dauphinee, Williams, & Maki, 1992) were obtained by the patient’s assigned physical therapist on admission and discharge. The total FIM™ on admission and discharge was used to determine the patient’s degree of disability and the progress he or she made during rehabilitation (Granger, Hamilton, Linacre, Heinemann, & Wright, 1993). The FIM™ scale is a reliable (Granger, 1998; Stineman et al., 1996) and valid measure of functional independence (Stineman & Maislin, 2000). The total score ranges from 18 (highest level of dependence) to 126 (highest level of independence). The team members who are FIM™ trained and certified recorded admission and discharge FIM™.


Fall prevention strategies consisting of wheelchair lap belts and bedrails were provided to all patients admitted to the unit. Two-hour timed voiding schedules are used regularly with all incontinent patients, and neuropsychiatric drugs (i.e., antipsychotics, sedatives, anticonvulsants) are used sparingly. Further measures, consisting of bed and chair alarms, enclosed beds, and rooms close to the nursing station where patients could be supervised closely between therapy and during meals, were undertaken in high–fall risk patients. Patients were considered at high fall risk on admission based on their history of prior falls in the acute referring hospital, impulsivity based on observed behavior, and poor insight into their impairments and disabilities.


The primary outcome variables were the fall frequency rate and the location and severity of fall-related injuries. A fall was defined as an unintentional landing on the floor or being eased to the floor. Both regulatory and hospital policies require all staff members to record and report all falls occurring in the hospital to the nursing administration. The following data were recorded in the fall incident reports: patient’s name and medical record number, date and location of the fall, the severity of the resultant injuries (Figure 1), and treatment rendered. Only falls occurring during the patient’s rehabilitation stay were recorded in this study. The term fall index (I) was calculated using the formula I = A/R x 1,000, where A is the number of fall incidents and R is the number of patient care days during the observation period.

Statistical Analysis

The data distribution of each variable was examined and outliers were checked. The dependent variable was fall occurrence. Continuous variables (parametric data) were compared between groups using Student’s t test, ordinal data were compared using the Mann-Whitney test, and nominal variables (nonparametric data) were analyzed using Pearson’s chi-square and the Fisher’s Exact tests. Alpha was set to .05. No adjustment was made at this stage for multiple comparisons.

Backward stepwise logistic regression analysis (a correlational analysis) was used to predict which variables were responsible for increasing the risk of falling in these patients despite the fall prevention strategies. The dependent variable was being a faller or a nonfaller. All statistical analyses were performed using the Statistical Package for the Social Sciences (Version 10.0, SPSS Inc., Chicago, IL).


The sample consisted of 754 stroke patients studied over a 24-month period. The mean age (± SD) of our patients was 70 ± 13 years. There were 361 men and 393 women. The stroke onset to rehabilitation admission time was 12 ± 7 days. The mean length of stay was 17 ± 9 days.

Of the 754 patients admitted to the unit during the study period, 117 (15.5%) fell. There were 159 fall incident reports for these 117 patients during their rehabilitation stay (range 0 to 36 days), or 0.21 falls per patient (Table 1). The fall index for the study population was 159/19,412 x 1,000 = 8.2, falls based on patient care days during the 24-month study period. The most frequent fall location was the patient’s own room (n = 111, 70%), followed by the bathroom or toilet (n = 24, 15%), corridors (n = 9, 5.6%), and various other locations (n = 10, 6%). Few falls occurred when patients were actively undergoing therapy (n = 9, 5.6%). No fall-related injuries were observed in 143 of 159 fall cases (90%). In 13 cases (8%) there were only minor injuries (bruises, soft tissue tenderness, or minor wounds requiring no sutures). Three falls (2%) resulted in serious injuries: two patients sustained hip fractures of the paretic side, and one patient sustained a basal skull fracture and died of an intracranial hemorrhage.

Most of the falls (n = 67, 42%) occurred during the first week of the patient’s stay, and the number of falls decreased with each subsequent week the patient was hospitalized (Table 1). Ninety-one patients were one-time fallers, 18 patients fell twice, 5 patients fell three times, and 3 patients fell four times. The average admission FIM™ score for one-time fallers was 50 ± 16, and this score declined to 40 ± 20 for those who had experienced three or more falls (p = .31).

The demographic and fall risk factors were not statistically different when fallers (N = 117) and nonfallers (N = 637) were compared for age (p = .40), male/female ratio (p = .22), and stroke location (p = .15) (Tables 2–6). However, there were differences in onset to admission time (p = .05), stroke type (p = .002), admission MMSE (p = .0001), visual field defects (p = .001), lower extremity MI (p = .03), Limb Placement Task (p = .004), stroke severity (p = .02), Berg Balance Scale (p = .009), and admission total FIM™ (p = .0001) between fallers and nonfallers (Tables 1–5). Thus, the fallers had more cognitive, visual, physical, and proprioceptive impairments, more postural instability, and a lower admission ambulation speed.

All the variables known to influence falls—cognitive, visual, physical, and proprioceptive impairments, postural instability, and the level of ambulation—were entered into a backward stepwise logistic regression. Only the admission ambulation speed (p = .004) and MMSE (p < .001) remained in the equation and predicted the fallers. Correct predictions were made for 68% of the cases, with fallers being predicted more often. Patients with impaired cognition (MMSE score of 24 or less) and a lower admission ambulation speed (0.49 m/s or less) were at higher risk for falls in the study population, with a prediction rate of 66%, p = .001. Patients who ambulated at 0.50 m/s or more had no falls regardless of their MMSE score. Of the 397 patients who ambulated at 0.49 m/s or less and had an MMSE score between 18 and 24, 50 (13%) fell, and of the 268 patients who ambulated at 0.49 m/s or less and had an MMSE below 18, 67 (25%) fell (Table 5).


During the 24-month retrospective study, fall prevention strategies decreased both the number of falls and the severity of fall-related injuries. The fall rate quoted in the literature for patients admitted to an acute stroke rehabilitation unit varies from 25% (Dromerick & Reding, 1994) to 39% (Nyberg & Gustafson, 1995) and from 0.76 falls per patient (Teasell, McRae, Foley, & Bhardwaj, 2002) to 0.95 falls per patient (Nyberg & Gustafson, 1995). The fall index (incidence) for the 24-month study period was 8.2 falls, based on our patient care days on the stroke unit, whereas the published fall rate varies from 9 to 19 per 1,000 patient care days for acute stroke inpatient rehabilitation (Mayo, Korner-Bitensky, Becker, & Georges, 1989; Nyberg & Gustafson, 1995, 1997). However, direct comparison between this study and other fall studies is not feasible because acute rehabilitation settings differ and none of these studies mention or undertake fall screening and prevention strategies.

The study results presented in this article are consistent with other studies reporting a low incidence of serious fall-related injuries (Fleming & Pendergast, 1993; Nyberg & Gustafson, 1995). Hip fracture rates of 4%–15% have been reported on the hemiplegic side in patients within 1 year of stroke. These hip fractures occur because of motor weakness and the development of disuse osteoporosis in the hemiplegic limb (Mulley & Espley, 1979).

In their study on a general medical rehabilitation unit, Mion and colleagues (1989) noted that the majority of patients fell during the first 2 weeks of hospitalization, and the fall frequency decreased linearly with the length of stay, whereas Mayo and colleagues (1989) found that 50% of first falls occurred within 13 days of admission to their rehabilitation hospital. Similarly, in this study most falls occurred during the first week of hospitalization. These patients had less motor weakness and proprioceptive impairment, whereas patients who fell during the third or fourth week of their hospital stay had more motor and proprioceptive impairment (p = .01). The admission FIM™ for one-time fallers was higher than for those who had multiple (i.e., three or more) falls but did not reach statistical significance because there were too few multiple fallers in this study. Teasell and colleagues (2002) also found a higher average admission FIM™ in one-time fallers than in those who experienced four or more falls.

It is not surprising that most falls occur in the patient’s own room, given that patients spend most of their time in their own rooms. In addition, patients undergoing therapy are closely supervised and therefore are at lower risk for falls during that time. In their study in an adult care facility, Fleming and Pendergast (1993) found the highest number of falls in the patient’s own room, followed by the bathroom. In their study of stroke patients on an acute rehabilitation unit, Nyberg and Gustafson (1995) noted that most falls occurred during transfers or when patients tried to climb over bedrails. Few falls occurred during supervised therapy.

In this study, although significant differences between fallers and nonfallers were noted on admission with respect to lesion type, neurological (e.g., cognitive, visual, motor, perceptual) impairments both individually and collectively, the patient’s ability to balance and ambulate, and a low score on total FIM™ for neurological disability, only the admission cognitive status (MMSE score less than 18) and the lower admission ambulation speed (99 ft/min or less) predicted the patients who were at an increased risk of falling. This is not surprising given that cognitively impaired patients who have limited ambulatory abilities often have decreased awareness of their functional limitations. These patients often attempt to ambulate when left alone in their rooms, thus increasing their risk of falling, irrespective of the fall prevention strategies instituted. Identification of these fall risk factors and the increased risk of falls during the first week of hospitalization have led to the implementation of additional preventive strategies, such as visiting patients’ rooms frequently, installing antitippers, using bed and chair alarms, and removing leg rests on wheelchairs to further decrease the number of falls. Furthermore, a fall risk screening tool called the Safety Promotion and Falls Prevention Plan is being implemented at the facility.

There were several limitations to this study. First, it was not possible to determine from the incident reports which activities were associated with falls while the patient was in his or her room. Second, data regarding the relationship between the time of day of a fall and nurse staffing levels and staff availability were not available. Third, the study was limited to the time period (5 weeks) during which the patient was being rehabilitated on the stroke unit. However, fall frequency is influenced by the time period over which it is studied. Finally, this was not a randomized controlled study because it is not ethical to withhold fall prevention strategies in order to ascertain effectiveness. Despite these limitations, our study sample is fairly representative of moderately to severely disabled stroke patients who need a longer hospital stay and further rehabilitation after the acute phase and in whom fall prevention strategies can reduce the number of falls and fall- related injuries without curtailing their level of activity.


In conclusion, fall prevention strategies may decrease both the number of falls and the severity of fall-related injuries, and patients are at an increased risk of falling if they are cognitively impaired and had a lower ambulation speed on admission.

About the Authors

Meheroz H. Rabadi, MD MRCPI, is an attending neurologist and assistant professor of clinical neurology at Weill Medical College of Cornell University, New York, NY. Address correspondence to Dr. Rabadi at 785 Mamaroneck Avenue, White Plains, NY 10605 or mrabadi@burke.org.

Freny M. Rabadi, BS (Hons), is a research assistant at Burke Rehabilitation Hospital, White Plains, NY.

Margaret Peterson, PhD, is a professor of statistics at Weill Medical College of Cornell University at Hospital for Special Surgery, New York, NY.


Berg, K., Wood-Dauphinee, S., & Gayton, D. (1989). Measuring balance in the elderly: Preliminary development of an instrument. Physiotherapy Canada, 41, 304–310.

Berg, K., Wood-Dauphinee, S., & Williams, J. I. (1995). The Balance Scale: Reliability assessment with elderly residents and patients with an acute stroke. Scandinavian Journal of Rehabilitation Medicine, 27, 27–36.

Berg, K., Wood-Dauphinee, S., Williams, J., & Maki, B. (1992). Measuring balance in the elderly: Validation of an instrument. Canadian Journal of Public Health, 83, 57–61.

Cameron, D., & Bohannon, R. W. (2000). Criterion validity of lower extremity Motricity Index scores. Clinical Rehabilitation, 14(2), 208–211.

Cheng, P. T., Liaw, M. Y., Wong, M. K., Tang, F. T., Lee, M. Y., & Lin, P. S. (1998). The sit-to-stand movement in stroke patients and its correlation with falling. Archives of Physical Medicine Rehabilitation, 79(9), 1043–1046.

Collin, C., & Wade, D. (1990). Assessing motor impairment after stroke: A pilot reliability study. Journal of Neurology, Neurosurgery, and Psychiatry, 53(7), 576–579.

Dromerick, A., & Reding, M. (1994). Medical and neurological complications during inpatient stroke rehabilitation. Stroke, 25, 358–361.

Fleming, B. E., & Pendergast, D. R. (1993). Physical condition, activity pattern, and environment as factors in falls by adult care facility residents. Archives of Physical Medicine Rehabilitation, 74, 627–630.

Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). Mini–Mental State: A practical method for grading the cognitive state of patients for the clinician.. Journal of Psychiatric Research, 12, 189–198.

Fugl-Meyer, A. R., Jääsko, L., Leyman, I., Olsson, S., & Steglind, S. (1975). The post-stroke hemiplegic patient, I: A method for evaluation of physical performance. Scandinavian Journal of Rehabilitation Medicine, 7, 13–31.

Granger, C. V. (1998). The emerging science of functional assessment: Our tool for outcomes analysis. Archives of Physical Medicine Rehabilitation, 79(3), 235–240.

Granger, C. V., Hamilton, B. B., Linacre, J. M., Heinemann, A. W., & Wright, B. D. (1993). Performance profiles of the Functional Independence Measure. Archives of Physical Medicine Rehabilitation, 72, 84–89.

Leipzig, R. M., Cumming, R. G., & Tinetti, M. E. (1999). Drugs and falls in older people: A systematic review and meta-analysis: I. Psychotropic drugs. Journal of the American Geriatric Society, 47(1), 30–39.

Mayo, N. E., Korner-Bitensky, N., Becker, R., & Georges, P. (1989). Predicting falls among patients in a rehabilitation hospital. American Journal of Physical Medicine & Rehabilitation, 68(3), 139–146.

Mion, L. C., Gregor, S., Buettner, M., Chwirchak, D., Lee, O., & Paras, W. (1989). Falls in the rehabilitation setting: Incidence and characteristics. Rehabilitation Nursing, 14(1), 17–22.

Mulley, G., & Espley, A. J. (1979). Hip fracture after hemiplegia. Postgraduate Medical Journal, 55(642), 264–265.

Nakai, A., Akeda, M., & Kawabata, I. (2006, October). Incidence and risk factors for inpatient falls in an academic acute-care hospital. Journal of Nippon Medical School, 73(5), 265–270.

Nyberg, L., & Gustafson, Y. (1995). Patient falls in stroke rehabilitation. A challenge to rehabilitation strategies. Stroke, 26, 838–842.

Nyberg, L., & Gustafson, Y. (1997). Fall prediction index for patients in stroke rehabilitation. Stroke, 28, 716–721.

O’Connor, D. W., Pollitt, P. A., Hyde, J. B., Fellows, J. L., Miller, N. D., Brook, C. P., et al. (1989). The reliability and validity of the Mini–Mental State in a British community survey. Journal of Psychiatric Research, 23(1), 87–96.

Ramnemark, A., Nilsson, M., Borssen, B., & Gustafson, Y. (2000). Stroke, a major and increasing risk factor for femoral neck fracture. Stroke, 31, 1572–1577.

Ramnemark, A., Nyberg, L., Borssen, B., Olsson, T., & Gustafson, Y. (1998). Fractures after stroke. Osteoporosis International, 8, 92–95.

Rapport, L. J., Webster, J. S., & Fleming, K. L. (1993). Predictors of falls among right hemisphere stroke patients in the rehabilitation setting. Archives of Physical Medicine and Rehabilitation, 74, 621–626.

Reding, M. J., & Potes, E. (1988). Rehabilitation outcome following initial unilateral hemispheric stroke. Life analysis approach. Stroke, 19(11), 1354–1358.

Stineman, M. G., & Maislin, G. (2000). Validity of functional independence measure scores. Scandinavian Journal of Rehabilitation Medicine, 32(3), 143–144.

Stineman, M. G., Shea, J. A., Jette, A., Tassoni, C. J., Ottenbacher, K. J., Fiedler, R., et al. (1996). The Functional Independence Measure: Tests of scaling assumptions, structure, and reliability across 20 diverse impairment categories. Archives of Physical Medicine and Rehabilitation, 77(11), 1101–1108.

Sunderland, A., Tinson, D., Bradley, L., & Langton-Hewer, R. (1989). Arm function after stroke. An evaluation of grip strength as a measure of recovery and prognostic indicator. Journal of Neurology, Neurosurgery, and Psychiatry, 52, 1267–1272.

Teasell, R., McRae, M., Foley, N., & Bhardwaj, A. (2002). The incidence and consequences of falls in stroke patients during inpatient rehabilitation: Factors associated with high risk. Archives of Physical Medicine and Rehabilitation, 83, 329–333.

Tombaugh, T. N., & McIntyre, N. J. (1992). The Mini–Mental State Examination: A comprehensive review. Journal of the American Geriatric Society, 40(9), 922–935.

Tutuarima, J. A., de Haan, R. J., & Limburg, M. (1993). Number of nursing staff and falls: A case-control study on falls by stroke patients in acute-care settings. Journal of Advanced Nursing, 18, 1101–1105.

Yarnell, P., & Boutell-Friedman, B. (1987). Left “hemi” ADL learning and outcome: Limiting factors. Journal of Neurologic Rehabilitation, 1, 125–130.