Risk Factors for Falls of Hospitalized Stroke Patients
Background and Purpose Patients with stroke are at a high risk for falling. We assessed the fall incidence and risk factors for patients hospitalized as the result of an acute stroke.
Methods We studied a cohort of 720 stroke patients from 23 hospitals in The Netherlands. The data were abstracted from the medical and nursing records.
Results We studied 346 women and 374 men with a median age of 75 years; 77% of the patients had had a cerebral infarct, 17% had had a hemorrhage, and 6% had had an undefined stroke. We recorded 104 patients (14%) who fell at least once; there were a total of 173 falls. The incidence of falls was 8.9/1000 patients per day. The daily incidence was 6.2/1000 patients for first falls and 17.9/1000 patients for second falls. Heart disease (relative risk [RR], 1.6; 95% confidence interval [CI], 1.0 to 2.4), mental decline (RR, 1.6; 95% CI, 1.0 to 2.4), and urinary incontinence (RR, 2.3; 95% CI, 1.3 to 4.1) were incremental risk factors for first falls, whereas the use of major psychotropic drugs lowered the fall risk (RR, 0.5; 95% CI, 0.3 to 0.8). The fall RR for patients with one previous fall was 2.2 (95% CI, 1.5 to 3.2), adjusted for the other risk factors. Most falls occurred during the day. Approximately 25% of the falls caused slight-to-severe injury, whereas three falls (2%) led to hip fractures.
Conclusions Stroke patients have a high risk of falling. The identification of patients at risk may be a first step toward the implementation of fall-prevention measures for these patients.
Stroke patients have a high risk of falling. In an earlier study, we found that 14% of the stroke patients fell at least once during their hospital admission.1 Other studies showed a considerable variation in the risk of falling among clinical departments. A fall incidence was reported of 5.2/1000 and 6.6/1000 patients per day on two neuroscience (both neurology and neurosurgery) wards, whereas the overall in-hospital fall incidence at these institutions was 3.1/1000 and 4.7/1000 patients per day, respectively.2 3 A very high incidence of falls (15.9/1000 patients per day) was reported for a geriatric unit for stroke rehabilitation.4 Studies on falls of stroke patients are rare and originate from settings such as acute care settings, rehabilitation facilities, and the community.1 4 5 6 7
For patients in the acute care setting, a few researchers have presented risk factors for falling,3 5 8 9 whereas only one study specifically focused on stroke patients. In that study, a history of falls and mental status impairment turned out to be incremental risk factors.5
The identification of risk factors associated with falls of stroke patients is an essential first step to the prevention of falling. The few studies on prevention programs in the acute care settings reported a relative reduction in the incidence of falls of 20% to 60%10 11 12 13 ; the number of fractures caused by falls was reduced by 80% in one study.12
None of these studies, however, were designed as experimental studies in the form of randomized controlled clinical trials. However, in one landmark study, a community-based study on fall prevention strategies among elderly people reported a relative reduction in falls of 30%.14
The objective of the present study was to investigate the incidence of falls and to identify risk factors in patients hospitalized as a result of an acute stroke. This study represents part of a nursing research program to reduce the number of in-hospital accidents.
Patients and Methods
The stroke patients described in this study were consecutively admitted between mid-1991 and mid-1992 to the neurological departments of 23 Dutch hospitals that participated in a study of the quality of care for stroke patients in The Netherlands.
The participating hospitals were randomly selected after stratification for number of beds and region. Nine hospitals had >400 beds (three of them were university-affiliated hospitals), nine hospitals were of an intermediate size (200 to 400 beds), and five hospitals were small (<200 beds). Seven hospitals were located in the northeastern, relatively rural region, and 16 were in the western region of The Netherlands. All hospitals admit patients with acute medical problems; the majority of the patients were referred by general practitioners. We did not observe any transfers of patients among the hospitals.
Patients were considered to have had a stroke if there was an acute focal neurological disturbance on an apparent vascular basis that lasted >24 hours or led to death. Patients with a subarachnoid hemorrhage were not included in the study. During the study period, 796 patients were admitted. Thirty-six patients (4.5%) declined to participate in this study; 39 patients (4.9%) were excluded because the nursing reports were not available; and one patient was excluded because of frequent deliberate falls. The final study group comprised 720 patients.
The study was approved by the medical ethical committees of the participating hospitals.
The data collection occurred soon after hospital discharge to minimize the influence on patient care of performance of the study. Four trained nonmedical research assistants collected the data from the medical and nursing records with the use of a prestructured record form. All completed record forms were checked by a neurologist (M.L.) for inconsistency and were revised if necessary.
The registered patient characteristics from medical and nursing records included sex and age of the patient,3 type of stroke (infarction or hemorrhage), stroke severity, medical history, comorbidity (history of stroke, heart disease, diabetes mellitus, impaired mobility, mental decline),8 9 15 16 neurological deficits (hemiplegia, cortical neurological impairments, visual impairments),5 8 17 and complications (confusional state, urinary incontinence).8 9 17 Furthermore, we registered the use of medications (diuretics, benzodiazepines, major psychotropic drugs)9 15 16 17 and the intensity of nursing care.8 18 19 If the patient had fallen, we collected data about the circumstances (time of day, place, patient's activity before the fall) and the consequences of the falls (injuries and physicians' and nursing personnel's actions). A fall was defined as any event described as “a fall” in the medical or nursing records.
An infarction was considered to be present if a CT (or an MRI) scan excluded an intracerebral hemorrhage. The type of stroke remained undefined if no scan had been performed.
Stroke was graded as severe if the patient had a depressed consciousness at admission. The level of consciousness was assessed with the Glasgow Coma Scale (GCS).20 Because of possible aphasia, the GCS was scored without the verbal component. A patient with a GCS score on the eye component of ≤2 as well as a score on the motor component of ≤3 was considered to be comatose. Heart disease included rhythm disturbances, including atrial fibrillation; congestive heart failure; valvular abnormalities; history of angina pectoris; history of myocardial infarction; or history of coronary artery bypass graft surgery. Mental decline referred to any form of cognitive dysfunctioning (eg, memory problems) before the present stroke on the basis of the history of the nearest relatives or the general practitioner. Hemiplegia referred to any degree of half-sided motor paresis. The cortical neurological impairments included aphasia, apraxia, neglect, and impaired spatial orientation and were considered to be present if the neurologist used these terms in the medical chart, physician's orders, or discharge summary. Visual impairments were considered to be present if the patient had hemianopia, glaucoma, cataract, or other serious ocular conditions. A “confusional state” was present if during some period there was agitation, unrest, or disorientation. We considered heart disease, diabetes mellitus, impaired mobility, mental decline, hemiplegia, cortical neurological impairments, visual impairments, confusional state, and urinary incontinence to be present if they were recorded in the medical or nursing records. The use of medications by the patients who had fallen related only to the period before the first fall. Use of medication of patients who had not fallen reflected the entire hospital stay. We assumed that the medication was considered to be given continuously. The major psychotropic drugs referred to antipsychotic, antidepressant, and hypnotic-anxiolytic medications. The intensity of nursing care was assessed according to the San Joaquin method21 and was graded as mild, moderate, high, or intensive care. We assessed the intensity of nursing care on the seventh day after admission or on the last hospital day if a patient was discharged or died earlier.
The incidences of all falls, first falls, and second falls were estimated as the number of falls that occurred during hospitalization divided by the observation time.
For all falls, the observation time is the sum of the total admission time. For first falls, the observation time is the sum of the admission time before the first fall or before discharge for the patients who did not fall. For second falls, the observation time is the sum of the admission time after the first fall and before the second fall or before discharge for the patients who did not fall twice. Univariate (crude) and multivariate (adjusted) relative risks (incidence rate ratios) with 95% confidence intervals (CIs) were calculated with Poisson regression.22 Age, sex, and the significant variables (P≤.10) identified from univariate analyses were forced into the multivariate model. The CIs of the fall incidences were based on exact Poisson limits, and CIs for the relative risks were the approximate confidence limits for the maximum likelihood estimate.
Of the 720 stroke patients, 104 (14.4%) fell at least once. Of these 104 patients, 69 patients (66.3%) fell only once, 19 (18.3%) fell twice, and 16 (15.4%) fell three or more times. We registered a total of 173 falls. The incidence of falls was 8.9/1000 patients per day. The incidences were 6.2/1000 patients per day (95% CI, 5.1 to 7.6) for first falls and 17.9/1000 patients per day (95% CI, 12.5 to 24.9) for second falls. The length of hospitalization ranged from 1 to 186 days (median, 21 days).
The incidences of first and second falls for each subsequent week of admission are depicted in the Figure⇓. The risk of falling does not seem to change during admission.
The characteristics of the study group, the fall incidences, and the relative risks of first falls are presented in Table 1⇓. The patients (346 women [48%] and 374 men [52%]) were usually admitted to general hospitals (76%) and ranged in age from 20 to 96 years; half were >75 years old, and ≈77% had had a cerebral infarction. Half of the study group had heart disease, and 80% had some degree of hemiparesis. Heart disease, mental decline, confusional state, and urinary incontinence were incremental risk factors in the univariate model, whereas patients using major psychotropic drugs were less likely to fall (P<.05). When entering these five significant factors and age and sex into a multivariate model, these characteristics remained statistical significant, with the exception of confusional state.
Even after adjustment for sex, age, heart disease, mental decline, confusional state, urinary incontinence, and major psychotropic drugs, the risk of falling a second time was increased for patients who had fallen once before. For patients with a previous fall, the univariate relative risk to fall a second time was 2.7 (95% CI, 1.9 to 4.0). The multivariate relative risk for patients with a previous fall was 2.2 (95% CI, 1.5 to 3.2).
The circumstances and outcomes of the 173 observed falls are presented in Table 2⇓. Most falls occurred during the day (45%), in the patient's room (51%), and during visits to the toilet or bathroom (20%). Many falls happened while the patient was in bed (23%) or sitting on a chair (24%). In approximately one fourth of the falls, the patients' activities were not registered in the nursing reports. Approximately 25% of the 173 falls caused slight-to-severe injuries. Hematoma or open wounds occurred in 14 patients (8%), and hip fractures occurred in 3 patients (2%). The remaining injuries were minor and did not require specific medical treatment or nursing attention.
We found an incidence of all falls of 8.9/1000 patients (ie, almost 1%) per day in stroke patients during admission. This is a high risk compared with hospitalized patients in general (range, 2.5 to 7.6/1000 patients per day)10 11 12 13 19 and with those admitted to neuroscience departments (range, 5.2 to 6.6/1000 patients per day) in particular.2 3
Thirty-five of the study patients (34%) who had fallen once fell twice or more. The risk for falling after the patient had fallen once doubled. Recently, it was reported that a fall during the stroke patient's hospital admission was an incremental risk factor for falling after discharge up to 6 months.7
We found hip fractures (2%) to be the most serious consequence of a fall. This percentage is in agreement with findings among nonstroke patients who have fallen.2
In this study, a fall was defined as any event registered as a fall in the medical or nursing records. This means that we also registered unwitnessed falls, self-reported falls, and events described as “found on the floor.” Some studies have shown that registration of falls on the basis of the patients' nursing charts increases the number of falls by ≤30% compared with use of the hospital incident reports.1 23 24 It is likely that the fall incidence as assessed with use of these formal hospital incident reports is considerably underreported. However, the incidence estimated on the basis of the medical and nursing records is based on unspecified definitions.
Another disadvantage of the retrospective design of our study is that the quality of the data regarding patient characteristics also varies because it is registered by different neurologists and nursing staff without the use of explicit rating scales to assess, for example, mental decline and confusional state. On the other hand, our method of data collection had no influence on the nurses' behavior and fall incidence.
Few studies have included a multivariate analysis of risk factors for falls of stroke patients. Among the previously identified risk factors for hospitalized stroke patients are a history of falls and mental impairment.5 Our study also identified a previous fall and mental decline as incremental risk factors, as well as heart disease and urinary incontinence. In addition, we found that the use of major psychotropic drugs was negatively associated with hospital falls, but we did not analyze the given doses and the characteristics of these types of drugs.25 Risk factors as reported in hospitalized patients in general, such as impaired mobility and confusional state, could not be confirmed in our study.8 9
Most falls happened during the day, when the intensity of nursing care is at a peak. Furthermore, we could not demonstrate an impact of nursing care intensity on the falls of stroke patients. Both these findings are in agreement with earlier studies in which it was shown that the number of patients per nurse was unrelated to the number of falls on the wards.1 2
A differentiation among clinical departments in risks for falling has been demonstrated, and neurological patients have been found to have a high risk of falling.2 3 It is worthwhile to investigate in another study the various mechanisms underlying the high risk for falling of patients admitted onto a neurological ward. With the knowledge of the incremental risk factors for falls and the assumption that some of the identified risk factors can be modified, this may lead to the development of interventions to reduce the number of falls.26 27
We conclude that stroke patients have a relatively high risk for falling. In the acute care setting in general, the efficacy of fall-prevention measures has been demonstrated.10 11 12 13 Examples for such measures are the assessment of the patient's risk for falling, provision of helpful information to the patient and the family, and implementation of specific nursing interventions in the form of putting things within hand's reach, putting the bed in a low position, encouraging the patient to ask for assistance, eliminating environmental hazards, and using additional night lighting. The identification of categories of patients at high risk for falling is an important step toward the implementation of these preventive measures and the improvement of the quality of hospital care.
Dr Limburg is a Clinical Investigator of the Netherlands Heart Foundation, The Hague. This study was supported by Ministerie van Welzijn, Volksgezondheid en Cultuur, (89-700); Dutch Heart Foundation (40.004); and National Committee on Investigative Medicine of the Health Insurance Executive Board of the Netherlands (O.G. 91-0370). We thank Mrs L. Kok-Noorman, Department of Clinical Epidemiology and Biostatistics, for her editorial guidance in preparing the manuscript.
- Received July 26, 1996.
- Revision received October 29, 1996.
- Accepted November 15, 1996.
- Copyright © 1997 by American Heart Association
Nyberg L, Gustafson Y. Patient falls in stroke rehabilitation: a challenge to rehabilitation strategies. Stroke. 1995;26:838-842.
Mayo NE, Korner-Bitensky N, Kaizer F. Relationship between response time and falls among stroke patients undergoing physical rehabilitation. Int J Rehab Res. 1990;13:47-55.
Foster A, Young J. Incidence and consequences of falls due to stroke: a systematic inquiry. BMJ. 1995;311:83-86.
Morse JM, Tylko SJ, Dixon HA. Characteristics of the fall-prone patient. Gerontologist. 1987;27:516-522.
Schmid N. 1989 Federal Nursing Service Award Winner−Reducing patient falls: a research-based comprehensive fall prevention program. Milit Med. 1990;155:202-207.
Prudham D, Grimley Evans J. Factors associated with falls in the elderly: a community study. Age Ageing. 1981;10:141-146.
Tinetti ME, Speechly M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med. 1988:319:1701-1707.
Myers AH, Baker SP, Van Natta ML, Abbey H, Robinson MG. Risk factors associated with falls and injuries among elderly institutionalized persons. Am J Epidemiol. 1991;133:1179-1190.
Tauton RL, Kleinbeck SVM, Stafford R, Woods CQ, Bott MJ. Patient outcomes: are they linked to registered nurse absenteeism, separation, or work load? J Nurs Adm. 1994;24(suppl 4):48-55.
Puetz K. Development of an incident reporting system. QRB. 1988;14:245-250.
McAthie M, Murphy L, Williams M, Van Aerman B. Development of Methods for Determining Use and Effectiveness of Nursing Service Personnnel: Final Report. Stockton, Calif: San Joaquin General Hospital; 1976.
Breslow NE, Day NE. The design and analysis of cohort studies. In: Breslow NE, Day NE, eds. Statistical Methods in Cancer Research. New York, NY: Oxford University Press Inc; 1987. IARC Scientific Publications, No. 82.
Overstall PW. Falls after stroke. BMJ. 1995;311:74-75.