Incidence and consequences offalls due to stroke: a systematic inquiry

BMJ 1995; 311 doi: (Published 08 July 1995) Cite this as: BMJ 1995;311:83
  1. Anne Forster, research physiotherapista,
  2. John Young, consultant physiciana
  1. aDepartment of Health Care for the Elderly, Saint Luke's Hospital, Bradford BD5 0NA
  1. Correspondence to: Dr Forster.
  • Accepted 27 April 1995


Objective: To undertake a systematic inquiry into the incidence and consequences of falls in a cohort of elderly patients with stroke after discharge from hospital.

Design: Administration of a questionnaire to patients and main carers at discharge from hospital and eight weeks and six months later.

Setting: Bradford Metropolitan District.

Subjects: 108 patients recruited to the Bradford community stroke trial. Patients were recruited to the trial if they were 60 years or over and resident at home with some residual disability.

Main outcome measures: Number of falls, motor club assessment, Barthel index, Frenchay activities index, and Nottingham health profile. Stress in carers was indicated by the general health questionnaire.

Results: Of 108 patients, 79 (73%) fell in the six months after discharge from hospital with a total of 270 falls reported. Patients who fell in hospital were significantly more likely to fall at least twice at home after discharge (x2=8.16; P=0.004). “Fallers” (two or more falls) were less socially active at six months and more had depressed mood. Carers of these patients were significantly more stressed at six months (53% v 18%; x2=8.5; P=0.003).

Conclusion: Stroke is associated with a risk of falling at home and affects the lives of patients with stroke and their carers. Falling and fear of falling is an important issue which needs to be dealt with by the multidisciplinary team.

Key messages

  • Falls are common in patients discharged home after stroke

  • Major injury is rare

  • Falling in hospital is a significant predictor for falling at home

  • Patients identified as being “at risk” should be given advice and guidance by the multi-disciplinary team before discharge


Falls are the commonest type of home accident among elderly people and are a major threat to their health and independence.1 There have been several studies into prevalence of falls in the community,2 3 acute care settings,4 rehabilitation hospitals,5 and nursing homes.6 Many medical and environmental factors have been identified. We examined specifically falls in older people at home who had been disabled by a stroke.

Subjects and methods

We undertook a systematic inquiry into falls as part of the Bradford community stroke trial. We studied the concerns of patients and carers about falling; the incidence of falls; the effects of falls; the characteristics of patients who fell; and whether a particular group of patients “at risk” of falling could be identified.

The trial was a randomised comparison trial of day hospital attendance or home physiotherapy in elderly patients with stroke discharged from hospital.7 Patients were recruited to the trial if they were 60 years or over, had some residual disability related to stroke (Barthel index8 score at discharge of less than 20), and lived at home. A questionnaire on falls was developed and administered by face to face interview of patients and carers at discharge from hospital and at eight weeks and six months after discharge home. Questions were asked about the occurrence of falls before the stroke, falls in hospital, falls at home after discharge, and if the patient had been taught how to get up from the floor. Details of the most recent fall (the cause, time, place, time spent on ground, injuries sustained, and assistance sought) were recorded. During this structured questioning the main carer was asked about concerns they had in relation to their partner or relative falling; if these concerns affected their own social activities; and for how long they were prepared to leave their partner or relative alone in the house.

Several assessments were incorporated into the randomised trial protocol. These included the Barthel index (a measure of disability), the motor club assessment9 (an indicator of balance and movement), the Frenchay activities index10 (an assessment of social activity), and the Nottingham health profile11 (perceived state of health). The main carers also completed the general health questionnaire-28 as an indicator of wellbeing.12 On recruitment to the trial patients were assessed for neglect (Albert's test13), proprioception, loss of sensation, and cognitive function.14 Using similar methodology to Blake et al, we recorded details of comorbidity on a dichotomous (present-absent) scale. These included self reported poor eyesight, heart disease, diabetes, chronic obstructive airways disease, previous stroke, and hypertension.

Statistical analysis was performed with the SPSS-X software package on the Bradford University mainframe. Overstall has suggested that patients most susceptible to falls, “true fallers,” should be so classified if they have fallen twice or more in an index year.15 The category of “non-fallers” therefore includes those patients who have fallen only once. By using this classification characteristics of fallers and non-fallers at discharge were compared with the X2 statistic and Mann-Whitney U test. A logistic regression analysis was also undertaken to identify factors which might predispose stroke patients to fall.


One hundred and twenty four patients were recruited to the trial, but 12 were unavailable for interview at eight weeks and a further four patients were unavailable at six months. Table I shows the characteristics of the 108 patients included in this survey.


Characteristics of patients available for reassessment at six months

View this table:

Number of falls reported--Only 23 (21%) patients reported that they had fallen in the year before their stroke, but 50 (46%) patients fell at least once while in hospital and 79 (73%) fell in the six months after discharge. Of these 79 patients, 31 were “new fallers” who had not reported falling previously; the remaining 48 had fallen either in hospital (40) or at home before their stroke (8). Table II records the numbers of falls after discharge, with just under half the patients (51) having fallen at least twice and a total of 270 falls being reported. The falls were not confined to the early period after discharge: 145 were reported in the first eight weeks and 125 in the eight week to six month period. Details of the most recent fall when the patients were interviewed at six months are reported below.


Number of falls reported by 108 patients in six months after hospital discharge

View this table:

Circumstances of the fall--Tables III and IV give details of the activities being undertaken and the reasons given for falling. Most falls occurred during the day in the lounge or bedroom, with only nine occurring at night. Patients fell while undertaking “basic” activities such as walking or transferring and not while undertaking “extended” activities (for example, reaching) in which body displacement is greater.16 Six patients reported falling while negotiating stairs or steps, and nine patients fell outside. Only 24 of the 79 who fell at all had been able to get up from the floor unaided, 36 needed assistance from a relative, and 19 called on help from a variety of services (police, ambulance staff, home help, warden, and community physiotherapist). Most patients were able to get off the floor fairly quickly, but one patient remained on the floor for three hours. Thirty two patients had been shown how to get up from the floor by a physiotherapist. Subsequently, 24 of these patients fell at home, but the specific training they had received did not influence their ability to arise from the floor unaided (X2=3.06; P>0.05).


Activities being undertaken when fall occurred among 79 patients who fell

View this table:

Reasons given by 79 patients for falling

View this table:

Consequences of falling--Despite the high number of falls reported, serious injuries were uncommon. Only four patients suffered a fracture (or 1% of falls), and only one patient was admitted to hospital as a direct result of a fall. Thirty eight patients, however, suffered soft tissue injuries. Many patients (34) felt the need to discuss the fall with someone other than their family and friends, and 21 called the doctor or attended the casualty department.

Impact on resident main carers--Many of the 74 carers were worried about the possibility of the patient falling irrespective of whether or not a fall had occurred. This concern decreased only slightly over time, with 42 carers expressing concern at eight weeks and 33 at six months. Only one third of the carers acknowledged that this concern limited their social activities, but of the patients with a resident main carer, half were not left alone for more than one hour.


With the classification suggested by Overstall15 there were 57 non-fallers who had fallen only once or not at all, and 51 fallers who had fallen at least twice (table V). There were no significant differences between the two groups in age, sex, mental state, proprioceptive loss, neglect (Albert's test13), previous stroke, living alone, or number of falls in the year before the stroke. Patients who fell in hospital were significantly more likely to fall at least twice at home after discharge (50 patients fell in hospital, 31 of whom subsequently fell more than once at home, X2=8.16; P=0.004).


Median (interquartile range) scores of fallers and non-fallers at discharge from hospital

View this table:

Balance--The motor club assessment is an index of functional movement and includes items which assess balance (sitting balance, standing balance, and standing on one leg). Fallers had significantly lower scores at discharge (P=0.005; table V).

Disability--There was a significant difference between the two groups in Barthel index at discharge, and this difference was maintained over the six months (medican score at six months 17 (non-fallers) v 16 (fallers); median of difference 2; 95% confidence interval 0 to 3; Mann-Whitney test P=0.01). Time taken to walk 5 metres was similar for the fallers and non-fallers at discharge from hospital, but when the patients were reassessed at six months a significant difference in walking speed had developed (12 v 24 seconds; −8; −14 to −3; P=0.003).

Comorbidity--No significant differences were found between fallers and non-fallers for the presence of additional health problems: poor eyesight, heart disease, diabetes, chronic obstructive airways disease, previous stroke, or hypertension.

Social activity and mood--Patients who had fallen at least twice were less socially active when assessed at six months (score for Frenchay activities index (10 v 4; 4; 2 to 7; P<0.001). More of the fallers reported depressed mood (a score of 30 or more on the Nottingham health profile17), and significantly more carers of fallers were stressed at six months (a score of 5 or more on the general health questionnaire12; table VI).


Category analysis of Nottingham health profile for fallers and non-fallers and general health questionnaire completed by main carer six months after discharge from hospital. Figures are numbers (percentages) of those able to answer questions

View this table:


A logistic regression analysis was undertaken by using fallers and non-fallers as grouping criteria and included sex, age, presence or absence of associated medical problems, mental state test score, Albert's test, proprioception (pass/fail), time taken to walk 5 metres, number of falls in hospital, and whether patients lived alone or with a carer. The Barthel index, Nottingham health profile, and motor club assessment scores at discharge were also included. Only “fall in hospital” was a significant predictor of falling at home with an odds ratio of 2.0 (95% confidence interval 1.2 to 3.5).


It is surprising that there has been no previous study which has specifically studied the issue of falls in survivors of stroke who live in the community. In our group of patients with stroke three quarters fell in the first six months after discharge from hospital. Previous community surveys of elderly people have reported rates of falling of 35%2 and 34%3 during an index year. Thus, and perhaps not unexpectedly, stroke as a condition is associated with a high risk of falling. Cummings et al have highlighted the difficulties that elderly people have in recalling falls,18 particularly for those with low mental test scores. Most of our patients, however, had high mental test scores and were interviewed in the presence of a carer who could confirm the accuracy of the events recalled.

The main limitations of our study lie in the selective nature of our sample population, which was dictated by the randomised trial comprising the main purpose of our work.7 Patients who had minimal physical disability from their stroke (Barthel index of 20) and those patients discharged to institutional care were excluded. Also excluded were those patients not admitted to hospital and younger patients. The subjects were therefore patients aged over 60 with mild to moderate disability (median Barthel index of 15) who had been admitted to hospital and subsequently returned home. This does, however, form a group of patients with which many rehabilitation staff will be familiar, and the high risk of falling in these patients needs to be better acknowledged in preparing patients and their carers to cope at home, especially as most of the falls had taken place after “basic” rather than “extended” activities.16

The reasons given for the falls reflect stroke related disability with most patients overbalancing while transferring or because their foot had become stuck. “Dizziness” preceding a fall was uncommon (only one patient) in contrast with results from community surveys of falls.3 One explanation for this unusually low finding is that some of the patients who did not know the reason for falling might have actually experienced a black out or dizzy turn.

Our separation of the patients into two groups (fallers or non-fallers) is as previously suggested.15 It acknowledges an important epidemiological aspect of falls in older people which occur as a non-random distribution and do not fit a Poisson function.19 Thus the category of non-fallers includes some people who will by chance have fallen occasionally, and true fallers are those who are most susceptible to falls.

Patients were most likely to fall if they had suffered a fall in hospital and had lower scores on the Barthel index and motor club assessment at discharge. The motor club assessment is similar to a previously described mobility score which was also a successful discriminator separating non-fallers from fallers.20

The fallers had significantly slower walking speeds at six months. Whether this was the cause or result of the falls cannot be inferred from our study, but the adverse functional effects of repeated falling have been reported elsewhere and are well known to geriatricians.21 22 Similarly, the direction of the relation between the observed lower scores for social activity and lower mood in the fallers is speculative. A possible and reasonable inter-relation, however, would be falls accelerating the background decline in mobility,23 which further increases the risk of falls and increases disability producing social restriction and low mood. Further work needs to be undertaken to test this specific interpretation.

It is a concern that so many patients fell in the months after discharge. Targeting of the patients at increased risk of falling (lower Barthel index scores, lower motor club assessment scores, and fallers in hospital) would seem sensible. Counselling and provision of pendant alarms needs investigation. A common approach is to prepare the patient with instruction from a physiotherapist on how to get up from the floor. This technique is inconsistently applied,24 and in our study, although just over a quarter of the patients received such training, most still required help to get up from the floor after their fall. Provision of physiotherapy both to prevent falls25 and after a fall26 has been shown to be beneficial. Physiotherapy applied as a late intervention in stroke has also been studied and has some effect,23 particularly in improving speed of gait, a factor associated with falling in our study.

The incidence and consequences of falls for patients with stroke and their carers seem to be important issues which need greater attention in clinical and research terms.


We thank the patients and families for their help and acknowledge the computer assistance and continued support given by Brian Howlett of the University of Bradford.


  • Funding Stroke Association.

  • Conflict of interest None.


View Abstract