Randomised controlled trial of a general practice programme of home based exercise to prevent falls in elderly womenBMJ 1997; 315 doi: http://dx.doi.org/10.1136/bmj.315.7115.1065 (Published 25 October 1997) Cite this as: BMJ 1997;315:1065
- A John Campbell, professor of geriatric medicinea,
- M Clare Robertson, research fellowa,
- Melinda M Gardner, research physiotherapista,
- Robyn N Norton, directorb,
- Murray W Tilyard, professorc,
- David M Buchner, professord
- a Department of Medicine, Dunedin School of Medicine, Otago Medical School, PO Box 913, Dunedin, New Zealand
- b Injury Prevention Research Centre, University of Auckland School of Medicine, Auckland, New Zealand
- c Department of General Practice, Dunedin School of Medicine, Otago Medical School
- d VA Puget Sound Health Care System, 1660 S Columbian Way, Seattle, WA 98108, USA
- Correspondence to: Professor Campbell
- Accepted 16 September 1997
Objective: To assess the effectiveness of a home exercise programme of strength and balance retraining exercises in reducing falls and injuries in elderly women.
Design: Randomised controlled trial of an individually tailored programme of physical therapy in the home (exercise group, n=116) compared with the usual care and an equal number of social visits (control group, n=117).
Setting: 17 general practices in Dunedin, New Zealand.
Subjects: Women aged 80 years and older living in the community and registered with a general practice in Dunedin.
Main outcome measures: Number of falls and injuries related to falls and time between falls during one year of follow up; changes in muscle strength and balance measures after six months.
Results: After one year there were 152 falls in the control group and 88 falls in the exercise group. The mean (SD) rate of falls was lower in the exercise than the control group (0.87 (1.29) v 1.34 (1.93) falls per year respectively; difference 0.47; 95% confidence interval 0.04 to 0.90). The relative hazard for the first four falls in the exercise group compared with the control group was 0.68 (0.52 to 0.90). The relative hazard for a first fall with injury in the exercise group compared with the control group was 0.61 (0.39 to 0.97). After six months, balance had improved in the exercise group (difference between groups in change in balance score 0.43 (0.21 to 0.65).
Conclusions: An individual programme of strength and balance retraining exercises improved physical function and was effective in reducing falls and injuries in women 80 years and older.
Modifiable risk factors for falls in elderly people have been well defined; they include loss of muscle strength and impaired balance
A programme to improve strength and balance in women aged 80 years and older can be set up safely with four home visits from a physiotherapist
This programme reduced falls and moderate injuries appreciably over the subsequent year in Dunedin, New Zealand
The benefit was most noticeable in elderly people who fell often
Prospective community studies have detailed risk factors for falls in elderly people and identified those old people who are likely to fall; they also provide the basis for preventive studies.1 2 3 The risk factors most commonly identified, which are possibly those most amenable to interventions that can be carried out in primary care, are loss of muscle strength and flexibility, and impaired balance and reaction time.4 However, some studies have shown that increased activity in very old people can mean more falls and injuries.5 6
Meta-analysis of seven studies in the “frailty and injuries: cooperative studies of intervention techniques” trials showed that strength and balance training reduced the frequency of falls.7 Three of the study sites showed an increased, but not statistically significant, risk of falling with the training programme. These studies used a variety of additional intervention strategies, and not all could be applied easily in a general practice setting.
A public health programme to reduce falls in elderly people needs to be simple, easy to implement, and affordable as well as effective. We developed a home based exercise and balance training programme which could be used in general practice. Age and female sex, the two most easily observable risk factors, were used to identify the study population.4 We report the effect of a randomised, single blind controlled trial of a home based strength and balance retraining programme on the frequency of falls, injury from falls, balance, and muscle strength in women aged 80 years and older.
Women aged 80 years and older living in the community were identified from the computerised registers of 17 general practices. They were invited by their general practitioner to take part in the study if they were able to move around within their own home and were not receiving physiotherapy.
Potential study subjects were visited at home by the research nurse. She obtained informed consent, filled in the mental status questionnaire in order to exclude those women unable to comply with the study requirements (score of <7 from 10), completed baseline questionnaires, and took a note of current medication.8 Subjects subsequently visited a clinic for a baseline assessment and were then randomised to the control group or to receive the exercise programme. Six months later the physical assessments were repeated at the clinic by the same physiotherapist. The same research nurse completed the questionnaires at the subject's home after one year. Falls, injuries from falls, and compliance with the exercise programme were monitored for one year.
Monitoring falls and injuries
Falls were the main outcome measure and were defined as “unintentionally coming to rest on the ground, floor, or other lower level.” Coming to rest against furniture or a wall was not counted as a fall.9 Each subject was given a calendar comprising 12 addressed, reply paid postcards on which she could record falls daily for each month. Postcards were mailed back at the end of each month throughout the year, and the participant was contacted by telephone if the postcard was not returned. When a subject reported a fall, she was telephoned by the research team and the date and circumstances of the fall and details of any injuries were recorded on a fall event form. Injuries were defined as “serious” if the fall resulted in a fracture or admission to hospital or if any wounds needed stitches and “moderate” if there was bruising, sprains, cuts, abrasions, or a reduction in physical function for at least three days, or if the woman sought medical help. The circumstances of “serious” injuries were confirmed from hospital records. An investigator (AJC) who did not know to which group individual women belonged reviewed all fall events to determine if they met the fall definition and to classify the injury. Falls and injuries were monitored until the date of death or withdrawal from the study.
At entry to the study, a medical history was taken; demographic information and details of social support, current medication, and health related behaviours were recorded; and weight, height, visual acuity, blood pressure while sitting and standing, and heart rate were measured. The instrumental activities of daily living scale, physical self maintenance scale, fear of falling, and the physical activity scale for the elderly questionnaires were completed at entry to the study and after one year.10 11 12
Physical assessment measures were chosen to evaluate balance, gait, strength, and endurance. The same tests were repeated at entry to the study and after six months. The assessments are shown in the box.
Assessment tests to estimate balance, gait, strength, and endurance
A functional reach test13
Standing with feet side by side, standing with one foot next to and halfway in front of the other, standing with one foot directly behind the other, and standing on one foot with the other raised for up to 10 seconds,14 scored as the 4-test balance scale8
Strength of the knee extensor muscle on the subject's dominant side, tested with an electronic dynamometer2
“Chair stand” test (time taken to rise from a chair and return to the seated position five times)14
Time taken to walk eight feet14 and 20 metres
Time taken to climb up and down a set of four steps in the assessment room
Distance walked in six minutes using the walking aid normally used outside the home15
The physiotherapist visited each subject who had been randomised to the exercise group four times over her first two months in the study. She prescribed a selection of exercises from the programme at appropriate and increasing levels of difficulty, and a walking plan. Each home visit took approximately one hour. Exercises included moderate intensity strengthening exercises with ankle cuff weights (0.5 kg and 1 kg) for the following muscle groups: hip extensor and abductor muscles, knee flexor and extensor muscles, inner range quadriceps, and ankle plantar and dorsiflexor muscles. Other exercises were standing with one foot directly in front of the other; walking placing one foot directly in front of the other; walking on the toes and walking on the heels; walking backwards, sideways, and turning around; stepping over an object; bending and picking up an object; stair climbing in the home; rising from a sitting position to a standing one; knee squat; and “active range of movement” exercises (for example, neck rotations and hip and knee extensions).
The exercises took about 30 minutes to complete. The women following the exercise programme were told to complete it at least three times a week and were encouraged to walk outside the home at least three times a week. Safety was ensured by prescribing each exercise appropriately, by giving the women adequate instructions on each exercise, and by providing an instruction booklet with illustrations. After the fourth visit, participants were encouraged to continue the exercise programme on their own and to telephone the physiotherapist with any problems. Subjects were telephoned regularly to maintain motivation.
Participants recorded whether they had completed the prescribed exercises or walked each day on a postcard calendar similar to the one used to record falls; they posted this back at the end of each month. The research nurse made a social visit to those in the control group four times during the first two months and telephoned them regularly during the year of follow up.
The sample size calculation was based on the proportion of elderly women who had fallen once or more during a 12 month prospective study in the community.2 Numbers of women in the groups were based on the expectation that the exercise programme would reduce the proportion of women who fell during the year by 20% and allowed for a significance level of 0.05, a power of 0.80, and a drop out rate of 20%. Data were analysed on an intention to treat basis using SPSS version 6.1.1. Baseline characteristics and changes from baseline to six months and baseline to one year were compared in the two groups using the χ2 test, Student's t test, or the Mann-Whitney U test as appropriate. The event rate was calculated as the mean of the number of falls divided by the time over which falls were monitored for each participant, and the 95% confidence interval of the difference was calculated assuming a negative binomial distribution.16 Proportional hazards models were used to determine relative hazards for the two groups for a first fall and a first fall with injury. A relative hazard was calculated to compare the two groups using the Andersen-Gill extension of the Cox model, which allows for multiple events per subject (SAS version 6.1).7 We used the first four falls for each participant in this analysis rather than all falls (maximum 10) to avoid overweighting by subjects who fell more than four times.
All subjects gave informed consent. Approval for the study was given by the Southern Regional Health Authority's ethics committee (Otago).
Group assignment and blinding
The group allocation schedule was developed by a statistician using computer generated random numbers and the list was held off site by an independent person. Group assignment was made by telephone contact after all baseline questionnaires and assessments were completed. The assessment physiotherapist and investigator classifying fall events remained blind to group allocation.
The progress of the participants through the trial is shown in figure 1.
Characteristics of women at entry to the study are given in table 1. Two measures differed between the groups, but these were not related to the risk of falling and did not influence the findings of the study.
After six months balance had improved in the exercise group compared with the control group (mean (SD) changes in the 4-test balance score were 0.42 (0.86) and −0.01 (0.80) respectively; difference 0.43; 95% confidence interval 0.21 to 0.65). A higher proportion of those in the exercise group had improved their performance in the chair stand test (relative risk 1.41; 1.07 to 1.87). There were no differences between the two groups for the remaining physical assessment measures.
After one year of follow up there had been 152 falls in the control group and 88 falls in the exercise group. The total follow up time was 113.4 person years for the control group and 108.8 person years for the exercise group. The mean (SD) rate of falls per year was lower in the group receiving the exercise programme than in the control group (0.87 (1.29) and 1.34 (1.93) falls per year respectively; difference 0.47; 95% CI 0.04 to 0.90). The number of falls for study participants is shown in table 2.
The hazard ratio for a first fall in the exercise group compared with the control group was 0.81 (0.56 to 1.16). With the Andersen-Gill extension of the Cox model, the hazard ratio for the exercise group compared with the control group for the first four falls was 0.68 (0.52 to 0.90).
Elderly people who had four or more falls during follow up had a higher risk of having fallen in the previous year than the remainder of the participants (13 of 17 v 81 of 195: relative risk 1.84; 1.35 to 2.51), but values at baseline did not differ for any other variables.
Eighty five falls resulted in moderate injury and 25 in severe injury. The hazard ratio for a first fall with injury was 0.61 (0.39 to 0.97). The proportion of subjects monitored for the full 12 months (n=213) who were injured from a fall was lower in the exercise group than in the control group (26.2% (27 of 103) v 39.1% (43 of 110); relative risk 0.67; 95% CI 0.45 to 1.00).
After one year 42% (48 of 114) of the survivors in the exercise group were still completing the programme three or more times a week. The control group became less active (mean (SD) change in the physical activity scale for the elderly score −11.0 (22.3) v −4.6 (22.9); difference 6.4; 0.2 to 12.6), and their fear of falling increased (mean (SD) change in falls self efficacy score −6.1 (12.2) v −2.5 (11.1); difference 3.6; 0.4 to 6.8). There were no differences between the group scores for the instrumental activities of daily living scale (median 8.0; range 0 to 8) or the physical self maintenance scale (6.0; 3 to 6) at baseline or after one year (7.0; 0 to 8 and 5.0; 2 to 6, respectively).
We have shown that a programme of strength and balance training exercises, which could be done at home and organised by general practices, reduced significantly the number of falls and injuries experienced by women aged 80 years and older. The reduction in the rate of falls was greater than that found in the combined frailty and injuries cooperative studies of intervention techniques studies, but similar to that achieved by combined interventions.3 7
Improvement in balance and strength
The exercise group showed improved balance and an improved performance in the chair stand test. Improvement in balance has been shown in previous studies, but has usually come about through group activities or by means of specific training equipment.17 18 19 Community programmes have also shown improvements in balance and reaction time in elderly people, but no reduction in the frequency of falls.20 The combined interventions of Tinetti and colleagues led to a significant improvement in balance and in transferring safely from one position to another.21
Prevention of falls
The balance retraining programme strengthened postural control mechanisms, but initially it put the elderly woman at risk of falling. Our strength retraining programme required an increase in activities such as daily walks and therefore increased the opportunities for falling. The time to first fall was used to assess this possible increase in the risk of early falls and was similar for the two groups.
Those who have one fall are more likely to have another.1 2 This differing tendency for recurrence was allowed for in the rigorous analysis used.16 One of the main differences between the two groups was in the prevention of several falls. At the Atlanta site of the “frailty and injuries: cooperative studies of intervention techniques,” Tai Chi had most benefit in preventing several falls.17 Although it could be argued that the programme should be directed at people who fall often, most older women are sufficiently susceptible to the risk of falling and sufficiently inactive to benefit from planned activity.22 It is difficult to predict those who may start to fall frequently. In our study a history of previous falls was the only factor which could have been used to predict which subjects were likely to have frequent falls.
Use of programme in general practice
The programme was designed to be used as part of a preventive plan based in general practice, but issues remain which need to be considered in the transition from a research project to a public health programme. The enthusiasm and commitment of the research physiotherapist may encourage greater compliance in the elderly people than is possible in the busy routine of a working practice. On the other hand, recruitment may improve when the programme is sponsored by the general practice. Our involvement was kept to a minimum but more regular encouragement during normal practice attendances may improve participation.
Although only 37% (233 of 622) of those aged 80 years and older on general practice lists participated in the programme, we did not have an elite, fit sample. Some of the women were very frail—the mean score on the physical activity scale for the elderly was only 51.5 within a possible range of 0 to 400. The high frequency of other risk factors for falls and the high fall rate overall in the study, similar to that observed in previous prospective studies,1 2 also indicated that this was not an unusually sprightly group. Although there was some improvement in physical activity in the exercise group, the improvement was small and the nurse who completed the 12 month questionnaire was aware of group allocation. The improvements must therefore be interpreted with caution.
Results from a study such as this depend on the completeness of the reports of falls. Tear-off calendars have been used successfully before, and 88% of falls were notified in this way. If a calendar postcard was not returned, or if a fall was noted, the participant was contacted by telephone and details of the fall recorded on a structured event form.
Falls remain a major public health problem and affect the lives of many older people. Not only may an individual programme of physical activity reduce the risk of falls, it may improve health in other ways.23 24 Younger people who have several falls may also benefit from the programme. Our study has shown that preventing falls through a home based programme in which strength and balance training is a key component can reduce the frequency of falls. The next step is to make the transition from trials of efficacy to trials of more general implementation and health promotion.
We thank the study participants; their general practitioners; the study research nurses Lenore Armstrong, Shirley Jones, and Rebecca Neill; assessment physiotherapist Edith Laba; statistician Sheila Williams; economist Paul Scuffham; and Jocelyn Thornicroft, adviser from the Accident Rehabilitation and Compensation Insurance Corporation.
Funding: Accident Rehabilitation and Compensation Insurance Corporation of New Zealand. DMB was sponsored by the Department of Veterans Affairs, United States.
Conflict of interest: None.