Methods

SEARCH METHODS

Six computerised databases were accessed in August 1999 for relevant articles in English. Medline, Psychlit, Index New Zealand, Current Contents, and the databases for Cumulative Index of Nursing and Allied Health Literature (CINAHL) were searched using Ovid. The databases were searched within the following periods: Medline 1990–1999; Psychlit 1991–1999; Current Contents 1995–1999, and CINAHL 1982–1999. The Web of Science was also searched from 1990 to 1999. Keywords for searching included controlled trials, public health, health promotion, elderly, old, exercise, physical activity, falls, fall injuries, fall prevention, costs, and cost effectiveness.

DATA EXTRACTION

Studies were reviewed if they met the following criteria: (a) included a control group; (b) participants were aged 60 years or older; (c) the intervention included an exercise component with details provided on exercise type, frequency, and duration; (d) prevention of falls and/or fall related injuries was an aim. The following factors were considered in each study: study design, eligible population, population agreeing to be randomised, age distribution, setting, inclusion and exclusion criteria, generalisability, use of blinding, form of intervention, duration of the intervention, co-intervention or contamination, measurement of outcomes, numbers lost to follow up, evidence of intervention effects, adverse effects, level of evidence, costs of the intervention, and effect on health care costs.

Results

Eleven controlled trials meeting the inclusion criteria were identified and reviewed. The results of one trial were reported both at one year and after two years of follow up.^10,11 In all 11 studies, participants were randomly allocated to an intervention or control group. Two randomised controlled trials were excluded because the article lacked sufficient detail about the exercise intervention, and one controlled health promotion trial was also excluded for the same reason.^12–14 One study was excluded because all the participants took part in the exercise programme.¹⁵ Four of the trials are from the frailty and injuries: cooperative studies of intervention techniques (FICSIT) group of studies concerning physical frailty and injuries in later life.¹⁶ Three other FICSIT trials had an exercise component and contributed to data reported in a preplanned meta-analysis of the trials.¹⁷ Information on intervention efficacy in reducing falls in these studies was not available from individual articles.

Tables 1 and 2 give the study aims, sample, interventions used, exercise compliance, intervention effects, and relevant comments based on the review of the studies. In nine of the 12 articles, exercise was a separate intervention (see table 1), and in three studies exercise was included with other interventions in a multifactorial or dual approach (see table 2). Three of the studies included costs of the intervention or total health care costs as outcome measures. Table 3 summarises these results.

STUDIES WITH EXERCISE AS A SEPARATE INTERVENTION

Reinsch et al¹⁸ assessed the effectiveness of two interventions: exercise and cognitive behavioural strategies on falls and fall related injuries. Men and women aged 60 years and older (n = 230) attending senior centres were recruited and randomised by centre to one of four groups. Participants in the exercise and exercise cognitive groups took part in “stand up/step down” exercise classes, first proposed by Liss¹⁹ for the prevention of falls. The exercise programme included standing up from a seated position and stepping up on to a six inch high stepping stool. At one year there was an attrition rate of 20%, with no age difference between the participants who dropped out of the programme and those who continued. There was no difference in the number of fallers, time to first fall, and fall rate among the four groups. The authors suggest that the exercise programme may not have been of sufficient intensity to reduce falls and that the cognitive behavioural group sessions may not have been frequent enough.

MacRae et al²⁰ also assessed the effectiveness of the stand up/step up routine in a sample of 80 community dwelling women aged 60 years and older. Participants were randomised by senior centre to an exercise (n = 42) or attention control group (n = 38). Initially participants stood up five times from a sitting position and carried out five step ups on to a six inch stall, with the number of repetitions increasing over the programme. At one year, 26% attrition rates were reported. Non-dropouts were similar to dropouts with regard to age and fall history. Randomly chosen intervention and control participants underwent assessments of gait at one year (n = 20). Maintaining quadriceps and ankle strength in the exercise group did not result in a reduction in the number of fallers or fall related injuries. This trial again provided no evidence for the use of the stand up/step up procedure for falls prevention in community dwelling people aged 60 years and older.

The San Antonio FICSIT trial targeted nursing home residents aged 60 years and older (mean (SD) age 79.7 (8.5) years in the intervention group, 81.4 (7.9) years in the control group), living in the rest home for at least three months and dependent in at least two activities of daily living.²¹ Only a small percentage of the long stay nursing home residents were eligible for participation (19%). Over half the participants (58%) used wheelchairs for locomotion, and 75% had at least three co-morbid conditions. The most common reason for ineligibility in the trial was impaired cognitive functioning (26%). Randomisation was in groups of four and stratified by nursing home. Most intervention participants (94%) received strength and endurance exercises. Physical therapy was progressive and also included transfer and gait training. Of the 14 people (7%) who did not complete the follow up assessments at four months, 12 had died. Modest improvements in physical functioning were reported, and there was no significant difference between the intervention and control groups in the number of fallers, falls, serious fall injuries, and falls requiring health care use.

Lord et al²² studied the effect of regular exercise on balance, strength, and falls in older women randomly selected from the community. Women (mean (SD) age 71.6 (5.4) years) were randomly allocated to an exercise group (n = 100) or a control group (n = 97). Exercise classes were held in two community sites easily accessible by public transport, and the classes emphasised enjoyment and social interaction. The exercise sessions incorporated warm up, conditioning, stretching, and cool down periods to music. Sensorifunction assessments provided quantitative measurements of systems contributing to balance that could be enhanced by exercise. The investigators reported that structured general exercise classes were effective in significantly improving leg muscle strength and other outcomes, without the use of specific strengthening equipment. There was no significant difference between the exercise or control groups in the proportion of fallers and recurrent fallers, although a trend towards lower fall frequency was observed in participants who attended 75% or more of the exercise classes. The authors suggested that incorporating interventions in addition to the exercise programme, such as checking and modifying vision, may be a more effective falls prevention strategy.

The effects of two different exercise approaches on physical functioning and falls were studied at the Atlanta FICSIT trial site.²³ Men and women aged 70 years and older living in the community were randomised to one of three arms: Tai Chi classes (n = 72; mean (SD) age 76.9 (4.8) years), computerised balance training (n = 64; mean (SD) age 76.3 (5.1) years), or an education control group (n = 64; mean (SD) age 75.4 (4.1) years). Tai Chi classes concentrated on components of movement that often become limited with aging, including the standing base of support, body and trunk rotations, and reciprocal arm movements. The participants were encouraged to practise the movements at least twice a day at home for 15 minutes. Computerised balance training took place on a moveable platform and under one on one supervision. Participants practised moving their centre of mass without moving their feet with their eyes open and then closed and also during floor movement. Both were 15 week interventions. Tai Chi was most effective in reducing falls in people who fell recurrently, and, compared with controls, Tai Chi participants were less afraid of falling.

The Seattle FICSIT trial targeted men and women, mean age 75 years, with impairments in balance and strength.²⁴ Eligible participants were those unable to complete eight tandem steps without errors and those below the 50th percentile in knee extensor strength for the subject's height and weight. Only 7% from a random sample of 13 866 health maintenance organisation enrollees were eligible to take part. The intervention participants attended supervised exercise classes for 24–26 weeks and were then given a discharge plan to continue exercising in supervised or unsupervised settings for a further three months. The study reported no significant effect of either strength or endurance training on gait and balance measures. One mechanism proposed by the authors to explain why exercise did not reduce fall rates but the fall rate in the control group increased was that people with mild deficits in strength and balance may be at high risk for further deterioration and exercise delays this decline.

Campbell et al¹⁰ targeted a group at high risk for falling, women aged 80 years and older. The women were invited by their general practitioner to participate. Participants were randomised to an exercise intervention group (n = 116; mean (SD) 84.1 (3.4) years of age) or a control group (n = 117; mean (SD) 84.1 (3.1) years of age). The exercise programme was designed as a public health intervention to prevent falls and injuries in older people. At six months there was a significant improvement in two measures of strength and balance in the intervention group compared with the control group, when assessed by an independent physiotherapist blind to group allocation. There were no significant differences between the two groups in six other tests of strength, gait, endurance, and function. Despite very modest improvements in physical functioning, falls and moderate injuries were reduced in the exercise group compared with the control group. Participants were invited to continue in the programme for a second year (summarised separately in table 1).¹¹ Of the 213 participants remaining at the end of one year, 71% agreed to continue for a second year. Those who continued were more active and less afraid of falling at the end of year one and took fewer medications at baseline compared with those who declined to take part. At the end of the second year, 31 (44%) of those remaining in the intervention group were still exercising at least three times a week. The year two follow up showed that the lower fall rate achieved in year one could be sustained over a second year. More frequent visiting from the physiotherapist and encouragement from the general practitioner to continue exercising may have improved exercise compliance. The intervention was designed for easy community implementation but has only been tested in women aged 80 years and older living at home.

In a companion study, Campbell and colleagues tested the same home based exercise programme and a second intervention, gradual withdrawal of psychotropic medication in men and women currently taking those medications.²⁵ Only 19% of eligible participants who were invited to take part by their general practitioner agreed to participate. The study compared the effects of exercise (n = 45) versus no exercise (n = 48), and psychotropic medication withdrawal (n = 48) versus continuing to take the original psychotropic medication (n = 45). Nearly half (45%) of the participants stopped taking the study capsules before the 44 week trial was completed. The exercise programme was modified so that there was no upper limit for the amount of ankle cuff weights used for leg strengthening exercises. The home based exercise programme was associated with significant improvements in tests of strength and balance at six months (unpublished). There was a 66% reduction in fall rate in the medication withdrawal group but no significant reduction in falls in the exercise group. The exercise programme was less effective in this study group than in the group of women 80 years and older.

MULTIPLE INTERVENTION TRIALS WITH AN EXERCISE COMPONENT

In a large community trial, participants were randomised by household to an intervention group (n = 1611) or minimal treatment control group (n = 1571).²⁶ Nearly a third of participants (32%) were 70–74 years of age. The intervention emphasised removal of home hazards, reducing risk taking behaviour, and improving physical fitness. The exercises selected for the intervention were designed to involve all body parts, maintain range of movement, provide strengthening, and improve posture and balance. After one supervised group session, the participants were given a manual and instructed to carry out the exercises at home. At one year, the odds of being a faller was significantly less in the intervention group. Statistical analysis did not address the fact that participants were randomised by household, but the unit of analysis was the individual. However, 75% of the households had only one participant.

At the New Haven FICSIT site, 301 community dwelling men and women aged 70 years and older with at least one risk factor for falling were studied (85% of the eligible study population).²⁷ The participants were randomised to either a multiple risk factor intervention group (n = 153; mean (SD) age 78.3 (5.3) years) or a control group (n = 148; mean (SD) age 77.5 (5.3) years). The intervention participants received specific interventions depending on a baseline assessment of falls risk factors. Participants in the intervention group received a mean (SD) of 7.8 (4.0) home visits. The participants were instructed to perform the exercises often (twice a day for 15–20 minutes each session). The physical assessor and falls assessor were blind to group allocation. At one year there was a significant reduction in the percentage of intervention participants compared with controls still taking four medications or more, with balance impairments and with impairments in transfers at baseline. There was also a significant reduction in the proportion of fallers in the intervention group compared with the control group at one year. Muscle strength did not improve, and the authors suggest that manual muscle assessing may be insensitive to change, or alternatively their strength training regimen was of insufficient intensity. This well designed study provides good evidence for the effectiveness of a targeted multifactorial falls prevention programme in community dwelling older people.

A volunteer sample of 118 women, mean age 64.5 (range 60–73) years, were randomised to a calcium supplementation or calcium supplementation plus exercise group.²⁸ The exercise component of this two year trial involved weight bearing exercises to music in a centre. Bone mineral density showed a significant increase at one of three sites in the exercise plus calcium supplementation group. The method used for monitoring falls and injuries was not specified. There were fewer falls in the calcium plus exercise group than the calcium group between 12 and 18 months but the difference over the two year period was not significant. With no report of intermediate outcomes it is not known if the exercise programme was associated with improvements in balance and strength.

ECONOMIC EVALUATION WITHIN THE STUDIES

Two of the studies reviewed reported the cost of the intervention and three studies included total health care service costs as outcome measures in the trial (see table 3).^21,24,27 One study reported the charge for the physical therapy intervention delivered to nursing home residents and estimated health care costs for all participants during the four month trial.²¹ Buchner et al²⁴ estimated health care use and costs after the first six months of the trial because exercise participants (but not controls) were asked to delay elective procedures until the end of the supervised exercise period. Hospital use was similar in both exercise and control groups, but control participants were more likely to spend more than three days in hospital.

There has been one report of a detailed economic evaluation of a falls prevention programme tested in these trials.²⁹ Rizzo et al investigated the cost effectiveness of the home based multifactorial programme which included an exercise component.²⁷ Table 3 shows a summary of results. The intervention was more cost effective for those at “high” risk, defined as having four or more of the eight targeted risk factors for falls. Health care costs resulting from falls during the study were also identified, and in each category, costs were lower for the intervention than the control group. No statistical comparisons were made for health care costs between the exercise and control groups. Sensitivity analyses indicated that cost effectiveness ratios were robust for minimum to maximum intervention costs, and for 25th to 75th percentile values of total health care cost distribution. The authors limited the time horizon to the year of the trial and did not attempt to forecast costs or consequences of the intervention into the future.

Discussion

SYNTHESIS

All 12 articles included in the review were reporting randomised controlled trials. Nine articles reported the effect of exercise only and one of these reported a second year of follow up. Three studies evaluated the effectiveness of exercise in combination with other interventions in preventing falls. Four studies investigated the effect of exercise in women only and all the studies, except one, involved independent, community dwelling older people rather than those in institutions. Five studies included people aged 60 years and older, two those aged 65 years and older, two those aged 70 years and older, one those aged 80 years and older, and one included participants aged between 68 and 85 years of age. The interventions included strengthening, endurance, balance and flexibility exercises, computerised balance training, Tai Chi, the stand up/step down procedure, and walking as well as combinations of these exercises. In five studies, the exercise intervention was delivered to a group,^{18,20,22,24,28} and in another four studies exercises were carried out in the home.^10,25–27 In one study both a group and home based approach was incorporated in one of the exercise interventions, and the second exercise intervention was not home based but required one on one supervision.²³ In the rest home, trial exercises were delivered one on one.²¹ Definitions of a fall and methods of measuring falls and testing effectiveness differed. Intention to treat analysis was stated in four studies.^10,21,24,25 Five studies showed a significant reduction in the rate of falls or risk of falling in the intervention group.^{10,23,24,26,27}

EXERCISE PROGRAMME COMPONENTS

There is a need to identify which components of an exercise programme are most effective in lowering falls risk. A wide variety of exercise interventions have been tried using different exercise frequencies, intensities, and duration periods. Studies successfully lowering falls have used strength and balance retraining, endurance training, and Tai Chi. A meta-analysis of the seven FICSIT exercise trials suggests that balance may be more effective in lowering falls risk than the other exercise components.¹⁷ Tinetti et al³⁰ investigated the effectiveness of a multifactorial intervention programme on the number of falls risk factors and concluded that a change in balance score of 1 (possible scores ranged from 0 to 12) was associated with an 11% reduction in fall rate. It is probable that exercise would have had the greatest effect on balance in this multiple intervention study. Three successful programmes have required the participants to exercise regularly against resistance using either therabands or weights.^10,24,27

COMPLIANCE

Programmes should be acceptable to older people to ensure compliance, and this needs to be considered at the exercise programme design stage. Definitions of exercise compliance differed, and two studies failed to report exercise monitoring and compliance.^18,20 Exercise compliance at one and two years will provide a better indication of programme acceptability than measures after shorter time periods. One trial reported 27% (31 of 116) of participants from the original sample still carrying out exercise sessions at least three times a week at two years.¹¹ It is not known whether a home based or group approach is more acceptable to older people. Programmes offering both approaches may enhance compliance.

ADVERSE EFFECTS

Four studies addressed adverse events.^21,22,24,27 Ten participants (6.5%) reported self limiting musculoskeletal symptoms in one home based programme, which the investigators attributed to the exercises.²⁷ No medical incidents occurred in another trial during group exercise sessions.²² One study reported that exercise related injuries were uncommon and not an important factor associated with dropout.²⁴ At the San Antonio FICSIT site, adverse effects were monitored by research assistants, blind to group assignment.²¹ Intervention participants reported moderate muscle soreness at 7% of the physical therapy sessions but physical therapists reported no injuries during the exercises. There were no significant differences in severe soreness, bruising, and fatigue between participants receiving physical therapy and those receiving friendly visits. Exercise can be carried out safely in older people with moderate disability and intact cognitive functioning, and also in frail institutionalised older people with intact cognitive functioning under one on one supervision from a physical therapist.

STUDY FACTORS DIMINISHING BENEFIT

Six studies reported no change in falls after the exercise intervention.^{18,20–22,25,28} We consider that the following factors contributed to this lack of effectiveness. Several studies used exercise of inadequate intensity to modify falls risk factors, and this was shown by the lack of change in intermediate variables.^18,20,26 Most negative studies lacked sufficient power to detect a reduction in falls, although reducing falls was not necessarily a primary outcome in some of these trials. Exercise may be less effective in fall prevention when there are other significant risk factors for falls present that are not influenced by exercise. For example, in a younger sample of men and women on psychotropic drugs, exercise was less effective in reducing falls than in an older frail population.²⁵ One exercise trial targeted frail rest home residents.²¹ This study reported modest improvements in physical function following one on one physical therapy and no effect on falls. While intermediate outcomes improve in frail institutionalised elderly following high intensity strength training,³⁴ falls may not decrease because other risk factors may not improve. Lastly, study compliance may be too low for the intervention to be effective across the sample as a whole when analysed on an intention to treat basis.

FALL RELATED INJURIES AND COSTS

Owing to the low number of serious fall injury events such as fractures, the studies, even in meta-analyses, lacked sufficient power to determine whether exercise had a beneficial effect on serious fall injury risk.¹⁷ One exercise study reported a significant reduction in moderate injuries in the exercise group compared with the control group at one¹⁰ and two years.¹¹ Similarly, studies looking at health care costs lacked adequate power to show cost savings. However, some falls prevention intervention studies have reported a reduction in health care use as a result of the intervention.^24,31,32 Reductions in falls should reduce the number of fall related injuries but there may be a difference in the degree of reduction. An exercise programme may improve protective responses at the time of the fall. A long term exercise programme may improve bone mineral density. On the one hand a fitter quicker group of elderly people may fall at greater speed while about their daily activities.³³ On the other hand active older people may spend less time in hospital.²⁴

CONCLUSIONS AND RECOMMENDATIONS

The wide variety of exercise interventions tried, some successful and others not, does enable us to draw some conclusions. Appropriate exercise programmes can decrease the number of falls and fall risk in randomised controlled trials but certain conditions need to be met.

For maximum effect the population needs to be right—not too fit and not too frail. Exercise interventions in people in institutions have not yet been shown to lower the risk of falling.²¹ The same exercises used in younger populations have not been as effective as in older groups.¹⁰ With increasing age there is a progressive loss of muscle strength and stability, but the weakness needs to reach a certain point or threshold before daily functions are affected. It is possible that around this point small increases in strength have a disproportionate effect on function, and exercise programmes are most effective.

The exercises need to be of sufficient intensity to improve muscle strength. We suggest that most investigators, including ourselves, initially underestimated the capacity of older people to manage weights. Balance retraining should be an important component of any exercise programme designed to decrease falls. This may consist of specific dynamic balance retraining exercises or be a component of a movement form such as Tai Chi. The exercises need to be regular and sustainable. There is no evidence of benefit beyond the period of the exercises but continued participation can lead to sustained lower fall risk at least up to two years.¹¹

The exercises may be performed at a centre or at home. Home exercises are suitable for a frail less mobile population without easy access to transport. They are safe if properly established by a trained therapist but the supervision is less than with a centre based programme. A centre based programme does have the additional value of social interaction which has important beneficial effects in its own right.³⁵

If the exercises are part of a public health programme to be introduced widely in the community, they should be simple, easily instituted, and low cost. Elderly people involved in fall prevention exercise programmes are prone to intercurrent illness, accident, and social change. Programmes need to have the resources to reassess and restart. They should also be planned for long term use. Repetitive programmes with little variety are unlikely to be sustained. If the exercises are part of a programme of fall prevention in a person presenting with falls, then the exercises must be part of a full assessment of the person's risk factors and treatment. Exercises are of value in falls prevention when part of a comprehensive package.²⁷

There is a need for further research on the most effective balance and strength retraining exercises, the groups most likely to benefit, the determinants of compliance, and the best ways to improve this.

Exercise programmes designed to prevent falls in older people have two important advantages. Falls are very common so programmes are likely to be cost effective when compared with other public health measures in this population. Exercise is also beneficial to the participants in additional ways such as decreasing fear of falling, improving functional reserve by increasing strength, and in improving other important health areas as varied as cardiovascular health,³⁶ sleep,³⁷ depression,³⁸ and mortality.³⁶

True or false?

1. Falls prevention exercise programmes work on which of the following premises:

   1. Muscle strength and balance are common risk factors for falls.

   2. Exercise must be continued to be effective.

   3. Only fit elderly people should take part.

   4. Strength training should be a gentle, optional extra exercise

2. Proven benefits of falls prevention exercise programmes to date include:

   1. Decreased fear of falling.

   2. Reduced admissions to rest home.

   3. Improved functional independence.

   4. Reduced hip fractures.

3. In a systematic review on falls which electronic databases would be searched?

   1. Web of science

   2. Ovid

   3. Generator

   4. Cochrane database of systematic reviews

(Answers p 76.)

True or false?—answers

(T = true; F = false)

p 7: Gardner MM, Robertson MC, Campbell AJ. Exercise in preventing falls and fall related injuries in older people

1 (a) T; (b) T; (c) F; (d) F. 2 (a) T; (b) F; (c) T; (d) F. 3 (a) T; (b) T; (c) F; (d) T.