Effectiveness and economic evaluation of a nurse delivered home exercise programme to prevent falls. 1: Randomised controlled trial

BMJ 2001; 322 doi: http://dx.doi.org/10.1136/bmj.322.7288.697 (Published 24 March 2001)
Cite this as: BMJ 2001;322:697
  1. M Clare Robertson (clare.robertson{at}stonebow.otago.ac.nz), research fellowa,
  2. Nancy Devlin, senior lecturerb,
  3. Melinda M Gardner, research physiotherapista,
  4. A John Campbell, professor of geriatric medicinea
  1. a Department of Medical and Surgical Sciences, Otago Medical School, PO Box 913, Dunedin, New Zealand
  2. b Department of Economics, University of Otago
  1. Correspondence to: M C Robertson


    Objectives: To assess the effectiveness of a trained district nurse individually prescribing a home based exercise programme to reduce falls and injuries in elderly people and to estimate the cost effectiveness of the programme.

    Design: Randomised controlled trial with one year's follow up.

    Setting: Community health service at a New Zealand hospital.

    Participants: 240 women and men aged 75 years and older.

    Intervention: 121 participants received the exercise programme (exercise group) and 119 received usual care (control group); 90% (211 of 233) completed the trial.

    Main outcome measures: Number of falls, number of injuries resulting from falls, costs of implementing the programme, and hospital costs as a result of falls.

    Results: Falls were reduced by 46% (incidence rate ratio 0.54, 95% confidence interval 0.32 to 0.90). Five hospital admissions were due to injuries caused by falls in the control group and none in the exercise group. The programme cost $NZ1803 (£523) (at 1998 prices) per fall prevented for delivering the programme and $NZ155 per fall prevented when hospital costs averted were considered.

    Conclusion: A home exercise programme, previously shown to be successful when delivered by a physiotherapist, was also effective in reducing falls when delivered by a trained nurse from within a home health service. Serious injuries and hospital admissions due to falls were also reduced. The programme was cost effective in participants aged 80 years and older compared with younger participants.

    What is already known on this topic

    Falls are the costliest type of injury among elderly people, and the healthcare costs increase with frequency of falls and severity of injuries

    An exercise programme delivered by a physiotherapist was successful in reducing falls and moderate injuries in elderly people

    What this study adds

    An exercise programme to prevent falls in elderly people worked well when delivered by a district nurse from a home health service in the suburbs of a large city

    Researchers, public health administrators, and health practitioners can work together to benefit elderly people in the community


    The frequency, serious consequences, and healthcare costs of falls in elderly people are well documented.1-5 Randomised controlled trials of single and multiple interventions have shown that falls can be reduced.6 The effectiveness of these programmes and their costs in usual healthcare settings have not been reported. Our research group developed a home based programme of strength and balance retraining, which was effective in reducing falls and falls resulting in moderate injuries when delivered by a research physiotherapist to a group of women aged 80 years and older living in the community. 7 8

    We have now tested in two healthcare settings the effectiveness and efficiency of the programme when delivered by health professionals previously untrained in prescribing exercise. This first paper reports on the effectiveness and cost effectiveness of the exercise programme in both men and women aged 75 years and older when delivered from an established home health service by a trained district nurse.

    Participants and methods

    Participant recruitment

    We identified potential participants aged 75 years and older from computerised registers at 17 general practices (30 doctors) in the West Auckland area, New Zealand. These patients received a letter from their doctor inviting them to take part in the study. The criteria for exclusion were inability to walk around own residence, receiving physiotherapy at the time of recruitment, or not able to understand the requirements of the trial. Recruiting took place over a six month period in 1998.

    Trial design

    This was a randomised controlled trial with one year's follow up. The sample size calculation was based on the proportion of elderly people who fell once or more in a 12 month prospective community study,9 an expected reduction from 0.50 to 0.30, and 20% allowance for dropouts. Our study was approved by the ethics committee of the Health Funding Authority Northern Division.

    Potential participants were informed there was an equal chance they would receive the exercise programme or act as a control. After written informed consent was obtained and baseline assessments (personal characteristics, health, and function) completed at home by an independent assessor, we randomised 240 participants: 121 to the exercise programme (exercise group) and 119 to usual care (control group). A statistician developed the schedule for group allocation using random numbers, and this was held at another centre. Participants were then informed of their group allocation by telephone.


    A district nurse who had had no previous experience in prescribing exercise attended a one week training course run by the physiotherapist from the research group. A series of site visits and regular telephone calls were made by the supervising physiotherapist to assess and ensure quality control.

    The implementation of the exercise programme was run from a home health service based in a geriatric assessment and rehabilitation hospital. The nurse delivered the exercise programme in conjunction with her work as a district nurse. The intervention consisted of a set of muscle strengthening and balance retraining exercises that progressed in difficulty, and a walking plan.7 The programme was individually prescribed during five home visits by the instructor at weeks 1, 2, 4, and 8, with a booster visit after six months. The number of repetitions of the exercise and the number of ankle cuff weights (1, 2, and 3 kg; range 0 to 6 kg) used for muscle strengthening were increased at each visit as appropriate. Participants were expected to exercise at least three times a week (about 30 minutes per session) and to walk at least twice a week for a year. Compliance was monitored with postcard calendars similar to those used to monitor falls. For the months when no home visit was scheduled the nurse telephoned participants to maintain motivation and discuss any problems.

    Measurement of falls and injuries and health status

    Falls were defined as “unintentionally coming to rest on the ground, floor, or other lower level.”10 Falls were monitored for one year in both groups by asking participants to return preaddressed and prepaid postcard calendars for each month. The independent assessor telephoned participants to record the circumstances of the falls and any injuries or resource use as a result of the falls. She remained blind to group allocation.

    Fall events were classified as resulting in “serious” injury if the fall resulted in a fracture, admissions to hospital with an injury, or stitches were required, “moderate” injury if bruising, sprains, cuts, abrasions, or reduction in physical function for at least three days resulted or if the participant sought medical help, and “no” injury. The circumstances of “serious” injuries were confirmed from hospital and general practice records. The investigator classifying fall events remained blind to group allocation. The SF-12 questionnaire was used to estimate self perceived health status at entry to the trial.11

    Methods used in economic evaluation

    We used cost effectiveness analysis to enable comparisons of programme efficiency with other interventions for preventing falls. We considered costs from the societal perspective because of the broad nature of the problems caused by falls, and we reported them in New Zealand dollars according to 1998 prices, exclusive of government goods and services tax. The control group was used as the comparator for the analysis. We measured cost effectiveness as the incremental cost of introducing the programme per fall event prevented during the trial.

    The concept of opportunity costs was kept in mind so that all relevant costs—that is, those resources that could have been employed elsewhere—could be included. We performed one way sensitivity analyses.

    Costs of the exercise programme

    We focused on the costs of implementing the exercise programme. Although there were costs associated with developing the programme, these costs were incurred before the trial and were not incremental to this programme. We did not include the research costs of evaluating the programme.

    Costs for implementing the programme were obtained from trial records and the financial records of the hospital and research group, using actual costs when available. We did not include the costs of recruiting the exercise instructor because existing staff in an organisation may deliver the exercise programme. We did not put a value on the time participants spent exercising or walking as it was assumed these activities were done in their leisure time; the opportunity cost was taken to be zero. Half of the recruiting costs for this first paper were allocated to implementation of the programme because half those recruited were randomised to the control group and did not receive the exercise programme. We estimated overhead costs as 21.9% of observed resource use because this was the sector average reported for all hospital and health services in New Zealand for operating costs and overhead expenses in 1998-9.12

    Resource use and healthcare costs resulting from falls

    In a previous trial of the exercise programme we found that 90% of the estimated healthcare costs resulting from falls were for hospital inpatient and associated health service costs.13 A further 4% were for those services used as a result of serious injuries and were not provided by the local hospital. Estimated costs for injuries we classified as moderate made up the remaining 6% of total healthcare costs resulting from falls.

    Therefore to estimate the costs resulting from fall injuries in this trial we restricted measurement to actual costs incurred by the hospitals admitting participants as a result of a fall. For each fall event these included costs for emergency room, theatre, ward, physician, radiology, laboratory, and blood services, pharmacy products, hospital social workers, physiotherapy, and occupational therapy. Each hospital cost item included overhead costs (cleaning, heating, lighting, telephone, laundry, food, administration, orderlies, computing, and depreciation on equipment) calculated by the accounting convention at each hospital.

    Costs for hospital items were identified as being associated with a fall by matching the date of the cost record with the date of a trial record for a fall event. Cost records were included only if the department and product description indicated that the item was likely to have been used as a result of the fall.

    Calculation of cost effectiveness ratios

    We measured cost effectiveness as the ratio ΔC: ΔE, where ΔC (incremental cost) was the change in resource use resulting from the exercise programme.14 This was taken as the total cost of implementing the exercise programme because the control group did not receive an intervention, plus the difference in hospital costs resulting from falls during the trial for the two groups. We planned to include estimates for hospital costs as a result of falls in ΔC only if these costs or the number of serious injuries proved to be significantly different between the two groups.

    We measured ΔE (incremental effect) as the difference between the number of falls and the number of falls resulting in moderate or serious injury in the two groups. We considered the actual number of fall events and a standardised measure, fall events per 100 person years. This measure takes into account the variable follow up times for individuals in the trial.

    Sensitivity analysis

    We carried out one way sensitivity analyses by calculating cost effectiveness ratios. We did this with a range of estimates of cost items for implementing the exercise programme to investigate robustness of the ratios to different delivery scenarios. We used the 125th centile of the total, the total, and the 75th centile of the total costs for implementation when calculating the cost effectiveness ratios to account for the possibility of different cost conditions when replicating the programme in different settings. Training and supervision of the exercise programme could be carried out from the same rather than a distant centre. We therefore calculated cost effectiveness ratios excluding travel costs between centres and the associated accommodation costs. We used the 125th centile of the costs for the home visits to give an indication of costs for delivering the programme in a more spread out community. The ankle cuff weights we used were manufactured cheaply in a non-commercial environment. Participants may well have been encouraged to use more weights. Therefore in the sensitivity analyses we used four times the actual cost of the weights. The home health service could not identify any extra overhead costs as a result of running the exercise programme. We included this scenario in the sensitivity analyses. For this part of our study we also calculated cost effectiveness ratios for those aged 80 years and older by apportioning the costs of the programme (on a pro rata basis) between the 75 to 79 year olds and those aged 80 and older, and using the number of fall events prevented in those aged 80 years and older.

    Time horizon

    Assuming that participants keep exercising, the benefits of the exercise programme would extend past the time individuals participated in the trial, but the extent of this benefit and longer term compliance rates are uncertain. We calculated cost effectiveness ratios for the duration of the trial only.

    Statistical analysis

    We analysed data on an intention to treat basis with Stata Release 6 and SPSS 6.1.1. No deviations occurred from random allocation—all those who were allocated to the exercise group received at least one home visit, and no participants in the control group received the programme. The mean (SD) time between baseline assessment and the first home visit was 11.5 (6.1) days.

    We compared the numbers of falls in the two groups using negative binomial regression models.15 These models estimate the number of occurrences of an event when the event has Poisson variation with overdispersion, and they allow for variable follow up times for participants and investigation of the treatment and interaction effects. We used Student's t test to compare means and Fisher's exact test or c2 test to compare proportions between groups.

    Flow of participants through trial


    Trial participants and follow up

    The mean (SD) age of participants was 80.9 (4.2) years, and ages ranged from 75 to 95 years. Table 1 shows the characteristics of participants at entry to the trial.

    View this table:
    Table 1

    Characteristics of participants at entry to trial. Values are numbers (percentages) unless stated otherwise

    The figure shows the flow of participants through the trial. We have not reported the results of assessments repeated after one year. More participants from the exercise group than the control group completed the trial (113 v 98, difference 11%, 95% confidence interval 3% to 19%). Those who died or withdrew were more likely to have had a fall in the year before the trial and took more drugs at entry to the trial (mean (SD) number 4.3 (2.4) v 2.8 (2.3), P=0.002).

    Overall, 43% (49 of 113) of participants who completed the trial carried out their prescribed exercise programme three or more times a week, 72% (n=81) carried it out at least twice a week, and 71% (n=80) walked at least twice a week during the year's follow up.

    Falls and fall related injuries

    Table 2 shows the actual and standardised numbers of falls and the numbers of falls resulting in injuries during the trial. We found a 46% reduction in the number of falls during the trial for the exercise group compared with the control group (incidence rate ratio from negative binomial regression model 0.54, 95% confidence interval 0.32 to 0.90). The number of falls was reduced in those aged 80 years and older (81 v 43 falls for control and exercise groups, respectively; P=0.007), and there was no difference in participants aged 75 to 79 years. One participant did fall while exercising according to instructions.

    Fewer participants in the exercise than control group had a serious injury resulting from a fall during the trial (2 v 9, relative risk 4.6, 95% confidence interval 1.0 to 20.7). Nine falls resulted in fractures (five required hospital admission) and three in lacerations requiring sutures. The same numbers of moderate injuries occurred in the two groups.

    View this table:
    Table 2

    Incidence of fall events and follow up times

    Economic evaluation

    Costs of implementing the exercise programme

    Table 3 shows the values for the cost items for implementing the exercise programme. The programme cost $NZ52 229 ($NZ432 per person) to deliver to the 121 participants for one year.

    View this table:
    Table 3

    Incremental costs of implementing exercise programme

    Resource use resulting from falls

    Overall, 44 of 189 (23%) falls resulted in the use of healthcare services (table 2). Medical care was sought for more falls in the control than exercise group, but the difference was not significant. The five people admitted to hospital were all from the control group and were aged over 80 years. The actual cost of these admissions and therefore the hospital cost averted by the exercise programme was $NZ47 818.

    Cost effectiveness measures

    The incremental cost per fall prevented was $NZ1803 (table 4). Estimates for the cost per fall with an injury prevented ranged from $NZ5603 to $NZ9437 for the different cost scenarios. When we included cost savings from hospital admissions in the calculation of cost effectiveness ratios, the estimates of the ratios were considerably lower (some indicated cost savings) than for those calculated using the exercise programme costs alone.

    View this table:
    Table 4

    Cost effectiveness ratios and sensitivity analysis: incremental cost per fall event prevented in exercise group compared with control group

    The exercise programme was considerably more cost effective for those aged 80 years and older than for the total sample. Estimates for cost effectiveness ratios for implementing the exercise programme in this age group were $NZ682 per fall prevented and $NZ1852 per injurious fall prevented. When hospital costs averted and costs for implementation were both used in the calculations of the cost effectiveness ratios, the net cost of the programme for those aged 80 years and older resulted in cost savings of $NZ576 per fall event prevented and $NZ1563 per injurious fall event prevented.


    An individually tailored exercise programme delivered at home can prevent falls. The programme can be delivered safely by a district nurse and is suitable for both men and women. Academic researchers are sometimes perceived as being remote from the day to day realities of delivering health care, and the results of research do not always reach those who could benefit.16 Our trial is an example of effective collaboration between researchers, public health professionals, and administrators, resulting in health benefits to elderly people in the community.

    Subgroup analysis showed that the programme was effective in those aged 80 years and older but not in those aged 75 to 79 years. Although our trial was not designed to test this, the finding is consistent with our previous finding that falls were not reduced by the exercise programme in a sample of women and men aged 65 years and older who were taking psychotropic drugs.17 The programme may be more effective in frailer, elderly people than younger, fitter people because the exercises increase strength and balance above the critical threshold necessary for stability.

    As with all age groups only a proportion will be prepared to join an exercise programme, but as shown by the characteristics at trial entry, the participants represented a general population of this age group. Follow up was good, although more people withdrew from the control than exercise group. This may have biased the results against effectiveness because those who withdrew were at a higher risk of falling.

    The exercise group had the same number of moderate injuries but fewer serious injuries as a result of a fall than the control group. Injuries resulting in hospital admissions are costly, and reducing injuries such as fractures and lacerations in our trial resulted in cost savings.

    We used hospital admission costs as a result of a fall injury as our estimate of the consequences of the exercise programme. We found the same number of moderate injuries resulting from falls in both groups. We also knew from an earlier study that the remaining medical and personal costs resulting from falls account for only 10% of the total healthcare costs for falls.

    We estimated the cost of implementing the exercise programme to serve as a guide for the cost of replicating the programme in the future. Costs may well differ in a different setting or be influenced by the reporting expectations of those who fund the programme, by the efficiency and experience level of the instructor, and by the age group enrolled. For example, some of the costs of implementing the programme would not be incurred if the programme was run in one urban area (see table 4 for the same centre scenario).

    Comparison with other interventions for preventing falls


    Implementing this single intervention proved as or more effective in reducing falls than other successful community based programmes reported in the literature.18-21 Withdrawing psychotropic drugs reduced the risk of falls by 66%, but there were difficulties in recruiting participants to the trial and a high dropout rate.16 Other community based interventions have not proved successful in reducing falls.22-25

    Economic efficiency

    Little information is available at present for comparing the efficiency of the exercise programme with other interventions aimed at preventing falls. We found only two publications reporting the cost effectiveness of implementing an intervention for preventing falls in the community. 26 27 The exercise programme in our trial was more cost effective than a home based, targeted, multifactorial intervention (total intervention implementation costs per fall prevented $US2668 (at 1993 prices; around $NZ6141) versus $NZ1803, although this figure did include “developmental” costs for the programme). A home assessment and modification programme, successful in reducing falls in those with a history of a fall in the previous year, cost an average of $A4986 (at 1997 prices; $NZ1.00=$A0.89 in 1997) per fall prevented. This cost effectiveness ratio incorporated all healthcare resource use during the trial.27

    Some other studies have shown reduced healthcare use or cost savings occurred as a result of a programme to prevent falls. 19 28 Benefits may result from early identification of health problems, earlier referrals, or physically fitter people spending a shorter time in hospital.


    In our previous trials, the exercise programme was delivered by a physiotherapist. 7 17 We conclude that a trained district nurse is also an appropriate person to implement the programme. Implementation of the programme worked well when run from an established home health service and required the minimum of input from other staff. We recommend that nurses are trained and supervised by a suitably qualified physiotherapist. Although supervision in the same centre would be less time consuming and less costly, long distance supervision combining site visits and telephone contact worked well. This trial studied one trained nurse in one health service delivering a home based exercise programme. Our second pragmatic trial studies practice nurses trained to deliver the programme from general practices.29


    We thank the participants; the West Auckland doctors and their receptionists; Gaye McKay, exercise instructor; Tania Roebuck, independent assessor; Lenore Armstrong, research nurse; Beth Cozens, manager, home health services; Margaret Devlin, Safe Waitakere; Toni Gibbins, clinical analyst; Peter Herbison, statistician; Molly Kavet, clinical information analyst; Professor Murray Tilyard and the General Practice Research Unit; Sheila Williams, statistician; and Gail Woollacott, locality manager.

    Contributors: All authors contributed to the study or protocol design, or both, interpreted the data, and wrote the paper. AJC directed the project. MCR managed the project and the data gathering, analysed and interpreted the data, and wrote the paper. MMG trained and supervised the exercise instructor. ND and Dr Paul Scuffham advised on the economic evaluation. AJC and MCR will act as guarantors for the paper.


    • Funding This project was funded by the Health Funding Authority Northern Division, New Zealand. MCR and MMG were part funded by Accident Rehabilitation and Compensation Insurance Corporation of New Zealand. MMG was also part funded by a Trustbank Otago Community Trust medical research fellowship.

    • Competing interests None declared.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    12. 12.
    13. 13.
    14. 14.
    15. 15.
    16. 16.
    17. 17.
    18. 18.
    19. 19.
    20. 20.
    21. 21.
    22. 22.
    23. 23.
    24. 24.
    25. 25.
    26. 26.
    27. 27.
    28. 28.
    29. 29.