Aim To assess the cost-effectiveness of exercise on prescription with ongoing support in general practice.
Methods Prospective cost-effectiveness study undertaken as part of the 2-year Women's lifestyle study randomised controlled trial involving 1089 ‘less-active’ women aged 40–74. The ‘enhanced Green Prescription’ intervention included written exercise prescription and brief advice from a primary care nurse, face-to-face follow-up at 6 months, and 9 months of telephone support. The primary outcome was incremental cost of moving one ‘less-active’ person into the ‘active’ category over 24 months. Direct costs of programme delivery were recorded. Other (indirect) costs covered in the analyses included participant costs of exercise, costs of primary and secondary healthcare utilisation, allied health therapies and time off work (lost productivity). Cost–effectiveness ratios were calculated with and without including indirect costs.
Results Follow-up rates were 93% at 12 months and 89% at 24 months. Significant improvements in physical activity were found at 12 and 24 months (p<0.01). The exercise programme cost was New Zealand dollars (NZ$) 93.68 (€45.90) per participant. There was no significant difference in indirect costs over the course of the trial between the two groups (rate ratios: 0.99 (95% CI 0.81 to 1.2) at 12 months and 1.01 (95% CI 0.83 to 1.23) at 24 months, p=0.9). Cost–effectiveness ratios using programme costs were NZ$687 (€331) per person made ‘active’ and sustained at 12 months and NZ$1407 (€678) per person made ‘active’ and sustained at 24 months.
Conclusions This nurse-delivered programme with ongoing support is very cost-effective and compares favourably with other primary care and community-based physical activity interventions internationally.
Statistics from Altmetric.com
Increasing physical activity levels long-term among sedentary populations can be challenging. Yet some physical activity interventions have been found to be effective as well as cost-effective.1 However, a recent systematic review noted that cost-effectiveness analyses are rare and more research in this area is needed, particularly to identify which characteristics are associated with favourable cost–effectiveness ratios.1 Physical activity interventions in primary healthcare, such as the Green Prescription in New Zealand, have been shown to be effective2 3 and cost-effective4 5 with a comparable cost–utility ratio to many currently funded pharmaceutical therapies. The original Green Prescription involved brief advice and an exercise prescription usually delivered by the general practitioner (GP), followed by monthly telephone support for 3 months delivered by regional sports trusts (RST). Despite favourable cost-effectiveness and cost–utility over 12 months, adherence rates to physical activity for those receiving a Green Prescription were low. Therefore, an ‘enhanced Green Prescription’ was designed, which was delivered by primary healthcare nurses and included more follow-up and added face-to-face interaction, to assess whether this ‘enhancement’ would improve adherence and cost-effectiveness of the original Green Prescription.6 7 This study assesses the cost-effectiveness over 24 months of the ‘enhanced Green Prescription’ intervention used to promote physical activity among less active women recruited through general practice.
What is already known on this topic
▶ Population increases in physical activity levels are associated with improved health outcomes.
▶ Community and primary healthcare-based physical activity interventions have been shown to be effective in increasing population physical activity levels.
▶ Cost-effectiveness of these interventions varies, and there is a scarcity of rigorously conducted cost-effectiveness studies of physical activity interventions.
What this study adds
▶ This exercise on prescription programme increased physical activity and was more cost-effective than most programmes assessed to date.
▶ Nurse delivery rather than doctor delivery and longer follow-up support (9 months) appeared to improve the cost-effectiveness of exercise on prescription programmes.
▶ Programmes involving primary care nurses and community-based telephone support could be set up elsewhere for efficient and cost-effective delivery of physical activity promotion.
The Women's lifestyle study was a 2-year randomised controlled trial (RCT) of the ‘enhanced Green Prescription.’6 7 A prospective cost-effectiveness analysis was undertaken from a societal perspective. Direct and indirect costs were calculated at baseline, 12 months and 24 months, including programme delivery costs, participant exercise costs, as well as primary and secondary care, allied healthcare and productivity costs.
Women between 40 and 74 years of age were screened for physical activity level from 17 general practices in the Wellington region. A validated one-question physical activity screening tool was used to identify women who were ‘less active’ (not achieving 30 min of at least moderate-intensity exercise such as brisk walking on 5 days or more per week).6 Participants were excluded if they had a medical condition that may be adversely affected by increasing physical activity or if their GP considered participation inappropriate. Between 2004 and 2006, 5913 women were sent invitations to be screened for current physical activity level. From those who responded (n=3036), 1089 were eligible (less active) and willing to participate.7
The intervention was 10 min of brief advice and a written exercise prescription given by a primary healthcare nurse, with telephone support for 9 months (average five calls lasting 15 min each) from an exercise facilitator from a RST and a half-hour face-to-face session with the nurse at 6 months. The recommended goal was at least 30 min of moderate-intensity physical activity five times per week. The intervention has been described in more detail elsewhere.6 7 Control participants received usual care from their GP.
The primary outcome was incremental cost of moving one person into the ‘active’ category over 24 months (achieving 150 min per week of at least moderate intensity physical activity) as assessed by the NZ Physical Activity Questionnaire.8 9 The secondary outcome was the incremental cost of increasing physical activity (cost per minute of moderate or vigorous physical activity gained per week) at 12 and 24 months. These outcomes allowed comparison with the cost-effectiveness of other primary-care-based physical-activity interventions.5 10 11
Direct costs of programme delivery were recorded. Other costs covered in the analyses included participant costs of physical activity participation, costs of primary and secondary healthcare utilisation, allied health therapies and time off work (lost productivity). In this paper, these ‘other’ costs are labelled as ‘indirect.’ Actual costs were collected wherever possible for each participant. Self-reported costs were collected by questionnaire administered by a research nurse during face-to-face interviews at baseline (prior to randomisation), 12 months and 24 months. Research nurses were blind to allocation of intervention at follow-up assessments.
The cost of delivering the programme by primary care nurses was calculated using an estimate of time spent issuing a Green Prescription during a health visit and records of visit duration for the 6-month face-to-face session. Information about costs and actual RST telephone support given to each participant was recorded by the RST.
Participant costs of physical activity participation
Self-reported costs included purchase of exercise or sports shoes, membership fees to exercise groups or gyms, costs of exercise equipment, travel to and from the location of exercise or physical activity, and any other costs associated with exercise, for the 12-month period prior to each assessment. Transport costs to and from exercise venues were calculated using self-reported distance to activity venue and car running costs for the period of interest (cost per km).
Primary healthcare utilisation
Unit costs were obtained from general practices, and actual utilisation was self-reported by participants at each assessment. Participant charges and subsidies vary according to the type of consultation (eg, accident or non-accident) and the socio-economic and health service utilisation status of the participant. This equates to four different charge rates: accident-related rate (Accident Compensation Corporation (ACC)), Community Services Card rate, High-User card rate and the rate for those holding no card (unsubsidised) (table 1). ACC is a government-funded accident insurance scheme, but small patient copayments are usually required at each consultation. Average rates were obtained for each type of consultation. To calculate the health funder component of GP consultations, data from previous studies that quantified the total cost of a GP consultation were used.12 13
After-hours medical centre visits and allied health visits
Unit costs were obtained from the only two after-hours medical centres in the region. Actual use was obtained by self-report. The unit costs for allied health providers, such as physiotherapists and chiropractors, were assigned based on the rate that ACC reimburse providers as specified in ACC regulations.14
Hospital and outpatient costs
Participants reported the reason for hospital admission and duration of stay for the previous 12 months at each study assessment. To calculate a cost weight for each event, costs for each type of admission were assigned using the Weighted Inlier Equivalent Separations method.15 A cost weight for each admission was calculated16 and multiplied by the national price for that financial year to produce an estimated cost for the event. Secondary care costs were based on public hospital costs for each admission type. Outpatient unit costs used average national costs for all categories of outpatient visit, excluding categories not relevant for this population, such as paediatric outpatient clinic costs. Unit costs for emergency department visits were obtained using national rates for level 5 and level 6 emergency department visits (Capital and Coast District Health Board, personal communication). These levels relate to the two emergency departments closest to where the study population resided. Actual use of outpatient and emergency department visits was obtained by self-report at each study assessment.
Costs of time off work (lost productivity)
Days of illness- and accident-related leave taken during the 12 months prior to each study assessment were obtained by self-report. Quarterly average wage data were obtained from Statistics New Zealand for women in 5-year age bands.17 The average daily wage for the four quarters relating to the relevant time period for each participant was applied to any days away from work.
A summary of unit costs and their sources is presented in table 1. All costs incurred in previous years were adjusted for inflation using the December Consumer Price Index (CPI) from each corresponding year to bring them in line with the June 2008 quarter.18 Discounting was not used, as the effects of discounting are small over such a short time-period (2 years). All costs are reported in New Zealand dollars (NZ$), and final results are also presented in Euros using conversion rate as at June 2008,19 and presented exclusive of goods and services tax.
Cost–effectiveness ratios were calculated for the cost of the primary and secondary outcomes at 12 and 24 months, with and without the inclusion of indirect costs. Repeated measures regression analyses were used to investigate whether there was any significant difference between the intervention and control groups in incremental change in indirect costs at 12 and 24 months, clustering by individual. Data were log-transformed for final analyses and presented back-transformed. An intention to treat analysis was undertaken according to allocation of randomisation regardless of adherence. Sensitivity analyses were also undertaken using the last value carried forward for missing data. Analyses were performed using SAS version 9.2 (SAS Institute, Cary, North Carolina).
Of the 1089 participants enrolled on the trial, 1008 (92.6%) attended the 12-month follow-up visit, and 974 (89.4%) attended the 24-month follow-up visit. Baseline characteristics are presented in table 2. Physical activity levels were significantly higher in the intervention group compared with the control group at 12 months (median 120 min/week vs 75 min/week) and 24 months (median 105 min/week vs 90 min/week) (p<0.01)7, and a greater proportion reached the target of 150 min of at least moderate intensity physical activity in the intervention group compared with the control group at 12 months (232/544, 43% vs 164/545, 30%) and 24 months (208/544, 39% vs 178/544, 33%) (p<0.001).7
All intervention participants (n=544) received the initial advice and a written Green Prescription from a primary care nurse. The estimated time taken by the nurse to administer the enhanced Green Prescription was 10 min per participant. Ninety-four per cent (n=514) received the 6-month face-to-face follow-up advice from the primary care nurse. The mean time for 6-month follow-up visit was 29 min.
The costs of nurse delivery of the intervention were estimated to be $3.62 for the initial brief advice and written exercise script and $10.47 for the 6-month follow-up visit, giving a total of $14.09 per participant for those who attended. Of the 544 intervention participants in the study, 535 (98%) were referred to the RST exercise specialists for telephone support over the following 9 months. The average cost of RST support was $75.22 per participant using the support. The total CPI adjusted enhanced Green Prescription programme cost was $93.68 per participant when costs were spread across all those allocated to intervention regardless of utilisation.
Health resource consumption is summarised in table 3, and indirect costs are summarised in table 4. There was no significant difference in indirect costs over the course of the trial between the two groups (rate ratios: 0.99 (95% CI 0.81 to 1.2) at 12 months and 1.01 (95% CI 0.83 to 1.23) at 24 months (p=0.9)) (table 5). A sensitivity analysis was conducted that carried forward the last value for all missing values (table 5), which did not change the results substantially.
The cost–effectiveness ratios were $687 (€331) per person made ‘active’ and sustained at 12 months, and $1407 (€678) per person made ‘active’ and sustained at 24 months (table 6). The cost–effectiveness ratios per minute per week of equivalent moderate intensity physical activity were $2.82 (€1.38) per minute sustained to 12 months and $8.58 (€4.20) per minute sustained to 24 months. When indirect costs were included in the calculation of cost–effectiveness ratios, the results did not change substantially at 12 or 24 months (table 6). Likewise, when last values were carried forward for missing data, results remained very similar (table 6).
The Women's lifestyle study enhanced Green Prescription programme cost $93.68 (€45.90) per participant enrolled to deliver. Cost–effectiveness ratios using programme costs were $687 (€331) per person made ‘active’ and sustained at 12 months, and $1407 (€678) per person made ‘active’ and sustained at 24 months, compared with usual care. Sensitivity analyses, including indirect costs and last value carried forward for missing values, did not change results substantially.
Therefore, the enhanced Green Prescription appears to be more cost-effective than the previously evaluated Green Prescription, which cost $1756 (€957) in programme costs to convert a sedentary person to an active category at 12 months.5 This may be explained by the fact that the enhanced Green Prescription used primary care nurses rather than doctors to deliver the advice, and the proportion made active at 12 months was slightly greater than in the previous Green Prescription trial.2 The extra telephone support and 6-month face-to-face follow-up with the nurse are likely to have contributed to the higher proportion of participants who were active at 12 months compared with the previous Green Prescription trial.2 The enhanced Green Prescription intervention also compares favourably with other primary care and community-based interventions reviewed, with better cost–effectiveness ratios for moving people to the active category than previously documented (figure 1).1 5 10 11 20 21
Strengths of this cost-effectiveness study include the prospective design, with the collection of costing variables alongside a RCT. Comprehensive costing data were collected, and in almost all cases, actual rather than estimated costs were used. Accordingly, few assumptions have been made. This contrasts with many of the previous cost-effectiveness studies conducted of lifestyle interventions, which estimated costs retrospectively21 22 or used modelling.23 Furthermore, participant follow-up rates were high in the RCT. This study fulfils the criteria set out by Drummond et al as being important for cost-effectiveness studies.24 For example, only primary outcomes found to be statistically significant in the RCT were included in the calculation of cost–effectiveness ratios. Sources of economic information were stated, and quantities, as well as unit costs, were reported. Appropriate adjustment for inflation was carried out, and currencies were converted to equivalent values. Unit costs were applied in a very precise manner; for some variables this was as accurate as quarterly time periods. Productivity changes were included in the analysis, incremental costs were used, and a sensitivity analysis was undertaken.
Limitations of this study are that the enhanced Green Prescription was compared with usual care, not with the standard Green Prescription, so we cannot be certain that it is more cost-effective. Although results from the two analyses suggest that it is more cost-effective, other factors may have varied between the two trials that may have influenced effectiveness and cost-effectiveness. Furthermore, secondary care costs were based on public hospital costs, and those who had procedures completed in private hospitals were valued at the public rate. Furthermore, in the absence of a health price index, the CPI was used to adjust for inflation for costs from previous years. The general inflation rate may differ from the inflation rate experienced in the health sector. Physical activity rather than health outcomes were used in this cost-effectiveness analysis. It is also difficult to assign quality adjusted life years or a monetary value to increases in physical activity for a cost–utility or a cost–benefit analysis. Despite this, the cost-effectiveness analysis does allow a comparison with other physical activity interventions that have calculated cost–effectiveness ratios for physical activity outcomes. The study was not powered to test whether indirect costs were significantly different between the groups over the 2-year follow-up. However, the cost ratios were very close to 1, so there is unlikely to be a difference. Including indirect costs in the calculation of cost–effectiveness ratios did not affect results markedly. While the incremental proportion of participants who became and remained active was quite small (13% at 12 months and 6.7% at 24 months), it is still a cost-effective intervention, as the cost of delivery is small.
Together with results from previous studies, the results of this cost-effectiveness analysis suggest that brief interventions (particularly those that involve exercise on prescription), brief face-to-face advice from a health professional and telephone follow-up or print-based support, can shift activity levels on a population basis. These shifts, while small, can have an impact on health outcomes and compare favourably with existing funded pharmaceutical therapies for cost-effectiveness.
Modifications to the original Green Prescription programme, including delivery of physical activity advice by primary care nurses, increasing telephone support and adding a brief face-to-face follow-up, were associated with improved cost-effectiveness of the Women's Lifestyle Study enhanced Green Prescription. Given the potential health benefit of population gains in physical activity, this cost-effective programme could be set up elsewhere. The reach of such an intervention would also increase if it was implemented systematically in everyday practice, using routine screening, delivery and follow-up.
The authors wish to thank Sport Wellington, particularly C Gordon for supplying information for the study, and Capital and Coast District Health Board, particularly R Rippon for supplying data on hospital costs. The authors also acknowledge the General Practices who participated in the RCT and supplied costing information. The authors would also like to thank E Robinson for biostatistical advice. Last, the authors thank all the participants who gave their time to participate in this study, and the team of research nurses for their role in data collection.
Funding The cost-effectiveness analysis was funded by Sport and Recreation New Zealand (SPARC). The randomised controlled trial was funded by National Heart Foundation of New Zealand (grant 1091 and grant-in-aid 1091 and 1222), the Lottery Health Research Grants Board and the Maori Health Directorate (Ministry of Health).
Competing interests None.
Ethical approval This work was approved by the Central Region Ethics Committee 2004 (WGT/04/08/061).
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.