BACKGROUND

It is well known that physical activity performed on a regular basis is associated with significant positive effects on health and mortality. It plays an important role in the prevention of various chronic diseases, such as cardiovascular diseases, ischaemic stroke, hypertension, obesity, diabetes mellitus, osteoporosis, colon cancers and fall-related injuries.1^–10 Although physical activity-related health benefits seem to increase with amount and intensity of physical activity levels, moderate levels have also proved to be beneficial.

Despite these well-known benefits of regular physical activity, sedentary or inactive lifestyle remains a common problem in industrialised countries and the overall levels of physical activity continue to be low or have even declined in recent years.11^–13 In addition, it has been shown that physical inactivity and overnutrition are associated with a substantial economic burden in most industrialised countries.11^–14 Promotion of physical activity behaviour has therefore become an important objective for the promotion of health and crucial in the prevention of the increasing burden of chronic diseases. Although current intervention strategies to promote physical activity have provided evidence that they can effectively increase physical activity in the long term, they need to be comprehensive and thoroughly conducted to achieve these effects.15^–17 Implementation of such strategies will consequently be associated with considerable costs. To address large population groups and result in meaningful public health benefits, the cost-effectiveness of intervention strategies will therefore be of major importance in addition to their clinical effectiveness.

The aim of the present systematic review was therefore to summarise the current evidence regarding the cost-effectiveness of physical activity interventions and to identify cost-effective intervention strategies for application to a wide population group.

METHODS

To identify all relevant current publications a structured search from August 2001 to June 2008 of the following databases was conducted: Cochrane Library, Medline, Embase and PsycInfo (table 1). In order to identify cost-effectiveness analyses of physical activity promotion interventions published before August 2001, Medline was searched using the original search strategy combined with the Mesh terms “costs” OR “cost and cost-analysis” OR “economics” (table 1).

We also conducted a manual search using reference lists of existing systematic reviews and identified literature. Titles and abstracts of identified publications were screened for relevance by one researcher without restrictions to study design. Two researchers independently assessed full-text articles of included studies with regard to relevance and methodological quality. Studies were included if they met the following criteria.

Inclusion criteria:

• study design: any primary study

• outcome measures: cost per increase in physical activity (physical fitness, energy expenditure, participant becoming more active)

• interventions: exercise programmes or interventions aimed to promote and maintain physical activity

• duration of the intervention/follow-up: no restrictions

• target population: healthy adults >18 years of age

• publication language: English or German

Exclusion criteria:

• physical activity interventions targeting populations with diagnosed diseases or selected on the basis of risk factor clustering

• cost-utility and cost-benefit analyses

One researcher extracted study characteristics and results of included studies. Abstracted data was cross-checked by a second researcher. The methodological quality of included studies was evaluated using standardised quality checklists developed by the German Institute for Documentation and Information (DIMDI). These checklists consider the following aspects: study objective, selection process of intervention and control participants, allocation method, comparability of (intervention) groups, consideration of relevant confounders, blinding (single/double) in case of RCTs, comparability of circumstances beyond the intervention, validity of outcome assessment, attrition rate/differential attrition, description of statistical methods, and intention to treat analysis. Depending on their risk of bias studies were then rated according to the grading system of the Scottish Intercollegiate Guidelines Network Group (SIGN) as high (minimal risk of bias), good (moderate risk of bias) and fair (high risk of bias) methodological quality.18 Disagreements related to study selection, quality rating and data extraction were resolved in discussion between both reviewers.

To compare the economic results of individual studies we converted all costs into Euros, using annual average exchange rates, referring to the year of cost estimation in individual studies.19 To improve comparability between cost-effectiveness results of interventions we calculated the annual costs per participant to become active at currently recommended levels (i.e. at least 30 min 5 times per week). This calculation was performed for studies with established intervention effectiveness. Interventions were than ranked according to their cost-effectiveness.

RESULTS

Our literature search identified 6543 references and we assessed full-text articles of 166 documents. Of those only eight studies met the selection criteria and were included in this review (fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Flow-chart of study selection.

The main reasons for exclusion from this review were:

• not reporting costs

• outcome not reported as cost per increase in physical activity

• participants selected on the basis of underlying disease or risk factors

Eight included studies reported the cost-effectiveness of 11 intervention strategies to promote physical activity behaviour. The majority of studies compared an innovative intervention strategy with a no-intervention control. One study only compared two different intervention strategies20 and one further study did not include a control group.21 While Wang et al21 reported the cost-effectiveness of an environmental intervention to promote physical activity, Sims et al22 investigated the cost-effectiveness of an intervention targeted at general practitioners to provide physical activity-related advice. The six remaining studies investigated the cost-effectiveness of individually targeted behavioural intervention strategies. Out of eight included cost-effectiveness analyses, six were conducted within the context of a randomised controlled trial, and one analysis used cross-sectional data. The remaining study by Sims et al22 performed an economic modelling based on results of a randomised controlled trial and therefore had to estimate the possible uptake of physical activity among primary care patients. Overall, two studies were rated to be of high,23 24 three of good20 25 26 and three of fair methodological quality.21 22 27

Most identified studies reported that the investigated intervention was effective in increasing physical activity behaviour. However, the reliability of these estimates of intervention effectiveness was frequently limited due to inadequate study design or study methodology. Among five studies rated with high or good methodological quality, only three studies provided sufficient evidence for the effectiveness of four different intervention strategies to increase physical activity behaviour compared with a no-intervention control. They reported an increase in the proportion of participants meeting physical activity targets of up to 10%23 25 or an increase of up to 80 minutes/week of moderate-intensity physical activity compared with the control group.24 In these studies, costs per participant becoming active at least 30 minutes on five days a week ranged between 773 and 5358 Euros. The two remaining studies judged to be of good methodological quality were not able to provide consistent evidence of intervention effectiveness.20 26 Studies reporting most favourable cost-effectiveness ratios did not attempt to target individual behaviour change, however. Sims et al22 reported costs of about 106 Euros per patient to become sufficiently active in an intervention targeted at GPs. Similarly, Wang et al21 calculated the cost-effectiveness of the only environmental intervention strategy to promote physical activity behaviour. They reported costs of about 126 Euros per participant becoming active for their health.21 Both estimates are compromised by methodological limitations, however, and respective studies could only be rated with fair methodological quality.

Lowest intervention costs per participant and month of 5.4 Euros were estimated for the environmental intervention by Wang et al.21 In studies with high or good methodological quality monthly intervention costs per participant were reported to range between 6.7 Euros in the green prescription intervention reported by Elley et al23 and 64 Euros in the phone-based intervention presented by Sevick et al.24 Due to an unclear time horizon it was not possible to estimate monthly costs in the study by Sims et al.22 Details of all included studies and results of respective cost-effectiveness analyses are presented in table 2 and fig 2.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Cost per participant and 12 months to achieve physical activity levels necessary to result in health benefits.

DISCUSSION

In sum, identified studies provide evidence that current interventions to promote physical activity can result in meaningful increases of physical activity at reasonable costs. Intervention strategies targeting individual behaviour directly were able to promote currently recommended levels of physical activity at a cost of about 800 Euros per participant23 24 over a 12 month period. Interventions targeted at GPs and environmental interventions seem to have the potential to achieve increases in physical activity at considerably lower costs.

Saying this, it should be recognised that the only study investigating environmental changes was retrospective and did not follow a control group design. Similarly, Sims et al22 performed economic modelling with rather optimistic assumptions regarding the difference in uptake of physical activity between patients of intervention and control group GPs. Confirmation of their findings in more rigorous studies will therefore be necessary in order to draw reliable conclusions with regard to their cost-effectiveness.

Among studies directly targeting individual behaviour change, two recent high-quality studies were particularly effective in increasing physical activity behaviour.23 24 Both were similarly cost-effective at monthly costs of 6.7 Euros and 32 Euros per participant.23 24 While Elley et al23 tested the effectiveness and cost-effectiveness of a primary care-based counselling and exercise prescription intervention, Sevick et al24 investigated phone and print-delivered individualised intervention messages and found superior cost-effectiveness for print-delivered messages. Compared with no intervention control, both strategies were reported to be highly cost-effective. Two previously tested interventions, a study investigating primary care-based counselling by Stevens et al25 and one comparing centre-based interventions by Sevick et al,20 reported less favourable cost-effectiveness estimates to achieve currently recommended levels of physical activity than were found for the print-based and green prescription interventions by Sevick et al24 and Elley et al,23 respectively.

Considering the broad health effects and long-term benefits of regular physical activity at currently recommended levels, presented costs are likely to be substantially lower than many alternative uses of healthcare resources. With regard to the green prescription intervention by Elley et al,23 Dalziel et al28 estimated in a cost-utility analysis based on the same data that costs per QALY gained would be approximately 10 000 Euros over the whole life expectancy, which could be considered highly cost-effective in most industrialised countries.

Due to the limited number of cost-effectiveness analyses, identification of most appropriate intervention strategies remains limited. There is evidence that exercise on prescription, in addition to other established strategies to promote physical activity, can result in long-term increases in physical activity behaviour.15 16 29 30 In addition to its efficacy, presented findings by Elley et al suggest that it also reflects a cost-effective approach.23 Exercise on prescription interventions, which additionally included exercise testing prior to prescription, were also found to be effective in previous studies.16 29 These would be associated with additional costs, however. Whether these more resource-intensive approaches would still be cost-effective can currently not be determined.

Previous studies have indicated that interventions including regular reminders or booster strategies seem to be associated with improved long-term increases in physical activity.30^–33 The mode of message delivery is controversial, however. Phone and print-based reminder contacts were frequently used in the past, but none was consistently reported to be associated with superior effectiveness. Now, Sevick et al24 reported that delivery of print messages is not only more effective in the long term, but also considerably lower in costs and thus considered highly cost-effective compared with phone delivery of intervention messages. Whether new information technologies can improve cost-effectiveness further has yet to be established. Marcus et al31 recently reported no difference in effectiveness between internet-based and print-based delivery of intervention messages. Due to potentially lower costs associated with the use of the internet, this could result in superior cost-effectiveness of internet-based strategies. However, firstly, cost-effectiveness outcomes have not been reported; secondly, effectiveness of both interventions was markedly lower than that of the presented study by Sevick et al.24 Further evidence is therefore required to answer this question.

In an earlier study Sevick et al20 compared the cost-effectiveness of centre-based behavioural training and centre-based supervised exercise sessions. They reported that the cost-effectiveness of supervised exercise classes was less favourable compared with behavioural interventions. However, cost-effectiveness estimates were based on a comparison with baseline physical activity levels. Because there is currently no consistent evidence that supervised or centre-based physical activity interventions can result in long-term increases in physical activity behaviour compared with usual care, this warrants further investigation.

When interpreting the positive cost-effectiveness of presented physical activity interventions it should be noted that most identified studies also reported positive intervention effects. In contrast to this finding, it has been shown in previous systematic reviews that the effectiveness of physical activity interventions is frequently limited.30 34 It is therefore possible that presented findings represent a selected set of studies which provided at least some evidence of intervention effectiveness prior to conduction of cost-effectiveness analysis. This could result in an overestimation of the true cost-effectiveness of physical activity interventions.

Some limitations of this systematic review should further be noted. Firstly, the results of this systematic review are limited by the scarcity of reliable cost-effectiveness analyses. It is likely that this is not the consequence of the presented literature search but reflects the current lack of appropriate studies. Secondly, comparability of included studies is substantially limited due to different study designs, participants and duration of follow-up. In addition, some economic aspects should be noted. A large number of factors can limit the transferability of cost-effectiveness results obtained in country-specific studies to another nation’s context. Examples include differing systems of physician reimbursement and related incentive schemes, as well as differences in pricing. Further, the perspective of the economic analysis could lead to different findings regarding the cost-effectiveness of an intervention, for example, whether an analysis conducted from society’s perspective considered all indirect costs or only those of specific healthcare participants (e.g. a third party payer). Even if an analysis was performed from the perspective of interest, transferability to other countries can be problematic. For example, the economic evaluation results of an intervention from the perspective of a social insurance system will be influenced by the structure of this system. Such a social insurance system could include healthcare insurance, pension insurance, nursing insurance etc. — or just part of these services.

Cost-effectiveness results could also be affected by the choice of outcome measure. In our review, we tried to calculate the intervention costs for one extra participant becoming sufficiently active. Comparability of results is thereby improved. However, most included studies evaluated the effectiveness using a range of different outcome measures, frequently reporting variable intervention effects between these measures. It is thus possible that use of other outcome measures would lead to different cost-effectiveness appraisals or changes in our cost-effectiveness ranking.

In recognition of these limitations and the obvious lack of cost-effectiveness studies, it can be concluded that there is currently no sufficient evidence to reliably compare cost-effectiveness results of individual studies and that the generalisability of presented findings is restricted considerably.

Results of identified studies provided evidence, however, that current behavioural physical activity interventions can increase physical activity in healthy sedentary adults at reasonable costs. In particular, well-designed interventions within high-quality studies were able to do so. Over a 12 month period, they were able to achieve increases in physical activity meeting current recommendations for as little as 800 Euros per participant. Presented findings further indicate that strategies targeting physicians or environmental changes could possibly be associated with superior cost-effectiveness compared with individual behavioural interventions. Methodologically rigorous studies will be essential, however, to confirm these findings. In addition, it needs to be emphasised that, despite the growing literature on physical activity interventions, the availability of appropriate cost-effectiveness analyses is currently limited considerably. High-quality economic evaluations are therefore warranted to evaluate the cost-effectiveness of behavioural and environmental physical activity interventions further.