Skip to main content

Table 2 Authors’ conclusions and reviewers’ comments on the approach to model-based analysis of physical activity MMCs

From: Economic evaluation of physical activity mass media campaigns across the globe: a systematic review

Author (year), Campaign name Authors’ conclusions about cost-effectiveness of the mass media campaign Comments
Roux (2008),Wheeling Walks [34] The intervention appeared to reduce disease incidence, to be cost-effective, and—compared with other well-accepted preventive strategies—to offer good value for money. The results support using the evaluated intervention as part of public health efforts to promote physical activity. Effectiveness measure appropriate: No. Effectiveness study targeted older adults (50–65 years) from a small community in West Virginia, effectiveness is being generalised to adults 25 to 64 years.
Attenuation of PA impact: The impact of the intervention was assumed to decline after the intervention had ended and a 50% decline in PA in year 2 was modelled. Estimate seems to be optimistic, and no evidence was given to support this assumption.
Intermediate measures reported: No but reported sensitivity analysis for time-horizon from 40 to 30, 20 or 10 years.
Uncertainty explored: Yes but results for sensitivity analysis exploring uncertainty in utility values, costs and effect of PA on health NR. Sensitivity analysis varying the dissipation of the effect size of the interventions had a marginal impact on results, but the details of the parameters used, and results NR.
Other: Community wide program with strong mass media campaign support. Additional components included: primary care initiatives, community programs, cross sectoral initiatives, website resources, worksites.
Cobiac (2009), Exercise, you only have to take it regularly not seriously [30] Physical activity promotion intervention is recommended as a public health measure. Despite substantial variability in the quantity and quality of evidence on intervention effectiveness, and uncertainty about the long-term sustainability of behavioural changes, it is highly likely that as a package, all six interventions could lead to substantial
improvement in population health at a cost saving to the health sector.
Effectiveness measure appropriate: Yes
Attenuation of PA impact: Base case analysis assumed that the intervention effects on PA are sustained for year 1, but decay exponentially (rate of 50% per annum) thereafter. Optimistic assumption that intervention effects would last for one year considering this was a 6-week one-off intervention. Assumption was not explored in a sensitivity analysis.
Intermediate measures reported: No
Uncertainty explored: Yes, but only for attenuation of PA impact (0 to 100% decay). Results were robust, intervention remained dominant. Sensitivity analysis exploring uncertainty in utility values, costs, and effect of PA on health not performed.
Other: Campaign was mostly a mass media campaign with some community support.
De Smedt (2011), 10,000 Steps Ghent [31] The community-based ‘10,000 Steps Ghent’ campaign is a dominant intervention. Sensitivity analyses have proved the robustness of the results; hence implementing this intervention on a population-based level could lead to improved health outcomes and reduced costs. Effectiveness measure appropriate: Yes
Attenuation of PA impact: Base case model assumes life-long programme with a life-long intervention effect. 5-year cycle used for the intervention where the intervention had to be implemented each year. Pedometers were implemented only in year 1 and the effects were expected to last for 5 years. In years 2-5 only the MM components were implemented. This cycle was repeated for the duration of the model. Effectiveness data used was from a 1-year intervention. It is unclear if the same level of effectiveness would sustain in years 2 to 5 where only the MM components were used. Possible increment over time with message reinforcement following a MMC not explored (as there were 6 segments of the campaign). Sensitivity analyses performed for 1-year intervention effects still showed favourable results, however there was a substantial decrease in QALY gain and cost savings.
Intermediate measures reported: scenario analysis presented for 1-year, 5-year and 20-year intervention effects.
Uncertainty explored: Yes, around intervention effects, utility, risk reduction, intervention costs and colon cancer costs. Uncertainty around risk reductions had the biggest impact on the results, but the intervention remained cost-saving. Other health care costs not explored. Probabilistic analysis showed 100% of the intervention being cost-saving and more effective.
Other: 1) QALY effect was small: average QALY gained: 0.16 (men) and 0.11 (women) at 20 years; 2) Unclear how intervention costs were calculated. Community walking intervention with some MM promotion which needs to be considered when interpreting the results.
Goryakin (2019), NR (Hypothetical campaign) [32] Investing in policies to promote active lifestyles is a good investment for Italy Effectiveness measure appropriate: Unclear. Used median effectiveness of 12 previous campaigns identified in a review conducted by the authors. However, these previous studies investigated different types of campaigns and various durations.
Attenuation of PA impact: Stated that model takes this into account but details not provided. 6 segments of three years each in 31 years. Effectiveness estimate changed over time (60% increase after 1 month, drop to 30% after 1 year, drop to 0 after 2 more years). Unclear how the 35.55% effect at the end of the intervention was used to calculate the attenuation. Median intervention duration of the 12 studies (18 wks) used by authors to calculate the effectiveness estimate. Therefore, the drop to 30% after 1 year might be optimistic. Authors did not consider or explore the possibility of effect increment over time with message reinforcement (as there were 6 segments of the campaign). For the less inactive individuals, probability of moving into the new PA category is higher. However, the opposite would be expected (i.e. the less active have a higher probability of moving into a new PA category).
Intermediate measures reported: ICER across the years are presented in a figure. Estimates are available from 2025, 2037 and 2050 (provided by the authors upon request from review authors).
Uncertainty explored: methods for dealing with uncertainty stated in the report describing the model, however no results presented. Therefore, it is unclear if the results are robust and how much each parameter would impact on the results.
Other: Study found a significant impact of the intervention on the burden of disease; however the intervention effect was small. Intervention cost items and details NR. Since this is a hypothetical MMC the extent to which other components contributed to the effectiveness estimate is unclear. Amongst the 12 studies used to estimate the effectiveness there was a range of types of MMCs, including stand-alone campaigns as well and community and cross-government programs. The lack of clarity of the content of the intervention impacts the interpretation of results.
Mizdrak (2020), NR (Hypothetical campaign to promote apps for physical activity) [33] A mass media campaign to promote smartphone apps for physical activity is unlikely to generate much health gain or be cost-effective at the population level. Other investments to promote physical activity, particularly those that result in sustained behaviour change, are likely to have greater health impacts. Effectiveness measure appropriate: No. Evidence for campaign effect on PA levels derived from a systematic review where the meta-analysis included studies investigating an app or wearable device with additional components (e.g. counselling, goal setting, education). Therefore, it is not possible to know the effect of app/pedometer use as a standalone intervention. Additionally, the studies included in the review were mostly small and conducted in controlled environments. The effectiveness measure used was 285 MVPA-MET min/wk.(equivalent to 1.6 hours of additional brisk walking per week). It is an optimistic assumption that such gains would result from using an app without any additional support.
Attenuation of PA impact: Considered change to app adherence over 1-year but did not consider reduction of PA effect over 1-year. The median intervention duration of the studies included in the meta-analysis used to estimate effectiveness was 8 weeks, therefore it is unclear if the effects would sustain for a year.
Intermediate measures reported: No
Uncertainty explored: Yes. Sensitivity analysis explored intervention parameters, sustainability of the intervention, narrowing target population and discount rate. Overall the analysis reviewed big uncertainty in the results. No sensitivity analysis was performed to explore the uncertainty of utility values, health costs, effect of PA on health.
Other: Source for utility values and how it was included in the model NR. Very small improvement in QALY.
  1. DALY Disability-adjusted life year, ICER Incremental cost-effectiveness ratio, MM mass media, MMC Mass media campaigns, PA Physical activity, QALY Quality-adjusted life year, MVPA-MET min/week Moderate to vigorous Physical Activity-Metabolic equivalent per minute per week, NR not reported