In this study, we have made some assumptions that need to be confirmed by further research. We used meta-analysis based results of the effect of pedometer use on the number of additional steps per day. However, the potential effect of pedometer use in the Dutch population is debatable. For the pedometer scenario, we used an additional 2491 steps per day after one year [4
], but walking is already a substantial part of the daily physical activity pattern in the Netherlands [21
]. It therefore remains to be seen if pedometer use in the Netherlands would indeed result in an additional 2491 steps per day. This may have led to an underestimation of the ICER. On top of that, the randomized controlled trials in Bravata et al. were mainly targeted at sedentary adults. The cohort of adults in our study is mostly not sedentary, and thus we may have overestimated the effects, also leading to an underestimation of the ICER. On the other hand, the additional 2491 steps found by Bravata et al. was a pooled effect of interventions with and without counseling, whereas our pedometer intervention always included counseling. This could mean that we underestimated the effects, and overestimated the ICER. We also had to make an assumption about the long-term effect of pedometers, which we estimated to be 25% of the effect after one year, namely 623 steps or 6
minutes per day. The accuracy of this estimation is unknown. If the long-term effect would amount to only three or less minutes of walking per day, the pedometer intervention would not be cost-effective.
A limitation of this study is that physical activity in the RIVM CDM is modeled via discrete classes: inactive, insufficiently active, and sufficiently active. Obviously, the real distribution of physical activity over the Dutch population is continuous, and even a single step per day more may have a (may it be minuscule) positive effect on health outcomes. Using discrete classes for physical activity implies a somewhat crude calculation of health outcomes.
In this study, we have chosen the health care perspective, focusing on health gains and health care costs but ignoring broader societal costs and consequences (higher labor productivity, less absenteeism, less environmental pollution, non-medical consumption in life years gained) of an increase in physical activity that fall outside the scope of the health care budget. This may have led to an overestimation of the ICER.
Comparing our results with the results of two previous economic evaluations, both Cobiac et al. [6
] and De Smedt et al. [7
] found the cost-effectiveness of their pedometer intervention to be dominant, i.e. health gain can be achieved while saving costs compared to current practice. A possible explanation for the more favorable result of Cobiac et al. [6
] may be that the average proportion of inactive persons is much larger in the Australian than in the Dutch population (about 35% vs. 17%), which implies a higher possible health gain. De Smedt et al. [7
] used much lower costs per person compared to our program costs: The total program costs per person accumulated to EUR 3.51 in the first year and EUR 0.23 in the second until fifth year. This is probably the main reason that their ICER was more favorable. Another difference between our and their study is that 65% of their target population group consisted of highly educated people. As prevention interventions are often more effective for high SES than low SES [22
], the “10,000 Steps Ghent” program may have been more effective and more cost-effective than our pedometer intervention.
The cost-effectiveness of the pedometer intervention in the Netherlands in this study is less favorable than the cost-effectiveness of other types of physical activity interventions: an internet-based intervention program, a GP physical activity prescription program and a program to encourage more active transport all have a high probability of being cost-effective [6
], and brief interventions in primary care [24
], exercise referral [24
], and mass media-based community campaign [6
] have found to be even dominant. This difference is probably not due to the nature of the intervention, but rather because 56% of the Dutch population aged 20–65 already satisfies the Dutch guidelines for the healthiest level of physical activity.