Main findings
In this population, the costs of cardiovascular prevention were higher in the intervention group, with annual costs per individual of €160 (control group) compared with €335 (intervention group). Costs per percent decrease in estimated 10-year cardiovascular mortality of €98 compared with €187, for the control and intervention group respectively. An added role for self-monitoring can be considered only for females and higher educated individuals. For both groups costs predominantly consisted of societal costs and staff time and not of medication.
Explanation and comparison with existing literature
The present study adds valuable new information compared with previous studies, as both costs and effects are based on an actual practice setting, which makes the outcomes more generalizable. In addition, societal costs were included, which is recommended to allow a broad perspective [
21]. Lost productivity due to practice visits caused the majority of costs and is usually not taken into account.
The time investment of both medical staff and participants not only caused the main cost driver in both groups, but also the difference between both groups (Table
1). The number of visits of intervention group participants was almost twice the number of the control group participants and also the duration per visit was slightly longer in the intervention group [
16]. All intervention group participants were offered lifestyle counselling and home collected self-monitoring results were discussed during the visits. The productivity cost estimates may have been on the high side with 30 minutes transportation time, and with some working individuals probably having part-time jobs with planned visits during spare time. Of minor influence were the increased costs of self-monitoring equipment and medication adjustments in the intervention group. Despite self-monitoring not being part of the control group treatment, some self-monitoring costs were made due to BP measurements (on participant’s initiative) at home.
Based on De Bekker-Grob et al. with €39 million spend on non-pharmaceutical cardiovascular preventive activities and €181 million on medication, we expected that costs of modified medication would be relatively high [
3]. In the present study however, these had only minor influence and consisted in both groups mainly of statins and thiazides. Medication costs in the SPRING study may be even overestimated, as we calculated costs as if all adjusted medication was prescribed for the whole study period whereas mostly participants were advised first to adjust their lifestyles. De Bekker-Grob et al. found a large difference in medication prescription between different GP’s. We did not study the differences in prescriptions between practices, but this probably is of minor influence in this study because medication adjustments were advised by the study protocol. As mentioned before, Kok et al. estimated for the Dutch situation annual costs as €293 for statin use and €258 for antihypertensives [
4]. Compared with the Kok et al. the costs in the SPRING-RCT appear to be much lower, despite in the SPRING-RCT societal costs were taken into account. The goal of Kok et al. was not to estimate annual costs however, it was one step in estimating the cost-effectiveness of a new guideline. It is hard to make exact comparisons with other studies due to differences in programs, perspective and whether statins had already run out of patent (which reduces costs significantly) or not.
With regard to exploration of the subgroups, the control group CER is lower for all subgroups. Higher educated participants and women seem to benefit most from the investment of extra time and immediate feedback and motivation from self-monitoring, as the ICER is most favourable for these two groups. Whether the intervention programme is preferable over control treatment for these groups, depends on how much a decision maker is willing to pay for a certain decrease in SCORE risk estimation. However, confidence intervals were very wide and not statistically significant.
On the other hand, especially lower educated participants seemed to be better off in the control group. A probable explanation is that for some lower educated individuals the instructions and feedback of the self-monitoring might have been too complex and might have had a discouraging effect. Higher social economic status is inversely related to cardiovascular risk [
22‐
27]. Some investigators suggest that screening and treating high risk individual patients might augment socio-economic health differences, compared with whole-population approaches [
28]. Self-monitoring probably enhances these differences. Individuals from a higher socio-economic background who are motivated for using self-monitoring might be asked to pay a contribution. During the SPRING-RCT, participants were offered self-monitoring free of charge, but self-monitoring devices are usually not reimbursed by health insurance companies in the Netherlands.
With respect to sex, the awareness of both the public and physicians is poor about the fact that -despite women having lower 10-year risk estimations compared with men- the annual death rate for CVDs is higher among females compared with males, due to a higher case fatality rate from coronary attacks [
1,
29]. Reduction of risk factors is also effective for cardiovascular risk in women [
30]. Our study indicates that self-monitoring may improve cardiovascular risk management in women.
There is controversy about cardiovascular risk management for specific age groups [
4,
6]. Both these studies estimated long term effects and our study only evaluates effects after one year. For this study, the intervention group CER is higher for participants aged 65 years and older, compared with younger participants, despite the fact that productivity losses will be present mostly in younger participants. Incremental cost-effectiveness ratio is more favourable for participants aged 65 years and older.
Strengths and limitations
Strengths of this study are the pragmatic protocol and the societal perspective. No information from the SPRING study was available on long term effects >1 year nor on adverse effects. No modelling was performed to estimate for example quality adjusted life years, which makes comparisons with some other studies difficult. On the other hand the lack of assumptions necessary for modelling makes the results more plain to interpret. The group size was too limited to allow reliable cost effectiveness analyses in subgroups, so conclusions about subgroups should be considered as preliminary.