This analysis reports that there is heterogeneity in the cost-effectiveness over the metabolic syndrome risk groups, due to the heterogeneity in patient characteristics, but this should not affect the overall conclusion on the cost-effectiveness of the KMSP. Cost-savings in the societal perspective was estimated for middle and high risk men, while the incremental cost vs do-nothing varied between EUR3,500 and 18,000 per QALY for other groups. Even the highest ICER (for high risk women) is moderately cost-effective in Sweden. In a healthcare perspective, no groups are estimated cost-saving, but the costs per QALY are low, below EUR20,000, for all groups.
Patients’ risk profile (low, middle, high) was defined as the number of metabolic syndrome risk factors (less than two, two, more than two). This graduation is very simple and does not require special medical investigations. This means that patients for whom treatment is most cost effective would be easily identified in routine clinical practice, which is an important consideration . This is one of the most important results of this study. The risk stratification sensitivity analyses show that the risk factor grouping is appropriate, as the analyses using the individual patient characteristics gave very similar results. No risk stratification, i.e. using average risk factor levels (analysis 1), would tend to overestimate the cost-effectiveness. This has also been asserted by UK NICE; “Average cost-effectiveness can, therefore, mask important sources of heterogeneity which may be important to reflect in decision making…” (, p. 3). Furthermore, medical practitioners in standard primary care are fully aware of the fact that patients exhibit marked heterogeneity in risk factor levels, which could affect treatment results and thus cost-effectiveness. The aspect is increasingly discussed in terms of personalized medicine  or stratified medicine .
Many studies have shown that there are considerable differences in ICERs over patient groups, in such diverse areas as children otitis media , tuberculosis screening , assertive community treatment for homeless people  and phosphate binders in end-stage renal disease . The Swedish health economic evaluation recommendations state that “The purpose of a health economic evaluation is to identify for which patient groups or indications a drug is cost-effective –it is never the medication itself which is cost-effective, rather the use of it” ( , p. 1). Nevertheless, the frequent use of subgroup analyses in clinical trials has been criticized . There are also ethical issues in subgroup analyses, in particular if socio-demographic characteristics such as age, sex, race and social class, determine which patient groups will be offered treatment and if these characteristics are not directly related to the treatment effect .
One of the aims of the study is to seek to reflect Swedish primary care conditions, to enable decisions based on cost-effectiveness analyses in a Swedish ordinary praxis context. A formal trial design was not chosen, because the study sought to evaluate the program in ordinary primary care practice with the patient groups normally found there. The study design with a before-after comparison requires the assumption that all changes of lifestyles and ensuing changes in risk factors depend on KMSP. Possible changes in medical treatment were eliminated by use of the inclusion criteria that the patients had no increase in drug treatment during the study period. Most of the patients had had some kind of medication during a long period, so the positive effects from these medications had already been achieved before the measurement at the start was done. Some patients even decreased their medication during the program period. Note that these decreases in pharmaceutical costs were not included in the analysis, as the comparator the do-nothing alternative assumed no changes in costs for the standard treatment. This implies that the cost-effectiveness of the program is underestimated. An indication of the effectiveness of the KMSP is that all groups exhibited positive changes in risk factor levels at 12 months, mostly statistically significant. The effectiveness might also be underestimated, as we do not consider the preventive effects of lifestyle changes of other health problems, such as cancer or chronic pain, nor the short-term increases in quality-of-life during the project period included in the previous cost-effectiveness analysis of the project .
The result of the cost-effectiveness analysis seems to be in line with previous similar studies. The average costs per individual in the cost-effectiveness analysis of the Finnish DPS applied to a Swedish cohort  is around EUR26,000, which is comparable with the estimates for high risk women in our study. That study, however, only models cardiovascular disease. The US DPP study [5, 31] reports average patient lifetime medical costs of around EUR50,000 (assuming year 2012 currency exchange rate), i.e. somewhat higher than our study estimates. Yet another comparison might be with the model simulations on US patients that fulfill the IDF criteria for metabolic syndrome , that resulted in 10 year healthcare costs of EUR37,000 for stroke patients and EUR18,000 for diabetes patients (assuming year 2012 exchange rate). These 10-year estimates seem more in line with our results.
Surprisingly, all risk factor groups were able to benefit from the lifestyle intervention. One might expect that lifestyle changes are more easily achieved among individuals that already lead comparatively healthy lives. However, all groups exhibit changes in risk factors, regardless of base line level. The risk factors that are most clearly related to changes in diet and physical activity, the weight and obesity parameters BMI and waist circumference, changed significantly in all risk groups. This indicates that even patients in the high risk group, with very high levels on risk factors, can change lifestyles if supported by an appropriate primary care program.