OBJECTIVES: We demonstrate an approach to select evidence-based preventive treatment thresholds by optimizing preferred outcomes illustrated with a study on preventive statin treatment based on 10-year coronary heart disease (CHD) risk predicted by the Framingham risk score (FRS). METHODS: A Markov decision-analytic model was used to simulate cohorts following usual care; preventive statin treatment in high-risk (FRS≥20%) individuals (ATPIII guideline), or, alternatively, an explorative approach of lowering treatment threshold T from 20.0% to 0.0% with 0.5% decrements. A population-based cohort (n=11,649) was used to recalibrate the FRS and calculate the distribution of individuals over the low (<0.5T%), intermediate (0.5T%-T%), and high (≥T%) risk category and corresponding observed CHD risks. Treatment complications, quality-adjusted life-years (QALYs), and Net Health Benefit (NHB) (willingness-to-pay of $50,000/QALY) were evaluated. Uncertainty was assessed through probabilistic sensitivity analysis. Furthermore, we assessed the balance between additional individuals treated and additional health benefits when incrementally lowering the threshold.

RESULTS: Over a 30-year time horizon, QALYs in men ranged from 12.679 at T=20.0% to 12.752 at T=0.0%, with a maximum of 12.753 at T=1.5%. For women QALYs ranged from 13.474 at T=20.0% to 13.587 at T=0.0%, with a maximum of 13.589 at T=1.5%. Lowering the threshold monotonically increased costs, whereas the incremental NHB was favorable for every T<20%. Incrementally lowering the threshold and comparing outcomes to the former threshold, for men and women marginal health effects achieved a maximum at T=10.0%, whereas marginal costs were highest at T=1.0% for men and T=2.5% for women. The marginal NHB was favorable down to T=2.0% for men and T=3.5% for women. CONCLUSIONS: Risk-stratified prevention is increasingly recommended, while current intuition-based treatment threshold selection leaves ample room for health gain and cost-savings. Evidence-based selection, including estimation of long-term (marginal) health effects and costs, is essential, whether the goal is to maximize health outcomes or optimize cost-effectiveness.