OBJECTIVES: When assessing the cost-effectiveness of interventions most studies do not consider the rollout of the intervention and instead assume immediate implementation and 100% uptake. We evaluate the cost-effectiveness of an opt-out HIV/STI testing policy for different dynamic rollout scenarios.
METHODS: We used a previously published stochastic agent-based network simulation model of HIV, gonorrhea, and chlamydia transmission among young men who have sex with men (YMSM) in Chicago to study different rollouts of an opt-out HIV/STI testing policy targeting YMSM. We modelled the dynamic rollout of this policy. Specifically, we model the fraction of YMSM automatically receiving an HIV test when testing for STIs, using a Bass diffusion model where the timing of growth was varied such that full rollout was achieved at 2.5 (A), 3.5 (B) and 4.5 years (C) while the imitation rate was kept constant. Natural diffusion after complete rollout was not considered based on empirical evidence. Costs in 2015 dollars included treatment and testing cost for both HIV and STIs; the health outcome considered was HIV infections averted (HIA); and the base case assumed no implementation.
RESULTS: For scenarios A, B and C the ICER was at 5 years $29,207, $54,073 and $100,230; at 10 years $19,634, $24,698 and $64,842; and at 15 years $4,412, $15,921 and $55,046. Differences in discounted cost and HIA between scenarios with the fastest (A) and slowest rollout (C) were $783,307 and 19.3 HIA at year 5; $633,634 and 27.7 HIA at year 10; and $460,527 and 46.4 HIA at year 15.
CONCLUSIONS: Dragging out the rollout has increasing ICERs and should be included in future models because beyond a certain point such a policy may not be cost-effective. Differences in cost decrease with increasing time horizon but increase in HIA showing that benefits accrue over the long term.