On time horizons in health economic evaluations

The issue of time horizons has received scant attention in discussions pertaining to health economic evaluations unlike discounting or translation of health outcomes into life-cycle measures (e.g. quality-adjusted life years or disability-adjusted life years). The available guidelines do not offer clear and consistent guidance for many problems addressed in health economic evaluations. In practice, variation of time horizons between studies for the same diseases is a matter of concern, as results on cost-effectiveness depend on the time horizon. Our paper contributes to establishing a consistent approach to setting time horizons across common types of health economic evaluations and mitigating potential bias where the choice of a time horizon may affect results of the evaluation. We find that available guidance is clear only for patient-focused interventions, but not in the presence of population-level effects owing to transmission of infections or other linkages. We distinguish between a policy period-over which an intervention is delivered or initiated-and an evaluation period over which the effects are measured. One important challenge in establishing a time horizon for evaluation is that, at least for infectious diseases, the state of the epidemic at the end of the policy period cannot be evaluated precisely and incorporated in the results of an economic evaluation. While longer policy periods partly mitigate this challenge, they are subject to greater uncertainty, and outcomes may not adequately reflect the cost-effectiveness of current policies because outcomes reflect an average over the policy period. Incremental analysis on interventions implemented in sub-periods of the policy period (especially at the beginning) potentially improves accuracy and helps to identify potential for improving cost-effectiveness by varying the path of implementation or the mix of interventions offered over time.