Presentation (CTI)

OBJECTIVES: Health economic models typically adopt a lifetime timehorizon horizon, requiring estimation of survival beyond the end of clinical trial follow-up. Traditionally, cohort-based models incorporate background mortality using life-table rates for the average age of the patient population at baseline. Lee et al. (2024) proposed an alternative method that accounts for the distribution of patient ages when estimating background mortality. In this study, we explore the impact of the method chosen to estimate background mortality.
METHODS: We generated hypothetical clinical trial populations varying by average age and by age heterogeneity. For each scenario, lifetime mortality was estimated using published life tables under two approaches: 1) applying rates based on the population’s average age, and 2) applying rates across the full age distribution. Mean life-years (LYs) were calculated under both approaches. A lifetime horizon was defined as modeling all patients to a maximum age of 100 years. The primary outcome was the difference in mean LYs between methods, assessed with and without a 3.5% annual discount rate.
RESULTS: Based on each combination of average age and heterogeneity, the difference in undiscounted mean LYs estimated using either method ranged from 0.47 in an older population with less heterogeneity, to 2.89 in a younger population with a broader age range. Discounting reduced the impact of the method, with the aforementioned range narrowing to 0.27-0.92.
CONCLUSIONS: The choice of method for estimating background mortality can meaningfully influence modeled survival, even in simple cohort models. Differences are most pronounced in populations with broader age heterogeneity, particularly when disease hazards tend towards background mortality rates and extrapolations continue significantly beyond the observed period. These findings highlight the importance of considering age distribution within health economic models where background mortality is expected to have a meaningful impact on results.