Modelled Distributional Cost-Effectiveness Analysis of Childhood Obesity Interventions: A Demonstration

Objective To demonstrate how distributional cost-effectiveness analyses of childhood obesity interventions could be conducted and presented for decision makers. Methods We conducted modelled distributional cost-effectiveness analyses of three obesity interventions in children: an infant sleep intervention (POI–Sleep), a combined infant sleep, food, activity and breastfeeding intervention (POI–Combo) and a clinician-led treatment for primary school-aged children with overweight and obesity (High Five for Kids). For each intervention, costs and socioeconomic position (SEP)-specific effect sizes were applied to an Australian child cohort (n = 4898). Using a purpose-built microsimulation model we simulated SEP-specific body mass index (BMI) trajectories, healthcare costs and quality-adjusted life years (QALYs) from age 4 to 17 years for control and intervention cohorts. We examined the distribution of each health outcome across SEP and determined the net health benefit and equity impact accounting for opportunity costs and uncertainty due to individual-level heterogeneity. Finally, we conducted scenario analyses to test the effect of assumptions about health system marginal productivity, the distribution of opportunity costs and SEP-specific effect sizes. The results of the primary analyses, uncertainty analyses and scenario analyses were presented on an efficiency–equity impact plane. Results Accounting for uncertainty, POI–Sleep and High Five for Kids were found to be ‘win–win’ interventions, with a 67% and 100% probability, respectively, of generating a net health benefit and positive equity impact compared with control. POI–Combo was found to be a ‘lose–lose’ intervention, with a 91% probability of producing a net health loss and a negative equity impact compared with control. Scenario analyses indicated that SEP-specific effect sizes were highly influential on equity impact estimates for POI–Combo and High Five for Kids, while health system marginal productivity and opportunity cost distribution assumptions primarily influenced the net health benefit and equity impact of POI–Combo. Conclusions These analyses demonstrated that distributional cost-effectiveness analyses using a fit-for-purpose model are appropriate for differentiating and communicating the efficiency and equity impacts of childhood obesity interventions.