Protocol for evaluating the cost-effectiveness of Mongolia's sugar-sweetened beverages tax using double machine learning.

Abstract

Elevated consumption of sugar-sweetened beverages (SSBs) has been associated with an increase in obesity, type 2 diabetes, and other non-communicable diseases (NCDs), a significant health and economic burden on Mongolia. To address this, the government has introduced a 20% SSB tax set to take effect in 2027. This study conducts a Cost-Effectiveness Analysis (CEA) using a Markov cohort model, incorporating Double Machine Learning (DML) to estimate price elasticity and assess policy-driven consumption changes while addressing potential confounding. The analysis integrates DML-estimated price elasticity and consumption shifts with disease transition probabilities, simulating outcomes for the 2023 Mongolian population, aged over 15 years old, over two time horizons of 20 years and a lifetime. The model estimates changes in obesity prevalence, healthcare costs, and disease burden, translating them into Disability-Adjusted Life Years (DALYs) averted, and Quality-Adjusted Life Years (QALYs) gained. Tax revenue projections and sensitivity analyses further assess the robustness of assumptions. By combining machine learning-based causal inference with economic modelling, this study provides policy-relevant evidence on the cost-effectiveness of SSB taxation, supporting data-driven decision-making for public health strategies in Mongolia, highlighting the tax's potential to reduce the burden of NCDs and promote healthier behaviours.