Small area estimates of co-occurring health risk factors: A test of two synthetic population methodologies to estimate small-area prevalence of smoking, harmful alcohol consumption and obesity in England.

INTRODUCTION: Small area estimates in health research currently capture single indicators but fail to reflect concurrent overlaps in health risk factors like smoking, increased alcohol consumption, and obesity. Local governments need detailed estimates of these risk factors to plan public health intervention strategies and tackle health inequalities. However, local data are often unavailable or highly restricted. This paper addresses this critical data gap by providing novel small area estimations of co-occurring health risk factors in England.

METHODS: To generate the local estimates, we employ a spatial microsimulation approach for synthetic population generation using two alternative population synthesis algorithms: Iterative Proportional Fitting (IPF) and Simulated Annealing (SA). These approaches integrated individual microdata from Health Survey for England (HSE) data with small-area demographic constraints (sex, age, ethnicity, qualifications, partnership status and socioeconomic classifications) derived from the 2021 Office for National Statistics Census and was representative of a single year (2022). Small area estimates of three health risk factors: tobacco smoking, obesity, and harmful alcohol consumption, were modelled and validated against official independent aggregated benchmark data to Local Authority and/or Region level.

RESULTS: Both methods produced small area estimates consistent with observed regional patterns. Analysis revealed smoking and obesity share similar spatial distributions, whereas alcohol consumption exhibits a distinct geographical profile. The resulting fine-grained analysis enabled the exploration of geographical areas of high estimated prevalence for co-occurring risk factors and overlaps between different sociodemographic characteristics.

CONCLUSIONS: This study demonstrates the utility of synthetic populations in estimating co-occurring health risk factors in small areas. Estimates from the synthetic populations can provide public health decision-makers and local governments with a vital tool for identifying local behaviour patterns, facilitating targeted, evidence-based public health interventions.