Water fluoridation as a population strategy for reducing oral health inequalities: high-dimensional effect heterogeneity analysis using machine learning.

Background: Dental caries is the most prevalent disease worldwide, associated with substantial health inequalities. Water fluoridation is a population strategy shown to effectively prevent dental caries; however, its impact on health inequalities remains inconclusive. We investigated the high-dimensional effect heterogeneity of water fluoridation in preventing dental caries among children in Australia.

Methods: Data from the National Child Oral Health Study-a national representative study conducted in 2012-14, linked to the information on lifetime exposure to fluoridated water-were analysed (n = 17 517 children aged 5-14 years). A doubly robust target minimum loss-based estimation was used to estimate the average treatment effect of lifetime exposure to fluoridated water (never exposed versus fully exposed) on the number of tooth surfaces with dental caries. The effect heterogeneity was evaluated by estimating the conditional average treatment effects (CATEs) using the causal forest algorithm, which integrated 47 child demographic, socioeconomic, and parental factors.

Results: In total, 58.1% were fully exposed to water fluoridation throughout their lifetime. Water fluoridation was associated with having -0.9 (95% confidence interval: -1.1, -0.8) fewer dental caries incidents. The estimated CATEs were mostly negative and the magnitude substantially varied (median CATE, -0.9; interquartile range, 0.7). Children from lower socioeconomic backgrounds exhibited greater benefits: i.e. the average CATEs ranged from -1.4 for children from a single parent with school-level education and low-income families to -0.8 for children from two parents with tertiary-level education and high-income families.

Conclusion: Water fluoridation was associated with lower dental caries and exhibited greater benefits for vulnerable subpopulations.