Men and women differ in weight excess and obesity patterns: evidence from a microarea-based analysis in Belo Horizonte

Weight excess and obesity are increasing in Brazil and may be unevenly distributed within cities. However, city-level estimates can mask intra-urban inequalities and provide limited evidence for local public health planning. This study estimated the prevalence of weight excess and obesity among adults living in Belo Horizonte, Brazil, from 2006 to 2018 and examined temporal and territorial inequalities according to sex and Health Vulnerability Index (HVI)-based microareas. This repeated cross-sectional ecological study used Vigitel, a national telephone-based surveillance system. Survey years were pooled into 2006–2009, 2010–2013 and 2014–2018. Weight excess (body mass index ≥ 25.0 kg/m²) and obesity (≥ 30.0 kg/m²) were analysed as binary outcomes. Estimates were stratified by sex and nine HVI-based microarea clusters. Conditional generative adversarial networks were used to generate synthetic observations, with greater augmentation in clusters with smaller observed samples. Weighted prevalence estimates and 95% confidence intervals (CI) accounted for the survey design; women-minus-men prevalence differences were estimated using survey-adjusted contrasts and design-based Wald tests. Weight excess and obesity increased over time in Belo Horizonte, with marked differences by sex and territorial vulnerability. Weight excess was generally more frequent among men in low-vulnerability clusters, whereas this difference was reduced or reversed in more vulnerable areas. In 2014–2018, women showed higher weight excess prevalence than men in HI-1, VH-0 and VH-1. Obesity showed a clearer territorial and sex-specific pattern, with the highest burden among women living in high- and very high-vulnerability clusters. In 2014–2018, obesity among women reached 21.2% in HI-1, 21.9% in HI-2 and 22.4% in VH-1. Sex differences in obesity also favoured women in more vulnerable clusters, with women-minus-men differences ranging from 6.7 to 9.7% points in high- and very high-vulnerability clusters. Weight excess and obesity in Belo Horizonte showed marked sex-specific and territorial inequalities. HVI-based small-area analysis identified patterns that would be diluted in city-level estimates, while synthetic population augmentation produced prevalence estimates closely aligned with those observed in Vigitel. Because synthetic data do not correct structural undercoverage in landline-based surveys, estimates in the most vulnerable areas should be interpreted cautiously. These findings support territorialised and gender-responsive obesity prevention strategies, particularly for women living in vulnerable urban areas.