Non-linear multivariate decomposition modelling of the predictors of menstrual product use among reproductive-aged women: evidence from the 2022 Ghanaian demographic and health survey

Le résumé fourni par la source

Background: Menstrual health is a critical yet under-prioritized component of women's sexual and reproductive health, with persistent inequities in access to menstrual products across socioeconomic and geographic contexts. In Ghana, these disparities are influenced by a combination of individual, household, and community-level factors. This study examined the determinants of menstrual product use and quantified the relative contributions of compositional and structural factors to residence-related disparities using nationally representative data. Methods: Data were derived from the 2022 Ghana Demographic and Health Survey among a weighted sample of 12,497 women aged 15-49 years. Single-and multilevel logistic and ordered logistic regression models were fitted to identify determinants of menstrual product use, adjusting for intra-cluster correlations and complex survey design. Predictive performance and model fit were evaluated using AIC/BIC, McFadden's pseudo-R², C-index, and ROC AUC. Non-linear multivariate decomposition analysis was conducted to disentangle the contributions of compositional differences (endowments) and differential effects (coefficients) to disparities by urban-rural residence. Results: Use of modern menstrual products was strongly associated with higher individual educational attainment, household wealth, urban residence, younger age, self-reported good health, and greater media exposure. At the community level, higher average education and media exposure within clusters were also positively associated with uptake. Multilevel modelling revealed that 13%-15% of the variance in product use was attributable to between-cluster differences, highlighting the influence of contextual factors. Predictive models exhibited excellent discrimination (C-index ≥ 0.82; ROC AUC = 0.87). Decomposition analyses indicated that compositional factors, particularly household wealth and education, accounted for the majority of rural-urban (63%) inequalities, while differences in the effects of these determinants further contributed to disparities. Conclusions: Structural inequities in wealth, education, and community resources are major determinants of menstrual product use in Ghana. Policy interventions that subsidize menstrual products, expand school-and community-based distribution programs, enhance mass media outreach, and improve rural WASH infrastructure are critical to reducing disparities. Integrating menstrual health into national health and gender equity strategies is essential to advancing Sustainable Development Goals related to health, education, and gender equality.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.