Exploring women’s empowerment in relation to contraceptive practices among married women in Bangladesh: a survey-weighted sequential logistic regression analysis

Background Women’s empowerment is multidimensional, yet contraceptive studies often combine distinct domains into one score. We examined whether household decision-making participation and rejection of wife-beating had different associations with current contraceptive use in Bangladesh. Methods We analyzed cross-sectional data from currently married or partnered, nonpregnant women aged 15–49 years in the 2022 Bangladesh Demographic and Health Survey. Tetrachoric principal component analysis derived standardized scores for household decision-making and rejection of wife-beating. Survey-weighted logistic regression estimated associations with current use of any contraceptive method across sequential adjustment models. Modified Poisson regression, marginal standardization, modern-method analyses, and alternative empowerment specifications assessed robustness. Results The complete-case sample included 17,345 women. Parallel analysis retained two empowerment components explaining 76.2% of total variance. In the fully adjusted model, each 1-standard-deviation higher household decision-making score was associated with higher odds of contraceptive use (OR = 1.068, 95% CI = 1.029–1.108; p < 0.001). The corresponding adjusted prevalence ratio was 1.023 (95% CI = 1.010–1.036), and the standardized prevalence difference comparing + 1 with − 1 standard deviation was 2.84% points (95% CI 1.25–4.42). Rejection of wife-beating was not independently associated with use (OR = 0.983, 95% CI = 0.946–1.021; p = 0.367). The domain-specific pattern persisted for modern contraceptive use and in reduced-adjustment analyses. Middle- and rich-wealth groups had lower adjusted odds than the poor group, showing that contraceptive use did not follow a simple positive wealth gradient. Conclusions Household decision-making participation, but not rejection of wife-beating, was associated with contraceptive use in this nationally representative sample. The findings show why empowerment domains should not be treated as interchangeable and support multidimensional measurement in family-planning research. Longitudinal studies are needed to clarify temporality.