Unmet need for contraceptives among adolescents and young women in Uganda: a multilevel weighted mixed-effects logistic regression modelling

Unmet need for contraception among adolescents and young women remains a major public health concern in Uganda, shaped by a complex interplay of individual, household, and contextual factors. This underscores the need to examine these factors and the magnitude of clustering effects in reproductive health data. Regrettably, many studies overlook cluster-level variation and apply final survey weights in multilevel modeling rather than appropriate level-specific weights. Such practices likely yield biased estimates, misguide statistical inference, and reduce study power. This study aimed to determine the sources of hierarchical variation and identify factors associated with unmet need for contraceptives among adolescents and young women in Uganda using a multilevel-weighted mixed-effects logistic regression model. This cross-sectional study analyzed data on 3,652 sexually active adolescent girls (15–19 years) and young women (20–24 years) from the 2022 Uganda Demographic and Health Survey (UDHS); “sexually active” respondents were those married/cohabiting or who had had sex in the 30 days preceding the survey. The outcome, unmet need for contraceptives, was binary. Variables with p < 0.25 at bivariate multilevel analysis were considered for multivariable modelling. A multilevel-weighted mixed-effects logistic regression model, incorporating individual-, household-, and community-level covariates with enumeration area as a random effect, was fitted stepwise. Adjusted odds ratios (aOR) with p < 0.05 were considered significant; model fit was assessed using AIC, BIC, and the likelihood ratio test (LRT), and the intra-class correlation coefficient (ICC) quantified clustering. We found that 970 (26.5%) had an unmet need for contraception. Lower odds were observed among those aged 20–24 years [aOR = 0.70, 95% CI: 0.53–0.91], with a history of pregnancy termination [aOR = 0.65, 95% CI: 0.44–0.96], with two or more sexual partners [aOR = 0.59, 95% CI: 0.38–0.91], who desired more children [aOR = 0.41, 95% CI: 0.27–0.61], who had ever used contraception [aOR = 0.16, 95% CI: 0.13–0.21], and who did not perceive distance to a facility as a big problem [aOR = 0.73, 95% CI: 0.56–0.95]. Higher odds occurred among those who had ever given birth [aOR = 1.75, 95% CI: 1.23–2.47], rural residents [aOR = 1.30, 95% CI: 1.12–1.64], and Eastern-region residents [aOR = 1.36, 95% CI: 1.21–2.66]. After full adjustment, about 30% of the total variation in unmet need remained attributable to differences between enumeration areas (ICC = 0.302; LRT p = 0.002). Unmet need for contraception among adolescents and young women in Uganda is shaped mainly by individual- and community-level factors, with limited contribution from household characteristics. Programmes should prioritize individual-level, behaviour-focused interventions alongside community-targeted approaches, particularly in rural areas and the Eastern region, rather than household-level targeting alone.