Abstract / Summary
Background: Unintended pregnancy is a global public health concern, yet its causal effects on adverse birth outcomes remain unclear. This study applied target trial emulation (TTE) to estimate its causal effects on low birth weight (LBW) and preterm birth (PTB). Methods: We analyzed Demographic and Health Surveys (DHS) data across 37 low- and middle-income countries (LMICs) from 1997 to 2021. Under the TTE framework, eligible pregnancies were assigned to intended or unintended groups. Inverse probability of treatment and censoring weighting were applied to address confounding and loss to follow-up (LTFU), respectively. A generalized linear mixed model (GLMM) was used to estimate risk ratios (RRs). For comparison, a conventional covariate-adjusted Poisson regression was conducted. Stratified and sensitivity analyses were performed to assess robustness. The E-value was estimated to evaluate the potential impact of unmeasured confounding. Results: Among 911,423 pregnancies from 635,139 women, 205,655 were unintended. In the TTE analysis, unintended pregnancy was significantly associated with an increased risk of PTB (RR = 1.267, 95% CI: 1.167-1.376), but not LBW (RR = 0.972, 95% CI: 0.914-1.035). In contrast, Poisson regression showed significant associations with both outcomes (LBW: RR = 1.090, 95% CI: 1.067-1.114; PTB: RR = 1.110, 95% CI: 1.084-1.137). Stratified and sensitivity analyses generally supported the primary findings. The E-value for PTB was 1.849. Conclusion: Unintended pregnancy may causally increase PTB risk, but is not associated with LBW. The findings support strengthening maternal healthcare, reproductive health education, and contraceptive access in LMICs.