Recorded gestational violence and neonatal outcomes in a cohort of deliveries requiring neonatal hospitalization: a 15-year hospital-based perinatal registry study.

BACKGROUND: Gestational violence is incompletely captured in routine perinatal registries, limiting prevalence estimation and etiologic interpretation. We evaluated associations between registry-recorded gestational violence, maternal characteristics, and neonatal outcomes in a selected hospital-based population. METHODS: This cross-sectional study included 26,519 delivery records from the Perinatal Information System at a tertiary referral hospital in Quito, Ecuador, from January 2009 to June 2024. Eligibility was restricted to deliveries whose newborns required neonatal hospitalization. Associations were estimated using maximum-likelihood, Firth-penalized, and Bayesian logistic regression. Neonatal analyses examined 12 prespecified outcome-trimester combinations, analyzed as exploratory. Sensitivity analyses addressed behavioral co-exposures, exposure misclassification, informative missingness, calendar time, prior specification, unmeasured confounding, and third-trimester ascertainment opportunity. RESULTS: Gestational violence was recorded in 1.19% of deliveries. The strongest maternal associations were observed for illicit drug use [Bayesian odds ratio [OR], 6.44; 95% credible interval [CrI], 3.78-10.74], passive smoking (OR, 3.56; 95% CrI, 2.62-4.79), and alcohol use (OR, 3.18; 95% CrI, 2.19-4.54). Third-trimester recorded violence showed the most consistent association with preterm birth (OR, 1.325; 95% CrI, 0.956-1.822; E-value, 1.57), corresponding to a risk difference of 6.5 percentage points (95% confidence interval, -0.8-14.2). The estimate weakened when restricted to births ≥34 weeks and when adjusted for calendar-related registry completeness, but increased modestly with adjustment for behavioral markers. These findings should not be interpreted as establishing causal associations. CONCLUSIONS: Routine perinatal registry data may be useful for hypothesis generation and for guiding clinical and epidemiologic research until stronger evidence from prospective studies using validated violence-assessment instruments becomes available.