Mode of delivery and neonatal outcomes in preterm twins less than 32 weeks of gestation or birthweight under 1500 g: a systematic review and meta-analyses.

Background: The mode of delivery for twins born before 32 weeks of gestation remains controversial. Our purpose is to conduct a meta-analysis of twin pregnancies less than 32 weeks or twin weight less than 1500 g, so as to find a suitable delivery mode.

Methods: We searched PubMed database, Cochrane Library database, and EMBASE database through December 2022. This protocol was registered with PROSPERO (CRD42023386946) prior to initiation. Studies that compared vaginal delivery to cesarean section for newborns less than 32 weeks of gestation or birthweight under 1500 g were included. The primary result was neonatal mortality rate. Secondary result was neonatal morbidity. The quality of literatures included in the research was evaluated in accordance with Newcastle-Ottawa Scale (NOS) literature quality evaluation scale. We use odds ratio (OR) as the effect index for binary variables. Point estimates and 95% confidence intervals (95% CI) were calculated. P < 0. 05 indicated statistically significant difference.

Results: Our search generated 5310 articles, and a total of 8 articles comprising a total of 14,703 newborns were included in the analysis. The odds ratios of neonatal mortality rate were for twins delivered by vaginal delivery compared to cesarean section were 0.84 (95% CI 0.57-1.24, P = 0.38). The 5-min Apgar score < 7 (95% CI 0.44-1.75, P = 0.72), necrotizing enterocolitis (95% CI 0.81-1.19, P = 0.82), intraventricular hemorrhage (95% CI 0.41-1.86, P = 0.71), periventricular leukomalacia (95% CI 0.16-4.52, P = 0.84), bronchopulmonary dysplasia (95% CI 0.88-1.36, P = 0.42), and respiratory distress syndrome (95% CI 0.23-2.01, P = 0.48) were not statistically significant between the two groups.

Conclusion: We have observed that vaginal delivery does not confer an increased risk of neonatal morbidity and mortality in twins born before 32 weeks of gestation. However, the current results are affected by substantial heterogeneity and confounding factors. We still need high-quality randomized-controlled studies require to address this important question.