Establishment of a prediction model for histological chorioamnionitis and its association with outcomes of premature infants.

Abstract

This study aims to construct a prediction model for histological chorioamnionitis (HCA) and analyze the associations between the predicted risk of HCA and adverse outcomes in preterm infants. In total, 673 subjects were included in this cohort study and divided into HCA group (n = 195) and non-HCA group (n = 478). A stepwise method was used to screen the predictors for HCA, binary logistic regression was used to construct the prediction model, and the associations between the predicted risk of HCA and adverse outcomes were analyzed. HCA occurred in 195 patients, accounting for 29.0%. The sensitivity of the prediction model was 0.821 [95% confidence interval (CI): 0.767-0.874)], the specificity was 0.684 (95% CI: 0.642-0.726), the positive predictive value was 0.514 (0.459-0.570), the negative predictive value was 0.903 (95% CI: 0.873-0.934), the area under the curve was 0.821 (95% CI: 0.786-0.855), and the accuracy was 0.724 (95% CI: 0.690-0.757). The predicted risk of HCA was associated with a higher risk of bronchopulmonary dysplasia (BPD) [odds ratio (OR) = 3.48, 95% CI: 1.10-10.95)], sepsis (OR = 6.66, 95% CI: 2.17-20.43), and neonatal infections (OR = 9.85, 95% CI: 3.59-26.98), but not necrotizing enterocolitis (OR = 0.67, 95% CI: 0.24-1.88), retinopathy of prematurity (OR = 1.59, 95% CI: 0.37-6.85), and brain damage (OR = 1.77, 95% CI: 0.82-3.83). After adjusting for confounders including gestational week at birth and birth weight, the risk of neonatal infections (OR = 5.03, 95% CI: 2.69-9.41) was increased in preterm infants' exposure to HCA. The model showed good predictive performance for identifying pregnant women with a higher risk of HCA. In addition, HCA was associated with the risk of BPD, sepsis, and infections in neonates.