A clinical predictive model of intrauterine inflammation for early single preterm birth

Objective: To develop clinical predictive model of intrauterine inflammation for single preterm births <= 34 weeks gestational age. Methods: Clinical parameters and placental pathology of 240 pregnant women were collected. According to placenta pathology, intrauterine inflammation was divided into maternal inflammatory response (MIR) and fetal inflammatory response (FIR). We used logistic regression to establish predictive models for MIR and FIR. Results: Among 240 births, 119 (49.58%) had MIR, while 54 (22.5%) presented FIR. The logistic model for MIR was: logit P = (0.133 x neutrophile counts) + (-1.473 x pregnancy hypertension) + (-1.302 x cesarean section) + (1.510 x prenatal antibiotics) -1.389, which yielded the area under the ROC curve of 0.845. The logistic model for predicting FIR was: logit P = (-0.703 x lymphocyte counts) + (0.193 x neutrophile counts) + (2.349 x PPROM) 3.951, and the AUC was 0.852. Conclusion: These predictive models are easily established and have much better performance than single factors.