A predictive model to calculate the probability of placental abruption in patients with initially low obstetric risk

Abstract

Aim. To develop a predictive model to calculate the probability of placental abruption development in patients with initially low obstetric risk by using a multidimensional mathematical analysis of anamnestic, clinical, and genetic data.
 Materials and methods. This logistic model is based on 212 observations. This study included two groups of healthy women aged 1835 years without a history of spontaneous abortion, preterm labor, stillbirth, preeclampsia, or other complications. The first group was comprised of 112 women with a history of a placental abruption (patient group), and the second group consisted of 100 women with non-complicated pregnancy (control group). Gene polymorphisms were detected using the polymerase chain reaction-real time technique. The data was analyzed using binary and multifactorial statistical mathematics. Our analysis of the predictive models was performed by using logistic regression. To determine the diagnostic value of the predictive models used, the ROC-curve is followed by determining the area under it (AUC).
 Results. The 24 anamnestic, clinical, and genotypic characteristics were examined as placental abruption predictors using multifactorial analysis. We were able to determine what the nine most informative predictors of placental abruption were according to Wald Chi-Square criteria (2, р): fetal growth restriction (13.143; р0.001), placental insufficiency (9.629; р=0.002), male sex of the fetus (6.965; р=0.008), preeclampsia (6.779; р=0.009), combination of genotypes PAI-1-5G4G/FV-1691GA (6.613; р=0.010) or PAI-1-5G4G/MTHFR-677CT (6.158; р=0.013), pathology of amnion (4.497; р=0.034), infectious diseases in pregnancy (4.277; р=0.039), tobacco smoking (3.963; р=0.047). ROC analysis identified a high specificity (82.64%) and sensitivity (77.42%) of the model, and the integral index of the effectiveness of predictive markers (AUC=0.836), according to the expert scale of values which is indicative of a very high quality model.
 Conclusion. It is recommended to use this elaborated predictive model in clinical practice for the purpose of individual risk assessment of placental abruption in healthy patients with initially low obstetric risk.