Mathematic modeling to predict abruptio placentae

Objective: This study was undertaken to identify correlates of abruptio placentae and to develop a mathematic model for the prediction of abruptio placentae. Study Design: A total of 170,258 singleton birth records from 1991 to 1996 contained in the Schleswig-Holstein perinatal database were analyzed. Fifty-two recognized obstetric risk factors were subjected to univariate analysis. Correlates of abruptio placentae then underwent stepwise forward binary logistic regression. A constant value B0, coefficients B1 through Bp, an odds ratio, and a 95% confidence interval were calculated for individual correlates. Results: Abruptio placentae occurred in 874 of 170,258 singleton gestations (0.5%). Of the 52 risk factors 31 proved to be correlates of abruptio placentae, with 16 among primiparous women and 25 among multiparous women. Ten correlates for primiparous, women and 13 for multiparous women emerged from the linear regression, with 7 correlates being shared by both primiparous and multiparous women. Conclusion: The probability that abruptio placentae will occur (p) can be calculated according to the following expression: p = ez/(1 + ez), where z = B0 + B1,…Bp. For example, for a primiparous woman who smokes with bleeding at >28 weeks’ gestation and a male fetus in the breech position, the following calculation would yield the chance of abruptio placentae:z = –2.25 + 2.51 + 0.41 + 0.24 + 0.60 = 1.51; p = e1.51/ (1 + e1.51) = 4.53/5.53 = 0.82, or 82%. (Am J Obstet Gynecol 2000;183:815-22.)