Objective: Prenatal prediction of neonatal survival and survival without short term major morbidity (STMM) in babies born at the borderline of viability.
Methods: Obstetric and neonatal records of 471 babies born from 22 0/7 to 25 6/7 weeks between the years 2000 to 2007 were reviewed. A total of 179 live-born nonanomaluos singleton infants with birth-weights of 300-1397 g, who underwent ultrasonographic examination < 7 days before delivery, qualified for the study. Ultrasound biometry, use of antenatal steroids, fetal gender, race, gestational age, estimated fetal weight, and maternal medical and obstetric characteristics were evaluated. Univariate analyses, logistic regression, and bootstrapping using 10,000 resamplings were performed to derive prediction formulas that projected neonatal outcomes. From these, quantitative patient-specific prenatal estimates of neonatal survival and survival without STMM were calculated.
Results: On the basis of the use of prenatally available parameters, neonatal survival was predicted with a sensitivity of 85.5% and specificity of 70.6%. Area under ROC curve (95%CI) = 0.835 (0.777-0.894). Corresponding values for survival without STMM were 94.7% and 34%. Area under ROC curve (95%CI) 0.738 (0.640-0.836). Among positive predictors of survival it is worth mentioning: intact membranes, femur length ≥ 39.1 mm and estimation of fetal weight > 493 g.
Conclusion: Prenatal prediction of neonatal survival and STMM was achieved. This approach has the potential improve counseling and obstetric decision making at the borderline of viability.