PROGNOSIS OF FIRST-TRIMESTER TWIN PREGNANCIES - POLYCHOTOMOUS LOGISTIC-REGRESSION ANALYSIS

PURPOSE: To determine which clinical and sonographic characteristics independently affect the prognosis of first-trimester twin pregnancies and to develop formulas for calculating the probabilities of the pregnancy resulting in two, one, or no liveborn infants. MATERIALS AND METHODS: The authors prospectively identified 137 twin pregnancies with two heartbeats at first-trimester sonography and with known pregnancy outcome. Stepwise polychotomous logistic regression analysis was used to identify characteristics that have an independent statistically significant relationship with pregnancy outcome and to develop outcome-prediction formulas. RESULTS: Of 137 patients, 110 (80.3%) had viable twins, 12 (8.8%) had one infant, and 15 (10.9%) had none. Gestational age, chorionicity, and sonographic findings were independent, statistically significant (P < .05) prognostic factors, while maternal age, method of conception, and indication for sonography were not. CONCLUSION: The logistic regression formulas involving the three independent prognostic factors can be used in the first trimester to predict pregnancy outcome.