The preterm prediction study: a clinical risk assessment system

Our aims were to develop a risk assessment system for the prediction of spontaneous preterm delivery using clinical information available at 23 to 24 weeks' gestation and to determine the predictive value of such a system.

A total of 2929 women were evaluated between 23 and 24 weeks' gestation at 10 centers. Demographic factors, socioeconomic status, home and work environment, drug and alcohol use, and medical history were evaluated. Information regarding symptoms, cultures, and treatments in the current pregnancy were ascertained. Anthropomorphic and cervical examinations were performed. Univariate analysis and multivariate logistic regression were performed in a random selection, constituting 85% of the study population. The derived risk assessment system was applied to the remaining 15% of the population to evaluate its validity.

A total of 10.4% of women were delivered of preterm infants. The multivariate models for spontaneous preterm delivery were highly associated with spontaneous preterm delivery (p < 0.0001). A low body mass index (<19.8) and increasing Bishop scores were significantly associated with spontaneous preterm delivery in nulliparous and multiparous women. Black race, poor social environment, and work during pregnancy were associated with increased risk for nulliparous women. Prior obstetric outcome overshadowed socioeconomic risk factors in multiparous women with a twofold increase in the odds of spontaneous preterm delivery for each prior spontaneous preterm delivery. Current pregnancy symptoms, including vaginal bleeding, symptomatic contractions within 2 weeks, and acute or chronic lung disease were variably associated with spontaneous preterm delivery in nulliparous and multiparous women. When the system was applied to the remainder of the population, women defined to be at high risk for spontaneous preterm delivery (> or = 20% risk) carried a 3.8-fold (nulliparous women) and 3.3-fold (multiparous women) higher risk of spontaneous preterm delivery than those predicted to be at low risk. However, the risk assessment system identified a minority of women who had spontaneous preterm deliveries. The sensitivities were 24.2% and 18.2% and positive predictive values were 28.6% and 33.3%, respectively, for nulliparous and multiparous women.

Although it is possible to develop a graded risk assessment system that includes factors that are highly associated with spontaneous preterm delivery in nulliparous and multiparous women, such a system does not identify most women who subsequently have a spontaneous preterm delivery. This system has investigational value as the basis for evaluating new technologies designed to identify at-risk subpopulations.