Evaluation of a risk-scoring system for prediction of preterm labor

A scoring system was used prospectively to screen 7329 patients for risk of preterm labor over a 6-year period. When 15.8% of patients (1156 of 7329) were scored as high risk, sensitivity was 42.4% (multiparas, 55.2%, nulliparas, 29.8%) and positive predictive value was 22.8%. The incidence of preterm labor increased with increasing risk score; gestational age at delivery was negatively correlated with risk score. Analysis of individual factors revealed that more than half were not of predictive value in the population under investigation. A simplification of the system, defining 12 of the factors as major (high risk) and six as minor (2 or more indicates high risk) yielded a high risk group of 14.1%, sensitivity of 41.0%, and positive predictive value of 24.6%. Thus, although risk scoring does predict a substantial proportion of patients with preterm labor, its sensitivity is less than ideal. No advantage was gained with a larger and more complex system. Planning for preterm birth prevention programs should include recognition of the limitations of risk scoring, consideration of simplified systems, and investigation of additional biochemical or biophysical screening methods.