11: The accuracy of fetal fibronectin and cervical length in women with signs of preterm labor before 34 weeks: a nationwide cohort study in The Netherlands (APOSTEL1 study)

We estimated the accuracy of cervical length (CL) and fetal fibronectin (fFN) measurement in predicting preterm delivery within 7 days among women with signs of preterm labor. We performed a nationwide cohort study in all 10 perinatal centers in the Netherlands between December 2009 and May 2012. We obtained fFN status and CL in women with threatened labor between 24 and 34 weeks gestational age with intact membranes. The study group consisted of women admitted directly to tertiary hospitals and women referred from secondary centers. Aim of the risk assessment was to correctly identify women who will not deliver, without too many unnecessary referrals. We estimated accuracy of fFN and CL separately, and then compared strategies that combine fFN and CL measurements in different ways, in which we varied CL cut-off values. The accuracy of different strategies and cut-offs was assessed using receiver operating curve analysis. We report on 559 of the 660 included women (85%). Full data will be presented at the congress. In total 87 (16%) women delivered within 7 days after inclusion. fFN only had a sensitivity of 76% (95%CI: 66-83%) and a specificity of 58% (95%CI: 54-62%). At the same sensitivity level as fFN, CL had a higher specificity of 80% (95%CI: 76-84%, p<0.001). Combining fFN with CL improved overall accuracy compared to single testing. CL measurement, and subsequent fFN testing in case of a CL between 15 mm and 30 mm had the same high negative predictive value as several other strategies (98%; 95%CI: 96-99%), at the highest specificity 67% (95%CI: 62-71%). The positive predictive value was 34% (95%CI: 28-40%), thus reducing unnecessary referrals from 209 of 472 with CL only to 157 of 472 (p<0.001). In women with signs of preterm labor, the optimal work-up is CL measurement, and fFN testing in case of a CL between 15 mm and 30 mm, reducing unnecessary referrals and treatment.Tabled 1Prognostic accuracy