Concordance-analysis and evaluation of different diagnostic algorithms used in first trimester screening for late-onset preeclampsia.

Objective: Concordance-analysis and evaluation of existing algorithms detecting late-onset preeclampsia during first trimester screeningMethods: Retrospective cohort study investigating risk algorithms of late-onset preeclampsia during first trimester screening in a German prenatal center. Three previously developed algorithms including anamnestic factors (Apriori) and biophysical markers (BioM) were investigated by using detection rates (DR) with fixed FPR 10% and fixed cutoff >1:100. Furthermore, we set up a concordance-analysis of test results in late-onset preeclampsia cases to examine the effect of influencing factors and to detect potential weaknesses of the algorithms. Therefore, we modeled the probability of discordances as a function of the influencing factors based on a logistic regression, that was fitted using a Bayesian approach.Results: 6,113 pregnancies were considered, whereof 700 have been excluded and 5,413 pregnancies were analyzed. 98 (1.8%) patients developed preeclampsia (79 late-onsets, 19 early-onsets). The Apriori-algorithm reaches a DR of 34.2%, by adding BioM (MAP and UtA-PI) the DR improves to 57.0% (FPR of 10%). In concordance-analysis of Apriori algorithm and Apriori+BioM algorithms, influencing factor BMI<25 increases the chance of discordances sigificantly. Additional, in the subgroup of late-onset preeclampsias with BMI<25 the DR is higher in Apriori+BioM algorithms than in Apriori algorithm alone. If both compared algorithms include BioM, influencing factor MAP decreases the chance of discordances significantly. All other tested influencing factors do not have a statistically significant effect on discordancesConclusion: Normal-weight patients benefit more from the integration of MAP and UtA-PI compared to overweight/obese patients.