Reliable antenatal identification of pre-eclampsia and small for gestational age is crucial to judicious allocation of monitoring resources and use of preventative treatment with the prospect of improving maternal/perinatal outcome. The purpose of this systematic review was to determine the accuracy of five serum analytes used in Down's serum screening for prediction of pre-eclampsia and/or small for gestational age.
The data sources included Medline, Embase, Cochrane library, Medion (inception to February 2007), hand searching of relevant journals, reference list checking of included articles, contact with experts. Two reviewers independently selected the articles in which the accuracy of an analyte used in Downs's serum screening before the 25th gestational week was associated with the occurrence of pre-eclampsia and/or small for gestational age without language restrictions. Two authors independently extracted data on study characteristics, quality and results.
Five serum screening markers were evaluated. 44 studies, testing 169,637 pregnant women (4376 pre-eclampsia cases) and 86 studies, testing 382,005 women (20,339 fetal growth restriction cases) met the selection criteria. The results showed low predictive accuracy overall. For pre-eclampsia the best predictor was inhibin A>2.79MoM positive likelihood ratio 19.52 (8.33,45.79) and negative likelihood ratio 0.30 (0.13,0.68) (single study). For small for gestational age it was AFP>2.0MoM to predict birth weight < 10th centile with birth < 37 weeks positive likelihood ratio 27.96 (8.02,97.48) and negative likelihood ratio 0.78 (0.55,1.11) (single study). A potential clinical application using aspirin as a treatment is given as an example.
There were methodological and reporting limitations in the included studies thus studies were heterogeneous giving pooled results with wide confidence intervals.
Down's serum screening analytes have low predictive accuracy for pre-eclampsia and small for gestational age. They may be a useful means of risk assessment or of use in prediction when combined with other tests.