A bivariate meta-analysis of sensitivity and specificity: accuracy of maternal serum human choriongonadotrophin (MShCG) determination in the prediction of pre-eclampsia

Article type
Authors
Cnossen J, van der Post J, Mol B, Khan K, ter Riet G
Abstract
Background: Pre-eclampsia, a syndrome of hypertension and proteinuria, is a leading cause of maternal and perinatal mortality and morbidity. Early prediction of pre-eclampsia may be used to target effective early treatment at high-risk women.

Objective: To systematically review the accuracy of MShCG determination in predicting pre-eclampsia.

Methods: We searched MEDLINE, EMBASE, Cochrane Library, Medion, reference lists of reviews and eligible primary articles, and contacted experts in the field. Reviewers working independently selected studies, abstracted data, and assessed validity according to eight criteria. We used a random-effects bivariate meta-regression model to meta-analyse estimates of sensitivity and specificity. Logit pairs of sensitivity and specificity of each study were analysed in a single model, accounting for their correlation due to differences in threshold between studies. This model acknowledges the difference in precision by which sensitivity and specificity have been measured in each study. We performed subgroup analyses by using clinical and, if possible, methodological covariates.

Results: We extracted 16 studies with a total of 72,732 women, including 2,258 women developing pre-eclampsia (median prevalence 3.5% (range 0.6%-26.8%)). Study populations consisted usually of cohorts undergoing foetal screening at 14-22 weeks, excluding multiple pregnancies and foetal anomalies. Details on relevant co-morbidity were poorly reported. Cut-off values varied mainly between 2.0 and 2.5 multiples of the median (MoM). Based on 14 studies, summary estimates of sensitivity and specificity were 23% (95% CI 16-33%) and 90% (95% CI 87-93%), respectively (See Figure on the following page). Subgroup analyses showed significantly higher sensitivity and lower specificity in studies that used an internationally accepted definition of pre-eclampsia, a cut-off value of 2.0 MoM, or an enzymatic type of immunoassay (table not shown). Limited (reporting of) data precluded controlling subgroup analyses for methodological items.

Conclusions: A negative MShCG test virtually excludes development of pre-eclampsia. Positive test results must be verified. Given the world-wide use of the triple test for Down-screening, the optimal use of its components (alpha-fetoprotein, estriol, MShCG) for prediction of pre-eclampsia should be investigated in individual patient data meta-analysis using multivariable methods. Future single test evaluation research should report according to STARD criteria.
Acknowledgements: Hans Reitsma for helping us with the statistical analysis.