METHODS: Systematic review and meta-analysis. PubMed, Embase and the Cochrane Library were searched. Methodological quality was rated using QUADAS and scientific evidence using GRADE. Summary measures of diagnostic accuracy were calculated using a bivariate random-effects model.
RESULTS: In a general pregnant population, there is moderate evidence that the pooled sensitivity is 0.993 (95% CI 0.955-0.999) and specificity was 0.999 (95% CI 0.998-0.999) for the analysis of T21. Pooled sensitivity and specificity for T13 and T18 was not calculated in this population due to the low number of studies. In a high-risk pregnant population, there is moderate evidence that the pooled sensitivities for T21 and T18 are 0.998 (95% CI 0.981-0.999) and 0.977 (95% CI 0.958-0.987) respectively, and low evidence that the pooled sensitivity for T13 is 0.975 (95% CI 0.819-0.997). The pooled specificity for all three trisomies is 0.999 (95% CI 0.998-0.999).
CONCLUSIONS: This is the first meta-analysis using GRADE that shows that NIPT performs well as a screen for trisomy 21 in a general pregnant population. Although the false positive rate is low compared with first trimester combined screening, women should still be advised to confirm a positive result by invasive testing if termination of pregnancy is under consideration.