Procedure-related risk of miscarriage following chorionic villus sampling and amniocentesis

Objectives: To estimate the procedure-related risks of miscarriage following chorionic villus sampling (CVS) and amniocentesis in a large unselected screened population, and to determine whether these risks are consistent with those reported in systematic reviews and meta-analyses.

Methods: This was a retrospective cohort study carried out on data obtained from a large fetal medicine unit in the UK between January 2009 and May 2018. We included all women with singleton pregnancy who booked for pregnancy care at our unit before 20 weeks' gestation, after excluding those with multiple pregnancy, major fetal defect, pregnancy termination and loss to follow-up. We estimated the risk of miscarriage in women who underwent a CVS or amniocentesis as well as in those who did not have an invasive procedure. The procedure-related risk of miscarriage was estimated as risk difference (95% CI) between the two groups. Univariate and multivariate regression analyses were used to derive odds ratios (95% CI) and determine which maternal and pregnancy characteristics provided a significant contribution in the prediction of miscarriage and whether CVS or amniocentesis provided a significant independent contribution.

Results: During the study period, 45 120 singleton pregnancies were booked for pregnancy care at our hospital, of which 1546 had an invasive procedure. We excluded 1429 (3.2%) pregnancies due to fetal defects, termination of pregnancy or missing outcomes. Of the 43 691 pregnancies included in the study population, 861 underwent CVS and 375 amniocentesis. In pregnancies that underwent CVS, the risk of miscarriage was 1.5% (13/861), compared with 1.2% (476/39 152) in pregnancies that had first-trimester combined screening and did not have an invasive procedure (P = 0.437). In pregnancies that underwent an amniocentesis, the risk of miscarriage was 0.8% (3/375), compared with 1.2% (491/42 463) in those that did not undergo an invasive procedure (P = 0.520). Univariate and multivariate regression analysis demonstrated that there was no significant contribution in the prediction of the risk of miscarriage from CVS (P = 0.399 and P = 0.592, respectively) or amniocentesis (P = 0.543 and P = 0.550, respectively). The risk of procedure-related loss attributed to CVS was 0.29% (95% CI, -0.53 to 1.12%) and that following amniocentesis was -0.36% (95% CI, -1.26 to 0.55%), which was not significantly different from the risk in women who did not have any procedure.

Conclusions: The procedure-related risks of miscarriage following CVS and amniocentesis in our study are considerably lower than those currently quoted and are consistent with the estimates of such risks reported by systematic reviews and meta-analyses. Copyright © 2019 ISUOG. Published by John Wiley & Sons Ltd.