Authors’ reply: comments regarding “Current evidence gaps to support systematic cytomegalovirus screening in pregnancy” Authors: Agathe Billette de Villemeur, Bruno Hoen, Louis-Rachid Salmi Published: 2026 We thank Périllaud-Dubois and colleagues for their comments on our Viewpoint, ‘Current evidence gaps to support systematic cytomegalovirus screening in pregnancy’.1 From their first comment, the authors state that we did not summarise fairly their study on the cost-effectiveness of screening and valacyclovir-based treatment strategies for first-trimester cytomegalovirus primary infection in pregnant women in France.2 We aimed to summarise each study as accurately and concisely as possible. We pointed out that Perillaud-Dubois et al.’s assertion that the trial conducted by Shahar-Nissan et al.3 demonstrates the effectiveness of screening is incorrect for several reasons. Firstly, this trial evaluated valaciclovir, not screening. Secondly, we emphasised that this trial had many flaws, such as lack of intent-to-treat analysis and failure to consider all terminations of pregnancies. Thirdly, and most importantly, we disagree with interpreting the results on mother-to-child transmission as evidence of a positive benefit-risk ratio, as this does not account for the sequelae associated with congenital CMV. Furthermore, we think that Périllaud-Dubois’ hypotheses on the benefit-risk ratio were overoptimistic, as they estimated the proportion of terminations of pregnancies at 42% when there was an ultrasound abnormality after a positive amniocentesis, resulting in a much lower number of terminations than reported by French authoritie.4, 5, 6 While Périllaud-Dubois et al. cite two additional articles to support the claim that valaciclovir is effective in preventing mother-to-child CMV transmission, we assume these studies do not provide sufficient evidence to justify systematic screening during pregnancy. One is an observational study that relies on a historical comparison of women who accepted treatment and those who did not or not treatment. Despite the use of propensity score adjustment, key confounding factors, such as smoking, parity, promiscuity and maternal age remain inadequately controlled.7 The other is a meta-analysis that combines observational studies with heterogeneous inclusion criteria, including HIV-infected women,8 alongside the only randomised trial ever conducted: the Shahar-Nissan trial,3 with all its limits discussed above and in our Viewpoint.1 Although we pointed out the limitations of CMV diagnostic assays, we did not deny that ‘the tools used for the diagnosis and timing of maternal primary infection are reliable’. Nevertheless, it is worth remembering that, given how rare the events in question are, there would be a high number of false positive results, even if the test had a specificity above 95%, the current recommended threshold for screening programs.9 Moreover, currently available tests are not able to detect reinfections/reactivations. Furthermore, we want to reiterate that we acknowledge that the subject is a matter of controversy. We finally agree with Périllaud-Dubois et al. that women deserve to be informed and to benefit from the most effective strategies for preventing congenital cytomegalovirus infection. However, we disagree with the statement that ‘screening and valaciclovir are the key to combatting CMV’, a claim that has yet to be proven. It is paramount to remind the clinical and public health community that the World Health organization,10 Public health England,11 and, in France, the Haute autorité de santé9 suggest that all primary prevention efforts should have been implemented before screening be considered. Moreover, the community randomised trial remains the reference design to assess the effectiveness and safety or a screening programme. As far as we could see from a quick update of our search of the literature and trial registries, there is no ongoing study testing their claim. Contributors ABV and BH contributed to the design and conduct of the work reported, and to the writing and reviewing of this article; they had direct access to and checked the accuracy of all the data reported in the manuscript. LRS contributed to the design and conduct of the work reported; he drafted the first version of this article and reviewed the final version; he had direct access to and checked the accuracy of all the data reported in the manuscript. All authors read and approved the final version of the manuscript. Declaration of interests All authors declare no conflict of interests.