No link between topical retinoids and birth defects in pooled pregnancy data

Oral retinoids carry well-established risks in pregnancy, and that reputation has cast a long shadow over their topical cousins — the retinoid creams and gels widely prescribed for skin conditions during the reproductive years. Whether applying a retinoid to the skin in early pregnancy actually carries any measurable risk to the fetus has remained a persistent concern, with individual studies too scattered to settle it. This review set out to pool the available evidence on two hard outcomes: major congenital malformations and spontaneous abortion after first-trimester exposure.

Following PRISMA 2020 guidelines, the authors searched PubMed, Scopus, the Cochrane Library, Web of Science and Google Scholar up to November 25, 2025, for studies comparing pregnancy outcomes in women exposed to topical retinoids against unexposed controls. Eight studies met the criteria, covering a combined 3,632,118 pregnancies. They ran a random-effects meta-analysis alongside Bayesian hierarchical modelling with multiple prior specifications and a Robust Bayesian Model-Averaged analysis, plus leave-one-out sensitivity checks and E-value analysis to probe the influence of individual studies and unmeasured confounding.

Topical retinoid exposure in the first trimester was not significantly associated with major congenital malformations (risk ratio 0.83, 95% CI 0.64–1.07) or spontaneous abortion (risk ratio 0.99, 95% CI 0.68–1.43), with no heterogeneity across studies (I² = 0%) for either outcome. Bayesian analyses under three prior specifications produced posterior estimates ranging from RR 0.86 to 0.89, all with credible intervals crossing unity, and the model-averaged analysis gave a 74.1% posterior probability favouring the null hypothesis. Leave-one-out analysis did flag one influential study: omitting a large 2025 registry study shifted the malformation estimate by 29.6%.

The worry about topical retinoids in pregnancy has largely been inherited from the established teratogenicity of oral retinoids rather than built on direct evidence of harm. This pooled analysis, spanning millions of pregnancies, found no signal for either malformations or miscarriage, which supports the relative safety the authors describe for early pregnancy. But they are explicit that the result is driven mainly by a single large registry study, and every included study is observational. Independent large cohorts would be needed to make the null result stand on more than one dataset.

For anyone using a topical retinoid who is pregnant or could become pregnant, this analysis offers a measure of reassurance: across a very large pooled dataset, first-trimester exposure showed no detectable association with major birth defects or miscarriage. It is not a green light — the evidence is observational, rests heavily on one registry, and says nothing about oral retinoids, whose risks are well documented. Decisions about any medication in pregnancy sit with a prescriber, but the data here point away from the worst fears about the topical forms.