StratumSkin science, cited
CIngredients·9 min·7 sources

Pregnant or breastfeeding? Here's what the evidence on your actives actually says

Retinoids, salicylic acid, hydroquinone and more, sorted by what's actually been studied, and where the honest answer is 'ask your doctor,' not 'the internet says.'

Disclosure

Nothing here is sponsored and we earn nothing from anything mentioned. If that ever changes, this box will say so before you read a word of the article.

Evidence at a glance

Topical retinoids
B Several cohorts, including one very large 2026 study, found no significant malformation signal, but no ethical trial can exist, and most guidelines still say avoid as a precaution.
Azelaic acid
C The active most often called 'pregnancy-safe', by absence of a red flag and low absorption, not by a dedicated safety study.
Hydroquinone
D 35–45% systemic absorption and essentially no human safety data. Consistently advised against.
Benzoyl peroxide, breastfeeding
C Low absorption and rapid clearance make it a generally accepted choice, but the low-risk conclusion is inferred from pharmacology, not a direct breastmilk study.
Salicylic acid, topical
C A different exposure route from oral aspirin. Low concentrations in cleansers are generally considered low risk, but concentration and surface area matter.
AHAs (glycolic, lactic)
D Guidance genuinely splits between sources, and what exists is animal data, not human trials.
Niacinamide, vitamin C
D Presumed fine by long OTC use and low absorption, not by a dedicated pregnancy study.

This is the article we were most nervous about writing, because it's the one where being wrong actually matters. Everywhere else, an overstated grade costs you money or patience. Here it could cost more than that. So we're going to be unusually blunt about where the evidence stops and where it's just consensus, habit, or caution dressed up as certainty.

The short version, before the long version: nobody can ethically run a randomised trial that deliberately exposes pregnant women to a skincare ingredient to see what happens. Everything below is built from observational cohorts, registries, and professional caution. That's a real and useful kind of evidence, just a fundamentally weaker one than what we usually cite elsewhere. Grade the confidence, not the safety.

Topical retinoids: the biggest fear, and the most reassuring data

Oral isotretinoin is a confirmed, unambiguous human teratogen. That means it causes birth defects, not a maybe. That part isn't in dispute, and nothing below applies to it. Topical tretinoin, adapalene and retinol are a completely different exposure: systemic absorption through skin is roughly 2% or less of what an oral dose delivers.

Several observational cohorts have followed women who used topical retinoids in early pregnancy, usually before they knew they were pregnant. A meta-analysis pooling the earlier ones found no significant increase in major malformations or miscarriage. The largest dataset by far is a 2026 Nordic registry study covering 3.87 million births, including 2,172 first-trimester topical retinoid exposures. The malformation rate was 3.3% in exposed pregnancies against 3.0% unexposed, an adjusted risk ratio of 1.10 that did not reach statistical significance. In plain terms, that's a difference small enough that it could plausibly be chance rather than a real effect of the cream.

There's one dissenting data point worth naming rather than burying: an old, small case-control surveillance report flagged a possible association between topical retinoid exposure and a rare brain malformation. It has never been replicated by any of the larger, later cohorts, including the 2026 study above. But it's real, and pretending the evidence is unanimous would be exactly the kind of quiet omission we're trying to avoid.

Put together: several independent cohorts, including a very large recent one, have not found the harm that oral isotretinoin causes. That is meaningfully reassuring. It is not the same as proof of safety, because the trial that would prove it can't be run. So almost every major dermatology and obstetric body still advises stopping topical retinoids in pregnancy as a precaution. That's a guideline-driven "avoid," not an evidence-driven "proven harmful." The distinction is worth knowing, even though the practical advice comes out the same either way: stop using it, and ask your doctor about resuming afterward.

Azelaic acid: the one everyone calls "safe," mostly by default

Azelaic acid is routinely described as pregnancy-safe, and it's often the first swap dermatologists suggest for acne or melasma during pregnancy. We went looking for the study behind that reputation and didn't find one: no human pregnancy-outcome cohort specific to azelaic acid turned up in our search. Its favourable reputation rests on low systemic absorption, animal reproduction data showing no fetal risk, and years of it being the go-to substitute in obstetric dermatology practice without a signal of harm turning up. That's genuinely useful information, just not the same category of evidence as a cohort study. We'd rather tell you that plainly than let "pregnancy-safe" imply a trial that doesn't exist.

Hydroquinone: the one to actually avoid

Hydroquinone is the clearest "don't" on this list, and the reasoning is more about what's missing than what's been found. Topical 2% hydroquinone is absorbed unusually well for a topical ingredient, around 35–45% systemically, far higher than most actives on our Ingredient Index, and there is essentially no human pregnancy-outcome data either way. High absorption plus a total absence of safety data is a bad combination, which is why teratology information services consistently recommend avoiding it in pregnancy. That's a precaution built on an evidence gap, not a demonstrated harm, but a sensible one given there's a genuine alternative (azelaic acid) that doesn't carry the same absorption profile.

Benzoyl peroxide while breastfeeding: the title's other half

This article's title promises breastfeeding as well as pregnancy, and this is the one point where that actually gets its own answer. Benzoyl peroxide has low topical systemic absorption and is broken down quickly once absorbed, which is why it's generally considered compatible with breastfeeding by reference sources that track medication safety during lactation. The honest caveat: that conclusion is inferred from how the drug behaves in the body, not measured directly, because no study has actually tested for it in breastmilk. Reasonable to use, standard advice either way, just built on pharmacology rather than a breastmilk assay.

Salicylic acid: route of exposure is the whole story

The concern people raise about salicylic acid comes from a real but different phenomenon: high-dose oral aspirin late in pregnancy has known risks, including premature closure of a fetal blood vessel. That's a systemic, high-dose, oral exposure. A leave-on serum or cleanser at 0.5–2% salicylic acid delivers a small fraction of that dose through skin. Professional guidance generally treats low-concentration daily use as low risk, while advising against higher concentrations and larger-surface-area professional peels, where the absorbed dose climbs. We could not find a dedicated cohort study isolating cosmetic-strength topical salicylic acid in pregnancy. This is a case of the guidance being built on comparative pharmacology, not a direct trial, and it's worth knowing that's what you're trusting.

AHAs, niacinamide, vitamin C: presumed fine, rarely tested

This is the most honest gap on the page. No cohort study or trial specific to cosmetic-strength niacinamide, vitamin C, or AHAs (glycolic, lactic acid) in pregnancy turned up in our search. The consensus that they're fine rests on decades of OTC use with no reported signal, and on these being naturally occurring, low-absorption compounds. That's reasonable, but not the same as evidence.

AHAs specifically show an actual, live disagreement between sources rather than a quiet consensus: some pregnancy-skincare guidance actively recommends glycolic acid for pregnancy pigmentation, while other sources advise avoiding AHAs, citing animal studies showing skin changes at high concentration. Both positions are working from animal data, not human outcomes. We're reporting the disagreement rather than picking a winner, because that's what's actually there.

Sunscreen: pick mineral if you'd rather not think about it

No trial has tested chemical UV filters for teratogenicity in humans. The consistent professional advice is to prefer mineral filters (zinc oxide, titanium dioxide) in pregnancy, on the basis that some chemical filters, oxybenzone in particular, show measurable systemic absorption in non-pregnant adults, not on the basis of any demonstrated pregnancy harm. This is the easiest swap on this whole list: mineral sunscreen works just as well, per our own sunscreen piece, and it sidesteps a debate that doesn't have a settled answer either way.

The one review worth reading in full

A 2024 two-part peer-reviewed review in the Journal of the American Academy of Dermatology covers dermatologic medication safety in pregnancy and lactation systematically. It's the single most current, most comprehensive source on this exact question. We could only confirm it exists and read its abstract. The full text sits behind a paywall we couldn't clear. If you or your doctor can access it, it's the best next step past this article.

The short version

Stop oral isotretinoin and hydroquinone. Azelaic acid, low-concentration salicylic acid, and mineral sunscreen are reasonable defaults with no signal of harm, even though the direct evidence behind "safe" is thinner than the word implies. Topical retinoids sit in the middle: several large cohorts are reassuring, but the standard advice is still to pause them and ask your dermatologist or obstetrician about timing. None of this is a substitute for that conversation. It's what to bring to it.

This is general education, not medical advice. Nothing here accounts for your history, your medication or your skin, and none of it replaces assessment by a qualified clinician. Where we describe what research found, that is a description of a study — not a promise about you.

Revision history

  • Added a breastfeeding-specific section on benzoyl peroxide. The article's title promised breastfeeding coverage and already carried a LactMed reference for it, but a citation-integrity pass found that reference was never actually cited anywhere in the body — the breastfeeding half of the title had no matching content. Added the missing section and glance row rather than just wiring up the existing reference to a throwaway line.

References

  1. Refsum E, Furu K, Cesta CE, et al. Topical retinoid use in women of reproductive age and risk of major congenital malformations in exposed pregnancies: a Nordic cohort study. British Journal of Dermatology, 2026.BRegistry-based cohort, ~3.87 million births, 2,172 first-trimester topical retinoid exposures·The largest dataset to date. Malformation rate 3.3% exposed vs 3.0% unexposed, with an adjusted risk ratio of 1.10, not statistically significant. Registry-based, so exposure and adherence misclassification is possible.PubMed 41365815DOI 10.1093/bjd/ljaf500
  2. Kaplan YC, et al. Pregnancy outcomes following first-trimester exposure to topical retinoids: a systematic review and meta-analysis. British Journal of Dermatology, 2015.BSystematic review and meta-analysis of observational cohorts·No significant increase in major malformations or spontaneous abortion pooled across the available cohorts. This is built entirely from observational data, since no randomised trial is ethical here.DOI 10.1111/bjd.14053
  3. Wals E, et al. Association between holoprosencephaly and exposure to topical retinoids: results of the EUROCAT survey. Paediatric and Perinatal Epidemiology, 1991.DCase-control surveillance report·The one dissenting signal in this literature. It's small, old, and not replicated by any later cohort, including the much larger 2026 Nordic study, but real and worth disclosing rather than smoothing over.PubMed 1754503
  4. UK Teratology Information Service Exposure to Hydroquinone in Pregnancy. UKTIS monograph, v3, 2024.BTeratology information service guideline, not a trial·States plainly that human data to assess teratogenic risk are lacking, and recommends avoidance in pregnancy as a precaution rather than on demonstrated harm. Cites ~35–45% systemic absorption from topical 2% hydroquinone.Read the source
  5. National Library of Medicine (LactMed) Benzoyl Peroxide — Drugs and Lactation Database. NCBI Bookshelf, NBK501421, 2024.CPharmacokinetic monograph, not a study·Reports low topical systemic absorption and rapid metabolic clearance, but no published data on actual breastmilk transfer exist, so the low-risk conclusion is inferred from absorption pharmacology, not measured directly.Read the source
  6. American Academy of Dermatology Dermatologist-approved pregnancy skin care. aad.org patient guidance, 2025.CProfessional-society patient education page, not a trial·The closest thing to organised guidance found on this topic. Recommends mineral sunscreen and limiting salicylic acid concentration, and flags that AHA caution is based on animal, not human, safety data.Read the source
  7. McMullan SC, et al. Safety of dermatologic medications in pregnancy and lactation: an update — Part I: Pregnancy. Journal of the American Academy of Dermatology, 2024.BPeer-reviewed expert review·The most current dermatology-specific review of this exact question. We could confirm it exists and read its abstract, but the full text sat behind a paywall we couldn't clear before publishing this piece.PubMed 38280679

Your turn

Did this hold up? Tell us honestly — especially if it didn’t.

Anonymous. Stored in your browser.

The Sunday study

One paper a week, explained properly

We pick a single new study, tell you what it measured, how many people it measured it in, and whether it should change anything you're doing — plus the two articles we published that week. No panic, and no hype.

We don't recommend products yet. When we do, they'll be things you can actually buy in India, any affiliate link will be labelled as one, and a commission will never buy a recommendation.

One email each Sunday. Unsubscribe in one click.