Rapid review · 27 Sep 2026

Does skin change across the menstrual cycle?

AI-assisted rapid review Following Cochrane rapid-review guidance. Directed by Sonali Krishnani; the protocol, screening, extraction and ratings were done by AI agents (Claude). No human second reviewer; not peer-reviewed; not medical advice.

Question 1In people with natural cycles, does facial skin differ between cycle phases?

Question 2Has anyone tested whether timing skincare to the cycle beats a steady routine?

484records screened from PubMed, ClinicalTrials.gov, Consensus and reference lists
47studies included
κ 0.87agreement between two independent AI screeners (97%)
0studies or registered trials of cycle-timed vs steady skincare

Answer to question 1

Only acne shows a fairly consistent pattern: many people with acne say it flares around their period, most often the week before. For oil, hydration, barrier, sensitivity, colour and skin bacteria, the studies are small and conflicting, and measured changes are small.

Answer to question 2

It has never been tested. No study or registered trial compares cycle-timed skincare with a steady routine. Untested isn't the same as harmful, but it isn't evidence either.

Summary of findings

Certainty uses GRADE. All studies are observational, so each outcome starts at low and drops for bias, inconsistency, indirectness or small samples.

Outcome	Studies (people)	What the evidence shows	Certainty	What the app says
Acne flares, self-reported	21 (~7,400)	Many women with acne report flares around their period; median about 60% (range 12–98%). The week before is most common.	Low	"Many people with acne break out more before a period"
Acne spots, counted	3 (152)	Two small studies found about 25% more inflamed spots before the period; a third saw no change.	Very low	Supporting evidence only
Oil (sebum)	7 (~100)	Conflicting. The best facial study (51 women) found no difference.	Very low	No oiliness claims by phase
Skin barrier (water loss)	4 (36)	Slightly more water loss after ovulation (9.9 vs 8.9 g/h/m²); mostly non-facial sites.	Very low	"In one small study…"
Hydration	6 (~150)	Conflicting: no difference, higher at ovulation, or driest days 1–6.	Very low	"In one small study…"
Sensitivity and irritation	5 (95)	4 of 5 found no difference; one (29 women) found a stronger irritant reaction on day 1.	Very low	"In one small study… treat this as a maybe"
Colour and redness	3 (29)	Redness varies, but by less than the eye can see.	Very low	"Ovulation glow" listed as a myth
Skin bacteria	3 (95)	No phase difference in the sequencing studies.	Very low	No claims
Cycle-timed vs steady skincare	0 direct	Never tested. One indirect study (30 women, peel timing) can't answer it.	No evidence	"Timing tips, not rules"

Method

Protocol first: question, eligibility, sources and analysis were written down before any screening, with a log of every deviation.
Search with life-science connectors: PubMed (473 records), ClinicalTrials.gov, Consensus and medRxiv, plus reference lists of key papers.
Two independent AI screeners: separate AI agents each judged all 484 records without seeing the other's decisions (97.1% agreement, Cohen's κ 0.87). A third AI agent settled the 26 undecided records.
Two independent AI extractors: separate AI agents each recorded design, size, how phase was measured, results and risk of bias for all 47 studies. They agreed on the direction of results 90% of the time; a third AI agent checked every difference against the paper.
Certainty: GRADE per outcome, drafted by an AI agent.
Sonali's role: asked for the review, set the evidence rules the app follows, and decided what the app should and shouldn't say based on the findings. She did not screen or extract studies herself.

How the studies were found

473PubMed

9Consensus

2ClinicalTrials.gov

0medRxiv

6Reference lists

↓ 6 duplicates removed

484Records screened by two independent reviewers

415Excluded by both

↓ 26 undecided → third reviewer

13Excluded after reading

9Couldn't be obtained

↓

47Studies included (43 agreed + 4 from the third reviewer)

What this changed in the app

Phase headlines about sensitivity and hydration now start "In one small study…", because only one study each supports them.
The acne headline stays: it's the one finding with consistent support.
The app's core message, "timing tips, not rules", is backed by the finding that cycle-timed skincare has never been tested.
Earlier checks the same day cut overstatements, fixed two wrong citations, and gave every tip at least one source.

Limits

AI agents screened and extracted; there was no human second reviewer.
33 of 47 studies could only be read as abstracts; only 4 confirmed cycle phase with hormone tests, and only 3 excluded people on the pill.
71 non-English records were excluded; Consensus was capped at 3 results per search; preprints were effectively unsearched.
The two screeners briefly shared a scratch file; this is logged in the protocol.

Data and files

Key studies

Nguyen et al. (2024). Physiological changes in women's skin during the menstrual cycle: a scoping review. PMC11703644
Stoll et al. (2001). The effect of the menstrual cycle on acne. PMID 11712049
Lucky (2004). Quantitative documentation of a premenstrual flare of facial acne in adult women. PMID 15096370
Geller et al. (2014). Perimenstrual flare of adult acne. JCAD
Agner et al. (1991). Menstrual cycle and skin reactivity. PMID 2033132
Nikoletić et al. (2025). Menopause, menstrual cycle, and skin barrier function. PMC12206585, with correction
Burriss et al. (2015). Facial skin colour over the ovulatory cycle is not detectable by the human eye. PLOS ONE
Garritty et al. (2021). Cochrane Rapid Reviews Methods Group interim guidance. J Clin Epidemiol

All 47 included studies, with links, are in the evidence table.

Cite this review: Krishnani, S. (2026). Skin Cycle: a skincare diary app and an AI-assisted rapid review of skin across the menstrual cycle. Zenodo. doi:10.5281/zenodo.23024631

See the demo → Read the case study