PCOS and Miscarriage: What the Rate Actually Is, and What Is Modifiable

What Is the Miscarriage Rate With PCOS?

PCOS raises the odds of miscarriage by about half: an odds ratio of 1.49 (95% CI 1.20–1.85) across 44 studies in a 2024 meta-analysis of 106,690 pregnancies — 17,384 in women with PCOS and 89,306 without. That is a relative figure, and relative and absolute are not the same number: applied to the widely-cited general-population estimate that roughly 10% of confirmed pregnancies end in early loss, a 49% relative increase works out to approximately 14–15% in absolute terms, not 59%. Most pregnancies in PCOS, including yours if you are reading this while pregnant, do not end in miscarriage.

This page also covers two questions that search as their own thing but are really the same question at a different resolution: recurrent miscarriage in PCOS, and what actually happens in an early pregnancy loss mechanically. Both are addressed below rather than split into separate pages, because the honest answer to any one of them needs the other two for context. Miscarriage is one piece of the site’s broader fertility guide, and it sits closest to the pregnancy-risk picture covered here, which handles the other three risks — gestational diabetes, hypertensive disorders and preterm birth — that the same 2024 dataset measured.

You may see PCOS written as polyendocrine metabolic ovarian syndrome (PMOS), after a 2026 global consensus of more than 50 organisations renamed it. Every figure on this page applies under either name — only the label changed.

How Common Is Recurrent Miscarriage in PCOS, Really?

Two studies at very different scales tell a more complicated story than “PCOS causes recurrent miscarriage.” In a UK cohort of 2,199 women with a history of recurrent miscarriage — three or more losses, median age 33 — 40.7% had polycystic ovaries on ultrasound. That sounds like strong evidence of a link, until the same study followed 486 of those women through an untreated, spontaneously-conceived next pregnancy and found the live birth rate was statistically no different: 60.9% in women with polycystic ovaries versus 58.5% in women with normal ovarian morphology. Elevated LH or elevated testosterone — the two markers earlier research had blamed — were not associated with a higher miscarriage rate either.

A separate, smaller UK study applied the full Rotterdam diagnostic criteria, not ultrasound appearance alone, to 300 women with recurrent miscarriage. Using hormone levels, cycle history and ultrasound together, only 8.3–10% actually met the criteria for PCOS — a much lower figure than the study’s own introduction says had “long been accepted,” and the authors concluded the prevalence of PCOS in recurrent miscarriage is considerably lower than previously assumed once a precise diagnosis, not just an ultrasound appearance, is required.

Read together, these two studies are not a contradiction — they are a lesson in vocabulary. “Ovaries that look polycystic on a scan” and “polycystic ovary syndrome” are not the same diagnosis, and the two studies above measured different things. Scan appearance alone (found in about 4 in 10 women with recurrent miscarriage) does not predict a worse outcome in a future pregnancy. A full PCOS diagnosis is present in a meaningfully smaller share of recurrent-miscarriage cases than the scan-appearance number implies.

What Is Actually Happening in an Early Pregnancy Loss

Chromosomal abnormality in the embryo remains the single most common cause of early miscarriage in the general population, and nothing in the PCOS literature above overturns that. What PCOS adds, per the mechanisms proposed across the studies above, is a modest additional burden from insulin resistance and androgen exposure on the uterine lining’s receptivity to an embryo implanting — plausible biology, not a settled, quantified mechanism, since no trial has isolated it directly from the elevated OR itself. The older theory that elevated LH hypersecretion drove PCOS miscarriage specifically was tested and not supported in the 2,199-woman UK cohort above — a good example of evidence overturning a mechanism that used to be taught as settled. The earliest-stage category of loss — a positive test that reverses before a pregnancy is ever seen on ultrasound — is common enough on its own to deserve separate treatment; how often a chemical pregnancy happens, and what one loss does and doesn’t predict for the next cycle, is covered here.

What Is Modifiable — and What the Evidence Actually Supports

Obesity is the modifiable factor with the clearest evidence, and it is worth separating from PCOS itself. A systematic review of 28,538 women pooled across six studies found a higher miscarriage rate in obese versus normal-weight women conceiving spontaneously — 13.6% versus 10.7%, an odds ratio of 1.31 (95% CI 1.18–1.46) — and, in a smaller subset, a higher rate of recurrent early miscarriage in obese versus normal-weight women, 0.4% versus 0.1% (OR 3.51, 95% CI 1.03–12.01), though that wide confidence interval reflects a genuinely small number of events and should be read as a signal, not a precise number. This evidence is not PCOS-specific — it applies to anyone conceiving spontaneously — but a large share of PCOS pregnancies also carry a higher average BMI, which is why the two are worth discussing as separate, additive factors with a clinician rather than assuming either one alone explains an individual’s risk. Insulin resistance is the plausible biological bridge between the two, and what that mechanism is doing outside pregnancy is covered here, but no trial has cleanly separated the weight effect from a PCOS-specific insulin effect on miscarriage risk.

Does Metformin Reduce Miscarriage Risk in PCOS? The Evidence Is Genuinely Contested

This is the clearest example on this page of two real studies pointing in different directions, and the honest answer is to report both rather than pick the tidier one.

A retrospective study of 99 women with PCOS at a single US academic endocrinology clinic — 68 pregnancies on metformin throughout pregnancy, 31 pregnancies without it — found an early pregnancy loss rate of 8.8% on metformin versus 41.9% without it (p < 0.001), and an even larger gap in the subgroup with a prior miscarriage: 11.1% versus 58.3%. Taken at face value, that looks like one of the largest effect sizes in this entire field.

It was not replicated at anything close to that scale when tested properly. The PregMet2 randomised, placebo-controlled trial — the type of study designed specifically to remove the selection bias a retrospective clinic study cannot rule out — tested metformin through pregnancy against a combined outcome of late miscarriage and preterm birth and found 5% on metformin versus 10% on placebo. That gap did not reach statistical significance in the trial on its own, though pooling it with two earlier, smaller trials did.

What metformin does and does not do for fertility more broadly, outside of this specific pregnancy-loss question, is covered in full here — the two questions get asked together often, and the evidence for each stands separately.

Who This Doesn’t Apply To — What a Loss Does Not Say About You

None of the numbers above are a verdict on an individual pregnancy, and this section exists because miscarriage and recurrent-loss content reaches people while they are grieving, not just while they are researching. A few things worth stating plainly:

• A miscarriage is not something you did. Chromosomal abnormality remains the most common cause of early pregnancy loss in the general population, PCOS or not, and nothing in the data above shifts that.
• An elevated relative risk is not a forecast. Most PCOS pregnancies — including recurrent ones after a loss — go on to a live birth; the 2,199-woman UK cohort above found essentially no difference in that outcome once women reached a treated or spontaneous next pregnancy.
• This page is not the place to work through grief. If you are processing a loss, that is a conversation for a GP, a midwife, or a bereavement-focused support service — not something to resolve by reading further statistics.

For a fuller picture of what else this 2024 dataset found — gestational diabetes, hypertensive disorders and preterm birth, all with the same relative-versus-absolute distinction applied — see PCOS in pregnancy: the four risks worth monitoring. And if the concern behind a miscarriage search is really about egg quality rather than the pregnancy itself once conceived, that is a separate question with its own evidence, covered in full there.