The Link Between PCOS and Autism: What the Research Shows

Population-level studies consistently find that children born to mothers with polycystic ovary syndrome (PCOS) have a modestly elevated risk of being diagnosed with autism, with the increase in odds typically falling in the range of 35 to 59 percent after adjusting for other factors. The association is real in statistical terms, but it is not large enough to be predictive for any individual pregnancy, and researchers are still working out which biological pathways actually drive it. What makes the relationship especially interesting is that it appears to run in both directions: women with autism themselves are more likely to have PCOS.

What the Large Studies Actually Found

The most cited study on this topic drew on Swedish national health registries covering over a million births. After adjusting for factors like parental age, parental psychiatric history, and birth complications, maternal PCOS was associated with a 59 percent increase in the odds of an autism diagnosis in the child. When the mother had both PCOS and obesity, the odds roughly doubled compared to the general population.¹ That finding held for both boys and girls, which was somewhat surprising given that autism diagnoses skew heavily male in the general population.

A separate set of studies published in Translational Psychiatry confirmed the direction of the effect but found a somewhat smaller adjusted estimate. After controlling for maternal psychiatric diagnoses, obstetric complications, and metabolic conditions, the odds of having a child with autism were about 35 percent higher for mothers with PCOS.² A later meta-analysis that pooled data across multiple studies landed on a similar figure, estimating the adjusted odds at about 40 percent higher for children of mothers with PCOS.³

These are not small studies. They are drawn from national health databases covering hundreds of thousands or millions of people, which gives them statistical power that smaller clinical studies lack. But the absolute risk remains low. Autism prevalence in the general population is roughly 1 to 2 percent depending on the country and diagnostic criteria. A 40 to 59 percent relative increase on that base rate means going from, say, 1.5 percent to roughly 2 to 2.4 percent. That distinction matters: a relative increase sounds alarming, but the actual numbers describe a modest shift in probability, not a likely outcome for any given child.

The Connection Goes Both Ways

One of the more revealing findings is that the link between PCOS and autism is not just about mothers and children. Women who are themselves autistic appear to have PCOS at higher rates than the general female population. In one analysis, about 2.3 percent of autistic women had PCOS, compared to 1.1 percent in a control group.⁴ The mirror image also held: women already diagnosed with PCOS were nearly twice as likely to have an autism diagnosis themselves, though the absolute rate was still quite low at 0.17 percent versus 0.09 percent.

Earlier work had also flagged this pattern. A survey-based study found that women with autism reported higher rates of several androgen-related conditions, including PCOS, hirsutism (excess body hair), severe acne, and irregular menstrual cycles.⁵ That finding was early evidence that something about androgen biology might overlap between the two conditions, beyond what you would expect by chance.

This bidirectional relationship is a strong hint that shared biology is involved, not just an environmental exposure during pregnancy. If the only explanation were that the in-utero environment created by PCOS somehow caused autism, you would not expect autistic women to have unusually high rates of PCOS themselves. The fact that both conditions cluster in the same individuals and families points toward overlapping genetic or hormonal pathways.

The Prenatal Testosterone Theory

The most widely discussed explanation centers on androgens, particularly testosterone. PCOS is defined in part by excess androgen production, and the fetus developing in the womb of a woman with PCOS may be exposed to higher-than-typical levels of these hormones during critical windows of brain development. This fits with a broader theory in autism research proposing that higher prenatal exposure to androgens nudges brain development in directions associated with autistic traits.

Some evidence supports this idea. A study measuring testosterone levels in amniotic fluid found that higher fetal testosterone was associated with more autistic traits in children later on, and this held within each sex, not just between boys and girls.⁶ In a Danish cohort, higher maternal testosterone in the third trimester was associated with increased autism-related traits in boys specifically.⁷ Animal research has also shown that exposing pregnant animals to excess testosterone alters placental function and steroid production, suggesting a plausible biological pathway from maternal androgens to fetal effects.⁸

But the picture is not clean. A study of adolescents found no statistical association between prenatal testosterone levels and autistic traits, in contrast to the childhood findings.⁹ And a broad systematic review that examined 25 outcomes related to the androgen hypothesis found that only 2 supported it, while 4 actually pointed in the opposite direction.¹⁰ The authors of that review concluded that genetics and confounding pregnancy factors are likely more important than a simple story about sex hormones.

In short, prenatal testosterone exposure is probably part of the picture for some individuals, but it does not come close to explaining the full association on its own. Researchers who study this connection generally acknowledge that the androgen theory was a useful starting point that has turned out to be too simple.

Metabolic and Inflammatory Pathways

PCOS is not just an androgen disorder. It comes bundled with insulin resistance, chronic low-grade inflammation, and a higher risk of gestational diabetes. Each of these metabolic features has its own potential route to affecting fetal brain development, independent of testosterone.

Insulin resistance is present in a large share of women with PCOS and drives many of the syndrome’s downstream effects. High insulin levels stimulate the ovaries to produce more androgens and suppress the liver’s production of the protein that binds and inactivates testosterone, so insulin resistance amplifies the androgen excess that is already a hallmark of PCOS.¹¹ But high glucose levels in the womb also appear to cause problems on their own, possibly by triggering inflammation in the developing fetal brain. This means insulin resistance could be influencing offspring neurodevelopment through two channels at once: by worsening androgen excess and by creating a high-glucose, pro-inflammatory uterine environment.

A Finnish cohort study found that when mothers had both PCOS and obesity, the range of neuropsychiatric outcomes in their children was broader than with PCOS alone, encompassing not just autism but also ADHD and other behavioral conditions. The researchers interpreted this as evidence that obesity compounds the hostile metabolic environment created by PCOS, adding lipid-related stress and chronic inflammation on top of hormonal disruption.¹² The Swedish registry study mentioned earlier found the same amplification effect: the combination of PCOS and obesity roughly doubled the odds of autism in the offspring, compared to about a 59 percent increase with PCOS alone.¹

This metabolic angle helps explain why the androgen theory alone falls short. PCOS creates a package of hormonal, metabolic, and inflammatory disturbances, and disentangling which one matters most in any given pregnancy may not be possible with the tools researchers currently have.

Genetics Likely Tie the Two Conditions Together

A purely environmental explanation, where something about the PCOS pregnancy causes autism in the child, has always been incomplete. The bidirectional overlap between the two conditions in the same individuals points strongly to shared genetic architecture. If the same gene variants increase susceptibility to both PCOS and autism, you would expect exactly the kind of co-occurrence and family clustering that the epidemiological studies report.

No single “PCOS-autism gene” has been identified, and given the complexity of both conditions, there probably will not be one. Both PCOS and autism are polygenic, meaning hundreds of genetic variants each contribute a small amount to risk. Some of those variants could plausibly affect overlapping biological systems, including androgen signaling, insulin sensitivity, and neurodevelopmental pathways. The meta-analysis that pooled studies across multiple countries noted that even after adjusting for a long list of confounders, the association persisted, suggesting it is not fully explained by shared environmental exposures like diet or socioeconomic status.³ That residual association could well be genetic.

Interestingly, one systematic review that tested whether sex hormones alone could explain the overlap between autism and conditions like PCOS concluded that genetic influences likely override simple hormonal associations in many cases.¹⁰ This does not rule out hormones as one contributing factor, but it suggests that genetics may be doing more of the heavy lifting than the prenatal testosterone theory originally assumed.

Epigenetic Changes in the Placenta

Between the mother’s blood and the fetus sits the placenta, and emerging research suggests it plays an active role in how PCOS affects fetal brain development. A study examining placentas from women with PCOS found altered patterns of DNA methylation, the chemical marks that influence whether genes are turned on or off, compared to placentas from unaffected women. Among the genes with altered expression were ones involved in neural function.¹³

Separately, researchers have proposed that excess androgens in PCOS may alter the function of BDNF, a signaling molecule critical for brain development and plasticity. Androgen excess could change how BDNF is produced or processed through epigenetic modifications, compromising the signaling that neurons depend on during development.¹⁴ This is still a relatively new line of research, and much of it involves animal models or small human samples. But it provides a plausible route for how the hormonal and metabolic disruptions of PCOS could leave lasting marks on fetal gene regulation without changing the DNA sequence itself.

PCOS Is Linked to More Than Just Autism

The meta-analysis that estimated the autism-specific odds also examined other neuropsychiatric outcomes. Children of mothers with PCOS had elevated odds of ADHD, anxiety disorders, and other behavioral symptoms, each with an increase roughly similar to the autism estimate, in the range of 33 to 45 percent.³ A more recent systematic review echoed this finding, concluding that the link extends broadly to neuropsychiatric and neurodevelopmental conditions rather than being specific to autism.¹⁵

This matters for interpretation. If PCOS uniquely increased autism risk and nothing else, you might look for a mechanism specific to the social-communication circuitry that autism affects. But the fact that the risk increase is broad, touching ADHD, anxiety, and behavioral symptoms generally, suggests a more general vulnerability in neurodevelopment rather than a targeted effect on one brain circuit. The metabolic and inflammatory disruptions described earlier are good candidates for that kind of diffuse influence on brain development.

Gender Diversity in Adolescents With PCOS

An unexpected thread in this research involves gender identity. A study of adolescents enrolled in a multidisciplinary PCOS clinic found that about 7.6 percent of PCOS patients self-identified as gender diverse, compared to 1.8 percent of non-PCOS youth in the same clinical setting. Gender-diverse PCOS adolescents also had significantly higher rates of depression and anxiety than their cisgender PCOS peers. However, gender diversity in these patients was associated with higher hirsutism scores but not with higher androgen levels, complicating any simple hormonal explanation.¹⁶

This finding sits at the intersection of several overlapping patterns. Autism is already known to be associated with higher rates of gender diversity, and PCOS appears to be as well. Whether these are connected through shared neurodevelopmental pathways, shared hormonal biology, or some other mechanism is an open question. But for clinicians seeing adolescents with PCOS, the finding highlights that screening for mental health concerns and being open about gender identity may be more important than in the general adolescent population.

What This Means If You Have PCOS and Are Planning a Pregnancy

The practical implications of this research are modest but worth understanding. A 35 to 59 percent relative increase in odds sounds dramatic, but it translates to a small absolute shift in risk. If the background rate of autism is around 1.5 percent, even the higher estimates would put the rate among children of mothers with PCOS at somewhere around 2 to 2.5 percent. The vast majority of children born to mothers with PCOS will not be autistic.

That said, managing the metabolic components of PCOS before and during pregnancy makes sense for many reasons, and the neurodevelopmental data add one more. Insulin resistance and obesity appear to amplify whatever risk PCOS creates, based on the studies that found substantially higher odds when both PCOS and obesity were present. Managing blood sugar, maintaining a healthy weight where possible, and treating gestational diabetes aggressively are standard recommendations for PCOS pregnancies, and the neuropsychiatric research aligns with rather than adding to those guidelines.

There is no established intervention to specifically reduce the neurodevelopmental risk associated with PCOS. One case study explored whether metformin, a drug commonly prescribed for insulin resistance in PCOS, might affect autistic traits, but found no change in social behavior scores after treatment.¹⁷ That was a single case, and no one would draw clinical conclusions from it, but it illustrates how far the field is from targeted prevention.

For women who have both PCOS and autism, or who have autistic children and are wondering about their own health, the bidirectional overlap is worth discussing with a doctor. Autistic women may benefit from being screened for PCOS-related symptoms, including irregular cycles, insulin resistance, and androgen-related conditions, since these are likely underdiagnosed in neurodivergent populations where healthcare access and communication barriers can be significant. The elevated rates of PCOS in autistic women are not something most general practitioners are aware of, and awareness alone can improve care.

Why the Research Has Been Slow to Mature

Given the size and consistency of the epidemiological signal, it is fair to ask why the mechanistic understanding remains so fragmented. Several factors explain the pace. First, both PCOS and autism are heterogeneous conditions. PCOS can present with or without obesity, with or without insulin resistance, and with varying degrees of androgen excess. Autism spans a wide range of presentations and severities. Studying the link between two conditions that each vary enormously within themselves makes it hard to isolate clean causal pathways.

Second, the exposures in question, hormonal and metabolic disruptions in the womb, are difficult to study directly in humans. Researchers rely on indirect measures like maternal blood testosterone levels in the third trimester or amniotic fluid samples collected during amniocentesis, neither of which captures the full hormonal environment the fetus actually experienced during the critical early weeks of brain development. The Danish cohort study, for instance, found that the link between maternal PCOS and offspring autism traits became non-significant once parental psychiatric diagnoses were excluded from the sample, which complicates the causal story further.⁷

Third, studies that rely on mental health screening through clinics that also happen to treat PCOS may capture higher rates of psychiatric diagnoses simply because those women are already in the healthcare system and being evaluated more thoroughly. Surveillance bias is a chronic problem in studies linking one medical condition to another. The large registry-based studies partly address this by including the entire population, but they depend on diagnostic codes that may undercount autism, especially in girls and women.