Study question: What is the prevalence in a normal population of polycystic ovary syndrome (PCOS) according to the Rotterdam criteria versus revised criteria including anti-Müllerian hormone (AMH)?

Summary answer: The prevalence of PCOS was 16.6% according to the Rotterdam criteria. When replacing the criterion for polycystic ovaries by antral follicle count (AFC) > 19 or AMH > 35 pmol/l, the prevalence of PCOS was 6.3 and 8.5%, respectively. WHAT IS KNOWN ALREADY?: The Rotterdam criteria state that two out of the following three criteria should be present in the diagnosis of PCOS: oligo-anovulation, clinical and/or biochemical hyperandrogenism and polycystic ovaries (AFC ≥ 12 and/or ovarian volume >10 ml). However, with the advances in sonography, the relevance of the AFC threshold in the definition of polycystic ovaries has been challenged, and AMH has been proposed as a marker of polycystic ovaries in PCOS.

Study design, size, duration: From 2008 to 2010, a prospective, cross-sectional study was performed including 863 women aged 20-40 years and employed at Copenhagen University Hospital, Rigshospitalet, Denmark.

Participants/material, setting, methods: We studied a subgroup of 447 women with a mean (±SD) age of 33.5 (±4.0) years who were all non-users of hormonal contraception. Data on menstrual cycle disorder and the presence of hirsutism were obtained. On cycle Days 2-5, or on a random day in the case of oligo- or amenorrhoea, sonographic and endocrine parameters were measured.

Main results and the role of chance: The prevalence of PCOS was 16.6% according to the Rotterdam criteria. PCOS prevalence significantly decreased with age from 33.3% in women < 30 years to 14.7% in women aged 30-34 years, and 10.2% in women ≥ 35 years (P < 0.001). In total, 53.5% fulfilled the criterion for polycystic ovaries with a significant age-related decrease from 69.0% in women < 30 years to 55.8% in women aged 30-34 years, and 42.8% in women ≥ 35 years (P < 0.001). AMH or age-adjusted AMH Z-score was found to be a reliable marker of polycystic ovaries in women with PCOS according to the Rotterdam criteria [area under the curve (AUC) 0.994; 95% confidence interval (CI): 0.990-0.999] and AUC 0.992 (95% CI: 0.987-0.998), respectively], and an AMH cut-off value of 18 pmol/l and AMH Z-score of -0.2 showed the best compromise between sensitivity (91.8 and 90.4%, respectively) and specificity (98.1 and 97.9%, respectively). In total, AFC > 19 or AMH > 35 occurred in 17.7 and 23.0%, respectively. The occurrence of AFC > 19 or AMH > 35 in the age groups < 30, 30-34 and ≥ 35 years was 31.0 and 35.7%, 18.8 and 21.3%, and 9.6 and 18.7%, respectively. When replacing the Rotterdam criterion for polycystic ovaries by AFC > 19 or AMH > 35 pmol/l, the prevalence of PCOS was 6.3 or 8.5%, respectively, and in the age groups < 30, 30-34 and ≥ 35 years, the prevalences were 17.9 and 22.6%, 3.6 and 5.6%, and 3.6 and 4.8%, respectively.

Limitations, reason for caution: The participants of the study were all health-care workers, which may be a source of selection bias. Furthermore, the exclusion of hormonal contraceptive users from the study population may have biased the results, potentially excluding women with symptoms of PCOS.

Wider implications of the findings: AMH may be used as a marker of polycystic ovaries in PCOS. However, future studies are needed to validate AMH threshold levels, and AMH Z-score may be appropriate to adjust for the age-related decline in the AFC.

Study funding/competing interest(s): None.

Trial registration number: Not applicable.