Understanding variation in prevalence estimates of polycystic ovary syndrome: a systematic review and meta-analysis

Research output: Contribution to journalArticleResearchpeer-review

128 Citations (Scopus)


BACKGROUND: Polycystic ovary syndrome (PCOS) prevalence estimates vary when different diagnostic criteria are applied. Lack of standardization of individual elements within these criteria may contribute to prevalence differences.

OBJECTIVE AND RATIONALE: A systematic review of studies reporting prevalence of PCOS, using at least one of the National Institutes of Health (NIH), Rotterdam or Androgen Excess Society (AE-PCOS) criteria, was conducted. The aim was to investigate the impact on prevalence reporting of different definitions of the clinical elements for PCOS diagnosis.

SEARCH METHODS: A systematic search of Ovid MEDLINE, EMBASE, Cochrane Central Register of Controlled Trials (CENTRAL), Emcare and BIOSIS was conducted. The search was limited to English language and studies published between January 1990 and January 2018. Included articles needed to define PCOS by at least one of the NIH, Rotterdam or AE-PCOS criteria, be of an unselected population and be published as a full text article. Risk-of-bias was assessed.

OUTCOMES: A total of 21 studies met the inclusion criteria. The random-effects pooled prevalence of PCOS in studies that used the NIH criteria (7% [95% CI: 6-7%]), was significantly different from that identified in studies that used the Rotterdam criteria (12% [95% CI: 10-15%], P < 0.0001) but not studies that used the AE-PCOS criteria (10% [95% CI: 6-13%], P = 0.075). The pooled estimates for Rotterdam and AE-PCOS were not significantly different from each other (P = 0.201). Pooled prevalence estimates were compared between studies separated on the basis of: oligo-amenorrhoea vs oligo-amenorrhoea plus short cycles, clinical androgen excess requiring hirsutism vs any clinical androgen excess, use of different versions and cut-offs for the Ferriman-Gallwey (F-G) score, and inclusion vs non-inclusion of oral contraceptive users. There were no statistically significant differences for any of these comparisons. There was insufficient information to allow subgroup analyses of definitions of polycystic ovaries.

WIDER IMPLICATIONS: Inclusion of ovarian morphology results in statistically significantly higher pooled prevalence estimates for PCOS. Heterogeneity in prevalence estimates for PCOS reflect the broad clinical spectrum of the condition, lack of standardization of the elements within each set of diagnostic criteria and the use of a range of diagnostic cut-offs, as well as potential differences between study populations. The use of different definitions for anovulation and clinical androgen excess did not appear to contribute to differences in the estimated prevalence of PCOS in this study. However, as the number of studies in most of the comparison groups was small, real differences may have been missed. Uncertainty surrounding the diagnosis of PCOS urgently needs to be addressed in order to provide clinicians and their patients with greater diagnostic certainty, and hence reduce inappropriate labelling and the potential psychological harm that may accompany misdiagnosis.

Original languageEnglish
Pages (from-to)694-709
Number of pages16
JournalHuman Reproduction Update
Issue number6
Publication statusPublished - 1 Nov 2018

Cite this