Driven by good intentions: why widening the diagnostic criteria for polycystic ovary syndrome may be harming women

The 2018 International Guidelines for polycystic ovary syndrome (PCOS) were intended to standardise diagnosis and improve care, but they also endorsed the controversial Rotterdam diagnostic criteria. These include a larger pool of individuals than alternative definitions and turn many more women into PCOS patients. A growing body of research shows that labelling some women with this condition can have harmful, lifelong consequences.

The previous 1990 National Institutes of Health definition of PCOS required both (a) oligo-/anovulation (eg, irregular menstrual cycles) and (b) signs of hyperandrogenism (eg, excess hair growth, acne, or excess male hormones/androgens). The 2003 Rotterdam definition added a third criterion: (c) polycystic ovaries on ultrasound (enlarged ovaries with lots of small follicles), but required only any two of the three criteria to meet the definition. This introduced four categories or subtypes that now fall within the scope of the PCOS label (ie, a+b, b+c, a+c, and a+b+c), dramatically increasing prevalence (e.g., estimates doubling from 9% to 18% in Australian women aged 27-34). The 2018 Guidelines made several minor modifications to the Rotterdam criteria. The ultrasound criteria have been tightened to reflect the greater sensitivity of modern imaging technology. Ultrasounds are also no longer recommended in adolescents because of the overlap with normal ovarian physiology. While these specific changes reduce the risk of overdiagnosis, the newly endorsed criteria still capture many women with few or mild symptoms. The lowered score for hirsutism (excess hair growth) in the 2018 guidelines has also raised concerns about overdiagnosis. 

PCOS is associated with an increased risk of infertility and metabolic complications including insulin resistance, obesity, metabolic syndrome, and type 2 diabetes. However, the potential consequences are not the same for all women diagnosed. Cross-sectional studies consistently show that the non-hyperandrogenic subtype (meeting criteria a+c) does not have the same increased risk of adverse metabolic complications found in the hyperandrogenic subtypes (a+b, b+c, a+b+c), indicating different clinical courses. Furthermore, diagnostic criteria of this subtype overlap with hypothalamic amenorrhea (absence of menstruation triggered by stress, overexercising, or undereating), resulting in women being misdiagnosed with PCOS and not receiving care for their actual condition.

Given evidence of the disparate metabolic risk profiles between subtypes and the potential for overdiagnosis, why did the recent guidelines endorse the broader Rotterdam criteria? No rationale was provided. Probably it is driven by good intentions—trying to ensure nobody misses out on potentially beneficial care (eg, counselling regarding fertility). However, we could find no acknowledgement in the 2018 guidelines of the evidence suggesting that the non-hyperandrogenic subtype has a lower risk metabolic profile. This leads women to believe their likelihood of future disease is higher than it actually is, causing undue distress and worry and potentially affecting insurance premiums. Furthermore, limited evidence has found that prevalence of all subtypes rapidly decreases after 25 years of age, suggesting symptoms of PCOS might be transitory for some younger women and spontaneously resolve over time. 

Some may ask: How is overdiagnosis of PCOS harmful? The benefits of a timely PCOS diagnosis are well described (eg, validation of experience and explanation of symptoms). However, research now highlights several unintended harms, such as altered self-perceptions and longlasting worry about future health. For example, one woman we interviewed reported that, despite having irregular periods, she was shocked by her diagnosis of PCOS after a magnetic resonance imaging scan for a hip injury incidentally detected polycystic ovaries. Blood tests confirmed high androgen levels despite no corresponding symptoms (eg, no acne, excess hair, or weight gain). Years later, the diagnosis still has a longlasting emotional impact, and she has concerns about infertility, cancer, and diabetes constantly playing on her mind. She sees herself as less healthy, despite experiencing minimal symptoms, negatively affecting her psychological wellbeing. As with many diseases on a spectrum of severity, online information, media reports, and support groups tend to focus on the severe end of the spectrum. Consequently, women on the milder end fear their condition will worsen, which threatens their perception of health and fertility. Our qualitative study showed fear of infertility can result in adverse psychological and behavioural consequences including anxiety, depression, lower self-worth, altered life and education goals, risk taking with contraception, and unintended pregnancies. Reduced contraceptive use has also been found in quantitative research

Others may ask: But doesn’t a diagnosis enable healthy lifestyle changes? Many clinicians we interviewed believed the diagnosis enabled early engagement in health promoting behaviour to prevent infertility, weight gain, and related metabolic consequences. However, multiple trials in various conditions have shown that disclosure of risk status or a diagnosis do not improve health related behaviours. Similarly in PCOS, our longitudinal analysis found young women newly diagnosed with PCOS did not increase their physical activity or increase their vegetable intake. This suggests that the assumed benefit of the diagnosis in motivating healthy lifestyle changes is overestimated. 

Ascribing a diagnosis should improve patient centred outcomes, not merely categorise and label more individuals as abnormal. To achieve this, disease definitions must be based on evidence, the consequences of definition changes should be considered thoroughly, and potential harms must be rigorously assessed. Stratifying subtypes into separate conditions may reduce the harms of trying to contain such a heterogeneous condition in one diagnostic label. The language we use matters. We must stop giving women with mild symptoms at low risk of adverse consequences a permanent PCOS label, implicating lifelong monitoring and evoking longlasting worry. We urge future updates to the criteria and guidelines to consider potential harms and balance them alongside potential benefits. 

Tessa Copp is a postdoctoral researcher at the University of Sydney. Twitter @TessaCopp

Jenny Doust is a general practitioner and clinical professorial research fellow at the University of Queensland. Twitter @jadoust

Kirsten McCaffery is a principal research fellow at the University of Sydney. Twitter @KirstenMcCaffer

Jolyn Hersch is an early career fellow at the University of Sydney. Twitter @jolynhersch

Jesse Jansen is a senior research fellow at Maastricht University and has an honorary associate position at the University of Sydney. Twitter @JesseJansen_

Competing interests: Tessa Copp receives funding from the NHMRC Centre for Research Excellence in Creating Sustainable Healthcare and NHMRC Wiser Healthcare Program Grant. She is also on the Junior Researcher Committee for the Preventing Overdiagnosis Conference. 

Jenny Doust receives funding from the NHMRC Centre for Research Excellence in Creating Sustainable Healthcare.

Kirsten McCaffery receives funding from the NHMRC Centre for Research Excellence in Creating Sustainable Healthcare and the NHMRC Wiser Healthcare Program Grant. 

Jolyn Hersch receives funding from NHMRC Early Career Fellowship #1112509. 

Jesse Jansen received funding from an NHMRC Career Development Fellowship APP1142094 and the NHMRC Centre for Research Excellence in Creating Sustainable Healthcare while conducting this work.

Commissioned and peer reviewed. 

Patient consent obtained