Polycystic ovary syndrome (PCOS) is a common endocrine condition affecting 12%–18% of reproductive-aged women [1
]. It is associated with reproductive (hyperandrogenism, anovulation, menstrual irregularly, infertility and pregnancy complications) [2
], metabolic (increased risk factors for and prevalence of impaired glucose tolerance, type 2 diabetes and cardiovascular disease) [3
] and psychological (worsened quality of life and increased risk factors for depression and anxiety) [6
] features. There is a proposed bidirectional relationship between obesity and PCOS [7
]. Women with PCOS have an elevated prevalence of obesity [8
] and increased longitudinal weight gain [7
]. Obesity also worsens the presentation and prevalence of PCOS [9
]. Mechanisms include increasing the pathophysiological factor insulin resistance, which increases hyperandrogenism through augmenting ovarian androgen production and decreasing hepatic production of the androgen binding-protein sex hormone binding globulin [10
]. Due to the key aetiological role of obesity and insulin resistance in PCOS, weight management, defined as prevention of excess weight gain or achieving and maintaining a modest weight loss, is a key treatment strategy in PCOS. Evidence based guidelines recommend achieving this through a combination of diet, exercise or behavioural management [12
The optimal dietary strategy as part of lifestyle management in PCOS remains controversial. We reported in a recent systematic review that the controlled clinical literature found no difference in the majority of anthropometric, reproductive, metabolic or psychological outcomes for a range of dietary approaches including higher protein, higher carbohydrate, lower glycaemic index or monounsaturated fat-enriched diets [13
]. Despite this, a range of dietary approaches may be prescribed by health professionals [14
]. While evidence-based National Health and Medical Council approved Australian guidelines outline the principles of dietary management for PCOS [12
], the effect of these guidelines on actual dietary prescription by health professionals and subsequent dietary intake by women with PCOS is not known. In the absence of specific recommendations by health professionals, women with PCOS may also often seek non-evidence based sources of information on dietary management [15
]. The effect of this on actual dietary intake is not known. We and others have reported subtle differences in dietary intake for women with PCOS compared to those without PCOS including a better dietary intake as indicated by elevated diet quality indices, fibre and micronutrient intake, lower glycaemic index and lower total fat or saturated fat intake or a poorer dietary intake indicated by poorer diet quality, increased fat, saturated fat and high glycaemic index food intake and decreased fibre intake compared to women without PCOS [16
]. There however remains uncertainty as to the quality of dietary intake in women with PCOS.
Assessment of dietary patterns offers an additional way of comprehensively assessing dietary intake. Rather than assessing single nutrients in isolation, dietary pattern analysis identifies underlying dietary characteristics of the study population in which the consumption of foods that are eaten together can be derived. In particular, exploratory approaches or posteriori dietary pattern analyses such as principal components analysis, which are not hypothesis driven, groups correlated food groups into uncorrelated factors termed dietary patterns [22
]. In pregnant populations, unhealthy dietary patterns in the pre-conception period were associated with increased risk for preterm birth [24
] or gestational diabetes [25
] and healthy, Mediterranean or prudent diet patterns were inversely associated with risk of developing hypertensive disorders during pregnancy [26
] or gestational diabetes [27
]. In non-pregnant populations, unhealthy/Western-type dietary patterns have been associated with increased risk of general and central obesity [29
] and type 2 diabetes [30
]; while a Mediterranean dietary pattern was associated with decreased prevalence of hypertension and metabolic syndrome [31
] and a healthy dietary pattern containing vegetables, fruits and whole grains was associated with reduced risk for diabetes [30
These findings are of potential relevance to PCOS given the increased prevalence of cardiometabolic conditions and pregnancy complications and the potential for clinical benefits with approaches such as the Mediterranean diet [32
]. However, there has been limited research examining dietary patterns in women with and without PCOS. This could provide an understanding on both the association of dietary intake with the pathophysiology of PCOS as well as of the dietary changes that occur following a diagnosis of PCOS. The aim of this study was therefore to examine dietary patterns in a large cohort of women, with and without PCOS, participating in the Australian Longitudinal Study on Women’s Health.
We report here for the first time that women with PCOS have different dietary patterns compared to women without PCOS, in a large population-based cohort of women. Women with PCOS were more likely to consume a dietary pattern consistent with the Mediterranean diet; however there were no differences in the other commonly consumed dietary patterns of unhealthy non-core foods or a pattern higher in meat.
The Mediterranean-style dietary pattern contains a number of foods similar to a Mediterranean diet which consists of fish, monounsaturated fats from olive oil, fruits, vegetables, wholegrains, legumes and nuts and moderate alcohol consumption [43
]. It is also consistent with previously defined “Mediterranean” patterns in prior research comprising vegetables, fish, fruits, poultry, low-fat dairy products, and olive oil [44
]. Surprisingly however, we found an inverse factor loading for poultry which is typically consumed in higher intakes in the Australian population compared to fish [46
], which loaded on this pattern in moderate amounts for both processed (i.e.
, tinned fish) and other fish (i.e.
, cooked fish), while fried fish was inversely associated. It is to be noted that intake of olive oil is not collected in the food frequency questionnaire used in this study. Another surprising finding was that both low fat and high fat dairy foods did not correlate to any pattern. This might reflect the overall low consumption of dairy in men and women in the adult Australian population; yet this is consistent with a previous study in pregnant women where low fat dairy did not load on any of the three dietary patterns, and high fat dairy only moderately correlated with the vegetarian-type dietary pattern [24
]. Nevertheless, non-core foods inversely loaded on this pattern such as take-away foods and crisps, as well as added sugar, which supports an overall healthier dietary pattern consisting of a number of Mediterranean foods. As we are the first to report that a Mediterranean-style dietary pattern was independently associated with increased likelihood of having PCOS, this discrepant finding may indicate the possible high level of women with PCOS seeking dietary knowledge with a subsequent adoption of healthy dietary patterns.
To date, there are few other studies reporting on the relationship between dietary patterns and other conditions co-existing with PCOS. In literature assessing infertile women, a large proportion of whom will likely have PCOS, a Mediterranean diet is associated with a higher chance of natural or assisted reproduction conception [44
]. The adoption of a Mediterranean-style diet in PCOS may therefore have positive implications for the appropriate lifestyle management of chronic diseases associated with PCOS. Further studies are needed to expand on our findings on the association of dietary changes in those with a diagnosis of PCOS, the optimal means of conveying dietary education at diagnosis and the long-term maintenance of positive dietary changes.
We observed here that the two other identified dietary patterns, namely those consisting predominantly of non-core foods or a higher meat intake from either take-away/processed or non-processed sources explained a moderate proportion of variability in food intake in all participants (13% and 8% respectively). However, neither pattern was associated with PCOS status in the adjusted analysis. In association with higher weight and BMI in PCOS, this is a positive finding that is also consistent with the diagnosis of PCOS contributing to an improvement of dietary habits in keeping with population-based dietary guidelines of minimising discretionary or non-core food intake, reducing processed meats and consuming a moderate intake of protein [48
While a Mediterranean diet is not a specifically recommended dietary intake for PCOS, emerging research suggests beneficial effects of certain components of this diet, such as elevated omega-3 fatty acids which are generally found in high amounts of oily fish. Although the specific types of fish consumed in our Mediterranean style dietary pattern cannot be extracted, both processed fish and cooked fish varieties contain some omega-3 fatty acids, likely contributing to a reasonable intake of omega-3 fatty acids in this population. The literature in PCOS focuses predominantly on omega-3 fatty acid supplementation studies which report improvements in outcomes including reductions in bioavailable androgens, triglycerides, blood pressure, glucose and surrogate markers of insulin resistance [49
]. One recent study found that a Mediterranean diet pre-pregnancy was associated with a 42% reduced likelihood of developing hypertensive related disorders during pregnancy [26
]; while higher consumption of sweets and seafood [25
] or high intake of red meat, processed meat, refined grain products and sweets [27
] during pregnancy was associated with a 23% and 63% increased risk of gestational diabetes. A Mediterranean dietary pattern has also been reported to be associated with improved health outcomes including decreased inflammation [53
] and prevalence of the metabolic syndrome [54
], abnormal glucose tolerance [55
] or depression [45
]. As adverse health outcomes are commonly associated with PCOS [3
], this dietary pattern may therefore result in health benefits. However, we have previously reported in this cohort that this improved diet quality occurred in conjunction with a modest increase in energy intake (+215 kJ/day) which could contribute to additional longitudinal weight gain [19
]. The potential benefits of an improved dietary pattern may not outweigh the effects of increased energy intake and consequent weight gain with regards to effects on reproductive, and potentially metabolic and psychological, parameters.
Strengths to our study include the large population of women with and without PCOS from a community-based population in contrast to the majority of the existing research assessing diet and PCOS. This minimises selection bias. This is also more likely to capture a lower proportion of women with PCOS with a more severe clinical phenotype and a higher BMI who typically present to clinical services and are captured in research studies [56
]. While the use of self-report PCOS is a limitation, the nature of this research means that it is not feasible to clinically verify PCOS or control status. It is also not possible to determine the PCOS phenotype or which diagnostic criteria were used in diagnosis. However, given that the Rotterdam criteria were first published in 2004 [57
], it is also most likely that the majority of women self-reporting diagnosed PCOS in Survey 4, conducted in 2006, would have been diagnosed based on NIH criteria. There are also some other limitations to our study. We report here 58% participant retention compared to baseline levels 13 years prior which may indicate bias and limit generalisability. However, no differences between completors and non-completors has previously been reported indicating a likely minimal effect of attrition on outcomes [36
]. Although the FFQ is a validated measure of assessing nutritional intake, we are not able to assess the contribution of dietary patterns to the development or severity of PCOS due to the study design and report here only associations between dietary patterns and PCOS status. Further, the total variance explained by each factor was intermediate compared with previous factor analyses conducted in different age groups [29
]; however, the Kaiser-Meyer-Olkin measure of sampling adequacy was 0.78, exceeding the recommended value of 0.6; and Bartlett's test of Sphericity achieved statistical significance indicating the correlations in the data set are appropriate for factor analysis. Moreover, the food groups loading on the factors were varied and many were greater than the 0.25 cut-off value suggesting that our population had a varied diet that was, nevertheless, still specific to the identified factors. As the present study is the first of its kind in this population, further studies are required to refute or support our findings and future work is warranted assessing the contribution of dietary pattern intake to the severity or incidence of PCOS.