Are Oral Contraceptives Risky or Protective in Women with Polycystic Ovary Syndrome?
Abstract & Commentary
By Jeffrey T. Jensen, MD, MPH , Leon Speroff, Professor and Vice Chair for Research, Department of Obstetrics and Gynecology, Oregon Health and Science University, Portland, is Editor for OB/GYN Clinical Alert.
Synopsis: Women with polycystic ovary syndrome (PCOS) are more likely to experience a venous thrombosis (VTE) than women without PCOS. Although obesity increases the risk of VTE associated with oral contraceptive use, it is not proven that PCOS has an independent effect.
Sources: Okoroh EM, et al. Is polycystic ovary syndrome another risk factor for venous thromboembolism? United States, 2003–2008. Am J Obstet Gynecol 2012;207:377.e1-8.
Bird ST, et al. Risk of venous thromboembolism in women with polycystic ovary syndrome: A population-based matched cohort analysis. CMAJ 2012 Dec. 3 [Epub ahead of print].
These two groups took slightly different approaches to assess the relationship between polycystic ovary syndrome (PCOS) and venous thromboembolism (VTE) using insurance claims databases. Okoroh and coauthors performed a cross-sectional analysis using the Thomson Reuters MarketScan Commercial databases for the years 2003 through 2008. For the study, the analysis was restricted to women between 18-45 years of age. The authors used International Classification of Diseases, Ninth Revision, (ICD-9) codes associated with each insured individual to classify women into four mutually exclusive PCOS phenotypes based on the three available criterion recommendations (National Institutes of Health, Rotterdam, and Androgen Society). Women were considered to have PCOS if their unique insurance identification numbers were linked with a diagnosis of hyperandrogenism, ovulatory dysfunction, and/or polycystic ovaries in the database. The authors also assessed the prevalence of potential confounders such as obesity, metabolic syndrome, and diabetes using applicable diagnosis codes. VTE was assessed using appropriate inpatient and outpatient codes for deep vein thrombosis (DVT) and pulmonary embolism. Information on demographic characteristics, medications (like oral contraceptives), and clinical comorbid conditions also was captured from the claims database. The authors calculated crude prevalence estimates of VTE for women with and without PCOS, and used multivariate logistic regression to adjust these estimates for age, pregnancy during study period, oral contraceptive use, region, obesity, and diabetes. A total of 23,941 VTE events were recorded among 12,171,830 women in the study group (overall 196 cases per 100,000 women). Compared with all women without PCOS (194 per 100,000), those with PCOS (374 per 100,000) were more likely to have VTE (adjusted odds ratio [95% confidence interval] 18-24 years, 3.26 [2.61-4.08]; 25-34 years, 2.39 [2.12-2.70]; 35-45 years, 2.05 [1.84-2.38]). Surprisingly, a reduction in DVT risk was found in PCOS women using oral contraceptives (adjusted odds ratio, 0.8 [0.73–0.98]).
The paper by Bird et al also used a population-based cohort from another U.S. insurance database (IMS Life-Link Health Plan Claims Database). This database contains paid claims data from more than 102 managed care plans, and covered a longer time period (2001- 2009). However, in contrast to the Okoroh’s study, these authors constructed cohorts (to calculate incidence data) based on exposure to an oral contraceptive and required a 1-year period of baseline enrollment in a plan to assess potential confounders (women with a history of cancer, cerebrovascular disease, cardiovascular disease, venous thromboembolism or prior anticoagulation [warfarin and heparin] were excluded). PCOS also was more narrowly defined by a single ICD-9 code (PCOS, 256.4). Women aged 18-46 years taking combined oral contraceptives and who had a claim for PCOS (n = 43,506) were matched to control women (n = 43,506) also taking oral contraceptives. To obtain a comparison to women with PCOS not using combined oral contraceptives, a random sample of 2 million women was used to create PCOS and non-PCOS cohorts of nonusers for the secondary analysis. Results from this study demonstrated that the incidence of VTE among women with PCOS using oral contraceptives (237 per 100,000 person-years) was more than two-fold higher than that of matched controls using oral contraceptives (109 per 10 000 person-years) (HR 2.14 [1.41-3.24]). Among non-users of oral contraceptives, the incidence of VTE was 63 per 100,000 person-years among women with PCOS and 41 per 100,000 in matched controls without PCOS (RR 1.55 [1.10-2.19]). To summarize, PCOS is an independent risk factor for VTE (1.5-fold increase in risk), and a diagnosis of PCOS doubles the risk of VTE associated with oral contraceptive use.
A tale of two papers with similar design coming to entirely different conclusions about the risk of thrombosis with oral contraceptives; but is this a surprise? The literature surrounding VTE risk is as clear as mud. The muddy water flows directly from inherent weaknesses in study design and needs to be carefully navigated. Since DVT is a rare event in otherwise healthy young women, large numbers of individuals must be studied to obtain statistically valid comparisons. Since most epidemiologists don’t have the funding to enroll subjects in large prospective studies (or the patience to wait for these results), shortcuts are used to obtain answers at a reasonable cost and in a manageable time period.
National and insurance claims databases seem like the perfect solution to the study of rare events like DVT, particularly when these databases link inpatient and outpatient diagnostic codes to prescriptions. Unfortunately, database studies are not truly prospective in nature, and cannot control for important baseline confounders such as obesity and preferential prescribing.1 The papers highlighted in this review illustrate some of these difficulties. The manuscript by Okoroh et al published in the American Journal of Obstetrics and Gynecology made a classic error by failing to control for the interaction between oral contraceptive use and PCOS. If the decision to treat or not to treat PCOS with combined oral contraceptives is an indication of disease severity, then any analysis of health effects due to oral contraceptive treatment would be subject to confounding by indication. The surprising conclusion from this paper that oral contraceptive use reduced the risk of VTE in PCOS women is certainly explained by this confounder (e.g., healthier PCOS patients are prescribed oral contraceptives). The authors attempt to explain this by postulating that the effect was due to the suppression of ovarian androgens; this might fool a few epidemiologists, but it should not fool any student of endocrinology.
The paper by Bird’s group tried to control for this bias by selecting a cohort of non-oral contraceptive users with the same characteristics of the oral contraceptive group, including PCOS. They also used a “prospective” approach to analysis — to calculate incident and not prevalent cases of VTE. Although this study is published in the more obscure Canadian Medical Association Journal, the findings are more plausible, robust, and consistent with known biologic mechanisms. The discussion is also much more relevant than the paper in the Gray Journal. The data demonstrate that PCOS is associated with an increased risk of DVT, and the use of oral contraceptives in PCOS patients increases the risk further. Keep in mind though that some of this interaction is due to obesity, and other factors such as abnormal lipids and insulin resistance certainly have an impact. So this study, while better, is also limited by the lack of baseline information on these important confounders.
The clinically important point from all of this is that the principle risk of combined hormonal contraception is thrombosis. The good news is that venous and arterial thromboses are rare events in otherwise healthy reproductive age women. Combined hormonal contraceptive methods are appropriate for most women, and the risk of VTE with pregnancy exceeds that of oral contraceptive use even in high-risk women.2 The bad news is that not all of our patients are healthy. Consider family history of thrombosis, metabolic syndrome, obesity, and lifestyle as important factors to discuss with your patients. Multiple risk factors are important to consider. Some PCOS women are relatively healthy, while others are not. Sometimes we have to balance potential benefits (reduction in androgens, menstrual regularity) with risks (increased VTE risk) in complicated conditions like PCOS. So look more carefully at your next patient with PCOS and use her overall health status to decide whether she is a good candidate for a combined hormonal method.
- Heinemann K, Heinemann LA. Comparative risks of venous thromboembolism among users of oral contraceptives containing drospirenone and levonorgestrel. J Fam Plann Reprod Health Care 2011;37:132-135.
- Heinemann LA, Dinger JC. Range of published estimates of venous thromboembolism incidence in young women. Contraception 2007;75:328-36.