Objective
We sought to determine prevalence and likelihood of venous thromboembolism (VTE) among women with and without polycystic ovary syndrome (PCOS).
Study Design
We performed a cross-sectional analysis using Thomson Reuters MarketScan Commercial databases for the years 2003 through 2008. The association between VTE and PCOS among women aged 18-45 years was assessed using age-stratified multivariable logistic regression models.
Results
Prevalence of VTE per 100,000 was 374.2 for PCOS women and 193.8 for women without PCOS. Compared with women without PCOS, those with PCOS were more likely to have VTE (adjusted odds ratio [aOR] 18-24 years, 3.26; 95% confidence interval [CI], 2.61–4.08; aOR 25-34 years, 2.39; 95% CI, 2.12–2.70; aOR 35-45 years, 2.05; 95% CI, 1.84–2.38). A protective association (odds ratio, 0.8; 95% CI, 0.73–0.98) with oral contraceptive use was noted for PCOS women.
Conclusion
PCOS might be a predisposing condition for VTE, particularly among women aged 18-24 years. Oral contraceptive use might be protective.
Venous thromboembolism (VTE) is a chronic condition that includes both deep vein thrombosis (DVT) and pulmonary embolism (PE). Initial presentation might be leg pain, tenderness, shortness of breath, or pleuritic chest pain, or there might be no symptoms. VTE is the third most common cardiovascular disease–after myocardial infarction and stroke–among the general population. The overall annual incidence of VTE is estimated to be 1-2 per 1000 adults per year. These incident estimates, however, might not represent the entire population because VTE incidence differs by age and race, and slightly by sex. VTE is a multifactorial disease with both inherited and acquired risk factors. In 26-47% of first-time VTE cases, the etiology is unknown ; a possible predisposing condition not yet assessed is polycystic ovary syndrome (PCOS).
For Editors’ Commentary, see Contents
PCOS is the most common endocrine disorder affecting women of reproductive age. Estimates of its prevalence vary and range from 4–6%. Its exact etiology is unknown, but it is characterized by a heterogeneous presentation of hyperandrogenism, ovulatory dysfunction, and polycystic ovaries. PCOS is also associated with insulin-induced elevations of plasminogen activator inhibitor (PAI)-1, which is a potent inhibitor of fibrinolysis. While some studies have found PCOS to be associated with coronary heart disease risk factors, other studies have shown that women with PCOS were 3 times more likely to have a family history of venous thrombosis. Mak and Dokras hypothesized that women with PCOS likely have an increased baseline risk for developing thrombosis compared with other women in the general population. However, no study has evaluated the prevalence of VTE among women with PCOS compared with women without PCOS. Given the potential for thrombotic disease among women with PCOS, the purpose of this study was to: (1) determine the prevalence of VTE among women with and those without PCOS; and (2) assess the association between PCOS and VTE.
Materials and Methods
Data source
Data for the study were derived from the 2003 through 2008 Thomson Reuters MarketScan Commercial databases. These databases comprise longitudinal, deidentified health insurance claims data from large employers and health plans across the United States and contain information on inpatient admissions, outpatient visits, and pharmaceutical claims. The data are derived from multiple US states that are geographically diverse. In 2008, >35 million individuals were enrolled, including employees and their dependents who were aged <65 years and geographically distributed throughout the United States.
This commercial database has met or exceeded requirements of the US Health Insurance Portability and Accountability Act of 1996 and, accordingly, does not require specific patient consent to participate in the study.
Population
We restricted the analysis to women 18-45 years of age. The upper limit of age was set at 45 years to keep within the age limit used by other prevalence studies. Women with PCOS were defined by the presence of hyperandrogenism, the presence of ovulatory dysfunction, and/or the presence of polycystic ovaries. International Classification of Diseases, Ninth Revision, Clinical Modification ( ICD-9-CM ) codes were used to identify these conditions in the database. Each condition was defined as follows.
“Clinical hyperandrogenism” was defined as the presence of ICD-9-CM codes for acne (706.0 or 706.1), alopecia (704.0x), or hirsutism (704.1). Elevated testosterone, another characteristic of clinical hyperandrogenism, was not included because no ICD-9-CM codes were available for this laboratory value in the database.
“Ovulatory dysfunction” was defined by the presence of any 1 of the ICD-9-CM codes for the 626 series, which are codes dealing with disorders of menstruation and other abnormal bleeding from the female genital tract: 626.0, 626.1, 626.2, 626.3, 626.4, 626.5, 626.6, 626.7, 626.8, or 626.9.
“Polycystic ovaries” were defined by ICD-9-CM code 256.4.
Exclusion of other conditions
We excluded women who met the criteria for any of the PCOS phenotypes and had ≥1 of the following conditions: adrenal hyperplasia (255.0 or 255.9), hyperprolactinemia (253.1), or thyroid disorder (244.0, 244.1, 244.2, 244.3, 244.8, 244.9, 242.90, or 242.91).
Combinations of these conditions were used to create 4 mutually exclusive PCOS phenotypes based on the 3 available criterion recommendations (National Institutes of Health, Rotterdam, and Androgen Society). Women with phenotype A, or classic PCOS, were defined by the presence of hyperandrogenism (704.1, 706.0 or 706.1, or 704.0x) and ovulatory dysfunction (626.0, 626.1, 626.2, 626.3, 626.4, 626.5, 626.6, 626.7, 626.8, or 626.9); but polycystic ovaries (256.4), adrenal hyperplasia (255.0 or 255.9), hyperprolactinemia (253.1), and thyroid disorder (244.0, 244.1, 244.2, 244.3, 244.8, 244.9, 242.90, or 242.91) were absent. Definitions of the remaining PCOS phenotypes are found in the Appendix .
Due to the heterogenic nature of PCOS and its complex presentation and definition, we also assessed the prevalence of associated conditions seen among women with PCOS (eg, obesity [278.0, 278.00-278.02, 783.1, or V77.8]; infertility [628.0 or 628.9]; syndrome X or metabolic syndrome [277.7]; and diabetes [250.00, 250.02, 250.90, or 250.92]) to provide additional information on the frequency of these conditions within the different PCOS phenotypes. For all conditions related to PCOS, diagnoses reported on inpatient claims were considered valid; diagnoses based solely on outpatient claims required that the diagnoses were reported on ≥2 claims that occurred >30 days apart.
Women with possible VTE during the study period also were identified using ICD-9-CM codes for DVT (671.3x, 671.4x, 671.5x, 671.9x, 451.11, 451.19, 451.2, 451.81, 451.9, 453.1, 453.2, 453.40-453.42, 453.8, or 453.9), PE (673.2x, 673.8x, 415.11, or 415.19), or both on any inpatient or outpatient claim. To reduce inaccurate reporting of DVT and PE diagnoses, we restricted outpatient diagnoses to those with a relevant Current Procedural Terminology code ( Appendix ) and who had a filled prescription for an anticoagulant medication within 90 days of the VTE diagnosis; for an inpatient diagnosis, only 1 appropriate ICD-9-CM code for DVT or PE was required. Diagnosis of VTE was counted only once, regardless of the number of times it was reported in the claims database.
Unique identification numbers were used to track all women longitudinally over the study period. Women were included if they were enrolled in any health plan, even if no health claims were submitted during the 6-year study period.
Covariates
Information on demographic characteristics and clinical comorbid conditions was captured from the claims database: age, geographic region of residence, pregnancy during the time frame of study, oral contraceptive use, diabetes, obesity, metabolic syndrome, thrombophilia (289.81), and history of a VTE (V12.51). Pregnancy was defined as any inpatient hospital admission for a pregnancy-related diagnosis or procedure (V27.x, 650, 72.0, 72.1, 72.21, 72.29, 72.31, 72.39, 72.4, 72.51-72.54, 72.6, 72.71, 72.79, 72.8, 72.9, 73.22, 73.59, 73.6, 74.0-74.2, 74.4, or 74.99), excluding records with codes or procedures for hydatidiform mole, ectopic pregnancy, other abnormal products of conception, or abortion (630, 631, 632, 633.x, 634.x-638.x, 639.x, 69.01, 69.51, 74.91, or 75.0). Contraceptive use was identified by ≥1 filled oral contraceptive pill (OCP) prescription during the study period; the use of other types of contraceptives such as the patch, the ring, and/or injections was excluded from the analysis.
Data analysis
To calculate VTE prevalence among women with PCOS, the number of VTE cases among women with PCOS was divided by the total number of women among the underlying MarketScan population with PCOS. For the individual phenotypes, prevalence was calculated by dividing the number of women with VTE and the particular PCOS phenotype by the total number of women with that type of PCOS. For women without PCOS, the prevalence of VTE was calculated by dividing the number of women with VTE and no PCOS by the total underlying MarketScan population without PCOS. Differences in the distribution of demographic and comorbid conditions among women with and those without PCOS were assessed using 2-tailed χ 2 tests.
Bivariate analyses were used to compare the distribution of demographic and potential clinical risk factors for VTE among women with and those without PCOS. Multivariate logistic regression analyses also were used to assess the likelihood of a VTE diagnosis among women with and those without PCOS. The final multivariable regression model had only those variables that have been shown to be associated with VTE risk and changed the estimate of the association between VTE and PCOS by >10%. These variables, which we adjusted for, were age, pregnancy during study period, OCP use, region, obesity, and diabetes. We also used logistic regression models to test for interactions between PCOS and the previously cited variables. Age was found to be an effect modifier, so we stratified the analysis by 3 age groups (18-24, 25-34, and 35-45 years). Even though OCP use was not found to be an effect modifier, we stratified the analysis by OCP use because, in 2003, Seaman et al demonstrated an increased risk of VTE among women with acne, hirsutism, or PCOS who used cyproterone acetate in combination with ethinyl estradiol. All data were analyzed and managed using software (SAS, version 9.2; SAS Institute Inc, Cary, NC). All analyses were conducted at a priori significance level ( P < .05).
Results
Prevalence of VTE
A total of 12,171,830 women were eligible to be in the study; of these, 23,941 (0.20%) had VTE; among the group with VTE, 14,321 (59.8%) had DVT only, 4967 (20.7%) had PE only, and 4653 (19.4%) had both VTE and PE. The overall prevalence of VTE among the PCOS population was 374.2 per 100,000, while the prevalence of VTE among women without PCOS was 193.8 per 100,000 ( Figure 1 ). Among the phenotypes, women with phenotype D had the highest prevalence of VTE, at 478.9 per 100,000 ( Figure 1 ). Figure 2 displays the prevalence of VTE per 100,000, stratified by age, among women with and those without PCOS. For both cohorts, the prevalence of VTE was highest among women 35-45 years of age ( Figure 2 ).
The characteristics of women with and those without PCOS are shown in Table 1 . Statistically significant differences existed in percentage distributions in VTE frequencies ( P < .0001), age ( P < .0001), region ( P < .0001), presence of metabolic syndrome ( P = .0002), obesity ( P < .0001), pregnancy ( P < .0001), and diabetes ( P < .0001). Other significant differences were noted between having inherited thrombophilia ( P < .0001) and having used OCPs ( P < .0001) ( Table 1 ). There was no significant difference in having a history of VTE ( P = .243) between women with and those without PCOS ( Table 1 ).
| Variables | PCOS, n = 192,936 (%) b | No PCOS, n = 11,978,894 (%) b | P value a |
|---|---|---|---|
| VTE frequencies | 0.37 | 0.19 | < .0001 |
| Age categories, y | < .0001 | ||
| 18-24 | 20.6 | 24.5 | |
| 25-34 | 45.3 | 32.6 | |
| 35-45 | 34.7 | 42.9 | |
| Geographic region c | < .0001 | ||
| Northeast | 10.3 | 9.8 | |
| North Central | 23.0 | 21.6 | |
| South | 47.5 | 45.4 | |
| West | 18.7 | 22.7 | |
| Unknown | 0.4 | 0.6 | |
| Comorbidities | |||
| Metabolic syndrome | 0.11 | 0.09 | .0002 |
| Obesity | 0.6 | 1.3 | < .0001 |
| Pregnancy | 0.4 | 7.4 | < .0001 |
| Diabetes | 0.3 | 1.4 | < .0001 |
| History of VTE | .2428 | ||
| Yes | 0.034 | 0.029 | |
| Inherited thrombophilia | < .0001 | ||
| Yes | 0.021 | 0.055 | |
| OCP use | < .0001 | ||
| Yes | 49.7 | 24.3 |
a χ 2 test comparing proportion between women with and without PCOS;
b Might not add up to 100% secondary to rounding;
Table 2 displays the age-stratified association between the PCOS phenotypes and VTE. We found that regardless of age and PCOS phenotype, the odds of having VTE were higher for women with PCOS compared with women without PCOS. We also observed that women 18-24 years of age with PCOS had the highest odds of having VTE (adjusted odds ratio [OR], 3.26; 95% confidence interval [CI], 2.61–4.08) when compared with women 18-24 years of age without PCOS. Lastly, we observed that women 18-24 years of age with phenotype B (hyperandrogenism and polycystic ovaries) had the highest probability of having VTE (adjusted OR, 5.23; 95% CI, 2.88–9.51) when compared with women of the same age without PCOS ( Table 2 ).
| PCOS types | Age groups | ||
|---|---|---|---|
| 18-24 y | 25-34 y | 35-45 y | |
| aOR (95% CI) a | |||
| No PCOS b | Reference | Reference | Reference |
| Any PCOS c | 3.26 (2.61–4.08) | 2.39 (2.12–2.70) | 2.05 (1.84–2.28) |
| Phenotype A d | 3.05 (2.29–4.07) | 2.25 (1.91–2.65) | 1.98 (1.75–2.25) |
| Phenotype B e | 5.23 (2.88–9.51) | 2.25 (1.35–3.74) | 1.87 (1.11–3.18) |
| Phenotype C f | 2.87 (1.80–4.58) | 2.35 (1.91–2.89) | 2.18 (1.73–2.73) |
| Phenotype D g | 3.11 (1.29–7.49) | 3.06 (2.06–4.54) | 2.37 (1.49–3.78) |
a Multivariate logistic regression model was adjusted for pregnancy, oral contraceptive pill use, region, diabetes, and obesity;
b Women without any PCOS phenotype;
c Women with any PCOS phenotype;
d Women with clinical hyperandrogenism and menstrual or ovulatory dysfunction, or both;
e Women with hyperandrogenism and polycystic ovaries;
f Women with menstrual or ovulatory dysfunction, or both, and polycystic ovaries;
g Women with clinical hyperandrogenism; menstrual or ovulatory dysfunction, or both; and polycystic ovaries.
In Table 3 , we present the adjusted relationships between PCOS phenotypes and VTE when stratified by age and OCP use. We found that women with PCOS still had increased odds of VTE compared with women without PCOS, regardless of age and OCP use. However, when we looked at the unadjusted effect of OCP use on VTE prevalence only among the cohort of women with PCOS (not shown in tables), we observed a protective effect for VTE (OR, 0.8; 95% CI, 0.73–0.97).
| PCOS type | Age groups | ||
|---|---|---|---|
| 18-24 y | 25-34 y | 35-45 y | |
| OCP use = yes, a aOR (95% CI) b | |||
| No PCOS c | Reference | Reference | Reference |
| Any PCOS d | 2.89 (2.17–3.85) | 1.98 (1.66–2.37) | 1.79 (1.51–2.12) |
| Phenotype A e | 3.11 (2.22–4.36) | 1.69 (1.33–2.16) | 1.83 (1.51–2.22) |
| Phenotype B f | 4.58 (2.17–9.66) | 1.53 (0.68–3.41) | 1.41 (0.58–3.39) |
| Phenotype C g | 1.07 (0.40–2.87) | 2.24 (1.66–3.03) | 1.65 (1.12–2.44) |
| Phenotype D h | 3.23 (1.20–8.64) | 2.89 (1.74–4.82) | 1.69 (0.80–3.57) |
| OCP use = no, i aOR (95% CI) b | |||
| No PCOS c | Reference | Reference | Reference |
| Any PCOS d | 3.76 (2.63–5.38) | 2.70 (2.29–3.18) | 2.18 (1.90–2.50) |
| Phenotype A e | 2.60 (1.47–4.59) | 2.79 (2.24–3.48) | 2.02 (1.71–2.38) |
| Phenotype B f | 6.37 (2.36–17.23) | 3.02 (1.57–2.84) | 2.20 (1.14–4.26) |
| Phenotype C g | 5.26 (3.09–8.94) | 2.36 (1.78–3.13) | 2.54 (1.91–3.36) |
| Phenotype D h | 2.36 (0.33–16.82) | 3.05 (1.63–5.70) | 3.03 (1.67–5.51) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
