Increased Prevalence of Gestational Diabetes Mellitus Among Women With Diagnosed Polycystic Ovary Syndrome : A population-based study

Kaiser Permanente of Northern California (KPNC) is a large integrated health care delivery system providing comprehensive medical care for >3,000,000 members in northern California. There are 14 delivery hospitals with >30,000 births/year. Comprehensive health plan databases of all hospitalizations, ambulatory visits, and laboratory tests are available and linked by unique patient identifiers (22). This study was approved by the KPNC Institutional Review Board.

As previously described (1), we used hospitalization databases to identify third trimester pregnancies between 1 January 2002 and 31 December 2004 among women aged 16–44 years, ascertaining the earliest pregnancy for each woman in this period. Maternal age, race/ethnicity, and single versus multiple gestation status were obtained from hospitalization databases. Approximately 95% of pregnancies within KPNC are screened for GDM (1). In accordance with national guidelines, if the random 50-g 1-h oral glucose tolerance test is abnormal, a diagnostic 100-g 3-h oral glucose tolerance test is performed following a 12-h fast (3,23). The KPNC GDM registry data (1) were used to ascertain GDM according to the Carpenter-Coustan criteria (two or more plasma glucose levels achieving the following: fasting ≥5.3 mmol/l, 1 h ≥10 mmol/l, 2 h ≥8.6 mmol/l, and 3 h ≥7.8 mmol/l) (3,23,24). Women with preexisting diabetes were excluded, as previously described (1).

We searched outpatient databases to identify women who received at least one outpatient diagnosis of PCOS (ICD-9 code 256.4) between 1 January 1994 and 31 December 2004 before their delivery date, in the absence of confounding conditions (e.g., congenital adrenal hyperplasia, Cushing’s syndrome, ovarian tumor, or prolactinoma) (25). To support the PCOS diagnosis, we identified whether women had any additional PCOS-related diagnoses (androgen excess, oligo/amenorrhea, anovulation, infertility, menstrual irregularity, insulin resistance, and acanthosis nigricans). Because of potential PCOS under ascertainment, we separately identified women without diagnosed PCOS who had any diagnosis of a PCOS symptom (“PCOS symptom only”), including oligo/amenorrhea, ovulatory dysfunction, hirsutism, and/or hyperandrogenism in the 4 years before delivery.

The clinical characteristics and proportion of GDM cases among women by PCOS status were compared using the χ² test and Student’s t test. Multivariable logistic regression was used to examine the association of PCOS, symptoms, and GDM. All analyses were conducted using SAS version 9.0 (Cary, NC).

Between 2002 and 2004, 92,933 pregnant women screened for GDM were identified. We excluded 4,335 (4.5%) women without GDM screening. The racial/ethnic distribution was 39.2% white, 6.5% black, 24.4% Hispanic, and 18.7% Asian. A total of 5,725 (6.2%) women met GDM criteria.

There were 1,542 (1.7%) women with diagnosed PCOS, of whom 97% had other PCOS-supporting diagnoses. An additional 6,509 (7.0%) women without diagnosed PCOS had diagnoses of oligo/amenorrhea, ovulatory dysfunction, hirsutism, and/or hyperandrogenism. Women with PCOS were slightly older, more likely to be white, and have multiple gestation compared with women with no PCOS or symptoms (Table 1). The prevalence of GDM was highest among women with PCOS and intermediate among those with PCOS symptoms only.

Multivariable logistic regression was used to investigate the association of PCOS and GDM, adjusting for maternal age, race/ethnicity, and multiple (versus single) gestation status. We found that women with PCOS had a more than twofold increased odds of GDM (adjusted odds ratio [OR] 2.44 [95% CI 2.10–2.83]), while women with PCOS symptoms only had an intermediate increased odds (1.40 [1.27–1.54]) compared with women with no PCOS or symptoms. Higher maternal age (1.10 [1.10–1.11]), multiple gestation (1.29 [1.07–1.55]), Hispanic ethnicity (1.81 [1.68–1.94]), and Asian race (2.25 [2.09–2.41]) were also associated with an increased odds of GDM in adjusted analyses.

Among a large population of pregnant women receiving health care within northern California, women with diagnosed PCOS had a 2.4-fold increased odds of GDM, independent of age, race/ethnicity, and multiple gestation. These data are consistent with the known association of PCOS and glucose intolerance among nonpregnant women (11–13) and support previous studies suggesting higher rates of GDM among women with PCOS (16,18,26,27). The intermediate increased odds of GDM among women with PCOS symptoms only is consistent with the overlap of metabolic perturbation and reproductive abnormalities and the possibility that some women actually had PCOS. Importantly, this study comprises the first population-based evaluation of the prevalence of GDM among women with PCOS, the strengths of which include the large sample size and identification of the outcome using objective laboratory diagnostic criteria.

Although we demonstrate a clear association between PCOS and GDM, the extent to which obesity contributes to these findings remains key, given the high prevalence of obesity among women with PCOS (25). There is also concern regarding PCOS under ascertainment, potential variability in PCOS diagnosis, and the possibility that some PCOS women did not manifest both androgen excess and anovulation. However, more accurate assessment of PCOS status would likely have magnified our results. Finally, these data pertain only to women achieving advanced gestation since pregnant PCOS women with early miscarriage were not captured.

In conclusion, we found that pregnant women with diagnosed PCOS have a more than twofold increased odds of GDM compared with women without PCOS or symptoms. Further research is needed to clarify the role of obesity in mediating this association, as well as the role of infertility treatment, glucoregulatory drugs, and other factors during preconception care that may impact GDM risk. Studies are also needed to determine whether PCOS women might benefit from management to reduce GDM risk and whether such measures affect pregnancy outcome.

This work was supported by the National Institutes of Health (National Institute of Child Health and Human Development, Office of Research on Women’s Health, and National Institute of Diabetes and Digestive and Kidney Diseases: 9 K12 HD052163-0 and R01 DK054834) and the Kaiser Permanente Research Program on Genes, Environment and Health funded by the Wayne and Gladys Valley Foundation, the Ellison Medical Foundation, and Kaiser Foundation Hospitals Health Plan.

The authors thank Isaac Ergas, MPH, Bix Swain, MS, and John Greene, MA, for their helpful technical assistance.