Obstetrics & Gynecology · Medical Literature
Prevalence and associated factors of gestational diabetes mellitus among women with polycystic ovary syndrome: a systematic review and meta-analysis.
Abstract
Polycystic ovary syndrome (PCOS), the most common endocrine disorder in reproductive-aged women, often leads to gestational diabetes mellitus (GDM). This study aimed to systematically evaluate GDM prevalence and its associated factors in PCOS. We searched PubMed, Embase, Web of Science, Cochrane Library, CNKI, Wanfang, Sinomed, and VIP from inception to December 31, 2025. Studies reporting GDM prevalence or associated factors in PCOS were included. A random-effects model pooled prevalence with odds ratios. Egger's test assessed publication bias, and subgroup analysis with meta-regression explored heterogeneity. A logit transformation was applied for pooling. Twenty-two studies with 5,507 PCOS patients were included. The pooled GDM prevalence was 24% (95% CI: 20%-28%; prediction interval: 10%-46%), with substantial heterogeneity (I² = 89.9%). The primary prevalence estimate, based on cohort studies (n=13), was 21.7% (95% CI: 17.7%-25.8%), while the overall estimate including case-control studies was 24% (95% CI: 20%-28%). By diagnostic criteria, GDM prevalence was 26.0% for one-step vs. 18.7% for two-step. Potentially associated factors included pre-pregnancy BMI (OR = 1.39), gestational weight gain (OR = 1.58), HOMA-IR (OR = 3.41), and family history of diabetes (OR = 2.88). Although the pooled GDM prevalence in PCOS was 24%, substantial heterogeneity and a wide prediction interval (10%-46%) suggest the true prevalence varies considerably across settings. Associated factors included HOMA-IR, family history, pre-pregnancy overweight, and excessive gestational weight gain. PCOS patients should be considered a high-risk group, and early monitoring of glucose metabolism and weight management may be beneficial, though these recommendations derive from observed rather than interventional evidence. https://www.crd.york.ac.uk/PROSPERO/view/CRD42026128476, identifier: CRD420261284765.
Clinical relevance
Among pregnant women with polycystic ovary syndrome (PCOS), what is the pooled prevalence of gestational diabetes mellitus (GDM), and which clinical and metabolic factors are significantly associated with its development? __SNAPSHOT__ {"studyDesign":"Systematic review and meta-analysis (random-effects model; logit transformation for prevalence pooling; subgroup analysis and meta-regression for heterogeneity exploration)","setting":"International; databases searched included PubMed, Embase, Web of Science, Cochrane Library, CNKI, Wanfang, Sinomed, and VIP — search period from inception through December 31, 2025","sampleSize":"n = 5,507 PCOS patients across 22 included studies (13 cohort studies, remainder case-control studies)","population":"Pregnant women with a confirmed diagnosis of PCOS; specific PCOS and GDM diagnostic criteria varied across included studies","intervention":"Exposure: diagnosis of PCOS during pregnancy (prevalence estimation and association analysis)","comparator":"Not applicable for prevalence arm; odds ratios reported for associated factors relative to absence of each factor within PCOS populations","primaryOutcome":"Pooled prevalence of GDM among women with PCOS; secondary outcomes included odds ratios for potentially associated clinical and metabolic factors","followUp":"Not reported (varied across included primary studies)","fundingSource":"Not reported"}
Key findings
Twenty-two studies enrolling 5,507 women with PCOS were included. The overall pooled GDM prevalence was 24% (95% CI: 20%–28%), with a prediction interval of 10%–46% and substantial statistical heterogeneity (I² = 89.9%). The primary estimate derived from cohort studies alone (n = 13) was 21.7% (95% CI: 17.7%–25.8%), modestly lower than the estimate that also incorporated case-control studies. GDM prevalence varied by diagnostic screening protocol: 26.0% with the one-step approach versus 18.7% with the two-step approach — a 7.3 percentage point absolute difference (calculated). No p-value or confidence interval for this subgroup difference was reported in the abstract. Potentially associated factors and their pooled odds ratios included: HOMA-IR (OR = 3.41), family history of diabetes (OR = 2.88), gestational weight gain (OR = 1.58), and pre-pregnancy BMI (OR = 1.39). Confidence intervals, p-values, and absolute risk differences for each individual associated factor were not reported in the available abstract. NNT/NNH could not be calculated from the data provided.
Practice implications
This meta-analysis establishes that roughly 1 in 4 to 1 in 5 pregnant women with PCOS will develop GDM, a prevalence substantially higher than reported GDM rates in the general obstetric population globally (approximately 7%–28% depending on region and diagnostic criteria). The identification of HOMA-IR, family history of diabetes, excessive gestational weight gain, and elevated pre-pregnancy BMI as associated factors is clinically intuitive and consistent with the known pathophysiology of insulin resistance in PCOS. However, several important limitations constrain direct clinical application. The prediction interval of 10%–46% is extremely wide, signaling that the 'true' prevalence in any individual clinical setting may deviate dramatically from the 24% point estimate. The I² of 89.9% reflects substantial heterogeneity that was not fully explained by subgroup analyses. Notably, the marked difference in GDM prevalence by screening strategy (one-step vs. two-step) underscores how diagnostic criteria — rather than underlying biology alone — substantially influence measured prevalence. The association data are reported as unadjusted or incompletely adjusted odds ratios without confidence intervals or p-values available in the abstract, precluding rigorous assessment of effect size precision or residual confounding. Because only the abstract was available for review, results should be interpreted with additional caution pending full-text verification of subgroup composition, risk-of-bias assessment scores, and adjusted analyses. The findings support current guideline recommendations that women with PCOS constitute a high-risk group warranting proactive GDM surveillance, but the data are insufficient to define a specific superior screening protocol for this population. __LIMITATIONS__ Extreme statistical heterogeneity (I² = 89.9%) with a prediction interval of 10%–46%, indicating that the pooled prevalence estimate may not be applicable across many clinical settings.,Variability in both PCOS diagnostic criteria (e.g., Rotterdam, NIH, AE-PCOS) and GDM diagnostic criteria (one-step vs. two-step) across included studies is a major source of heterogeneity and limits direct comparisons.,Pooling of cohort and case-control studies in the primary estimate is methodologically heterogeneous; the overall 24% estimate is likely influenced by case-control study design, which may overestimate prevalence.,Associated factor odds ratios are reported without confidence intervals or p-values in the available abstract, making it impossible to assess statistical precision or significance for individual predictors.,Small total sample size (n = 5,507 across 22 studies) relative to the breadth of the research question limits precision and statistical power for subgroup analyses.,Publication bias was assessed with Egger's test; results were not detailed in the abstract, leaving its potential impact on the pooled prevalence unclear.,Generalizability is uncertain: database inclusion of Chinese-language sources (CNKI, Wanfang, Sinomed, VIP) suggests probable overrepresentation of East Asian populations; ethnic and geographic diversity of the sample is not reported.,Only abstract available for review; risk-of-bias assessments, adjusted confounder analyses, and full subgroup data could not be independently verified.,Funding source not reported, precluding assessment of potential sponsorship bias. __BOTTOM_LINE_CATEGORY__ potentially_practice_informing __BOTTOM_LINE__ Among pregnant women with PCOS, the pooled GDM prevalence is approximately 21.7%–24%, representing a clinically meaningful burden that supports universal GDM screening in this population. Key potentially modifiable risk factors include pre-pregnancy BMI, gestational weight gain, and insulin resistance (HOMA-IR), alongside the non-modifiable factor of family history of diabetes. However, extreme heterogeneity (I² = 89.9%, prediction interval 10%–46%) and variable diagnostic criteria across studies mean that this estimate should inform risk stratification rather than replace individualized clinical assessment; practice should not change beyond existing guideline-based screening recommendations pending higher-quality, more homogeneous primary data.