BackgroundPolycystic ovary syndrome is the most common endocrine disorder in women of reproductive age. This disease is also a metabolic disorder because of the high prevalence of accompanying metabolic disturbances, which might pose a high long-term risk of diabetes. The aims of this study were to analyse the prevalence, prognosis, and risk factors of insulin resistance, impaired glucose tolerance, and diabetes in Chinese patients with polycystic ovary syndrome. MethodsData recorded from July, 1998, to December, 2015, of patients with polycystic ovary syndrome were extracted from the database of our department; patients without relevant data were excluded. Retrospective analysis of data was done from first visit versus last follow-up. Polycystic ovary syndrome was diagnosed according to the Rotterdam 2003 criteria. Homeostatic model assessment-insulin resistance was 2·14 mIU/L or higher and fasting insulin of 11·87 mIU/L or higher. Abnormal glucose metabolism included impaired glucose tolerance (IGT) and type 2 diabetes diagnosed according to the 2006 WHO criteria. Disease regression was patients with baseline insulin resistance reversed to non-insulin resistance, or those with baseline IGT reversed to non-IGT at follow up. Cox regression analysis was used for hazard ratios (HRs) of variables for prognosis of insulin resistance or abnormal glucose metabolism. The cutoff value of insulin or glucose concentration was set to the 75th percentile of the value in referential groups. Findings2931 patient records with all the relevant data were selected for analysis. The mean patient age at baseline was 26 years (SD 5·28, range 12–43). The mean follow up time was 2·18 years (SD 2·25; range 0·1–16·0). There was no effect on prognosis of insulin resistance and abnormal glucose metabolism with clinical polycystic ovary syndrome phenotypes, however polycystic ovary syndrome decreased the risk of regression of insulin resistance being temporary (HR 0·30 [95% CI 0·09–0·94]). Baseline fasting insulin concentrations (HR 1·11 [95% CI 1·07–1·15]) and change in fasting insulin concentration from baseline to follow up (1·08 [1·06–1·11]) increased the risk of development of insulin resistance. Glucose concentration 1 h after oral glucose (HR 1·17 [95% CI 1·05–1·31]), glucose concentration after 2 h (1·48 [1·22–1·79]), change in 1 h glucose concentration from baseline (1·18 [1·12–1·25]), and change in 2 h glucose concentration from baseline (1·09 [1·04–1·15]) increased the risk of development of IGT. Glucose concentration after 1 h (HR 1·25 [95% CI 1·05–1·49]) and after 2 h (1·47 [1·19–1·81]), change in 1 h glucose concentration (1·22 [1·11–1·36]), and in 2 h (1·20 [1·08–1·32]) increased the risk of development of diabetes. Development of insulin resistance was more likely in patients with a baseline fasting insulin concentration of more than 8·82 mIU/L (p=0·001) or change in fasting insulin concentration of more than 2·91 mIU/L (p<0·0001). Development of glucose intolerance was more likely in patients with a baseline 1 h glucose concentration of 8·4 mmol/L (p=0·0002), a 2 h glucose concentration of more than 6·6 mmol/L (p<0·0001), a change in 1 h glucose concentration of more than 1·7 mmol/L (p<0·0001), or change in 2 h glucose concentration of more than 1·1 mmol/L (p<0·0001). Development of diabetes was more likely in patients with a baseline 1 h glucose concentration of more than 9·4 mmol/L (p=0·0026), a 2 h glucose concentration of more than 7·6 mmol/L (p<0·0001), a change in 1 h glucose concentration of more than 1·9 mmol/L (p<0·0001), or a change in 2 h glucose concentration of more than 1·4 mmol/L (p<0·0001). InterpretationPatients with polycystic ovary syndrome seem to have high prevalence of insulin resistance and abnormal glucose metabolism, whereas clinical polycystic ovary syndrome phenotypes seem to have no effect on the prognosis of insulin resistance and abnormal glucose metabolism. Subclinical insulin resistance, abnormal glucose metabolism, baseline impaired insulin or glucose concentrations, or worsening of these factors during follow-up seem to be predictors for development of insulin resistance or abnormal glucose metabolism in patients with polycystic ovary syndrome. FundingNational Natural Science Foundation of China and Sun Yat-sen University.