Statistical analysis
Linear regression models were used to assess differences in mean biomarker levels between PCOS and control groups, adjusting for age, BMI and smoking wherever the latter were found to have explanatory value. Best fitting models were selected on the basis of Akaike’s Information Criterion (AIC). A two-tailed p-value of <0.05 was considered significant. All analyses were performed with the GraphPad Prism version 5 Windows and the R language and environment for statistical computing.