Statistical Analysis

SPSS 20.0 software (IBM Corporation, Armonk, NY, USA) was used to perform all the data analysis unless specified. Normal distribution was checked using Kolmogorov-Smirnov test. Continuous data were presented as mean ± standard deviation (SD) if their normality was not rejected, or median (interquartile range) otherwise. Categorical variables were expressed as number (percentage). For variables with normal distribution, comparisons between groups were performed using independent Student t-test. Mann-Whitney U test was used to compare differences of variables whose normality was rejected. χ² test or Fisher's exact test where appropriate was used to compare differences in categorical variables between groups. Effect sizes of comparisons between groups were calculated as previously reported (21, 22). After checking P-P plots of standardized regression residuals for normality, logistic regression analysis was performed to obtain odds ratios (ORs) and their 95% confidence intervals (CIs) of types of diabetes and other factors for MetS. A structured adjustment scheme was used to adjust for confounding effects of other variables. First, we performed univariable analysis. Second, we performed multivariable analysis with stepwise selection of confounders from traditional and potential MetS risk factors (P = 0.05 for entry and P = 0.10 for exit), including age, gender, region, education attainment, family history of diabetes, smoking and alcohol consumption habits. Third, we further adjusted for HOMA2-IR to check whether the increased or decreased risks of MetS in T2DM and T1DM was attributable to different levels of insulin resistance in those types of diabetes. A two-sided P < 0.05 was considered statistically significant.