In sensitivity analyses, we found no substantive change for any of these findings when alternative definitions of metabolic syndrome or alternative thresholds for BMI or HbA1c were used. In analyses stratified by age, the HR for incident diabetes associated with rosuvastatin as compared with control was 1·26 (95% CI 1·02–1·56) for participants aged 50–69 years and 1·25 (0·90–1·74) for those aged 70 years and older.8