Fig. 3

A The dose–response of fasting blood glucose on the incidence of silent myocardial infarction. This association was estimated after adjusting for age, gender, race, studies, smoking status, diastolic blood pressure, blood low-density lipoprotein, BMI, and blood creatinine. B Subgroup analysis and sensitivity analysis between glucose and the incidence of silent myocardial infarction. All hazard ratios and 95% confidence intervals are estimated in fully adjusted models (adjusted for age, gender, race, studies, smoking status, diastolic blood pressure, blood low-density lipid, BMI, and blood creatinine). ¶: repeated measures fasting glucose (per 25 mg/dL); †: average glucose over time of each subject (per 25 mg/dL); ‡ variability of glucose (per 1 sd)