The role of fasting plasma glucose (FPG) levels below diabetes “thresholds” in predicting
mortality or coronary heart disease (CHD) is unclear. This study examines whether FPG
predicts mortality or CHD in subjects without diabetes (historical or undiagnosed) or in
those with undiagnosed diabetes (or lesser degrees of glucose intolerance). We have
analyzed all-causes mortality and CHD incidence from a 16-year follow-up in a cohort of
Australian senior citizens, 60 years and older, first examined in 1988–89. Diabetes was
defined on historical grounds or by use of medication; undiagnosed diabetics were those
without history but with FPG >124 mg/dl. Hazard ratio and 95% confidence intervals of
the specified outcomes were obtained from Cox models, with FPG being entered as a
continuous variable. Mortality and CHD incidence rates in subjects with previous cardiovascular
disease (CVD) and diabetes were substantially higher than in nondiabetics, but
CHD rates were disproportionately higher in diabetic women. FPG did not significantly
predict any outcome in men in the absence of diabetes. In women, FPG was a significant
predictor of death (hazard ratio = 1.30, 95% confidence interval 1.09 to 1.56) and CHD
(hazard ratio 1.24, confidence interval 1.02 to 1.51) in the cohort, which included previous
CVD but excluded all diabetes. In women with undiagnosed diabetes, FPG predicted death
independently of previous CVD presence but did not predict CHD. In conclusion, FPG in
the range of 95 to 108 mg/dl in a nondiabetic woman is still of prognostic importance for
survival or CHD if she has previous CVD, whereas FPG is of prognostic importance for
survival if she has undiagnosed diabetes. No similar findings were made in men.