The efforts of Paynter et al1 are a very useful addition to the growing drum beats calling for appropriate cardiovascular disease (CVD) risk estimation among people with diabetes mellitus. We also share the view that individualized risk estimation should form the basis for implementing CVD risk-reducing strategies in people with diabetes. Indeed, there is robust evidence that diabetes status may not be a coronary heart disease (CHD) risk equivalent in all circumstances2,3 and that the distribution of global CVD risk among people with diabetes follows a gradient, just like in the general population. Inherent to the concept of “CHD equivalent” is the 10-year risk threshold greater than 20% with no assumption about risk distribution beyond, which obviates the need for individualized risk estimation. Therefore, we are uncertain about the validity of using such a concept as a basis for model building and comparison. With the increasing attention to hemoglobin A1c (HbA1c) measurement for diabetes diagnosis, HbA1c values will be acquired increasingly in nondiabetic persons. Assuming that HbA1c would be less available in nondiabetic population, the authors set HbA1c values to zero in their nondiabetic subcohort. This generates a new predictor whose properties may differ from those of HbA1c measures contributing predictive information across the entire cohort.