Coronary heart disease risk assessment in diabetes mellitus - a comparison of PROCAM and Framingham risk assessment functions

Aims To assess any differences between coronary heart disease (CHD) risks calculated by the Framingham equation and those calculated by the PROCAM equation in men with and without diabetes mellitus, and whether any such differences are associated with the hypertriglyceridaemia of diabetes mellitus. Methods Clinical and biochemical data collected from 1774 men seen in either general practice, a hospital diabetes or lipid clinic. CHD risks were calculated by both the Framingham and PROCAM functions and comparisons made between those patients with and those without diabetes. Results Of the 1774 men only 996 fulfilled the criteria for assessment by the PROCAM equation and thus further analysis. Patients with diabetes mellitus had significantly higher serum triglyceride levels than those without (1.9 mmol/l vs. 1.7 mmol/l). Median annual CHD risks calculated by the Framingham function were 1.7% in the patients with and 1.32% in the patients without diabetes mellitus, whereas those calculated by the PROCAM function were 0.77% and 0.6%, respectively. Bland-Altman difference plots showed that in both groups of patients the PROCAM equation systematically underestimated risk in comparison with the Framingham equation at low levels of risk but overestimated at higher levels of risk. The shape of the plots in each group of patients was, however, similar. Conclusion There were no systematic differences between CHD risks calculated by the two different equations in patients with diabetes compared with those without, despite the higher serum triglyceride levels associated with diabetes. Restrictions in the use of the PROCAM function meant that only 56% of the original cohort could be assessed in this way. Thus the Framingham equation remains the most suitable method of CHD risk prediction for UK patients with and without diabetes mellitus.