This topic was discussed at the 2009 IDF World Diabetes Congress. You can read very good summaries of two presentations there, one based on ADVANCE data and another on a study done in Greece, in this article:
Heartwire: Current risk equations may overestimate CVD risk in diabetic patients
The IDF presentation which was based on the ADVANCE data reported,
...the Framingham and UKPDS risk equations overestimated four-year CHD risk by as much as almost 300%. [i.e. it predicted three times as many cases of heart disease as were actually diagnosed.]The IDF presentation which was based on the study of a Greek population reported, according to the Heartwire article,
... that in a five-year study of over 900 diabetic patients in Greece, the Framingham and UKPDS risk equations predicted about a 10% incidence of CHD, whereas the actual incidence was significantly lower, at 6.8%.You can read the abstract of the study based on Hoorn data which was just published in Diabetes Care here:
Prediction of Coronary Heart Disease Risk in a General, Pre-Diabetic, and Diabetic Population During 10 Years of Follow-up: Accuracy of the Framingham, SCORE, and UKPDS Risk Functions: The Hoorn Study van der Heijden, et al., doi: 10.2337/dc09-0745
Diabetes Care November 2009 vol. 32 no. 11 2094-2098
This study compared the Framingham, UKPDS, and a third formula, SCORE, and concluded
The Framingham and UKPDS prediction models overestimated the risk of first CHD event in all glucose tolerance groups. Overall, the prediction models had a low to moderate discriminatory capacity. Further analysis revealed that the SCORE equation did the best job of predicting heart disease in people with normal glucose tolerance and that the UKPDS did a slightly better job in the group of people diagnosed with elevated blood sugars, though the UKPDS still overestimated risk in this group too.
The findings of these studies conflicts with the finding published n 2007 of the DECODE study which found the Framingham equation underestimated the risk of heart disease in it's population.
What The Risk Formulas Ignore
After using online calculators to assess my own risk using all three formula cited in the above studies, I noticed that several important parameters are completely missing.
1.The Calculators Rely on Heavily on Cholesterol Ratios However, most interestingly, none of these calculators has factored in statin use or looked at whether long term statin use decreases predicted mortality using any of these risk formulas. My guess is that if they did, you would have heard about it from the drug companies pushing statins.
2.The Calculators Ignore C-Reactive Protein (CRP) Given that there seems to be a strong link between the presence of inflammation as measured by CRP and heart attack, this omission is important.
3.The Calculators Ignore BMI or Other Measures of Obesity. One wonders, again, if this is because they were found NOT to correlate to risk in the studies used to define these formula. This is quite possible, since in my own experience heart disease strikes people of all weights.
Calculate--and Overestimate--Your Own Risk
SCORE uses slightly different formulas for different European populations, but not all appear to be available online. You can find an online SCORE calculator based on data from the UK populations Note that this calculator uses fasting glucose to determine whether someone is diabetic and refuses to calculate risk if they are, because of the assumption--disproven by the research cited above, that everyone with diabetes is very high risk. To use this calculator, use the category that corresponds to your fasting glucose.
SCORE Risk Calculator: UK
The UKPDS calculator can be downloaded here:
UKPDS Risk Engine Download Page
The Framingham calculator can be found here: 10-Year CVD Risk Calculator (Framingham)