Predicting coronary risk in the general population--is it necessary to measure high-density lipoprotein cholesterol?

Background The Joint British Societies Coronary Risk Prediction Charts recommend the use of a high-density lipoprotein cholesterol value of 1 mmol/l where actual values have not been measured. It is important to quantify the impact of this advice if risk assessments are to be sufficiently accurate to guide treatment decisions. Design The risks of 5005 individuals from the Health Survey for England 1998 were calculated using the Joint British Societies charts. Each individual's risk was recalculated assuming a high-density lipoprotein cholesterol value of 1 mmol/l. These risk estimates were compared with those derived from the Framingham equation. Methods Using the Framingham equation as the gold standard, the positive and negative predictive values, sensitivity and specificity with 95% confidence intervals of the Joint British charts with actual and estimated high-density lipoprotein cholesterol values were calculated. Results At the 30% 10-year coronary heart disease risk threshold using measured high-density lipoprotein cholesterol values, the charts had a sensitivity of 83% and specificity of 99%. Using an estimated high-density lipoprotein cholesterol value of 1 mmol/l reduced the sensitivity to 58% with a specificity of 98%. Conclusions In the presence of measured high-density lipoprotein cholesterol values there was good agreement between the Framingham equation and the Joint British Societies charts. The use of a fixed high-density lipoprotein cholesterol value of 1 mmol/l introduced important and significant errors into the risk assessment. This study reinforces the need to measure both total and high-density lipoprotein cholesterol when assessing coronary risk. (C) 2003 Lippincott Williams Wilkins.