Decision rules for predicting future lipid values in screening for a cholesterol reduction clinical trial

Recent large clinical trials have required screened patients to have serial measurements of an entry criteria variable, eliminating patients form further consideration if the average value is not in the eligibility range specified by the trial protocol. The increasing costs of large clinical trials required that they be executed efficiently. One way to improve efficiency would be to reduce the number of required screening measurements for a patient likely to be ineligible. A procedure is proposed that predicts the value of an average based on n measurements serially obtained on a patient during the screening phase when only m < n measurements are available. The employment of this procedure in a large clinical trial that uses low-density lipoprotein (LDL) cholesterol, total cholesterol, and triglycerides as entry criteria during the screening process is described. As a second example, this procedure is applied to population screening for lipid levels above a treatment threshold. The National Cholesterol Education Program recommends that the average of two LDL cholesterol measurements be used to determine whether LDL cholesterol is above 130 mg/dl, the threshold for treating patients with coronary heart disease. However, data from a sample of patients from a postinfarction population suggest that, if a single LDL cholesterol is above 146 mg/dl, the probability is greater than 95% that the average of the two LDL cholesterol measurements will be above 130 mg/dl.