Identifying patients with hypercholesterolemia. More than one blood sample is needed.

OBJECTIVE

To compare the use of one and two blood samples for diagnosing hypercholesterolemia

DESIGN

A test-retest substudy conducted as part of a randomized control trial designed to compare the effectiveness of different counseling strategies for lowering serum cholesterol, dietary fat, and dietary cholesterol in patients with moderate hypercholesterolemia.

SETTING

Thirty urban family practices.

PARTICIPANTS

One hundred forty-two patients provided two blood samples for total cholesterol (TC) level determination at two different times (test results were being used as an eligibility criterion for enrollment in the main trial).

MAIN OUTCOME MEASURES

Number of subjects correctly classified to cholesterol risk category (normal < 6.2 mmol/L; moderate 6.2 to 6.9 mmol/L; high > 6.9 mmol/L) on the basis of one TC value and on the average of two TC values.

RESULTS

Overall misclassification rate on initial TC level was 22.5%. Overall false-positive rate was 19.0%, but false-positive rate for those initially assigned to the high category was 50%. Overall false-negative rate was 3.5%. Misclassification rates did not differ statistically on the basis of age, sex, blood pressure, smoking status, family history of coronary heart disease, presence of diabetes, obesity, the laboratory used, or whether the patient had fasted before giving blood.

CONCLUSIONS

Single TC levels are too unreliable for diagnostic purposes, even if the subjects fast before testing. Family physicians should base their treatment decisions on the average of two cholesterol readings taken at different times 1 to 8 weeks apart.