Impact of within-person variability on identifying children with hypercholesterolemia: Framingham Children's Study

The National Cholesterol Education Program guidelines recommending that children be screened to detect abnormal levels of blood lipids depend on accurate assignment to risk categories. Our objective was to assess the impact of within-person variability of blood lipids on such risk classification.

We measured total cholesterol (TC) and high-density lipoprotein-cholesterol (HDL-C) at three successive weekly visits in 24 children aged 6 to 9 years.

Using within-person variance estimates, we constructed 95% confidence intervals to evaluate how well one, two, and three lipid measurements per child allowed assignment of risk as recommended by the National Cholesterol Education Program.

The mean, overall, and within-person standard deviations were 145.7, 24.5, and 9.2 mg/dl, respectively, for TC, and 42.8, 10.1, and 4.6 mg/dl, respectively, for HDL-C. One TC measurement allowed confident assignment to the "acceptable" (less than 170 mg/dl) category only if less than 154.9 mg/dl, and to the "high" (greater than or equal to 200 mg/dl) category only if greater than 215.1 mg/dl. The average of two TC values allowed assignment to these categories if less than 159.3 or greater than 210.7 mg/dl, respectively. For the average of three TC values, these ranges were less than 161.3 and greater than 208.7 mg/dl, respectively. With one TC measurement, no value allowed assignment to the "borderline" (170 to 199 mg/dl) category. The average of two TC values allowed assignment to the "borderline" category if between 180.7 and 189.3 mg/dl. Using the average of three values enlarged this range by 4 mg/dl. For HDL-C, multiple measurements improved risk assignment in a similar fashion.

The magnitude of within-person variability of TC and HDL-C limits the ability to classify children into risk categories recommended by the National Cholesterol Education Program.