Variability in cholesterol measurements in a worksite cholesterol screening program

Worksite cholesterol screening programs are becoming increasingly prevalent. Variability in cholesterol measurements frequently renders the interpretation of single or repeated cholesterol determinations problematic, complicating employee counseling within screening programs. In the setting of a corporate cholesterol screening program, this study quantifies variability in cholesterol measurements and calculates a 90% confidence interval (CI) for cholesterol measurements. We chose 15 men and 15 women to represent a broad range of cholesterol values. We obtained duplicate samples of blood from each subject at one-week intervals for a total of three weeks. We performed total cholesterol measurements in duplicate twice weekly on each blood sample for three weeks, yielding a total of 36 measurements per subject. Components of variability were estimated using a hierarchical random effects analysis of variance (ANOVA) model. The coefficients of variation for biologic and analytic variation were 5.2% and 3.4% for men and 4.7% and 3.2% for women. CIs were +/- 10.2% for men and +/- 9.3% for women. As a result of this variation, we recommend the routine use of CIs when discussing cholesterol measurements with patients.