Objective: To calibrate Reflotron-measured capillary total cholesterol (cTC) relative to a laboratory-measured venous total cholesterol (vTC) criterion standard for monitoring and screening hypercholesterolemia.
Study design and setting: Quasi-simultaneous assessment in 1999-2002 of Reflotron cTC and laboratory vTC in a random sample of 4,269 adult residents of Geneva, Switzerland (calibration development subsample n=3,067; validation subsample n=1,172), by means of bias, precision, correlation, sensitivity, and false positive percentage (calculated as 100-specificity) analyses of Reflotron cTC vs. laboratory vTC measures for predicting hypercholesterolemia.
Results: Total bias was small (-0.26 mmol/L), but there was substantial negative drift in Reflotron cTC (annual biases +0.08, -0.17, -0.27, and -0.60 mmol/L in 1999-2002). Overall, 57% of Reflotron cTC measurements for 894 hypercholesterolemic patients underestimated laboratory vTC (2%, 57%, 71%, and 98% in 1999-2002). Prior to calibration, sensitivity was 73% for the development and 35% for the validation subsample, with false positive at 4% (development) and 0.1% (validation). After calibration, sensitivity was 78% for the development and 37% for the validation subsample, with false positive at 5% (development) and 0.2% (validation). Using 95% upper prediction limits (UPL) for individual vTC values increased sensitivity to 99% and 83% and false positive percentage to 30% and 7% for the development and validation subsamples, respectively.
Conclusion: Crude results of Reflotron-measured cTC have poor sensitivity. Instead, 95% UPL can be used for monitoring and screening. Simply adding 0.8 mmol/L to a patient's observed Reflotron cTC value provides a very good approximation to the 95% UPL.