Total and high-density lipoprotein cholesterol measurements. Hazards in clinical interpretation.

Duplicate plasma specimens from 24 persons were sent to a community hospital, a commercial laboratory, and a university lipid research laboratory at two separate times to assess the intralaboratory and interlaboratory variations of total and high-density cholesterol measurements. For all three laboratories, the 95% confidence limits for the reproducibility of total cholesterol levels are about +/- 5.5%. For high-density lipoprotein (HDL) values, they are +/- 7%, +/- 14.45%, and +/- 9%. Interlaboratory differences for total cholesterol at the 95% confidence level are +/- 7.7% and for HDL +/- 18.4%. Construction of a "cardiac risk ratio" of total to HDL cholesterol levels is subject to confusion because small errors in HDL cholesterol levels produce large errors in the ratio.