PubMed1997
Norway has a national programme for quality assurance of laboratory analyses performed in general practice. A central laboratory distributes control material, and a local team gives practical assistance in each county. To perform laboratory tests properly, the use of quality control material ("internal control") is essential. The users' control results provide a basis for finding both systematic error and the analytical coefficient of variation. The analytical and biological coefficients of variation can be used to calculate the confidence intervals around laboratory test results, and around medical decision thresholds. It is also possible to calculate the magnitude of significant changes in serial results ("critical difference"). In Rogaland county, 11-14 general practitioners showed an analytical coefficient of variation ranging from 0.5 to 5.0% on haemoglobin determination, from 1.2 to 9.9% on glucose, and from 3.1 to 10.5% on prothrombin time. Five general practitioners showed an analytical coefficient of variation from 3.3 to 9.7% for cholesterol. This gives a critical difference ranging from 7.8 to 15.8% for haemoglobin and 18.8 to 31.6% for cholesterol. We present a diagram that can be used to find the critical difference for any test. Such an evaluation of laboratory analyses should become routine in all near-patient testing.