Patients as their own controls: use of the computer to identify "laboratory error".

I describe a system of quality control based on computer detection of changes in individual patient test results. This system, called "delta check," was used to follow all the tests performed by the clinical chemistry laboratory in a 1200-bed hospital. Analysis of 22 months' experience indicates that specimen misidentification is a serious problem in the clinical chemistry laboratory. Over a nine-month period, errors were most frequent in the results for total thyroxine, total calcium, and total protein. Instances of laboratory error detectable by the delta check system are not detected by other currently used methods of quality control. This system therefore appears to be a valuable asset to the clinical laboratory.