Long-term retrospective control procedures for monitoring analytical performance of laboratory instruments.

We propose a statistical procedure for long-term quality-control of laboratory instruments, including daily, day-to-day, and monthly evaluations. The procedure is based on the unique and unequivocal interpretation of five results for control sera by calculation of a Reliability Index and further manipulations of this unitless parameter. This method, which we have tested during the past two years, allows for monitoring analytical performance and making comparisons with results of interlaboratory surveys. The monthly analytical variability, expressed as "total error," is an indicator of the clinical usefulness of analytical results.