Analytical goals developed from the inherent error of medical tests

Abstract

Biological variation is inherent in all medical tests and is a major source of error in clinical test interpretation. We use inherent test error due to biological variability as the touchstone by which to validate the clinical acceptability of analytical goals. Analytical goals should consist of sets of complementary limits for total error, bias, and imprecision and a medical usefulness criterion (MUC) to ensure that the total error goal is met with stated certainty. We propose analytical goals that are limits for the total error of a measurement system, the joint capability of the process control procedures and the measurement process. Thus, the goals are the allowable analytical error for proficiency evaluation, internal process control, and the total error budget for design of method performance characteristics. We define MUC as the allowable maximum proportion of test diagnostic efficiency lost due to analytical error near medical decision points. MUCs < 0.03 are not routinely clinically useful; MUCs > 0.08 allow method performance characteristics that are undesirable for general clinical use. Using MUC = 0.05 and recommendations that imprecision not exceed one-half of the biological CV, we propose that fixed bias should not exceed one-fourth to one-third of the intraindividual biological CV of the analyte and that total analytical error should not exceed 1.25 to 1.40 times the biological CV. Modification of these analytical goals may arise for reasons of documented clinical practice or process stability. Our model quantifies in medical utility terms the results of such modifications.