Limited usefulness of the proportion of tests with normal results in review of diagnostic services utilization.

We tested the hypothesis that the proportion of normal results in tests ordered by a physician correlates directly with the proportion of tests that are disapproved by a utilization review audit for inappropriate test use. Auditors reviewed 3497 charts for 13 067 repeated determinations of 10 diagnostic tests ordered by 81 medical residents to determine if test use had been appropriate, according to explicit criteria. By the Pearson product-moment correlation coefficient, there was a significant (p less than 0.001) correlation between the proportion of normal results in tests ordered by residents and the residents' audit-disapproval rates, a previously undocumented relationship. Although statistically significant, the correlation coefficient of 0.25 means that the proportion of normal tests results predicted only 6% of the variance in the audit-disapproval rates. This relationship is not strong enough to use the proportion of normal results as a measure of a physician's inappropriate test use, except to select physicians for utilization review.