Do we know what inappropriate laboratory utilization is? - A systematic review of laboratory clinical audits

Objective.-Laboratory utilization has steadily increased, and some studies suggest inappropriate utilization. Therefore, we wished to assess studies that measure inappropriate laboratory use in light of methodological criteria. Design.-Systematic review of published studies. Data Sources.-MEDLINE, HEALTHSTAR, and EMBASE databases were searched from 1966 to September 1997 using a broad and inclusive strategy with no language restriction. In addition, the references of all retrieved studies and 3 textbooks on diagnostic testing were hand-searched. Study Selection.-All studies that provided and applied criteria for inappropriate laboratory use. Data Extraction.-Studies were categorized based on whether the criteria were implicit (objective criteria for inappropriate utilization not provided or very broad) or explicit. Guidelines for evaluation were applied to each study by a single reviewer. Data Synthesis.-Forty-four eligible studies were identified. Eleven studies used implicit criteria for inappropriate laboratory utilization and contained small numbers of patients or physicians. Most did not adequately assess the reliability of the implicit criteria. Thirty-three studies used explicit criteria based on the appropriateness of test choice, frequency, and timing, as well as the probability of a positive result. There were large variations in the estimates of inappropriate laboratory use (4.5%-95%). Evidence supporting the explicit criteria was frequently weak by the standards suggested for therapeutic maneuvers, but was nonetheless compelling based on principles of physiology, pharmacology, and probability. Conclusions.-Many studies identify inappropriate laboratory use based on implicit or explicit criteria that do not meet methodological standards suggested for audits of therapeutic maneuvers. Researchers should develop alternative evidentiary standards for measuring inappropriateness of laboratory test use.