An algorithm for the operational assessment of adverse drug reactions. III. Results of tests among clinicians

J. M. Leventhal, T. A. Hutchinson, M. S. Kramer and A. R. Feinstein

To determine how practicing clinicians use a recently developed algorithm for the diagnostic assessment of suspected adverse drug reactions (ADRs), eight clinicians--four board-certified, practicing physicians and four interns--rated the likelihood of 30 suspected ADRs. Each physician reviewed the case summaries, first using implicit clinical judgment and two months later by means of the ADR algorithm. The algorithm significantly improved the reproducibility of the senior clinicians' assessments as compared with their implicit assessments; however, the improvement in the interns' assessments with the algorithm was not significant. The validity of the physicians' assessments, which was measured by comparing their ratings with a consensus rating of the three algorithm developers, was also significantly improved by the use of the algorithm. When used by practicing clinicians, the algorithm improves the reproducibility and validity of their assessments of ADRs and should provide a more precise diagnostic approach to these complex clinical phenomena.