Controlled trial using computerized feedback to improve physiciansʼ diagnostic judgments

Abstract

The goal of this study was to test an innovative method to improve physicians' diagnostic judgments by integrating the use of a computer program (employing cognitive feedback to teach a clinical rule that predicts the probability of streptococcal pharyngitis), a traditional lecture, and periodic disease-prevalence reports. In a controlled trial using pre- and postintervention measures involving 885 patients, the authors compared the effects of the integrated method on the diagnostic judgments of seven experienced physicians at a university health service (from 1982 to 1985) with the effects of the lecture alone on the judgments of seven experienced physicians at a different university health service (1986 to 1987). The integrated method significantly improved the quality of the physicians' judgments as measured by calibration curves and Brier scores, and increased the level of agreement between the physicians' judgments and those made by the clinical prediction rule. The lecture alone produced less improvement in the quality of the physicians' judgments, and decreased the level of agreement with the rule. The authors conclude that this method, based on cognitive psychology, is a promising educational tool.

Le raisonnement clinique: données issues de la recherche et implications pour l’enseignement

“…Lorsque l'enseignant met à jour de façon explicite les éléments discriminants d'une situation clinique donnée, les étudiants les retiendront davantage et les utiliseront mieux ensuite, phénomène appelé rétro-action cognitive par ces auteurs. L'effet d'apprentissage sera d'autant plus bénéfique si l'enseignant note également la probabilité a priori des diagnostics entretenus [82][83][84] . Les objectifs de base de l'enseignement du raisonnement clinique et quelques moyens pour les réaliser sont proposés dans le tableau 3.…”

Social Judgement Theory and Medical Judgement

“…This cue is one that was highly weighted overall, and was emphasised by the textbooks of the time. In another example, physicians who worked in a university student health clinic used widely varying cue weights for each of seven variables in diagnosing strep throat in vignettes of patients with pharyngitis (Poses et al, 1992). Kirwan asked rheumatologists to estimate the change in disease activity in a series of "paper cases" about patients with rheumatoid arthritis.…”

“…In addition to recording patient data, such studies often ask physicians seeing the patients to record their predictions of the outcome. This provides the physician's judgements (the right side of the lens) and, along with the outcome data, allows investigators to measure the accuracy of the judgements (Goldman et al, 1988;Poses et al, 1992;Pozen et al, 1984;Speroff et al, 1989a;Tape et al, 1991).…”

“…In these modified lens model studies, the cases are produced using a balanced, orthogonal design, and the clinicians' judgements are compared to outcomes as calculated from the statistical rule, not the actual outcome of each case (Tape, Kripal, & Wigton, 1992;Poses et al, 1992;Wigton et al, 1986b). Here achievement is the correlation between the physicians' judgments (J s ) and the predicted outcome (J¢ e ), not the actual outcome (J e ).…”

Bayesian Communication: A Clinically Significant Paradigm for Electronic Publication

“…Can readers express their belief in terms of probabilities? There is a fair literature on calibration that suggests that physicians can be calibrated through training 51,52 or through experience. Our own data (unpublished) on 825 microbial culture predictions by pediatric residents suggest that, untrained, residents' predictions between 20 and 80 percent are calibrated, but, as others have shown, they are overconfident for high probabilities and underconfident for low.…”