Efeito da ferramenta cognitiva slow como estratégia para reduzir viés de confirmação diagnóstica em estudantes de medicina

Abstract

Introduction: Cognitive bias is an important source of error of diagnostic, but it is a challenging area to learn and teach. Up to the moment, over than 100 cognitive errors have been described in the literature and at least 38 in the medical field. Among these biases, we highlight confirmation bias, also called confirmatory bias or confirmation bias, which is the tendency to remember, interpret or search for information in order to confirm or confirm initial beliefs or hypotheses. Goals: To verify the potential effect of a cognitive strength tool, known as SLOW in reducing the cognitive bias of diagnostic confirmation in the medical students of 11th period, comparing the diagnostic accuracy in the resolution of seven biased clinical cases. Methods: Experimental study in 11th-period medical students, comparing the diagnostic accuracy in solving seven clinical cases between a group that used the SLOW cognitive tool (GS) and a control group (CG). Using only cases elaborated with the objective of inducing confirmation bias, both groups should answer whether or not they confirmed the initial diagnosis in step 1 and indicating a new diagnosis if the initial diagnosis was not confirmed in step 2.Results: The verall diagnostic accuracy, considering the final score, was significantly different between students GS and students CG (2.50 ±1.47 vs 1.63 ± 1.28, respectively, p = 0.037), with students GS presenting final mean scores higher than those of the CG. When analyzing the diagnosis confirmation score, we observed that confirmation bias occurred in 36.3% of the diagnoses initially given in the SG and in 39.2% in the CG. After the analytical stage of the experiment, the CG students exchanged the initial biased diagnosis for an alternative diagnosis in only 9.1% of the cases; among GS students, this exchange occurred in 25.0% of the cases analyzed, this difference being statistically significant (p = 0.019) in the GS. Conclusion: Higher diagnostic accuracy was evidenced among students who used the cognitive tool SLOW when compared to students in the control group, in solving clinical cases constructed in order to induce confirmation bias.