Failure to demonstrate effects of interruptions on diagnostic reasoning: three experiments

• Main Authors: Alajaji, Mai, Saleh, Nada, AlKhulaif, Ali Hassan, Mamede, Silvia, Rotgans, Jerome Ingmar, Sukkarieh, Hatouf, AlHarbi, Nouf, Magzoub, Mohi Eldin, Schmidt, Henk G.
• Other Authors: Lee Kong Chian School of Medicine (LKCMedicine)
• Format: Article
• Language: English
• Published: 2022
• Subjects: Science::Medicine; Diagnostic Error; Cognition
• Online Access: https://hdl.handle.net/10356/161357
• Institution: Nanyang Technological University

Background: Diagnostic error is a major source of patient suffering. Research shows that physicians experience frequent interruptions while being engaged with patients and indicate that diagnostic accuracy may be impaired as a result. Since most studies in the field are observational, there is as yet no evidence suggesting a direct causal link between being interrupted and diagnostic error. The experiments reported in this article were intended to assess this hypothesis. Methods: Three experiments were conducted to test the hypothesis that interruptions hurt diagnostic reasoning and increase time on task. In the first experiment (N=42), internal medicine residents, while diagnosing vignettes of actual clinical cases were interrupted halfway with a task unrelated to medicine, solving word-spotting puzzles and anagrams. In the second experiment (N=78), the interruptions were medically relevant ones. In the third experiment (N=30), we put additional time pressure on the participants. In all these experiments, a control group diagnosed the cases without interruption. Dependent variables were diagnostic accuracy and amount of time spent on the vignettes. Results: In none of the experiments interruptions were demonstrated to influence diagnostic accuracy. In Experiment 1: Mean of interrupted group was 0.88 (SD=0.37) versus non- interrupted group 0.91 (SD=0.32). In Experiment 2: Mean of interrupted group was 0.95 (SD=0.32) versus non-interrupted group 0.94 (SD=0.38). In Experiment 3: Mean of interrupted group was 0.42 (SD=0.12) versus non-interrupted group 0.37 (SD=0.08). Although interrupted residents in all experiments needed more time to complete the diagnostic task, only in Experiment 2, this effect was statistically significant. Conclusions: These three experiments, taken together, failed to demonstrate negative effects of interruptions on diagnostic reasoning. Perhaps physicians who are interrupted may still have sufficient cognitive resources available to recover from it most of the time.