A Priori Diagnosis and Diagnostic Errors

Methods:

Study design:

This is a prospective randomized controlled single-blind trial, the participants not being aware of the purpose of the study. The trial was approved by the regional ethical committee (EKNZ 85/04) and all participants gave written informed consent.

Participants:

Medical students in their 4th year at the University of Basel, Switzerland were offered to participate in voluntary workshops at the simulator centre of the medical intensive care unit of university of Basel hospital. The present study was conducted as part of these workshops.

At the beginning of the workshop participants were asked to fill in a questionnaire. The questionnaire included a) the Big Five personality traits (neuroticism, extraversion, openness, agreeableness, conscientiousness) (McCrae 1992); b) Rosenberg's self-esteem scale(Colani); c) the task to diagnose the cardiac rhythm in 12 one-lead ECG rhythms strips (each correct diagnosis was counted as one point resulting in a maximum of 12 attainable points); d) the task to list the typical symptoms of an acute myocardial infarction in bullet point form (each of the following items were counted as one point resulting in a maximum of 6 attainable points: chest pain; radiation of the pain; pain described as constricting, compressing, or squeezing; at least one accompanied vegetative symptom; dyspnoea; and fear); and e) the task to list the typical signs symptoms of 4 further medical emergencies (anaphylactic shock, trauma management, cardiac arrest, syncope) in bullet point form (answers were coded using a predefined checklist).

Simulator and Scenario:

The study was conducted using a patient simulator. All participants received a standardised instruction of the patient simulator (human patient simulator, Leardal SimMan®). This simulator's features include among others palpable pulses, spontaneous breathing with visible thoracic excursions, spontaneous movements of the eye lids and a speaker in the mannequin's head that broadcasts the voice of a remote operator, allowing for verbal interaction with the participants. For the present trial, the operator was an experienced physician.

For the study scenario participants were given the role of a resident-on-call in the emergency department and were instructed by a senior physician (confederate) to triage a newly arrived patient. Upon entering the simulator room, the participants encountered the patient (simulator) lying on a bed and dressed in a hospital gown. Participants were assisted by a confederate nurse. The nurse was instructed to display a helpful manner, to act only on command, but not to interfere in any way with the assessment and history taking.

Intervention:

All participants performed the scenario on their own and faced a patient with a simulated acute myocardial infarction. Participants were randomized (computer-generated numbers) to three versions of the scenario. The three versions differed only in the reply of the patient to the initial question about the reason for his visit. The patient volunteered that he had phoned his family physician because of acute chest pain and dyspnoea. The family physician had recommended urgent visit of the emergency department because he judged the situation as 1) medical emergency (control group with no à priori diagnosis); 2) acute myocardial infarction (correct à priori diagnosis group); or, 3) acute pulmonary embolism (wrong à priori diagnosis group). Apart from this randomly allocated initial information, the scenario was identical in all groups: until the end of the scenario, the patient only spoke when asked a question and all answers were given according a pre-scripted answer check-list (table) Participants of the groups with no à priori diagnosis and correct à priori diagnosis were given 4 minutes study time. Participants of the group with wrong diagnosis were given one additional minute (i.e. 5 minutes study time) to compensate for probing the given diagnosis. At the end of the study time, the senior physician entered the room and announced that he required the participant to temporarily leave his/her patient to help with another emergent case. The participants were then asked on their presumptive diagnosis and which measures, if any, the nurse should take during their temporary absence during the next couple of minutes.

Data analysis:

Data analysis was performed using video recordings obtained during simulations. Video recordings were analysed independently by two trained raters who noted all questions asked and all measures taken by the participant. Inter-observer differences were solved by jointly re-watching the video recordings.

Statistics:

The primary endpoint was a correct presumptive diagnosis of an acute myocardial infarction or acute coronary syndrome when asked at the end of the scenario by the senior physician or when previously revealed to the patient and/or the nurse. Mentioning the correct diagnosis as possible differential diagnosis was rated as correct diagnosis. A co-primary endpoint was the composite endpoint of a correct presumptive diagnosis OR ordering a 12-lead ECG, the rationale being that a timely 12-lead ECG will most likely prevent harm to a patient not correctly triaged as suffering from acute myocardial infarction.

Secondary endpoints were the number and kind of questions asked during history taking; probing of the à priori diagnosis; measures taken during assessment; and the results of the questionnaire. Results of the knowledge part of the questionnaire are reported as percentages of the attainable maximum.

A 20% difference in the primary outcome was considered to be of medical relevance- A power analysis revealed that approximately 150 teams (50 per group) had to be included to detect this difference with a power of 0.9 at the 0.05 significance level.

Statistical analysis was performed by SPSS (version 25). Non-parametric ANOVA, Mann-Whitney test, chi-square test, Univariate and Multivariate regression analysis, and Cohen's kappa were used as appropriate. A p < 0.05 was considered to represent statistical significance. Data are reported as median [Interquartile range {IQR}] unless otherwise stated.