Study design
We conducted a prospective cluster randomized trial on multidisciplinary teams that work in an emergency department (ED) in Paris, France. Each team was randomized before the simulation session into a life group or in a death group. Randomization was performed using the “RANDOM” function of MS Excel (Microsoft, Redmond, WA, USA). Participants were blinded about study objectives and were advised that they were participating in a study designed to assess emotions whilst managing life-threatening situations. We followed CONSORT guidelines for the reporting of cluster randomized trial.
Objective and endpoints
Our primary objective was to study the effect of a simulated patient’s unexpected death on the change in learners’ anxiety caused by a life-threatening situation. We also studied participants’ feeling and appreciation of the training. We anticipated that the simulation-based courses will decrease anxiety amongst participants when facing a LTS regardless of group allocation. Our primary endpoint was the change in anxiety when facing a LTS between baseline and 1 month after the training. Secondary endpoints include satisfaction regarding the training and willingness to undertake further similar simulation-based courses. We also studied 3-month anxiety reduction.
We evaluated anxiety through the validated French version of the Spielberger state part of the State-Trait Anxiety Inventory (STAI-S) [18]. The state elements of the score measures the transitional emotional status evoked by a stressful situation (namely caring for a patient with a LTS), using 20 items each rated by a 4-point Likert scale (possible score range 20–80). Higher scores are positively correlated with higher levels of anxiety. A score >37 for men and >42 for women reflects high anxiety, and a score >48 for men and >55 for women corresponds to anxiety liable to interfere with performance [19]. The level of anxiety was evaluated in all subjects before the training and then assessed at 1 and 3 months using the state elements of the STAI. We also assessed baseline anxiety of subjects using the trait elements of the STAI. The state element of STAI assess the anxiety resulting from a situation or confrontation, whereas the trait element may be seen as an assessment of “baseline anxiety,” the anxiety caused by normal situation in day-to-day life [20]. These measures have been reported to have excellent interrater agreement for their assessment and good correlation with previously used methods to assess anxiety [19–24].
Electronic questionnaires were submitted to all participants to record anonymously their feeling and feedback of the training on a numeric scale that ranges from 0 (lower) to 10 (higher).
Selection of participants
Participants were physicians, medical students, nurses, and healthcare assistants employed in EDs in four academic hospitals in Paris, France. Physicians and medical students were enrolled from all four study sites. Nurses and healthcare assistants were enrolled from Pitie-Salpetriere hospital, where every nurse and healthcare assistant had a mandatory course of multiprofessional simulation-based training. We enrolled participants in this context, with informed consent. Participants that refused to participate and those with no follow-up at 1 month were excluded. We included participants during a 4-month period, from March to June of 2015. Each simulation team included a core consisting of one resident, two nurses and one healthcare assistant. To include all volunteers and keep the core composition, some teams included a fifth participant (healthcare assistant or medical student). Our institutional review board (Comite de protection des personnes—Paris Ile de France 6, Paris, France) authorized the study as it involved no patients. The Ethics Committee of the French society of intensive care (SRLF, Paris, France) approved the study. All subjects were followed up with a face-to-face interview and anonymous electronic questionnaire at 1 and 3 months.
Description of the training
All subjects participated in a simulation-based training (SBT) that comprised pre-briefing, scenario, and debriefing. During the pre-briefing, subjects were told that they were participating in a study investigating emotional responses in the settings of life-threatening situations. The scenario included a 35-year-old man in the ED with ventricular fibrillation subsequent to a Brugada syndrome. We randomly assigned each team in two groups: after three electrical shocks, the patient returns in spontaneous cardiac activity (life group), or the patient ends in asystole (death group). In the death group, an instructor had the role of the intensivist that admitted the patient to ICU for ongoing advanced life support. Participants were told immediately after the 10- to 15-min session that the patient underwent 45 min of unsuccessful advanced life support and was declared dead. During the debriefing, participants were encouraged to talk about the patient’s death. The pre-specified educational objectives of our scenarios were: to identify cardiac arrest, to initiate and provide basic life support then advanced life support (including calling the intensivist). Our non-technical skills educational objectives were mainly teamwork communication. We use a Laerdal low-fidelity manikin to practise simulation, with audio–video recording and debriefing (Laerdal, Stavanger, Norway, and Sim Station, Wien, Austria).
Statistical analysis
Continuous data are expressed as mean (standard deviation—SD) when normally distributed and as median [25–75 interquartile range—IQR] if not. Categorical variables are expressed as number (percentage). Normality was tested using the Kolmogorov–Smirnov test. Comparison of data was made using Student’s t test and presented as mean differences and their 95 % exact confidence interval (CI) that included the analysis of the primary endpoint of change in STAI score between life group and death group. The reduction in STAI (score at day 0 minus score at 1 or 3 months) was compared using the paired Student’s t test for comparison of paired variable. Categorical variables were compared using Chi-square test or Fisher exact test when appropriate.
With no pre-existing literature or pilot study on this subject, we evaluated the first 35 included participants and found a mean reduction of five points for the primary endpoint [25]. With a non-inferiority a priori hypothesis, we estimated that we needed to report a lower bound of the 95 % CI of the primary endpoint in the death group that should not be less than 3 points below the mean reduction in the life group. With a hypothesis of a mean reduction in anxiety of 5 in the life group, the lower bound of the 95 % CI of the mean reduction in anxiety in the death group should be higher than 2. With a mean cluster size of 4, an alpha of 2.5 %, a power of 90 % and accounting for an intraclass correlation of 0.01, we calculated that we needed 60 subjects in each group (PASS 14, Statistical Solution Ltd, Cork, Ireland).