Ethics approval / license / Registration

The study protocol was approved by the Institutional Review Board of Claude Bernard University Lyon 1, Lyon, France (n°IRB 2019_07_09_03, July 2019) and informed consent was obtained from all subjects before participating in the trial. The research was performed in accordance with the Declaration of Helsinki. The trial was retrospectively registered at (Marc Lilot, NCT04141124, Date of registration: 28/10/2019). The results were reported using the Consolidated Standards of Reporting Trials guidelines [21].

Population and setting

This study involved all fifth-year anaesthesiology and critical care residents from Lyon University who participated in high-fidelity simulation at the Lyon medical simulation centre in 2019. No exclusion criteria were applied. Simulations were part of the resident’s educational program and each session lasted four to five hours. Each session was composed of the same four different scenarios. Each resident participated individually in one scenario. Simulations were always structured as follows: briefing (one to five min), scenario (10 to 20 min), and debriefing (30 to 45 min) [22,23,24]. The scenarios dealt with crisis situations in the intensive care unit, operating room, and delivery room (Tamponade, Neonatal, Amniotic Fluid, Pacemaker), no residents performed these scenarios beforehand (App. A). For each scenario, the instructor playing the embedded nurse acted as neutrally as possible. SimMan Essential® and SimNewB® manikins (Laerdal Medical AS, Stavanger, Norway) were used.


This prospective randomized controlled study involved three parallel arms and a hypothesis of superiority (1:1:1 allocation). Two sessions of simulation were performed daily (one in the morning and one in the afternoon). A blinded investigator assigned each session of simulation to one intervention (Fig. 1) (Rb, Bfb + Rb, Control).

Fig. 1
figure 1

Timeline of the experimental design. VAS: Visual Analogue Scale. The resident received the briefing of the scenario first, followed by the intervention, the scenario, and the debriefing. The breathing intervention consisted of a relaxing breathing exercise (iterative sequence of 4 s of inspiration and 6 s of expiration). The biofeedback + relaxing breathing intervention corresponded to the relaxing breathing exercise paired with the viewing of real-time cardiac parameters. Overall performance corresponded to both technical (clinical specific evaluation grid) and non-technical skills (Ottawa scale) performance

Then, following simple computerized randomization procedures, residents were randomly allocated to a session of simulation (M.L.). Upon arrival, residents were informed about the study and told that the research was interested in well-being during simulation (M.L./S.S.). They were briefly informed that they will perform a breathing exercise at various times during the session of simulation (before the scenario or after the debriefing). Participants were blinded to working hypotheses and randomization procedures. Each intervention was performed by the main investigator (S.S) in an isolated room immediately after the briefing, so that all instructors leading the simulation were blinded to the group allocation. The active participant sat on a chair, while the investigator stayed in the room and verified that residents performed their interventions.

The three interventions were:

  1. A)

    Rb: Residents in the relaxing breathing (Rb) group received standardized relaxing breathing. The resident was asked to follow, for five minutes, a standardized rhythm of breathing, with an inspiration for four seconds and expiration of six seconds. The standardized relaxing breathing was guided by looking at a moving breathing cursor on a computer. No visual cardiac biofeedback was provided.

  2. B)

    Bfb + Rb: The residents in the heart rate variability biofeedback paired with relaxing breathing group (Bfb + Rb), were asked to follow the same standardized breathing as in the Rb group. The standardized relaxing breathing was guided by looking at a smartphone screen (Iphone 5S ™, Apple, cupertino, CA, USA) and a heart rate variability-biofeedback was provided on the same screen through a connection to a cardio frequency meter placed on the resident’s ear lobe (Stress control™, My Mercurochrome®, Paris, France). Through the help of the interface, residents were asked to try to increase their heart rate variability.

  3. C)

    Control: Residents from the control group reviewed normal printed laboratory test results. They were informed that those tests were unrelated to the scenario. The resident was asked to read the results for five minutes. This control has been used in previous studies, and might be seen as a normal standardized clinical practice [25,26,27].

Each intervention was conducted in a standardized manner for 5 min. Afterwards, the participant went directly to the simulation room and the scenario started. All had received a formal training in Tactics to Optimize the Potential in 2017 [25]. These tactics combined specific tools of mental preparation such as mental imagery and projection of success with cognitive toolboxes.

Performance evaluation

Two assessors (M.L./C.T.) blinded to group allocation evaluated performance independently using video recordings. For each scenario, a checklist was established beforehand to assess specific aspects of clinical performance. Each item on the checklist was associated with a number of points so that the total reached 100. These checklists have been extensively described previously [26, 27]. For each scenario, the Ottawa Crisis resource management Global Rating Scale (Ottawa GRS) was used to assess non-technical skills. Ratings for the six criteria were summed, scores ranging from 6 to 42 points [28]. Then, for each resident, the average of the two assessors’ scores for each performance (clinical specific or non-technical skills) was computed.


Before and after the intervention, each resident completed a 10 cm Visual Analogue Scale (VAS) of stress (VAS-Stress), participants also answered the Thayer questionnaire (Activation-Deactivation Adjective Check List) [29]. The Thayer questionnaire is a multidimensional questionnaire of transitory arousal states [5 to 20 points], including energetic (activation, deactivation) and tense (tension, relaxation) arousals. All residents filled out a questionnaire on demographic data and the BIG-5 personality inventory assessing 5 dimensions of personality (Openness, Consciousness, Extraversion, Agreeableness, Neuroticism) [30]. Neuroticism is known to confer a particular vulnerability to stress [31]. At the end of the experiment, all residents judged if their intervention could be used in their professional or personal practice (VAS use 10 cm, Supplemental Fig. 1).


The primary endpoint was the mean overall performance during simulation. This endpoint was calculated as the sum of the clinical performance score out of 100 points plus the Ottawa GRS score (6 to 42) adjusted to a scale of 100 points (score /42*100). The sum was then divided by two to obtain an overall score between zero and 100 points. Secondary endpoints were clinical performance, Ottawa GRS scores, VAS-Stress, Thayer questionnaire scores (relaxation, tension, activation, deactivation).

Statistical analyses

The Consolidated Standards of Reporting Trial is presented in Fig. 2. Statistical analyses were performed with R studio version 1.2.1335 (R Foundation, Vienna, Austria). All tests were two-tailed. Visual assessments were used to confirm normality of data distribution with histograms and quantile-quantile plots. For all performance measures, inter-rater reliability of investigators was assessed by calculating absolute interclass correlation coefficients and 95% CI for individual measures (package DescTools, ICC function); agreement was interpreted according to Cichetti [32]. We assessed the treatment effect on performance (overall, technical skills and non-technical skills) using an analysis of variance including the main effects of group (Rb, Bfb + Rb, Control) and scenario, and assessing the group-by-scenario interaction. Results are presented as differences in mean between groups with 98.3% confidence interval (CI) and P < 0.017 was the significance criterion when the 3 groups were compared. An outlier detection test was first performed on all variables of performance (± 1.5 inter-quartile range rejection threshold). Effect sizes are reported using eta squared (η2). The effect size can be classified as small (0.01), medium (0.06) or large (> 0.14). We assessed the effects of the group (Rb, Bfb + Rb, Control) on psychological variables (VAS-Stress, Thayer scores) using linear regression controlled for the scenario, the basal level, the level of neuroticism. Normality of residuals of the models were checked. Results are presented as estimate (standard error). For the multiple regression, the adjusted R2 was provided. No data was available to calculate a priori the sample size requested. Therefore, this pilot study included all the fifth-year anaesthesiology and critical care residents who participated in high-fidelity simulation at the Lyon medical simulation centre in May 2019 in order to assess the interest for a further study allowing a deeper exploration of the psychophysiological effects of these stress management techniques.

Fig. 2
figure 2

The study flow chart described this prospective randomized controlled study involved three parallel arms and a hypothesis of superiority (1:1:1 allocation). The figure follows the guidelines of Moher et al. 2010 (Moher et al., CONSORT explanation and elaboration: updated guidelines for reporting parallel group randomised trials. BMJ, 2010)

Source link