Study of Virtual Simulated Resuscitation in Junior Clinicians

Sample size: For a pilot study, a sample size of at 24-50 participants is advisable to be informative regarding population characteristics such as expected means and standard deviation, to facilitate future study, where sample size calculation will target a small standardised effect size of 0.2, where standardised effect size would be the difference in means divided by standard deviation. About 50 junior clinicians are anticipated to rotate through NUH emergency department for at least 3 months in 2022.

Randomisation:

• After informed consent and recruitment, participants will be randomised to virtual simulation versus in-situ simulation groups, through the use of opaque sealed envelopes. The randomisation sequence will be generated by one investigator (A), with block randomisation in varying block sizes, while another investigator (B) will open envelopes to allocate learners to their groups, to maintain allocation concealment.

Interventions:

• Two investigators (C and D) will schedule participants for assigned interventions, namely virtual simulation or team-based in-situ simulation. As team-based in-situ simulation sessions are dispersed across each month, virtual simulation sessions will be matched to in-situ simulation days, to prevent confounding by time from outcome assessment. All interventions will be scheduled after didactic materials for resuscitation have been delivered to learners as part of standard departmental curriculum.
• Virtual simulation group: Participants will log in to the virtual simulation software from home and play 2 cases (1 case of sepsis, 1 case of trauma). Participants are instructed to play each case as many times as personally desired within 70 minutes, to reach the highest score possible in that time. This process is proctored, with participants sharing their screen over video conferencing with two investigators (C and D) to ensure adherence to time limit and cases played.
• Team based in situ simulation group: Faculty observing and debriefing learners at the mock code will be variable and consist of a wide range of NUH emergency department faculty outside of this study, due to logistical constraints in having the same faculty being present consistently in the entire year. The in situ simulation will cover 1 case of sepsis, and 1 case of trauma, with case content matched to the cases in the virtual simulation group, and time matched at 70 minutes.

Outcome assessment:

• Scores obtained based on number of correct actions completed in a timely manner, during simulated resuscitation with a high fidelity manikin around the end of the 3rd month of their posting, as compared to a checklist of required actions derived from expert consensus among NUH emergency department clinicians, and aligned to established international guidelines for resuscitation in sepsis or trauma (e.g. Advanced Trauma Life Support 10th edition).
• Two trained and standardised assistants, playing the roles of one nurse and one doctor, will be present to assist the medical officer, who will take the role of resuscitation team leader.
• Each study participant will be observed and marked on two scenarios: one scenario on major trauma resuscitation, and one scenario on resuscitation of a patient with sepsis. Marking will be performed independently by two investigators (E and F) who will calibrate and standardise their ratings by watching one investigator (A) perform simulated resuscitation at different levels of competence, and using that as a stimulus to calibrate their ratings.
• All assistants and assessors doing marking will be blinded to participant allocation, and will not be involved in team based in situ simulated resuscitation during the study period.

Other data collection:

• Learner demographic data, level of post graduate experience generally and in emergency medicine, and views regarding assigned intervention (i.e. Kirkpatrick level of reaction), will be collected via an online questionnaire.
• The following virtual simulation metrics will be automatically captured: Number of attempts for each unique case, time spent on each attempt, scores obtained during each attempt as marked by a virtual patient simulator

Data analysis:

• Descriptive statistics of the following will be tabulated:
• Baseline characteristics of learners in each group, to determine effectiveness of randomisation
• Learner reaction data, to determine extent of acceptability of virtual patient simulation based on learner perception
• Mean, median, and measures of spread (either standard deviationor interquartile range, depending on whether data is normally distributed) of scores obtained

Analysis for associations:

• Primary outcome: Association between exposure to virtual patient simulation (categorical variable) and marks obtained during observed in-situ resuscitation using a high fidelity manikin (likely to be a continuous variable with normal distribution, and this pilot study would help to determine if that is true), as measured by number of correct actions completed on a checklist derived from local expert consensus, rated by two independent observers (Investigators E and F) with statistical testing using a 2-tailed student t-test. --> Interobserver agreement will be measured using intra class correlation
• The investigators acknowledge that as a pilot, the analyses are unlikely to be adequately powered. Nonetheless, understanding the distribution of marks in each group, including mean/median/measures of variance, would be useful for planning further study.