Few issues in simulation seem to generate as much discussion as the question of when (or whether) to allow the mannequin to die in response to learner actions.1–3 Proponent argue that this is necessary to assure realism, whereas opponents opine that the emotional stress caused by the mannequin's death distracts from key learning points. There is often more heat than light generated when this issue is discussed, and what evidence exists in the literature is often conflicting and inconsistent. Some studies indicate that the additional stress generated enhances recall, whereas others suggest that at least for certain learner groups, this stress may inhibit long-term retention.4,5 Published learner surveys and commentary are similarly divided, with several suggesting that learners desire to experience the outcome of their actions in a realistic manner and others raising the possibility of eroded trust in the team as a response.6–8 The question thus remains; do we as simulation-based healthcare educators have a license to kill?
In this issue of Simulation in Healthcare, we present a study that significantly advances this discussion. Building on their previous work addressing stress and learning, Goldberg and colleagues sought to examine the effects of different levels of simulated mortality on anxiety and nontechnical skills.9 They begin with the hypothesis of a positive correlation between the volume of death experienced and learner improvement, basing this on the dose–response curves often used to model drug effects. Three groups of anesthesiology residents were then assessed after engaging in a simulation-based course centered on 12 cases of known, but uncommon and often unexpected, life-threatening perioperative adverse events. The first group experienced no mannequin death during the course, whereas the second group experienced a death after each case regardless of learner action similar, in deference to those science fiction fans in the audience, to the Kobayashi Maru training scenario of Star Trek fame (For those unfamiliar with this scenario, a Starfleet Academy cadet is given the helm of a simulated starship and tasked with retrieving a civilian ship from enemy territory. Regardless of the decisions made by the cadet, the scenario ends in failure and is, thus, an excellent pop culture example of a programmed “no-win” situation in simulation).10 The third group experienced a variable level of mannequin death that was tightly correlated to the perceived appropriateness of the learner's actions. What they found was fascinating: although no statistically significant differences existed between the “never-death” and “always-death” groups, the “variable-death” group demonstrated a measureable improvement in nontechnical skills such as task management, teamwork, and situational awareness. Although effect sizes were not large and our ability to generalize the findings is limited by the scenario types and learner populations studied, the clear presence of this difference provides an intriguing window into how learners may process these experiences.
Data Versus Interpretation: The Intersection of Stress, Fidelity, and Learner Agency
So how, then, do we interpret these findings? The authors themselves appropriately refer back to the model with which they began the study, speculating that the reduced nontechnical skills noted in the always-death group might be related to a stress overdose effect. To push the pharmacological analogy a bit further, we are reminded of drugs such as dopamine that have different effects on the cardiovascular system at various dosage levels. Such an effect also bears resemblance to the inverted u-shaped relationship between stress and performance that exists for complex skills, in which increasing stress produces a corresponding increase in performance until a certain threshold is reached, after which diminishing returns obtain.11 It is also possible, however, that a degree of habituation occurred among members of the always-death group, which caused them to become partially tolerant to the effects of stress, diminishing its effect on learning.12 Thus, although stress may indeed be the best explanation for these findings, the relationship is unlikely to be a straightforward one. All these being said, we think that it is also necessary to consider other potential contributors, such as the effects of the intervention on case fidelity and learner perceptions of agency.
Unlike stress, which one could presume increases steadily with exposure to simulated death, case fidelity in this context is much more variable and is related to how tightly the instructor correlates the actions of learners with the clinical outcomes (including death) portrayed by the mannequin. A separate but associated concept is agency, which, in the social sciences, refers to the capacity of an individual to act in a way that results in specific results.13 Taken together, these concepts encompass both how real the progression and outcome of a simulation feels to the learner and what that progression and outcome imply to those learners about their ability to make a difference during the simulation and in similar real-world situations.
Although learner stress levels in the never-death and always-death groups differed greatly, the relationship between learner actions and case outcome was deliberately nullified in both cases. This leads to the possibility that participants in these groups might have intuited (without being told) that the mannequin would either survive or die no matter what they did. In the variable-death group, however, learner actions were tightly connected to outcomes, opening up the possibility that it was this connection (and the increased sense of case fidelity and personal agency that may have resulted from it) that led to the enhanced nontechnical skills observed. Conversely, perhaps the violation of this relationship in the always-death and never-death groups diminished the learners' sense of fidelity and agency in ways that made their actions feel unhelpful and unrealistic, decreasing retention of the learning points. Other recent studies of learner reactions to mannequin death highlight the link in subjects' minds between the death of the mannequin and inadequate resuscitation or performance on their part, lending support to this possiblity.7,14
The design of the study by Goldberg et al,9 unfortunately, does not allow us to distinguish between the various potential causes for their findings, a fact openly acknowledged by the authors. The overall approach they used could easily be altered to investigate this further, however, by the addition of a second variable-death control group. This group would experience mannequin death at the same rate and in the same sequence as the original variable-death group, but in a manner completely disconnected from their actions. Thus, for example, if the original variable-death group's actions resulted in mannequin death in the 1st, 3rd, 4th, 8th, 9th, and 10th simulations but resulted in survival in the 2nd, 5th, 6th, 7th, 11th, and 12th simulations, then the variable-death control group's mannequin would die or live in the exact same sequence. The difference is that in the variable-death control group, this pattern would be unrelated to how the learners performed within the scenarios. Termed a yoked-control design, this approach assures that both groups experience identical stress patterns, isolating the potential effects attributable to fidelity and agency and allowing them to be more readily detected.15
A further question concerns the optimal balance between fidelity and agency. Even when an underlying diagnosis is known with high certainty, perfect actions do not always lead to recovery and highly flawed actions can sometimes still result in success (as per the old saying “It's better to be lucky than good”). A case can, thus, be made that one-to-one correlation without the incorporation of some random variation may not best represent reality. It can even be argued that employing such a one-to-one approach could result in a degree of long-term clinician psychological distress as our trainees enter practice and painfully learn that the linkage between clinician intervention and outcomes is often indeterminate.7 All of these issues are ripe for further study.
Moving Forward: Constructing a Theory
Thus, does allowing the mannequin to die leave our learners shaken psychologically or stirred to improve their performance? Although the Goldberg article9 significantly advances our knowledge in this area, a substantial amount of work remains to be conducted. Although we have discussed the effects of stress and the effects of perceptions of fidelity and agency as if a sharp dichotomy existed between them, in reality, these curves overlap, and a comprehensive approach to these emotionally challenging situations will involve simulation educators learning to weigh the contribution of multiple factors in real time and devising methods to train future educators in how to navigate these issues reliably. The development of theoretical frameworks encapsulating the considerations discussed previously represents one promising means of accomplishing this. One such framework focused specifically on mannequin death as it affects experienced learners was published by our team in 2016.14
Organized according to the logical flow of a simulation, the framework begins with factors such as preparation, emotional responsiveness, and previous experience that may potentially impact the learner's response. After mannequin death has occurred, the model first considers the immediate effect of that event on team and self-perception before proceeding to the effects of debriefing on emotion, learning, and ultimately long-term practice.14 The model is currently limited, however, by the study population used in its construction, and significant further development is required before it could be applied to a wider and more diverse group of learners.14 Until we know more, we are left with a combination of evidence, judgment, and personal experience, a reality similar to that which clinicians experience daily as they seek to parse the complex interactions of physiology, pathology, and pharmacology occurring within their patients to make the best medical decisions.
As our field continues to move forward, ethical and psychological issues such as this will continue to arise, prompting ongoing discussion and debate among simulation educators. With this in mind, we praise the study by Goldberg et al9 as an excellent example of the kind of empirical work that it will take to slowly convert the heat of debate to the light of understanding. It is our hope that others will also take up these questions and improve that understanding, as there is indeed much here to be explored.
As a tribute to the recently deceased Sir Roger Moore and to the seminal work of 86-year-old Sean Connery (both stars of many classic James Bond movies), we have embedded several covert references to the film series in this editorial.
1. Corvetto MA, Taekman JM. To die or not to die? A review of simulated death. Simul Healthc
2. Leighton K. Death of a simulator. Clin Simul Nurs
3. Gaba DM. Simulations that are challenging to the psyche of participants: how much should we worry and about what? Simul Healthc
4. Demaria S Jr., Bryson EO, Mooney TJ, et al. Adding emotional stressors to training in simulated cardiopulmonary arrest enhances participant performance. Med Educ
5. Fraser K, Huffman J, Ma I, et al. The emotional and cognitive impact of unexpected simulated patient death: a randomized controlled trial. Chest
6. Bruppacher HR, Chen RP, Lachapelle K. First, do no harm: using simulated patient death to enhance learning? Med Educ
7. Lizotte MH, Latraverse V, Moussa A, et al. Trainee Perspectives on Manikin Death During Mock Codes. Pediatrics
8. Phrampus P, Dorfsman M, Cole J. Death During Simulation Training: Feedback From Trainees [abstract]. In: International Meeting on Medical Simulation: 2005
. Miami (FL): Society for Technology in Anesthesia; 2005:95, Abstract nr 16.
9. Goldberg A, Samuelson S, Khelemsky Y, Katz D, Weinberg A, Levine A, Demaria S. Exposure to Simulated Mortality Affects Resident Performance During Assessment Scenarios. Simul Healthc
10. Star Trek 2: The Wrath of Khan. Dir. Nicholas Meyer. Paramount Pictures, Hollywood, CA; 1892. Film.
11. Dobson CB. Stress: The Hidden Adversary
. Lancaster, UK: MTP Press; 1982.
12. Deinzer R, Kirschbaum C, Gresele C, et al. Adrenocortical responses to repeated parachute jumping and subsequent h-CRH challenge in inexperienced healthy subjects. Physiol Behav
13. Burkitt I. Relational agency: relational sociology, agency and interaction. Eur J Soc Theory
14. Tripathy S, Miller KH, Berkenbosch JW, et al. When the Mannequin dies, creation and exploration of a theoretical framework using a mixed methods approach. Simul Healthc
15. Passer MW. Research Methods: Concepts and Connections
. 1st ed. New York: Worth Publishers; 2014.