In medical emergency situations such as cardiac arrests, it is important that leaders establish a structure within the team to assure high performance. However, there is substantial variance in quality and effectiveness of leadership in emergency situations.1–7 Previous research shows that even in groups with a clearly established hierarchy, it is not guaranteed that more qualified members will take the lead.1,7 The problem of insufficient leadership is thus likely to be even more pronounced in teams of equal-status peers, with no previous leadership assignment. For such teams, the question that arises is who will assume a leadership role. For medical emergency situations, there is no data on why some healthcare workers take the lead while others refrain from it. However, better knowledge of factors associated with leadership could have important implications for medical practice, training, and medical education.
Psychologic research on “emerging leadership”8 has shown that personality traits, gender, physical height, and general abilities predict leadership emergence, particularly in short-term groups.9–12
Personality is often assessed using the five-factor model (“the Big Five”) of personality.13 It refers to the five personality traits of neuroticism (vs. emotional stability), extraversion (vs. introversion), openness to experience, agreeableness, and conscientiousness. In a comprehensive meta-analysis, extraversion and conscientiousness were the strongest predictors of leadership emergence.11 With regard to gender, previous research showed that men emerge as leaders more often than women in short-term groups and particularly in groups carrying out more technical tasks. On the other hand, women emerge more often as leaders in tasks that require more complex social interactions or are typically female.10 Furthermore, for both genders, physical height has been found to be positively related with leadership emergence.12 General abilities and specific task-related knowledge were related to emergence of leadership in some cases.9
The aim of this study was to determine predictors of leadership in teams of fourth-year medical students dealing with cardiac arrests. Because of lacking evidence in the medical setting, the study is exploratory in nature and assesses the influence of the Big Five personality factors, gender, height and task-specific knowledge, and experience on emergent leadership. We operationalized leadership as the number of verbal leadership statements in the first 3 minutes by coding verbal interactions.
Fourth-year medical students at the University of Basel, Switzerland, were invited to participate in a workshop using a patient simulator in 2006 and 2007. The study protocol was approved by the local ethical committee, and written informed consent was obtained from all participants.
Before the simulation exercise, all participants were given 45 minutes to complete a questionnaire containing the following aspects: (1) Demographic information (age, gender, and height). (2) A short German version of the NEO Five-Factor Inventory14,15 to assess the Big Five personality traits (neuroticism, extroversion, openness, agreeableness, and conscientiousness). (3) Experience in 10 medical skills including 3 skills relevant in resuscitation: (a) cardiac massage, (b) defibrillation, and (c) mask ventilation; rated on a 5-point Likert scale (1 = “I never saw or performed this skill;” 5 = “I master this skill”). Context experience was defined as the participants' prior practical experience with the treatment aspects of a cardiac arrest and quantified as the cumulative score for the three skills related to resuscitation (minimum score 3, maximum score 15). (4) Knowledge of symptoms and treatment algorithms was assessed with nine questions related to four different medical emergencies. These included one question related to the symptoms of a cardiac arrest and two questions on treatment algorithms for cardiopulmonary resuscitation. Answers were coded according to a key provided by a panel of experts from intensive care and cardiology. (5) Students were also asked to diagnose 12 different electrocardiogram strips including four strips relevant for cardiopulmonary resuscitation. Context knowledge was defined as the participants' theoretical knowledge related to the diagnosis and treatment of a cardiac arrest and quantified as the cumulative score for (a) the question on the symptoms of a cardiac arrest, (b) the two questions on treatment algorithms of cardiopulmonary resuscitation, and (c) the score of the four electrocardiogram strips relevant for cardiopulmonary resuscitation (minimum score 0, maximum score 15).
A patient simulator (Human Patient Simulator, METI, Sarasota, FL) was used in the study. A cannula was placed in a peripheral vein to allow for intravenous administration of drugs. A commercially available manual defibrillator was placed next to the bed. All participants received a 15-minute structured instruction on the technicalities of the simulator before the simulation.
This is a prospective observational study. Participants were randomly assigned to teams of three students each. Each team was aided by a nurse instructed to only respond to participants' requests, to speak only if addressed, and to act only if requested during the time of intended data collection, ie, the first 3 minutes of the cardiac arrest. Before the simulation, teams were told that they were the responsible physicians for the “patient” and that a nurse, fully familiar with the simulator and the technical equipment, would help them upon request.
The patient (simulator) was a 66-year-old man who felt dizzy after an uneventful bicycle stress test. On entering the simulator room, the participants encountered a talkative patient connected to a monitor showing normal sinus rhythm. The patient did not feel dizzy anymore and volunteered a detailed account of that episode. The patient also complained of stiff muscles in both thighs. Two minutes after the participants had entered the simulator room, the patient suffered a cardiac arrest due to ventricular tachycardia displayed on the monitor. With the onset of the cardiac arrest, the patient closed his eyes, ceased to speak and to breathe, and pulses were no longer palpable. Regardless of any measures taken, the patient stayed in cardiac arrest for at least 3 minutes. Thereafter, sinus rhythm could be achieved by defibrillation. To avoid a potentially traumatic experience, the death of the patient was prevented by the nurse who, after 6 minutes, suggested appropriate measures. Note that this intervention of the nurse was after data collection.
Using frame-in-frame technology, the teams' performance and the monitor displaying the patient's vital signs were simultaneously recorded. Data were coded by two independent observers based on the videotapes recorded during the simulation. Differences between the two coders were resolved by jointly watching the videos and negotiating a common solution. As time is a critical parameter in the treatment of cardiac arrests and because the quality of early intervention is pivotal, we restricted our analyses to the first 3 minutes after the onset of the cardiac arrest.
All verbal statements were noted and coded according to a predefined classification system (Table 1). Leadership emergence is assessed typically (a) by investigating who achieves leadership positions or is perceived as leader and (b) by analyzing actual behavior.10 We assessed leadership emergence based on behavior in counting the number of leadership statements for each participant for the first 3 minutes. The observational system was based on nine-item adapted Leadership Behavior Description Questionnaire used in real cardiac arrests in a study by Cooper and Wakelam.1 Findings of subsequent simulator-based studies suggest that only four of the nine leadership items of the adapted Leadership Behavior Description Questionnaire are necessary to code unambiguously with a high inter-rater agreement3,4,16: (1) decision about what should be done; (2) decision on how things should be done; (3) direction/command; and (4) task assignment. Moreover, leadership instruction resulted in both an increase in leadership statement and better team performance.4 Thus, we assessed leadership behavior using the abridged four items of the adapted Leadership Behavior Description Questionnaire.
The primary outcome was the predictors of leadership statements. The variability in the number of leadership statements was the secondary outcome. Multilevel analysis was performed using MLwiN 2.0, a commercially available statistical software. SPSS 17.0, a commercially available statistical software, was used for all other statistical analyses. Pearson's correlation was used for bivariate analyses, and stepwise regression analysis was performed for multivariate analysis. Cohen's kappa was used to assess inter-rater reliability. Descriptive results are provided as means ± 1SD unless otherwise indicated. A P value <0.05 was considered to represent statistical significance.
Two hundred thirty-seven of 280 (85%) medical students approached agreed to participate in this study and were randomly allocated in 79 teams of three students each. The composition of the teams was as follows: 25 teams with all females, 31 teams with two females and one male, 19 teams with one female and two males, and four teams with all males. The study could be conducted as intended, and no protocol violations occurred. All 237 participants (158 females and 79 males) completed the study.
The inter-rater reliability for leadership statements (kappa 0.91, P = 0.001) and total number of statements (kappa 0.96, P = 0.001), respectively, was high. Table 2 displays all verbal statements during the first 3 minutes of the cardiac arrest according to classification and gender. The individual variability in context experience was substantial and covered the entire range from minimum to maximum values. Of 15 possible points, the participants achieved a mean of 7.0 ± 2.3; males students had higher values than female students (7.8 ± 2.9 vs. 6.6 ± 1.8; P = 0.002). With a range from 1 to 13 (out of possible 15) points, a large individual variability was also observed for context knowledge; our participants achieved a mean of 6.2 ± 2.5 points with no significant gender difference.
Primary Outcome: Predictors of Leadership Statements
Multilevel analysis revealed that the team level did not explain a significant amount of variance of the number of leadership statements, indicating that the analyses can be conducted on the individual level. Table 3 shows the results of all bivariate correlations.
To assess significant predictors of leadership statements, gender, height, the five personality factors (neuroticism, extraversion, openness to experience, agreeableness, and conscientiousness), context knowledge, and context experience were entered into the analysis. Multivariate stepwise regression analysis revealed that gender, extraversion, and agreeableness (negative) are the only predictors significantly related to the number of leadership statements (Table 4): the regression model shows that being female results in 1.9 fewer leadership statements than being male, and that a difference of one point on the personality traits extraversion and agreeableness (each measured on a scale from 1 to 6) results in a difference of approximately one leadership statement.
The excluded variable context experience (P = 0.13) showed no significant effect on leadership statements in the multivariate analysis. The significant bivariate effect of context experience (Table 3) may best be explained by the significant correlation between gender and context experience (r = −0.26, P < 0.001). Context knowledge failed to achieve statistical significance in the multivariate analysis, albeit by a small margin (P = 0.09). Height had no significant effect on leadership statements in the multivariate analysis. The strong bivariate correlation of height with leadership (Table 3) is explained by the high correlation between height and gender (r = 0.73, P < 0.0001). A subsequent post hoc analysis revealed that height also did not have a significant effect on leadership within each gender (P = 0.72 for females; P = 0.41 for males).
Secondary Outcome: Variability in the Number of Leadership Statements
In the first 3 minutes of the cardiac arrest, participants communicated a median of 20 verbal statements (range, 1–45; interquartile range, 7) including a median of five leadership statements (range, 0–22; interquartile range, 2). Thirteen of the participants (5.5%, all females) showed no leadership statements, and 19 participants (8%, 15 females) showed only a single leadership statement in the first 3 minutes. The 32 participants with ≤1 leadership statement differed from the remaining 205 participants in gender (P = 0.008), extraversion (3.7 ± 1.0 vs. 4.2 ± 0.8, P = 0.004), and context knowledge (4.9 ± 2.5 vs. 6.4 ± 2.5, P = 0.0014). There was no difference in context experience (11.5 ± 2.6 vs. 10.8 ± 2.4) and other personality traits.
Although the importance of leadership in emergency situations is undisputed,1–3,5–7 there are only limited data on why some individuals take the lead while others do not. To the best of our knowledge, this study is the first to address this topic in a medical emergency setting: gender, extraversion, and negative agreeableness predicted the number of leadership statements during the first 3 minutes of a cardiac arrest, whereas context knowledge, context experience, and other personality traits had no effect.
Previous research in other than the medical domain reported relationships between some of the Big Five personality traits and leadership.11 Our results are in agreement with a meta-analysis, showing that extraversion has the most consistent relation with emergent leadership.11 By contrast, conscientiousness, regarded as the second strongest predictor of leadership in the meta-analyses,11 showed no significant relationship with leadership statements in this study. One explanation could be the comparatively short duration of the cardiac arrest scenario that may have prevented typical features of conscientiousness such as organizing activities or dependability to emerge and show their leadership promoting effect. Contrary to our finding of an effect of agreeableness on the number of leadership statements, in previous research, agreeableness was found to be only a weak predictor of leadership.11 This discrepancy may be related to the specific settings of our study: typical features of an agreeable person-like sensitivity, tact, modesty, and altruism may be an obstacle to active and directive leadership needed in the early phase of an emergency. This may favor individuals with low agreeableness, ie, with a lesser tendency to “conform to others” to emerge as leaders in short and rapidly evolving scenarios.
An important finding of this study is the observed gender difference. Despite having the same opportunity and the same level of knowledge as their male colleagues, female students made significantly fewer leadership statements. As male and female students showed a similar number of overall communication statements (Table 2), the difference in the number of leadership statements cannot be explained by female students not daring to speak up. With an effect size of d = 0.38 (ie, the difference in means between males and females, divided by the pooled standard deviation), the difference between male and female students for leadership statements is in the moderate range, defined as an effect size of ≥0.36.17 Less than 20% of all gender differences reported so far are above this threshold.17 However, an effect size of 0.38 implies that the within-gender variance exceeds the between-gender difference.
A very recent study demonstrated that in teaching sessions among first-year medical students, disproportionately fewer women than men volunteered to become small-group leaders.18 Research outside the medical field reported that men tend to emerge as leaders more often than women.10,19 The tendency for males to emerge as leaders weakens when the task requires complex social interactions and it also weakens over time.10 Eagly and Karau attribute this finding to the fact that complex social interactions require more relationship-oriented (“female”) leadership, which is in accordance with gender stereotypes. This makes leadership for females more accepted by males as well as females. Moreover, gender stereotypes become less salient over time while relationship issues become more important for the team.10 A cardiac arrest is a short emergency event that requires an immediate and clear focus on the task rather than on social relationships.1,6 In addition, handling a defibrillator adds a technical (“stereotypically male”) component to the task. Thus, based on previous research, this situation should favor males as emergent leaders, and this is supported by our results.
A troubling aspect of our results concerns the lack of knowledge and experience effects on leadership. One would expect that more knowledgeable and more experienced people take the lead. The large variation in both context knowledge and context experience in the studied sample indicates that the nonsignificant effect on leadership in our study is not simply due to too small interindividual differences. There is good evidence that surface-level attributes, ie, attributes immediately obvious to other group members such as age, race, or gender are more influential in the early phase of group interaction. By contrast, the effects of deep-level attributes, ie, attributes not immediately obvious to other group members such as attitudes and aptitudes, strengthen with the length of time groups work together.20,21 Thus, the lacking effects of knowledge and experience might be explained by the comparatively short group interaction during the treatment of a medical emergency. Nevertheless, the fact that personality and gender effects override knowledge and experience is a matter of concern; teaching students to accurately judge their own knowledge and to behave according to these judgments seems critical.
Limitations of our findings relate mainly to our study population. Although typically fourth-year students will have been instructed about cardiac arrests and the necessary treatments, they are quite inexperienced in dealing with such situations. However, previous research has demonstrated substantial differences in leadership behavior even among experienced medical professionals.1,6,7 Although we cannot conclude from the current study that our results are generalizable to more experienced medical professionals, this might be the case and thus deserves further investigation. Our results were obtained in a homogeneous cultural cohort and thus may not be necessarily extrapolated to settings with a different social, cultural, or educational background.
Some researchers used video camera recordings or defibrillators capable of event recording to evaluate the performance during cardiopulmonary resuscitation.22,23 However, such recording equipment is usually made functional during, rather than before, resuscitation and therefore may miss the initial phase of a cardiac arrest where leadership is most important. Strengths of our study were (1) the recording of objective data from both patient and participants at the start of the cardiac arrest; (2) standardized conditions across a high number of participants; and (3) that no interventions were necessary in case of a poor performance to protect the patients. Thus, simulation allowed investigation of issues that for a variety of medical, practical, and ethical reasons are impossible to investigate in real patients.
Leadership can be regarded as an additional and important medical skill necessary to achieve and maintain patients' safety in specific circumstances.1–4,6,7 Thus, a reasonable goal for medical education is that every physician is capable of adequate leadership in a medical emergency. Our results demonstrate an impressive variability across individuals in leadership statements during a medical emergency. Moreover, our results demonstrate that leadership behavior is predicted by personality traits and gender rather than knowledge and experience. Behavior associated with gender and personality, although deep-seated, can still be modified: one may well learn to behave counter to these tendencies, especially in specific situations.24 Thus, it might be feasible to train people to identify the need for directive leadership, and to exercise that leadership accordingly in medical emergencies, regardless of their predisposition. Future research should investigate what kind of educational interventions are necessary to reach competency in leadership, taking into consideration personality traits and gender. Analogous to “individualized medicine” that takes into account patient-specific features, such research may lead to “individualized medical education” that takes into account learner-specific features. Medical simulation seems to be ideally suited to conduct such research and to develop, test, and implement such education.
For the time being, teaching of medical emergencies should stress the importance of communication and leadership. However, as mentioned above, leadership is an additional skill and not a replacement of knowledge or skills. Thus, promoting confidence and leadership skills in the absence of the prerequisite knowledge and skills could be as detrimental to patient care as promoting technical skills without highlighting the importance of leadership.
During the initial phase of a medical emergency, there is a substantial variability across individuals in the amount of verbal leadership. Leadership was determined by gender and personality traits and not by knowledge or experience. Future research should address the importance of learner-specific features in the teaching on leadership.
The authors thank all the staff of the intensive care unit, notably Marc Breuer, Martin Spychiger, and Sabine Schweitzer, for their support during the study. The authors thank all involved students from the University of Basel for participating in this study. The authors thank Jennifer Stevens, MD, for her critical review of the article.
1. Cooper S, Wakelam A. Leadership of resuscitation teams: “Lighthouse Leadership.” Resuscitation
2. Cooper S. Developing leaders for advanced life support: evaluation of a training programme. Resuscitation
3. Hunziker S, Tschan F, Semmer NK, et al. Hands-on time during cardiopulmonary resuscitation is affected by the process of teambuilding: a prospective randomised simulator-based trial. BMC Emerg Med
4. Hunziker S, Bühlmann C, Tschan F, et al. Brief leadership instructions improve cardiopulmonary resuscitation in a high-fidelity simulation
: a randomized controlled trial. Crit Care Med
5. Kunzle B, Kolbe M, Grote G. Ensuring patient safety through effective leadership behaviour: a literature review. Saf Sci
6. Marsch SC, Müller C, Marquardt K, Conrad G, Tschan F, Hunziker PR. Human factors affect the quality of cardiopulmonary resuscitation in simulated cardiac arrests. Resuscitation
7. Tschan F, Semmer NK, Gautschi D, Hunziker P, Spychiger M, Marsch SU. Leading to recovery: group performance and coordinative activities in medical emergency driven groups. Hum Perform
8. Hogan R, Curphy GJ, Hogan J. What we know about leadership. Effectiveness and personality. Am Psychol
9. Kickul J, Neuman G. Emergent leadership behaviors: the function of personality and cognitive ability in determining teamwork performance and KSAs. J Bus Psychol
10. Eagly AH, Karau SJ. Gender and the emergence of leaders: a meta-analysis. J Pers Soc Psychol
11. Judge TA, Bono JE, Ilies R, Gerhardt MW. Personality and leadership: a qualitative and quantitative review. J Appl Psychol
12. Judge TA, Cable DM. The effect of physical height on workplace success and income: preliminary test of a theoretical model. J Appl Psychol
13. Goldberg LR. An alternative “description of personality”: the big-five factor structure. J Pers Soc Psychol
14. Ostendorf F. Sprache und Persönlichkeitsstruktur. Zur Validität des Fünf-Faktoren-Modells der Persönlichkeit. Regensburg, Germany: Roderer; 1990.
15. Schallberger U, Venetz M. Kurzversion des MRS-Inventars von Ostendorf (1990) zur Erfassung der fünf ”grossen“ Persönlichkeitsfaktoren. Berichte aus der Abteilung Angewandte Psychologie. Nr. 30. Zurich, Switzerland: University of Zurich; 1999:1–51.
16. Lüscher F, Hunziker S, Gaillard V, et al. Proficiency in cardiopulmonary resuscitation of medical students at graduation: a simulator-based comparison with general practitioners. Swiss Med Wkly
17. Hyde JS. The gender similarities hypothesis. Am Psychol
18. Wayne NL, Vermillion M, Uijtdehaage S. Gender differences in leadership amongst first-year medical students in the small-group setting. Acad Med
19. Stelter NZ. Gender differences in leadership: current social issues and future organizational implications. J Leader Organ Stud
20. Harrison DA, Price KH, Bell MP. Beyond relational demography: time and the effects of surface- and deep-level diversity on work group cohesion. Acad Manage J
21. Harrison DA, Price KH, Gavin JH, Florey AT. Time, teams, and task performance: changing effects of surface- and deep-level diversity on group functioning. Acad Manage J
22. Abella BS, Alvarado JP, Myklebust H, et al. Quality of cardiopulmonary resuscitation during in-hospital cardiac arrest. JAMA
23. Wik L, Kramer-Johansen J, Myklebust H, et al. Quality of cardiopulmonary resuscitation during out-of-hospital cardiac arrest. JAMA
24. Zaccaro SJ. Trait-based perspectives of leadership. Am Psychol