Study of Team Leadership Training Effect on Patient Care

Detailed Description: This study will implement an individual randomized controlled trial. Eligible trainee subjects (second or third year emergency medicine or general surgery residents) will be randomized within each month in a 1:1 proportion to either the intervention (team leadership training) or control (no training) conditions. Subjects will be randomized each month to ensure that outcomes are not biased by opportunities for concurrent learning throughout the residency year. The primary outcome is team leadership during actual trauma resuscitations. The secondary outcome is patient care during actual trauma resuscitations. Primary hypotheses include that the intervention group will demonstrate improved leadership (H1) and patient care (H2) during actual trauma resuscitations. The investigators will collect baseline data (pre-intervention, first 2 weeks of the month) and post-intervention/control data (last 2 weeks of the month).

Control Condition: Subjects randomized to the control condition will receive a standard orientation to the trauma team leader role at the start of their month in the ED. This orientation involves review of key administrative information, including departmental policy on chart completion, attendance policy, and available resources. No further formal trauma or team leader training will be provided for the control group.

Intervention Condition: Subjects randomized to the intervention arm of the study will also receive the standard orientation to the trauma leader role at the start of their trauma leader month in the ED. The study coordinator will then schedule these subjects for the leadership training intervention to start 11-13 days after the start of the rotation. Subjects will attend a session where they will complete the demonstration-based training and simulation-based training (simulation + debriefing + additional practice, 4 hours total) curriculum. A Laerdal SimMan 3G patient simulator and trained confederates (team members) will operationalize the resuscitation scenario. The investigators will record all simulations to allow for video-assisted debriefing and opportunity for self-reflection.

Trauma resuscitations will take place at Harborview Medical Center, an urban, tertiary care center affiliated with the University of Washington Medical Center that serves as the sole trauma center for a 5-state region (AK, WA, WY, MT, ID). Dedicated trauma resuscitation bays are equipped for all levels of trauma resuscitations. Data collection during clinical assessments will involve video recording of ongoing trauma resuscitations within two resuscitation bays. All patients presenting to the ED with a trauma-related complaint will be screened using inclusion and exclusion criteria (below). Criteria were chosen to ensure a broad representation of trauma patient care and adequate opportunity to observe team leader behaviors. Certain exclusion criteria (c-g) were selected to try to identify those patients with near 100% mortality or those inappropriate for resuscitative efforts. Additionally, pediatric and pregnant patients will be excluded as their team composition differs markedly from those involved in standard adult trauma resuscitations. Criteria will be further honed and tested prior to data collection to ensure eligible resuscitations are reliably identified and appropriate for the proposed research.

Measures Primary Outcome: Team leadership Trained raters will code video recordings of actual trauma resuscitations using a team leadership rating scale that will produce a single leadership score for each resuscitation. Raters will be blinded to study arm and will have no prior history of interaction with the subject as a physician provider. Leadership rating dimensions will be summed and averaged to create a reliable overall indicator for each trauma case. Aggregation of the multiple trauma case assessments is necessary to both create reliable indicators across trauma cases and to align the level of the data with the unit of observation and the level of analysis (i.e., the subject team leader).

Secondary Outcome: Patient care Trained raters will code video recordings of actual trauma resuscitations using a patient care rating scale that will produce a single patient care score for each resuscitation. Raters coding patient care will be distinct from those coding team leadership to minimize bias. Raters will be blinded to study arm and will have no prior history of interaction with the subject as a physician provider. Patient care rating dimensions will be summed and averaged to create a reliable overall indicator for each trauma case. Aggregation of the multiple trauma case assessments is necessary to both create reliable indicators across trauma cases and to align the level of the data with the unit of observation and the level of analysis (i.e., the subject team leader).

Patient Outcomes (Pilot) are defined below.

Analysis (Data Preparation)

Leadership and patient care rating dimensions will be summed and averaged to create a reliable overall indicator for each primary dependent variable for each trauma case. The sub-dimension ratings for each factor are on the same metric, so the combination of the rating dimensions into an overall composite indicator is straightforward.

Control composites for the subject (team leader) and patient will be examined. The investigators will first determine whether candidate control variables are associated with the primary criteria. Those candidate control variables that significantly relate to leadership behavior or patient care will be retained as controls. For retained variables, the investigators will first transform the raw control variable metrics into standardized scores to equalize their means and standard deviations. The standardized indicators will then be summed and averaged to compute control composites for the subject (team leader) and patient. Aggregation of the multiple trauma case assessments and controls (patient) is necessary to both create reliable indicators across trauma cases and to align the level of the data with the unit of observation and the level of analysis (i.e., the subject team leader). Aggregation of the patient outcome assessments will be accomplished by summing and averaging the pre- and post-interventions separately. Aggregation of the patient controls will be accomplished by summing and averaging the patient control composites.

Descriptive statistics will be used to provide univariate summaries of the basic characteristics of the team leader subjects and patients. These summaries will also be used to identify outliers and to determine if randomization resulted in the composition of study groups with similar characteristics.

Analysis (Primary and Secondary Outcomes) With the data prepared and aligned to the unit of observation, the experimental design is structured as a fixed effects two-condition pre- and post-intervention model. The analyses will contrast an intervention group receiving leadership training with a control group receiving no training, controlling for relevant subject (team leader) and patient factors and for pre-intervention subject performance. To provide a measure of control for the overall Type I experiment-wise rate the investigators will first conduct a Multivariate Analysis of Covariance (MANCOVA) that will assess the significant effect of training condition on the set of dependent variables (leadership and patient care). A significant effect at the designated alpha level (.05), will trigger follow up univariate analyses for each hypothesis (i.e., dependent variable) using Analysis of Covariance (ANCOVA).

The study was powered to address the impact of leadership training on leadership. The ability to detect significant effects in an experiment depends on the target level of statistical significance (alpha), sample size, and effect sizes of the experimental treatments. Given that our research foci are innovative, estimating anticipated effect sizes represents the major challenge for conducting a power analysis. A meta-analysis of leadership training/development programs was consulted to estimate the effect size for leadership training. Based on 35 studies with a total sample size of 3389, a mean effect size (Cohen's d, corrected for sampling error and measurement error) of .65 was found for leadership training and all outcome criteria. Using this effect size, a power analysis indicated a sample size of 60 (30 per condition) would provide .80 power in detecting true effects.

Pilot Outcome Analysis

Linkage: Assessed as the proportion of patients treated by study subjects (control and intervention groups) linked to vital status at discharge after traumatic injury. Assessed will be proportions linked to each of these groups as well as overall.

Interim Analyses: As this is a feasibility study to characterize linkage, there will be no formal interim analyses that would allow stopping the study for either efficacy or futility.

Efficacy Analyses: The linkage rate will be summarized as a binomial proportion. Secondary measures will be summarized descriptively, as recommended for pilot studies. Sample Size: A sample size of 160 eligible patients will allow estimation of linkage from simulation intervention to hospital outcome with a precision (as measured by half-width of 95% CI) of 7.7%.