Live Versus Delayed Video-Based TEAM Assessment During Real-World Cardiac Arrest: Agreement and Exploratory Differences by Rater Experience

To evaluate agreement between immediate live and delayed video-based assessment of nontechnical skills during real-world adult cardiac arrest resuscitation, and whether agreement differed by rater experience.

We conducted a prospective observational study of adult cardiac arrest resuscitations in a tertiary emergency department in Taiwan in 2024. For each event, one trained physician or nurse who participated in the resuscitation completed the Team Emergency Assessment Measure (TEAM) immediately afterward and re-rated the same event from video approximately 30 days later. Live-video agreement was assessed using intraclass correlation coefficients and Bland-Altman analysis.

Of 105 recorded resuscitations, 80 had complete paired assessments. Mean global ratings were similar between modalities (8.25 versus 8.22; mean paired difference, -0.03; 95% confidence interval [CI] -0.25 to 0.20), whereas teamwork scores were modestly lower on video review (mean paired difference, -1.00; 95% CI -1.77 to -0.23). Event-level agreement was limited: intraclass correlation coefficients were 0.21 (95% CI 0.00 to 0.41) for total TEAM score and 0.16 (95% CI -0.06 to 0.37) for global rating. Total TEAM video ratings averaged 1.24 points lower than live ratings, with limits of agreement from -12.28 to 9.80. Agreement was higher among senior raters.

Delayed video-based TEAM assessment produced similar group-level scores but limited event-level agreement during real-world cardiac arrest. Agreement appeared higher among senior raters. Video review may support debriefing and monitoring, but appraisal of a specific resuscitation or other higher stakes use likely requires experienced raters and structured safeguards.