Trauma video review outperforms prospective real-time data collection for study of resuscitative thoracotomy

Background: A major challenge in the study of high-impact, low-frequency procedures in trauma is the lack of accurate data for time-sensitive processes of care. Trauma video review offers a possible solution, allowing investigators to collect extremely granular time-stamped data. Using resuscitative thoracotomy as a model, we compared data collected using review of audiovisual recordings to data prospectively collected in real time with the hypothesis that data collected using video review would be subject to less missingness and bias. Methods: We conducted a prospective cohort study of patients undergoing resuscitative thoracotomy at a single urban academic level 1 trauma center. Key data on the timing and completion of procedural milestones of resuscitative thoracotomy were collected using video review and prospective collection. We used McNemar's test to compare proportions of missing data between the 2 methods and calculated bias in time measurements for prospective collection with respect to video review. Statistical analyses were performed using Stata v. 15.0 (College Station, TX). Results: We included 51 subjects (88% Black, 82% male, 90% injured by gunshot wounds) over the study period. Missingness in resuscitative thoracotomy procedural milestone time measurements ranged from 34% to 63% for prospective collection and 0 to 8% for video review and was less missing for video review for all key variables (P < .001). When not missing, bias in data collected by prospective collection was 10% to 43% compared with data collected by video review. Conclusions: The data collected using video review have less missingness and bias than prospective collection data collected by trained research assistants. Audiovisual recording should be the gold standard for data collection for the study of time-sensitive processes of care in resuscitation.