A metric of our own: Failure to rescue after trauma. Issue 4 (October 2017)

• Record Type: Journal Article
• Authors: Holena, Daniel N.; Kaufman, Elinore J.; Delgado, M. Kit; Wiebe, Douglas J.; Carr, Brendan G.; Christie, Jason D.; Reilly, Patrick M.
• Is Part Of: Journal of trauma and acute care surgery. Volume 83:Issue 4(2017)
• Publication Date: 2017-10
• Subjects: Failure-to-rescue -- metrics -- outcomes; Surgical intensive care -- Periodicals; Surgical emergencies -- Periodicals; Wounds and injuries -- Surgery -- Periodicals; 617.026
• DOI: 10.1097/TA.0000000000001591
• Languages: English
• ISSNs: 2163-0755

Abstract

Abstract : BACKGROUND: Failure to rescue (FTR) is defined as death after an adverse event. The original metric was derived in elective surgical populations and reclassifies deaths not preceded by recorded adverse events as FTR cases under the assumption these deaths resulted from missed adverse events. This approach lacks face validity in trauma because patients often die without adverse events as a direct result of injury. Another common approach simply excludes deaths without recorded adverse events, but this approach reduces the reliability of the FTR metric. We hypothesized that a hybrid metric excluding expected deaths but otherwise including patients without recorded adverse events in FTR analysis would improve face validity and reliability relative to existing methods. METHODS: Using 3 years of single-state adult trauma registry data from 30 trauma centers, we constructed 3 FTR metrics: (1) excluding deaths not preceded by adverse events (FTR-E), (2) reclassifying deaths not preceded by adverse events (FTR-R), and (3) including deaths not preceded by adverse events in FTR analysis except those with predicted mortality or greater than 50% (FTR-T). Mortality, adverse event, and FTR rates were calculated under each method, and reliability was tested using Spearman correlation for split-sample center rankings. RESULTS: A total of 89, 780 patients were included (median age, 57 years [interquartile range, 26–73 years]; 85% were white; 59% were male; 92% had blunt mechanism of injury; median Injury Severity Score, 9 [interquartile range, 5–14]). The FTR rates varied by metric (FTR-E, 11.2%; FTR-R, 31.2%; FTR-T, 21.4%), as did the proportion of deaths preceded by adverse events (FTR-E, 28%; FTR-R, 100%; FTR-T, 60%). Spit-sample reliability was higher FTR-T than FTR-E (ρ = 0.59 vs. = 0.27, p < 0.001). CONCLUSIONS: A trauma-specific FTR metric increases face validity and reliability relative to other FTR methods that may be used in trauma populations. Future trauma outcomes studies examining FTR rates should use a metric designed for this cohort. LEVEL OF EVIDENCE: Retrospective cohort study, outcomes, level III.