Debunking the survival bias myth: Characterization of mortality during the initial 24 hours for patients requiring massive transfusion

Joshua B. Brown, Mitchell J. Cohen, Joseph P. Minei, Ronald V. Maier, Micheal A. West, Timothy R. Billiar, Andrew B. Peitzman, Ernest E. Moore, Joseph Cushieri, Jason L. Sperry*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

82 Scopus citations


BACKGROUND: Controversy surrounds the optimal ratios of blood (packed red blood cell [PRBC]), plasma (fresh frozen plasma [FFP]) and platelet (PLT) use for patients requiring massive transfusion (MT) owing to possible survival bias in previous studies. We sought to characterize mortality during the first 24 hours while controlling for time varying effects of transfusion to minimize survival bias. METHODS: Data were obtained from a multicenter prospective cohort study of adults with blunt injury and hemorrhagic shock. MT was defined as 10 U of PRBC or more over 24 hours. High FFP/PRBC (≥1:1.5) and PLT/PRBC (≥1:9) ratios at 6, 12, and 24 hours were compared with low ratio groups. Cox proportional hazards regression was used to determine the independent association of high versus low ratios with mortality at 6, 12, and 24 hours while controlling for important confounders. Cox proportional hazards regression was repeated with FFP/PRBC and PLT/PRBC ratios analyzed as time-dependent covariates to account for fluctuation over time. Mortality for more than 24 hours was treated as survival. RESULTS: In the MT cohort (n = 604), initial base deficit, lactate, and international normalized ratio were similar across high and low ratio groups. High 6-hour FFP/PRBC and PLT/PRBC ratios were independently associated with a reduction in mortality risk at 6, 12, and 24 hours (hazard ratio [HR] range, 0.20-0.41, p < 0.05). These findings were consistent for 12-hour and 24-hour ratios. When analyzed as time-dependent covariates, a high FFP/PRBC ratio was associated with a 68% (HR, 0.32; 95% confidence interval [CI], 0.12-0.87, p = 003) reduction in 24-hour mortality, and a high PLT/PRBC ratio was associated with a 96% (HR, 0.04; 95% CI, 0.01-0.94, p = 004) reduction in 24-hour mortality. Subgroup analysis revealed that a high 1:1 ratio (≥1:1.5) had a significant 24-hour survival benefit relative to a high 1:2 (1:1.51-1:2.50) ratio group at both 6 hours (HR, 0.19; 95% CI, 0.03-0.86, p = 003) and 24 hours (HR, 0.25; 95% CI, 0.06-0.95, p = 004), suggesting a dose-response relationship. A high FFP/PRBC or PLT/PRBC ratio was not associated with development of multiple-organ failure, nosocomial infection, or adult respiratory distress syndrome in a 28-day Cox proportional hazards regression. CONCLUSION: Despite similar degrees of early shock and coagulopathy, high FFP/PRBC and PLT/PRBC ratios are associated with a survival benefit as early as 6 hours and throughout the first 24 hours, even when time-dependent fluctuations of component transfusion are accounted for. This suggests that the observed mortality benefit associated with high component transfusion ratios is unlikely owing to survivor bias and that early attainment of high transfusion ratios may significantly lower the risk of mortality in MT patients.

Original languageEnglish
Pages (from-to)358-364
Number of pages7
JournalJournal of Trauma and Acute Care Surgery
Issue number2
StatePublished - Aug 2012
Externally publishedYes


  • Cox proportional hazards regression
  • survival bias
  • transfusion ratios


Dive into the research topics of 'Debunking the survival bias myth: Characterization of mortality during the initial 24 hours for patients requiring massive transfusion'. Together they form a unique fingerprint.

Cite this