Simulation of survivorship bias in observational studies on plasma to red blood cell ratios in massive transfusion for trauma

Abstract

Observational studies on injured patients requiring massive transfusion have found a survival advantage associated with use of equivalent number of units of fresh frozen plasma (FFP) and packed red blood cells (RBCs) compared with use of FFP based on conventional guidelines. However, a survivorship bias might have favoured the higher use of FFP because patients who died early never had the chance to receive sufficient FFP to match the number of RBC units transfused. A Markov model using trauma data from local hospitals was constructed and various FFP transfusion scenarios were applied in Monte Carlo simulations in which the relative risk of death associated with exposure to high FFP transfusion was set at 1.00, so that the FFP : RBC ratio had no influence on mortality outcome. Simulation results showed that the relative risk associated with exposure to high FFP transfusion was less than 1.00 (0.33-0.56 based on programmed delays in achieving an FFP : RBC ratio of 1 : 1-2), thus demonstrating a survivorship bias in favour of FFP : RBC equal to or more than 1 : 1-2 in certain observational trauma studies. This bias was directly proportional to the delay in achieving a FFP : RBC ratio of 1 : 1-2 during resuscitation. Some observational studies comparing low and high FFP administration in injured patients requiring massive transfusion probably involve survivorship bias that inflates or creates a survival advantage in favour of a higher FFP : RBC ratio.