• M L A Heldeweg, T T R Heldeweg, J A H Stohlmann, P Freire Jorge, R Boer, L A Schwarte, and P Schober.
• Department of Anesthesiology, Amsterdam University Medical Centers, Vrije Universiteit Amsterdam, Location VUmc, Amsterdam, The Netherlands. M.heldeweg@amsterdamumc.nl.
• Scand J Trauma Resus. 2024 Oct 25; 32 (1): 104104.

BackgroundThe Advanced Trauma Life Support classification (ATLS) of hypovolemic shock is a widely used teaching and treatment reference in emergency medicine, but oversimplifies clinical reality. A decade ago, a landmark study compared vital parameters to base deficit (BD) in trauma patients. The investigators concluded that BD had higher accuracy to detect the need for early blood product administration. BD was subsequently introduced in the ATLS shock classification and has since been widely accepted as a laboratory standard for hypovolemia. The aim of this study is to investigate whether a methodological bias may have inadvertently contributed to the study's results and interpretation.MethodsIn the current study, we replicate the original study by simulating a cohort of trauma patients with randomly generated data and applying the same methodological strategies. First, a predefined correlation between all predictor variables (vital parameters and BD) and outcome variable (transfusion) was set at 0.55. Then, in accordance with the methods of the original study we created a composite of ATLS parameters (highest class amongst heart rate, systolic blood pressure, and Glasgow Coma Scale) and compared it with BD for resulting transfusion quantity. Given the preset correlations between predictors and outcome, no predictor should exhibit a stronger association unless influenced by methodological bias.ResultsApplying the original imbalanced grouping and composite allocation strategies caused a systematic overestimation of shock class for traditional ATLS parameters, favoring the association between BD and transfusion. This effect persisted when the correlation between BD and transfusion was set substantially worse (rho = 0.3) than the correlation between ATLS parameters and transfusion (rho = 0.8).ConclusionsIn this fully reproducible simulation, we confirm the inadvertent presence of methodological bias. It is physiologically reasonable to include a metabolic parameter to classify hypovolemic shock, but more evidence is needed to support widespread and preferred use of BD.© 2024. The Author(s).

24 keywords...

hide…