Estimating attributable fraction of mortality from sepsis to inform clinical trials.

•We test whether understanding the attributable fraction of deaths from sepsis (AFsepsis) could improve sepsis trial design?•We estimated AFsepsis in a cohort of critically ill sepsis patients, using two controls•First, by propensity matching to critically ill non-sepsis patients and, second, by age and sex matching to the general population controls.•We found that AFsepsis varied from 15%-93% and septic shock had higher attributable fraction of deaths•Using AFsepsis to calculate sample sizes for clinical trials results in significantly larger values than existing trials have used.•Sample size estimations based on the trial specific eligibility criteria derived AFsepsis could help improve sensitivity of sepsis trials.