Understanding the underlying cause of mortality in sepsis has broad implications for both clinical care and interventional trial design. However, reasons for death in sepsis remain poorly understood. We sought to characterize reasons for in-hospital mortality in a population of patients with sepsis or septic shock.Materials and methods:
We performed a retrospective review of patients admitted to the intensive care unit with sepsis or septic shock who died during their index admission. Reasons for death were classified into 6 categories determined a priori by group consensus. Interrater reliability was calculated and Fleiss κ reported. The associations between selected patient characteristics (eg, serum lactate) and reason for death were also assessed.Results:
One hundred fifteen patients were included. Refractory shock (40%) and comorbid withdrawal of care (44%) were the most common reasons for death. Overall interrater agreement was substantial (κ = 0.61, P < .01). Lactate was higher in patients who died because of refractory shock as compared with those who died for other reasons (4.7 vs 2.8 mmol/L, P < .01).Conclusion:
In this retrospective cohort, refractory shock and comorbid withdrawal of care were the most common reasons for death. Following prospective validation, the classification methodology presented here may be useful in the design/interpretation of trials in sepsis.