Objective: To measure the association between intensive care unit (ICU) admission and both hospital and long-term mortality, separate from the effect of hospital admission alone.
Design: Retrospective cohort study.
Setting: All hospitals in British Columbia, Canada, during 3 fiscal years, 1994 to 1996.
Patients: A total of 27,103 patients admitted to ICU and 41,308 (5% random sample) patients admitted to hospital but not to ICU.
Intervention: None.
Measurements and main results: Although ICU admission was an important factor associated with hospital mortality (odds ratio: 9.12; 95% confidence interval: 8.34-9.96), the association between ICU admission and mortality after discharge was relatively minimal (hazard ratio: 1.21; 95% confidence interval: 1.17-1.27) and was completely overshadowed by the effect of age, gender, and diagnosis.
Conclusions: After controlling for the effect of hospital admission, admission to ICU has minimal independent effect on mortality after discharge.