Long-term survival of intensive care and hospital patient cohorts compared with the general Australian population: a relative approach.

Rationale Long-term post-hospital survival of intensive care cohorts has been poorly characterized. The relative survival of septic and non-septic intensive care and general hospital patient cohorts, compared with the Australian population, was determined Methods A retrospective cohort study in a tertiary-level adult intensive care. Index intensive care admissions, July 1993 to June 1999, with sepsis and surviving hospital, constituted the intensive care sepsis cohort; residual patients, the intensive care non-sepsis cohort. Hospital cohorts, infected and non-infected, and Charlson Comorbidity Score (CCS) were obtained electronically, from ICD-9 codes. Follow-up was until death, or for a minimum of 4.2 years, to a maximum of 9.6 years. Time-to-death was sourced from the State registry. Relative survival was determined using the Esteve method and excess hazard modelled by covariate adjusted generalized linear models. Results The ICU sepsis (n = 224) and non-sepsis (n = 1798) cohorts were of mean (standard deviation, SD) age of 63.2 (15.6) and 59.8 (18.9) years; with co-morbidity score 1.2 (1.3) and 0.5 (0.9) respectively. Hospitalized infected (n = 8455) and non-infected (n = 51 152) cohorts were of age 56.5 (22.2) and 52.2 (20.9) years; co-morbidity score 0.4 (0.9) and 0.3 (0.9) respectively. Relative survival of all cohorts was less than the Australian population; for the two intensive care cohorts, progressive relative survival decline suggested a perpetuating excess mortality. Both age and CCS increments were associated with progressive increases in excess hazard. There was a reduced hazard for intensive care sepsis versus non-sepsis cohorts; 0.42 [95% confidence interval (CI): 0.25–0.71, P = 0.001] and surgical versus medical patients, 0.64 (95% CI: 0.50–0.84, P = 0.001); and an excess hazard for men, 1.38 (95% CI: 1.08–1.74, P = 0.009). Conclusions Adverse long-term survival of intensive care and hospital patients was demonstrated. For hospital patients there was additional infection-related mortality risk, not evident for ICU patients after case mix control.Dhaval Ghelani, John L. Moran, Andy Sloggett, Richard J. Leeson and Sandra L. Peak