Abstract

PURPOSE:

To determine the epidemiology, in-hospital mortality, trends, patient characteristics and predictors of intensive care unit (ICU) readmission in Australia.

METHODS:

A retrospective longitudinal study of data for 38 Australian ICUs extracted from the Australian and New Zealand Intensive Care Society Adult Patient Database (ANZICS-ADP) for the years 2000-2007. Demographic, diagnostic, physiological and outcome data were analysed. A multivariate model was constructed to identify risk factors for ICU readmission. Outcomes examined included observed and risk-adjusted in-hospital mortality.

RESULTS:

A total of 247,103 patients were discharged alive from their first ICU admission; 13,598 (5.5%) were readmitted at least once. Variables associated with an odds ratio greater than 1.05 for readmission (p < 0.001) were an initial ICU admission source other than elective surgery, any chronic health variable on severity scoring, tertiary hospital ICU and discharge between 6 p.m. and 6 a.m. Five initial diagnoses were associated with an odds ratio (OR) greater than 2 for readmission (p < 0.001). In-hospital mortality in readmitted patients was 20.7% compared with 4.4% in those not readmitted. Readmission rates have not changed over the study period. After adjustment for illness severity and readmission propensity, ICU readmission remained significantly associated with in-hospital mortality (OR 5.4, 95%, confidence interval (CI) 5.1-5.7).

CONCLUSIONS:

Many risk factors for increased ICU readmission were identified in this study including ICU discharge between 6 p.m. and 6 a.m. This was the only modifiable variable studied. Prospective studies are required to identify other factors and to determine whether interventions may reduce ICU readmission and its high associated in-hospital mortality.