Patients Retrieved to Intensive Care via a Dedicated Retrieval Service do not Have Increased Hospital Mortality Compared with Propensity-Matched Controls

Abstract

This study was performed to estimate the effect of the retrieval process on mortality for patients admitted to a mixed adult intensive care unit (ICU) compared with propensity-matched, non-retrieved controls. Patients retrieved to the Royal Adelaide Hospital (RAH) ICU between 2011 and 2015 were propensity-score matched for age, gender, Aboriginal and Torres Strait Islander status, Acute Physiology and Chronic Health Evaluation (APACHE) III score and diagnostic group with non-retrieved ICU patients to estimate the average treatment effect of retrieval on hospital mortality. Factors associated with mortality in those retrieved were assessed by multiple logistic regression. Retrieved patients comprised 1,597 (14%) of 11,641 index ICU admissions; this group were younger, mean (standard deviation) 53 (18.5) versus 59 (17.7) years, had higher APACHE III scores, 61 (30.3) versus 56 (27.5), were more likely to be Indigenous (5.1% versus 3.7%) and to have sustained trauma (34% versus 9%). The average treatment effect for retrieval on hospital mortality, risk difference (95% confidence interval), was −0.7% (-2.8% to 1.3%), P=0.50. Variables independently associated with hospital mortality in those retrieved included age, APACHE III score and diagnostic category. Time from retrieval team activation to arrival with the patient, rural location, radial distance from the RAH and population size at the retrieval location were not significantly associated with mortality. The hospital mortality for retrieved patients was not significantly different when compared with propensity-matched controls. Mortality in those retrieved was associated with increasing age, APACHE III score and diagnostic category; however, was independent of time from team activation to arrival with the patient.