The APACHE II-derived standardised mortality ratio (SMR) is widely used as a measure of quality of care. Treatment and resuscitation of a patient prior to intensive care unit (ICU) admission may reduce a patient's APACHE II score without necessarily reducing the risk of death. We investigated whether treatment on a high-dependency unit (HDU) immediately prior to admission to an ICU in a university teaching hospital influences (a) outcome in the HDU treatment group and (b) the overall SMR of an ICU.
A retrospective study of all patients admitted to an adult general ICU eligible for APACHE II predicted mortality from 1996 to 2002. The place of care immediately prior to intensive care admission was identified. The APACHE II-derived SMR, risk ratio and confidence intervals were calculated for patients with prior treatment on the HDU and for all other ICU admissions (controls). In addition, because SMR was not constant across all strata of predicted mortality, patients were stratified by APACHE II-predicted mortality. An estimated odds ratio and the Mantel–Haenzsel chi-squared test were used to compare hospital mortality for each group.
A total of 2045 admissions with eligible APACHE II mortality prediction were studied. The overall hospital mortality was 720 (35.2%). Two hundred (9.7%) admissions received prior treatment on the HDU with a mortality rate of 44.5%. The SMRs of the HDU and of the control group were 1.64 and 1.45, respectively. Overall, the risk ratio of the APACHE II SMR between groups was not significant (risk ratio [confidence interval] 1.13 [0.84–1.52]). The estimated odds ratio for hospital mortality in the HDU group was not significant (odds ratio [confidence interval] 1.36 [0.99–1.88]).
Treatment on the HDU immediately prior to ICU admission did not appear to influence the SMR in the HDU group or overall. We could not find evidence that the APACHE II-predicted mortality was less reliable in patients treated in the HDU immediately prior to ICU admission.