Comparison of intensive care unit mortality performances: standardized mortality ratio vs absolute risk reduction

Introduction

The aim of this study was to assess the role of absolute risk reduction (ARR) to measure ICU performance as an alternative to the standardized mortality ratio (SMR).

Methods

This retrospective study involves patients admitted to three ICUs of a single tertiary medical center from January 2003 through December 2005. Only the first ICU admission of each patient was included in the study. The ICUs were staffed similarly. We abstracted data from the APACHE III database. For each ICU, the SMR and ARR with their 95% confidence intervals (CI) were calculated. ICU performance was categorized as shown in Table 1. When comparing ICUs, if the 95% CI of the SMR or the ARR overlap between the units, the performances were considered similar. If there was no overlap, the differences in performance were considered statistically significant.

Table 1 (abstract P470)

Results

During the study period, 12,447 patients were admitted to the three ICUs: 4,334 to the medical ICU, 3,275 to the mixed ICU and 4,838 to the surgical ICU. The predicted mortality rates were 19.5%, 16.0% and 9.0% and the observed mortality rates 14.8%, 9.7% and 4.3% for the medical, mixed and surgical ICUs, respectively. The SMR and ARR in mortality for each ICU are presented in Table 2.

Table 2 abstract P87

Conclusion

ICU mortality performances assessed by SMR and ARR give different results. The ARR may be a better metric when comparing ICUs with a different case mix.