Related Experiment Videos
Performance of APACHE III models in an Australian ICU
1Intensive Care Unit, Princess Alexandra Hospital, Woolloongabba, Australia. d.cook@mailbox.uq.edu.au
Chest
|December 15, 2000
Summary
The APACHE III model accurately predicts intensive care unit mortality in Australia. Hospital mortality predictions showed some underestimation, but adjusted models performed well for quality surveillance.
Area of Science:
- Critical Care Medicine
- Health Services Research
- Biostatistics
Background:
- The APACHE III (Acute Physiology and Chronic Health Evaluation III) scoring system is widely used for predicting mortality in intensive care unit (ICU) patients.
- Independent validation of such predictive models is crucial for assessing their generalizability and performance in different healthcare settings.
Purpose of the Study:
- To evaluate the performance of the APACHE III ICU and hospital mortality prediction models in an Australian tertiary adult ICU.
- To assess the accuracy of APACHE III predictions against observed mortality rates in a diverse patient population.
Main Methods:
- A noninterventional, observational study was conducted at a metropolitan Australian tertiary referral medical/surgical ICU.
- Data from 3,398 consecutive eligible admissions between January 1, 1995, and December 31, 1997, were prospectively collected.
- Demographic, diagnostic, physiologic, laboratory, and outcome data were analyzed to assess model performance.
Main Results:
- The study population differed from the APACHE III development sample, being younger, more male, with more comorbidities.
- Receiver operating characteristic curve areas for ICU (0.92) and hospital mortality (0.90) indicated excellent discrimination.
- Observed ICU mortality (9.9%) closely matched APACHE III predictions (8.9% and 10.5% with adjustments). Hospital mortality (16.0%) was underestimated by the unadjusted APACHE III model (13.6%), but adjusted models showed closer agreement.
Conclusions:
- This Australian study demonstrates that the APACHE III model performs well on independent assessment, contrasting with some other non-American studies.
- While perfect discrimination and calibration are challenging, APACHE III models with proprietary adjustments offer a good compromise for quality surveillance.
- The findings support the utility of APACHE III for performance evaluation in Australian ICUs.