Dr Tsang Hin Hung
Intensive Care Unit, Kwong Wah Hospital (December 2001 Critical Care Medicine Exit Assessment Exercise)
Abstract
Objective To evaluate the use of daily APACHE II score and its dynamic change in prediction of
hospital mortality in adult intensive care unit (ICU) patients.
Design Prospective cohort analysis
Methods The study included a consecutive sample of 709 ICU admissions. Daily APACHE II
scores and the dynamic changes of the scores (daily risk score) during the first 7 days of ICU stay
were collected. Multivariable logistic regression analysis was used to select variables
independently predict hospital mortality. Daily APACHE II score and daily risk score with 90%
and 95% specificity in prediction of hospital mortality were selected.
Results The current or most recently measured APACHE II score was the single most important
variable in prediction of hospital mortality. Sensitivity, positive predictive value, negative
predictive value and accuracy of prediction with daily APACHE II score selected at 90%
specificity in prediction of hospital death were around 45%, 80%, 80% and 78% respectively.
Daily risk score assessment did not improve the prediction accuracy.
Conclusion These mortality prediction models can be utilized to improve communication among
clinical staff and between clinicians and family members. Clinical judgment, especially on
withdrawal of intensive treatment, will still require integration of a wide range of medical facts.