• Jennifer A Burkmar and Rajesh Iyengar.
• Jackson Park Hospital Department of Geriatrics, Wound Care, and Palliative Care, Chicago, IL, USA. jenniferburkmar@hotmail.com
• Am J Hosp Palliat Care. 2011 Aug 1; 28 (5): 321-7.

BackgroundThe Acute Physiology and Chronic Health Evaluation (APACHE) IV and Palliative Performance Index (PPI) are scales commonly used to assess prognosis in intensive care units (ICUs) and acute care units (ACUs).ObjectiveTo compare the utility of APACHE IV, PPI, and combined APACHE IV with PPI for predicting overall and disease-specific mortality.DesignThis is a prospective cohort study using admission data during the first 24 hours. Chi-square contingency tables were used to analyze mortality data for each scale.SettingThis study was conducted at a community hospital.PatientsParticipants were admitted between December 24, 2008 and April 2, 2010.ResultsThe APACHE IV, PPI, and APACHE IV plus PPI (n = 599) were significant for predicting overall mortality (P < .0001 each). The APACHE IV was also significant in predicting mortality in patients with congestive heart failure (CHF), pulmonary edema (PULEDEM), stroke (cerebrovascular accident [CVA]), terminal or metastatic cancer (CA), and dementia. The PPI was significant for predicting mortality in PULEDEM, CA, and dementia but not CVA or CHF, while the APACHE IV with PPI was significant for all diseases but CVA. The APACHE IV was the most robust in predicting ICU/ACU mortality. The combined APACHE IV and PPI improved the specificity of the PPI to predict mortality but caused a decline in sensitivity.LimitationsLimitations are due to the subjective nature of the PPI and Glasgow Coma scale (GCS), differences in illness trajectories, and a lack of reliable follow-up of all participants.ConclusionThe benefits of combining scales were best exemplified in participants with dementia. Inconsistencies in the predictive value of specific participant populations are likely due to difference in the illness trajectories of disease processes.

23 keywords...

hide…