Related Experiment Videos
Automatic calculation of a modified APACHE II score using a patient data management system (PDMS).
Axel Junger1, Sebastian Böttger, Jörg Engel
1Department of Anesthesiology and Intensive Care Medicine, University Hospital Giessen, Rudolf-Buchheim-Strasse 7, Germany. axel.junger@chiru.med.uni-giessen.de
International Journal of Medical Informatics
|June 8, 2002
Summary
Automated calculation of a modified APACHE II score using patient data management systems shows good predictive performance for ICU mortality. Further validation is needed due to patient and method variations.
Area of Science:
- Critical Care Medicine
- Health Informatics
- Predictive Analytics
Background:
- The APACHE II score is a widely used severity-of-illness measure in intensive care units (ICUs).
- Manual calculation can be time-consuming and prone to errors.
- Integrating automated scoring into patient data management systems (PDMS) can streamline this process.
Purpose of the Study:
- To develop and validate a fully automated method for calculating a modified APACHE II score.
- To assess the predictive accuracy (discrimination and calibration) of this automated score.
- To utilize routine data from an ICU's PDMS for score calculation.
Main Methods:
- Developed SQL scripts to automatically calculate modified APACHE II scores for 524 ICU patients.
- Treated unavailable data as 'not pathological' during calculation.
- Evaluated predictive performance using Receiver Operating Characteristic (ROC) curves for discrimination and the Hosmer-Lemeshow test for calibration.
Main Results:
- The automated modified APACHE II score demonstrated good discriminative power, with an Area Under the ROC Curve of 0.790 (95% CI: 0.712-0.825).
- The score showed good calibration, as indicated by the Hosmer-Lemeshow goodness-of-fit test (P=0.70).
- Average scores differed significantly between survivors (17.8) and deceased patients (22.7).
Conclusions:
- A prediction model using automatically calculated modified APACHE II scores can be effectively built from PDMS data.
- The automated system provides a viable alternative for routine scoring in ICUs.
- Results require validation in different patient populations and settings due to potential variations in methodology and patient characteristics.