LAUSR.org creates dashboard-style pages of related content for over 1.5 million academic articles. Sign Up to like articles & get recommendations!

32: RECALIBRATION OF APACHE II FOR MORTALITY PREDICTION IN A NEUROINTENSIVE CARE UNIT PATIENT POPULATION

Photo by thinkmagically from unsplash

Learning Objectives: The Acute Physiology and Chronic Health Evaluation II (APACHE II) severity score is widely used for inhospital mortality prediction in general intensive care units (ICU). The aim of… Click to show full abstract

Learning Objectives: The Acute Physiology and Chronic Health Evaluation II (APACHE II) severity score is widely used for inhospital mortality prediction in general intensive care units (ICU). The aim of this study is to recalibrate APACHE II for a specialist neurological ICU (NICU). Methods: Prospectively collected audit data from November 2007 to June 2017 in a Scottish tertiary head and neck centre was used in a retrospective study. Data from 3,110 patients was suitable for calculation of APACHE II scores and logistic regression modelling to generate inhospital mortality predictions. APACHE II calibration was evaluated on the complete dataset. The dataset was then divided into a training set (n = 2,073, 67%) and test set (n = 1,037, 33%). APACHE II was recalibrated on the training set, creating new diagnostic coefficients for 7 neurological, 4 trauma and 1 major maxillofacial surgery category, from the original APACHE II diagnostic categories. Patients admitted to our centre and subsequently admitted to NICU for a reason out with these categories were combined into an ‘other’ category. The recalibrated model discrimination was assessed using the concordance (C) statistic. In the test set, using bootstrapping, calibration was assessed and compared to the calibration of original APACHE II in the test set, for median absolute error between actual and predicted mortality. The Le Cessie-van HouwelingenCopas Hosmer global goodness of fit test validated the model. Results: APACHE II calibration on the complete dataset gave a median absolute error of 14.1% between actual and predicted mortality. The recalibrated model on the test set achieved a median absolute error of 1.4% compared to 11.1% on the original APACHE II model, between actual and predicted mortality. The recalibrated model had very good discrimination (C statistic = 0.842) and calibration (Le Cessie-van Houwelingen-Copas Hosmer global goodness of fit test z = -0.90715, p = 0.3643). Conclusions: The recalibrated APACHE II model improved predictive accuracy for in-hospital mortality in our NICU. External validation of the model in patients admitted with neurological diagnoses to general and neurological ICUs is required.

Keywords: apache; mortality; model; calibration; test; care

Journal Title: Critical Care Medicine
Year Published: 2018

Link to full text (if available)


Share on Social Media:                               Sign Up to like & get
recommendations!

Related content

More Information              News              Social Media              Video              Recommended



                Click one of the above tabs to view related content.