Short-term mortality prediction for acute lung injury patients: External validation of the Acute Respiratory Distress Syndrome Network prediction model*
- 1 May 2011
- journal article
- research article
- Published by Ovid Technologies (Wolters Kluwer Health) in Critical Care Medicine
- Vol. 39 (5), 1023-1028
- https://doi.org/10.1097/ccm.0b013e31820ead31
Abstract
An independent cohort of patients with acute lung injury was used to evaluate the external validity of a simple prediction model for short-term mortality previously developed using data from Acute Respiratory Distress Syndrome Network (ARDSNet) trials. Data for external validation were obtained from a prospective cohort study of patients with acute lung injury. Thirteen intensive care units at four teaching hospitals in Baltimore, MD. Five hundred and eight nontrauma patients with acute lung injury. None. Of the 508 patients eligible for this analysis, 234 (46%) died inhospital. Discrimination of the ARDSNet prediction model for inhospital mortality, evaluated by the area under the receiver operator characteristic curves, was 0.67 for our external validation data set vs. 0.70 and 0.68 using Acute Physiology and Chronic Health Evaluation II and the ARDSNet validation data set, respectively. In evaluating calibration of the model, predicted vs. observed inhospital mortality for the external validation data set was similar for both low-risk (ARDSNet model score = 0) and high-risk (score = 3 or 4+) patient strata. However, for intermediate-risk (score = 1 or 2) patients, observed inhospital mortality was substantially higher than predicted mortality (25.3% vs. 16.5% and 40.6% vs. 31.0% for score = 1 and 2, respectively). Sensitivity analyses limiting our external validation data set to only those patients meeting the ARDSNet trial eligibility criteria and to those who received mechanical ventilation in compliance with the ARDSNet ventilation protocol did not substantially change the model's discrimination or improve its calibration. Evaluation of the ARDSNet prediction model using an external acute lung injury cohort demonstrated similar discrimination of the model as was observed with the ARDSNet validation data set. However, there were substantial differences in observed vs. predicted mortality among intermediate-risk patients with acute lung injury. The ARDSNet model provided reasonable, but imprecise, estimates of predicted mortality when applied to our external validation cohort of patients with acute lung injury.Keywords
This publication has 27 references indexed in Scilit:
- Assessing the Performance of Prediction ModelsEpidemiology, 2010
- A simple clinical predictive index for objective estimates of mortality in acute lung injury*Critical Care Medicine, 2009
- Recent trends in acute lung injury mortality: 1996–2005*Critical Care Medicine, 2009
- Has Mortality from Acute Respiratory Distress Syndrome Decreased over Time?American Journal of Respiratory and Critical Care Medicine, 2009
- Predictors of mortality in acute lung injury during the era of lung protective ventilationThorax, 2008
- Deciding who to admit to a critical care unitBMJ, 2007
- High Tidal Volume and Positive Fluid Balance Are Associated With Worse Outcome in Acute Lung InjurySocial psychiatry. Sozialpsychiatrie. Psychiatrie sociale, 2005
- Outcome prediction for individual intensive care patients: Useful, misused, or abused?Intensive Care Medicine, 1995
- A new method of classifying prognostic comorbidity in longitudinal studies: Development and validationJournal of Chronic Diseases, 1987
- APACHE IICritical Care Medicine, 1985