Background: There is mounting confusion as to which anatomic scoring systems can be used to adequately control for trauma case mix when predicting patient survival.
Methods: Several Abbreviated Injury Scale (AIS) and International Classification of Disease Clinical (ICD-9CM) -based methods of scoring severity were compared by using data from the Pennsylvania Trauma Outcome Study. By using a design dataset, the probability of survival was modeled as a function of each score or profile. Resulting coefficients were used to derive expected probabilities in a test dataset; expected and observed probabilities were then compared by using standard measures of discrimination and calibration.
Results: The modified Anatomic Profile, Anatomic Profile, and New Injury Severity Score outperformed the International Classification of Disease-based Injury Severity Score. This finding remains true when AIS values are obtained by means of a conversion from International Classification of Disease to AIS.
Conclusion: Results support the integrity of the AIS and argue for its continued use in research and evaluation. The modified Anatomic Profile, Anatomic Profile, and New Injury Severity Score, however, should be used in preference to the Injury Severity Score as an overall measure of severity.
From ThinkSharp, Inc. (W.J.S.), Bel Air, Johns Hopkins University, School of Hygiene and Public Health (E.J.M., E.G.D.), Baltimore, University of Maryland at Baltimore (H.R.C.), Army Research Laboratory (E.G.D.), Aberdeen Proving Ground, Maryland, and Temple University Hospital (R.F.B.), Philadelphia, Pennsylvania.
This research was supported in part by Grant R49/CCR302486 from the Centers for Disease Control and Prevention.
Two proprietary software products are mentioned in this paper and used in the research. TRI-CODE software is a product of Tri-Analytics, Inc. None of the authors are in a position to benefit financially from the distribution of TRI-CODE. Dr. Sacco has exited from Tri-Analytics and no longer benefits from the company's revenue. Tri-Analytics, Inc., is the exclusive licensee of the computer software, ICDMAP-90, under a license granted by The Johns Hopkins University, employer of Dr. MacKenzie. Both the University and Dr. MacKenzie receive royalty under this license.
Address for reprints: Howard R. Champion, FRCS (Edin), University of Maryland at Baltimore, 22 South Greene Street, Baltimore, MD 21201; email: firstname.lastname@example.org.