A Comparison of the Injury Severity Score and the Trauma Mortality Prediction Model

Published in: Journal of Trauma and Acute Care Surgery, Volume 76, Issue 1, pages 47–53 (January 2014). doi: 10.1097/TA.0b013e3182ab0d5d

Posted on RAND.org on January 28, 2021

by Alan Cook, Jo Weddle, Susan P. Baker, David W. Hosmer, Laurent G. Glance, Lee S. Friedman, Turner M. Osler

Read More

Access further information on this document at Journal of Trauma and Acute Care Surgery

This article was published outside of RAND. The full text of the article can be found at the link above.

Background

Performance benchmarking requires accurate measurement of injury severity. Despite its shortcomings, the Injury Severity Score (ISS) remains the industry standard 40 years after its creation. A new severity measure, the Trauma Mortality Prediction Model (TMPM), uses either the Abbreviated Injury Scale (AIS) or DRG International Classification of Diseases—9th Rev. (ICD-9) lexicons and may better quantify injury severity compared with ISS. We compared the performance of TMPM with ISS and other measures of injury severity in a single cohort of patients.

Methods

We included 337,359 patient records with injuries reliably described in both the AIS and the ICD-9 lexicons from the National Trauma Data Bank. Five injury severity measures (ISS, maximum AIS score, New Injury Severity Score [NISS], ICD-9–Based Injury Severity Score [ICISS], TMPM) were computed using either the AIS or ICD-9 codes. These measures were compared for discrimination (area under the receiver operating characteristic curve), an estimate of proximity to a model that perfectly predicts the outcome (Akaike information criterion), and model calibration curves.

Results

TMPM demonstrated superior receiver operating characteristic curve, Akaike information criterion, and calibration using either the AIS or ICD-9 lexicons. Calibration plots demonstrate the monotonic characteristics of the TMPM models contrasted by the nonmonotonic features of the other prediction models.

Conclusion

Severity measures were more accurate with the AIS lexicon rather than ICD-9. NISS proved superior to ISS in either lexicon. Since NISS is simpler to compute, it should replace ISS when a quick estimate of injury severity is required for AIS-coded injuries. Calibration curves suggest that the nonmonotonic nature of ISS may undermine its performance. TMPM demonstrated superior overall mortality prediction compared with all other models including ISS whether the AIS or ICD-9 lexicons were used. Because TMPM provides an absolute probability of death, it may allow clinicians to communicate more precisely with one another and with patients and families.

Research conducted by

This report is part of the RAND Corporation External publication series. Many RAND studies are published in peer-reviewed scholarly journals, as chapters in commercial books, or as documents published by other organizations.

Our mission to help improve policy and decisionmaking through research and analysis is enabled through our core values of quality and objectivity and our unwavering commitment to the highest level of integrity and ethical behavior. To help ensure our research and analysis are rigorous, objective, and nonpartisan, we subject our research publications to a robust and exacting quality-assurance process; avoid both the appearance and reality of financial and other conflicts of interest through staff training, project screening, and a policy of mandatory disclosure; and pursue transparency in our research engagements through our commitment to the open publication of our research findings and recommendations, disclosure of the source of funding of published research, and policies to ensure intellectual independence. For more information, visit www.rand.org/about/principles.

The RAND Corporation is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.