TY - JOUR
T1 - A comparison of prognosis calculators for geriatric trauma
T2 - A Prognostic Assessment of Life and Limitations after Trauma in the Elderly consortium study
AU - Madni, Tarik David
AU - Ekeh, Akpofure Peter
AU - Brakenridge, Scott C.
AU - Brasel, Karen J.
AU - Joseph, Bellal
AU - Inaba, Kenji
AU - Bruns, Brandon R.
AU - Kerby, Jeffrey D.
AU - Cuschieri, Joseph
AU - Mohler, M. Jane
AU - Nakonezny, Paul A.
AU - Clark, Audra
AU - Imran, Jonathan
AU - Wolf, Steven E.
AU - Paulk, M. Elizabeth
AU - Rhodes, Ramona L.
AU - Phelan, Herb A.
N1 - Publisher Copyright:
Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
PY - 2017/7/1
Y1 - 2017/7/1
N2 - BACKGROUND The nine-center Prognostic Assessment of Life and Limitations After Trauma in the Elderly consortium has validated the Geriatric Trauma Outcome Score (GTOS) as a prognosis calculator for injured elders. We compared GTOS' performance to that of the Trauma Injury Severity Score (TRISS) in a multicenter sample. METHODS Three Prognostic Assessment of Life and Limitations After Trauma in the Elderly centers not submitting subjects to the GTOS validation study identified subjects aged 65 years to 102 years admitted from 2000 to 2013. GTOS was specified using the formula [GTOS = age + (Injury Severity Score [ISS] × 2.5) + 22 (if transfused packed red cells (PRC) at 24 hours)]. TRISS uses the Revised Trauma Score (RTS), dichotomizes age (<55 years = 0 and ≥55 years = 1), and was specified using the updated 1995 beta coefficients. TRISS Penetrating was specified as [TRISS P = -2.5355 + (0.9934 × RTS) + (-0.0651 × ISS) + (-1.1360 × Age)]. TRISS Blunt was specified as [TRISS B = -0.4499 + (0.8085 × RTS Total) + (-0.0835 × ISS) + (-1.7430 × Age)]. Each then became the sole predictor in a separate logistic regression model to estimate probability of mortality. Model performances were evaluated using misclassification rate, Brier score, and area under the curve. RESULTS Demographics (mean + SD) of subjects with complete data (N = 10,894) were age, 78.3 years ± 8.1 years; ISS, 10.9 ± 8.4; RTS = 7.5 ± 1.1; mortality = 6.9%; blunt mechanism = 98.6%; 3.1 % of subjects received PRCs. The penetrating trauma subsample (n = 150) had a higher mortality rate of 20.0%. The misclassification rates for the models were GTOS, 0.065; TRISS B, 0.051; and TRISS P, 0.120. Brier scores were GTOS, 0.052; TRISS B, 0.041; and TRISS P, 0.084. The area under the curves were GTOS, 0.844; TRISS B, 0.889; and TRISS P, 0.897. CONCLUSION GTOS and TRISS function similarly and accurately in predicting probability of death for injured elders. GTOS has the advantages of a single formula, fewer variables, and no reliance on data collected in the emergency room or by other observers. LEVEL OF EVIDENCE Prognostic, level II.
AB - BACKGROUND The nine-center Prognostic Assessment of Life and Limitations After Trauma in the Elderly consortium has validated the Geriatric Trauma Outcome Score (GTOS) as a prognosis calculator for injured elders. We compared GTOS' performance to that of the Trauma Injury Severity Score (TRISS) in a multicenter sample. METHODS Three Prognostic Assessment of Life and Limitations After Trauma in the Elderly centers not submitting subjects to the GTOS validation study identified subjects aged 65 years to 102 years admitted from 2000 to 2013. GTOS was specified using the formula [GTOS = age + (Injury Severity Score [ISS] × 2.5) + 22 (if transfused packed red cells (PRC) at 24 hours)]. TRISS uses the Revised Trauma Score (RTS), dichotomizes age (<55 years = 0 and ≥55 years = 1), and was specified using the updated 1995 beta coefficients. TRISS Penetrating was specified as [TRISS P = -2.5355 + (0.9934 × RTS) + (-0.0651 × ISS) + (-1.1360 × Age)]. TRISS Blunt was specified as [TRISS B = -0.4499 + (0.8085 × RTS Total) + (-0.0835 × ISS) + (-1.7430 × Age)]. Each then became the sole predictor in a separate logistic regression model to estimate probability of mortality. Model performances were evaluated using misclassification rate, Brier score, and area under the curve. RESULTS Demographics (mean + SD) of subjects with complete data (N = 10,894) were age, 78.3 years ± 8.1 years; ISS, 10.9 ± 8.4; RTS = 7.5 ± 1.1; mortality = 6.9%; blunt mechanism = 98.6%; 3.1 % of subjects received PRCs. The penetrating trauma subsample (n = 150) had a higher mortality rate of 20.0%. The misclassification rates for the models were GTOS, 0.065; TRISS B, 0.051; and TRISS P, 0.120. Brier scores were GTOS, 0.052; TRISS B, 0.041; and TRISS P, 0.084. The area under the curves were GTOS, 0.844; TRISS B, 0.889; and TRISS P, 0.897. CONCLUSION GTOS and TRISS function similarly and accurately in predicting probability of death for injured elders. GTOS has the advantages of a single formula, fewer variables, and no reliance on data collected in the emergency room or by other observers. LEVEL OF EVIDENCE Prognostic, level II.
KW - Geriatric
KW - elderly
KW - prognosis
KW - score
KW - trauma
UR - http://www.scopus.com/inward/record.url?scp=85017619070&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85017619070&partnerID=8YFLogxK
U2 - 10.1097/TA.0000000000001506
DO - 10.1097/TA.0000000000001506
M3 - Article
C2 - 28422904
AN - SCOPUS:85017619070
SN - 2163-0755
VL - 83
SP - 90
EP - 96
JO - Journal of Trauma and Acute Care Surgery
JF - Journal of Trauma and Acute Care Surgery
IS - 1
ER -