Hazard Regression Models of Early Mortality in Trauma Centers

David E. Clark, Jing Qian, Robert J. Winchell, Rebecca Betensky

Research output: Contribution to journalArticle

Abstract

Background: Factors affecting early hospital deaths after trauma can be different from factors affecting later hospital deaths, and the distribution of short and long prehospital times can vary among hospitals. Hazard regression (HR) models might therefore be more useful than logistic regression (LR) models for analysis of trauma mortality, especially when treatment effects at different time points are of interest. Study Design: We obtained data for trauma center patients from the 2008-2009 National Trauma Data Bank. Patients were included if they had complete data for prehospital times, hospital times, survival outcomes, age, vital signs, and severity scores. Patients were excluded if pulseless on admission, transferred in or out, or had an Injury Severity Score <9. Using covariates proposed for the Trauma Quality Improvement Program and an indicator for each hospital, we compared LR models predicting survival at 8 hours after injury with HR models with survival censored at 8 hours. Hazard regression models were then modified to allow time-varying hospital effects. Results: A total of 85,327 patients in 161 hospitals met inclusion criteria. Crude hazards peaked initially and then declined steadily. When hazard ratios were assumed constant in HR models, they were similar to odds ratios in LR models associating increased mortality with increased age, firearm mechanism, increased severity, more deranged physiology, and estimated hospital-specific effects. However, when hospital effects were allowed to vary by time, HR models demonstrated that hospital outliers were not the same at different times after injury. Conclusions: Hazard regression models with time-varying hazard ratios reveal inconsistencies in treatment effects, data quality, and/or timing of early death among trauma centers. Hazard regression models are generally more flexible than LR models, can be adapted for censored data, and potentially offer a better tool for analysis of factors affecting early death after injury.

Original languageEnglish (US)
Pages (from-to)841-849
Number of pages9
JournalJournal of the American College of Surgeons
Volume215
Issue number6
DOIs
StatePublished - Dec 1 2012

Fingerprint

Trauma Centers
Proportional Hazards Models
Logistic Models
Mortality
Wounds and Injuries
Survival
Injury Severity Score
Vital Signs
Firearms
Quality Improvement
Statistical Factor Analysis
Odds Ratio
Regression Analysis
Databases

ASJC Scopus subject areas

  • Surgery

Cite this

Hazard Regression Models of Early Mortality in Trauma Centers. / Clark, David E.; Qian, Jing; Winchell, Robert J.; Betensky, Rebecca.

In: Journal of the American College of Surgeons, Vol. 215, No. 6, 01.12.2012, p. 841-849.

Research output: Contribution to journalArticle

Clark, David E. ; Qian, Jing ; Winchell, Robert J. ; Betensky, Rebecca. / Hazard Regression Models of Early Mortality in Trauma Centers. In: Journal of the American College of Surgeons. 2012 ; Vol. 215, No. 6. pp. 841-849.
@article{06375897925d45c095a63c935e0123ce,
title = "Hazard Regression Models of Early Mortality in Trauma Centers",
abstract = "Background: Factors affecting early hospital deaths after trauma can be different from factors affecting later hospital deaths, and the distribution of short and long prehospital times can vary among hospitals. Hazard regression (HR) models might therefore be more useful than logistic regression (LR) models for analysis of trauma mortality, especially when treatment effects at different time points are of interest. Study Design: We obtained data for trauma center patients from the 2008-2009 National Trauma Data Bank. Patients were included if they had complete data for prehospital times, hospital times, survival outcomes, age, vital signs, and severity scores. Patients were excluded if pulseless on admission, transferred in or out, or had an Injury Severity Score <9. Using covariates proposed for the Trauma Quality Improvement Program and an indicator for each hospital, we compared LR models predicting survival at 8 hours after injury with HR models with survival censored at 8 hours. Hazard regression models were then modified to allow time-varying hospital effects. Results: A total of 85,327 patients in 161 hospitals met inclusion criteria. Crude hazards peaked initially and then declined steadily. When hazard ratios were assumed constant in HR models, they were similar to odds ratios in LR models associating increased mortality with increased age, firearm mechanism, increased severity, more deranged physiology, and estimated hospital-specific effects. However, when hospital effects were allowed to vary by time, HR models demonstrated that hospital outliers were not the same at different times after injury. Conclusions: Hazard regression models with time-varying hazard ratios reveal inconsistencies in treatment effects, data quality, and/or timing of early death among trauma centers. Hazard regression models are generally more flexible than LR models, can be adapted for censored data, and potentially offer a better tool for analysis of factors affecting early death after injury.",
author = "Clark, {David E.} and Jing Qian and Winchell, {Robert J.} and Rebecca Betensky",
year = "2012",
month = "12",
day = "1",
doi = "10.1016/j.jamcollsurg.2012.08.023",
language = "English (US)",
volume = "215",
pages = "841--849",
journal = "Journal of the American College of Surgeons",
issn = "1072-7515",
publisher = "Elsevier Inc.",
number = "6",

}

TY - JOUR

T1 - Hazard Regression Models of Early Mortality in Trauma Centers

AU - Clark, David E.

AU - Qian, Jing

AU - Winchell, Robert J.

AU - Betensky, Rebecca

PY - 2012/12/1

Y1 - 2012/12/1

N2 - Background: Factors affecting early hospital deaths after trauma can be different from factors affecting later hospital deaths, and the distribution of short and long prehospital times can vary among hospitals. Hazard regression (HR) models might therefore be more useful than logistic regression (LR) models for analysis of trauma mortality, especially when treatment effects at different time points are of interest. Study Design: We obtained data for trauma center patients from the 2008-2009 National Trauma Data Bank. Patients were included if they had complete data for prehospital times, hospital times, survival outcomes, age, vital signs, and severity scores. Patients were excluded if pulseless on admission, transferred in or out, or had an Injury Severity Score <9. Using covariates proposed for the Trauma Quality Improvement Program and an indicator for each hospital, we compared LR models predicting survival at 8 hours after injury with HR models with survival censored at 8 hours. Hazard regression models were then modified to allow time-varying hospital effects. Results: A total of 85,327 patients in 161 hospitals met inclusion criteria. Crude hazards peaked initially and then declined steadily. When hazard ratios were assumed constant in HR models, they were similar to odds ratios in LR models associating increased mortality with increased age, firearm mechanism, increased severity, more deranged physiology, and estimated hospital-specific effects. However, when hospital effects were allowed to vary by time, HR models demonstrated that hospital outliers were not the same at different times after injury. Conclusions: Hazard regression models with time-varying hazard ratios reveal inconsistencies in treatment effects, data quality, and/or timing of early death among trauma centers. Hazard regression models are generally more flexible than LR models, can be adapted for censored data, and potentially offer a better tool for analysis of factors affecting early death after injury.

AB - Background: Factors affecting early hospital deaths after trauma can be different from factors affecting later hospital deaths, and the distribution of short and long prehospital times can vary among hospitals. Hazard regression (HR) models might therefore be more useful than logistic regression (LR) models for analysis of trauma mortality, especially when treatment effects at different time points are of interest. Study Design: We obtained data for trauma center patients from the 2008-2009 National Trauma Data Bank. Patients were included if they had complete data for prehospital times, hospital times, survival outcomes, age, vital signs, and severity scores. Patients were excluded if pulseless on admission, transferred in or out, or had an Injury Severity Score <9. Using covariates proposed for the Trauma Quality Improvement Program and an indicator for each hospital, we compared LR models predicting survival at 8 hours after injury with HR models with survival censored at 8 hours. Hazard regression models were then modified to allow time-varying hospital effects. Results: A total of 85,327 patients in 161 hospitals met inclusion criteria. Crude hazards peaked initially and then declined steadily. When hazard ratios were assumed constant in HR models, they were similar to odds ratios in LR models associating increased mortality with increased age, firearm mechanism, increased severity, more deranged physiology, and estimated hospital-specific effects. However, when hospital effects were allowed to vary by time, HR models demonstrated that hospital outliers were not the same at different times after injury. Conclusions: Hazard regression models with time-varying hazard ratios reveal inconsistencies in treatment effects, data quality, and/or timing of early death among trauma centers. Hazard regression models are generally more flexible than LR models, can be adapted for censored data, and potentially offer a better tool for analysis of factors affecting early death after injury.

UR - http://www.scopus.com/inward/record.url?scp=84869507408&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84869507408&partnerID=8YFLogxK

U2 - 10.1016/j.jamcollsurg.2012.08.023

DO - 10.1016/j.jamcollsurg.2012.08.023

M3 - Article

VL - 215

SP - 841

EP - 849

JO - Journal of the American College of Surgeons

JF - Journal of the American College of Surgeons

SN - 1072-7515

IS - 6

ER -