Date on Master's Thesis/Doctoral Dissertation

5-2022

Document Type

Doctoral Dissertation

Degree Name

Ph. D.

Department

Civil and Environmental Engineering

Degree Program

Civil Engineering, PhD

Committee Chair

Kluger, Robert

Committee Co-Chair (if applicable)

Sun, Zhihui

Committee Member

Sun, Zhihui

Committee Member

Li, Richard

Committee Member

Gentilil, Monica

Author's Keywords

Crash-related data linkage; data linkage application; police-reported crash data; emergency medical services data; trauma registry data; selectivity bias; injury severity misclassification

Abstract

Introduction: Traffic crash reports lack detailed information about emergency medical service (EMS) responses, the injuries, and the associated treatments, limiting the ability of safety analysts to account for that information. Integrating data from other sources can enable a better understanding of the characteristics of serious crashes and further explain variance in injury outcomes. In this thesis, first, a heuristic approach is proposed and implemented to link crash data to EMS run data, patient care reports, and trauma registry data. Next, the method was adapted through larger datasets in a statewide linkage effort. The performance of the heuristic method was compared with the Bayesian probabilistic linkage method. Further, EMS times, along with other crash-related explanatory variables, were used to investigate influential factors on injury severity. The level of consistency in injury severity estimation among medical experts based on trauma registry data was investigated and factors that contribute to misclassification of injury severity in crash reports were identified. Methods: A heuristic framework was developed to match EMS run reports to crashes through time, location, and other indicators present in both datasets. A comparative bias analysis was implemented on several key variables. Bayesian record linkage was also performed, and the results were compared with the heuristic one. A random-effects ordered probit approach was implemented by employing crash-EMS runs linked data to study the impact of crash-related effective factors along with EMS times on injury severity. Three models of (1) crash-related variables, (2) crash-related and EMS times, and (3) crash-related, EMS times and interaction effects of EMS times and injury location on the body were developed. Furthermore, the discrepancy between police-reported injury severities and physicians’ evaluations of corresponding trauma records was modeled using crash-related linked data. The trauma data were reviewed and classified by a panel of emergency physicians. Analysis of Variance was applied to model variation within the panel. An ordered probit model was used to model factors contributing to misclassification between police reports and emergency physicians. Results: 72.2% of EMS run reports matched to a crash record, and 69.3% of trauma registry records matched with a crash record. Females, individuals between 11 to 20 years old, and individuals involved in single-vehicle or head-on crashes were more likely to be present in linked data sets. The heuristic linkage method performs better compared to Bayesian linkage, and the reasons behind the linkage rate gap were discussed. In EMS times impact on injury severity analysis, although the outcome could not find the impact of faster EMS times on injury severity in the general model, but when the interaction effects were considered, faster EMS response time was associated with decreasing the severity of entire-body injuries. According to the discrepancy analysis results, age, internal injury, and a proposed field - injury visibility- were found to be contributing factors to injury severity discrepancy. Internal injury and injury visibility were among the trauma-related factors that were developed to explore their impact on injury severity discrepancy. Results show inconsistent physicians’ injury severity evaluation based on injuries’ detailed information. Conclusions: Linking data from other sources can significantly enhance the information available to address road safety issues, data quality issues, and more. Linking data can result in biases that should be investigated as they relate to the use-case for the data. Based on the EMS times association with injury severity outcome, although a significant relationship between EMS times and injury severity in all types of injuries was not found, EMS times based on injured body locations shed light on the relationship between EMS times and injury severity. In discrepancy analysis, findings indicate officers tended to underestimate injuries associated with a high gore factor, increasing age and the presence of an internal injury, specifically among trauma patients. Practical Applications: Linked crash-related datasets provide a valuable opportunity to evaluate the impact of prehospital care and emergency department care on crash outcomes. In general, policy steps could be taken to require cross-reporting and linkage of the data sets as the events occur to better monitor outcomes of injury crashes without requiring post-hoc linkage. This method can also realistically be integrated into a tool or software to undergo record linkage automatically. The findings of this study could act as a base for further investigation of EMS impact on injury severity, particularly with respect to effective use of EMS times in the evaluation of service quality. Further research should also be devoted to developing field tests that support officer injury assessment and identifying the factors leading to underestimating injuries identified in this study. Also, results suggest that injury visibility is important and should be investigated further for reporting purposes.

Share

COinS