Determinants of Economic Costs Following Road Traffic Injuries in Canada: A Quantile Regression Forests Machine...

Area of Science:

• Health Economics
• Injury Prevention
• Machine Learning in Healthcare

Background:

• Road traffic (RT) injuries represent a significant global health and economic burden.
• Understanding the determinants of injury costs is crucial for effective resource allocation and policy development.
• Traditional cost analysis models often fail to capture the heterogeneity of costs among survivors.

Purpose of the Study:

• To identify and rank the major determinants of road traffic (RT) injury costs.
• To assess the impact of these determinants across different cost quantiles (low, medium, high).
• To inform targeted interventions for high-cost RT injury survivors.

Main Methods:

• Analysis of data from 1372 Canadian RT survivors (2018-2020).
• Estimation of healthcare and lost productivity costs over one year post-injury.
• Application of quantile regression forests and classical quantile regression to analyze 24 potential cost determinants.
• Grouping determinants into sociodemographic, psychological, health, crash, and injury factors.

Main Results:

• The 10th, 50th, and 90th cost quantiles were $1,141.9, $7,403.1, and $49,537.5, respectively.
• Injury Severity Score (ISS), Glasgow Coma Scale (GCS), and age were top determinants across cost levels.
• Common determinants included sex, employment, and living situation; ethnicity was significant for higher cost quantiles.
• Factors like education, psychological distress, and specific injury types were uniquely important for high-cost patients.

Conclusions:

• High-cost RT injury survivors are often older, retired, less educated, and present with poorer clinical and psychological indicators.
• Quantile regression methods reveal disproportionate impacts of predictors across cost distributions.
• Findings support targeted prevention and resource allocation strategies to mitigate the economic impact of RT injuries.