Crashes and Injuries on Rural Roads in Alaska - Toward a Better Understanding of Rural Safety Issues through Linked Data

Abstract

In this study, motor-vehicle crash records (MVCD) are linked with hospital-based trauma data from the Alaska Trauma Registry (AKTR) using a probabilistic method using open-source record linkage software (FRIL) with the aim of developing a more comprehensive and accurate depiction of transportation-related injuries across the state. The study achieved a successful match for 66% of AKTR cases using demographic, temporal, and geographic identifiers. Injury classifications from both datasets were harmonized through a reclassification scheme to evaluate the accuracy and completeness of MVCD reporting relative to trauma data. Key findings reveal that while most records were either adequately classified or slightly overclassified, approximately 17% of MVCD records underestimated injury severity. Discrepancies were particularly evident in rural and remote communities. In addition, over 10% of linked records involving alcohol-related trauma were either not suspected of impairment or categorized as “Unknown” in MVCD, despite confirmation of intoxication in AKTR. This research demonstrates the viability and utility of data linkage to identify and address critical gaps in transportation safety records as well as support more effective injury prevention strategies, particularly in underserved and remote communities.