Linking crash and breathalyzer data in Connecticut

Area of Science:

• Public Health
• Transportation Safety
• Data Linkage

Background:

• Motor vehicle crashes (MVCs) pose a significant public health concern.
• Alcohol impairment is a major contributing factor to MVCs.
• Linking diverse datasets can enhance understanding of crash dynamics and prevention.

Purpose of the Study:

• To document the methodology for linking breathalyzer and MVC data in Connecticut.
• To utilize a unique identifier to protect personal information while enabling data linkage.
• To assess the impact of data linkage on key variables and dataset characteristics.

Main Methods:

• Deterministic linkage was employed using Microsoft SQL Server.
• Breathalyzer and MVC driver records were joined for the period 2017-2022.
• Data quality was assessed by comparing linked and original dataset distributions.

Main Results:

• Over 80% of breathalyzer records were successfully linked to MVC driver records.
• Minor changes were observed in breathalyzer data proportions; however, linked MVC data showed an overrepresentation of males and drivers aged 26-40.
• The linked dataset exhibited lower proportions of severe injury records compared to the original MVC data.

Conclusions:

• The linkage methodology is effective for creating quality matches without personal data.
• Observed demographic and injury severity differences in the linked data align with existing research.
• Future research will incorporate additional toxicology data to build a comprehensive transportation research database.