Background Although blood alcohol concentration (BAC) is undoubtedly associated with increased risk of injury among driver victims involved in motor vehicle crashes (MVCs), some studies noted that high BAC was associated with reduced risk of mortality after injury. In addition, most of the previous studies included only injured patients admitted, which may lead to potential selection bias arising from exclusion of those with minor injury and those who died at the accident scene of MVC.

Method The population-based design included 2586 driver victims with BAC equivalent >0 and 10 307 matched controls (BAC equivalent =0) selected from the Police-reported Traffic Accident Registry from 1 July to 31 December 2016 in Taiwan. The hospital-based design comprised a subset sample, which included 517 driver victims with BAC equivalent >0 and 662 with BAC equivalent =0 hospitalised on the same day the MVCs occurred. Conditional logistic regression models with adjustment for potential confounders were used to estimate the ORs and 95% CIs of 30-day mortality associated with BAC equivalent level.

Results In the population-based design, a positive dose–gradient relationship was observed between BAC equivalent level and 30-day mortality, with a covariate-adjusted OR of 3.77 (95% CI 1.84 to 7.72), 6.19 (95% CI 3.13 to 12.26) and 7.75 (95% CI 4.51 to 13.32) for low, moderate and high BAC equivalent levels, respectively. By contrast, the hospital-based design revealed no significant association between 30-day mortality and alcohol concentration regardless of the BAC equivalent level.

Conclusion The association between BAC equivalent level and short-term mortality could have been overlooked in hospital-based studies that excluded MVC-related deaths outside hospital settings.