Brazil · American journal of epidemiology 2016 · longitudinal difference-in-differences panel study · n=100 metropolitan areas

Uber and Metropolitan Traffic Fatalities in the United States.

Cited 133 times in the scientific literature.

Level 3 - non-randomized controlled study

Longitudinal panel study across US metropolitan counties (graded by design analogy, not clinical CEBM).

PubMed 27449416 · doi:10.1093/aje/kww062 · record verified 2026-08-31

What was done

The authors analyzed variation in the timing of Uber's rollout across US metropolitan counties from 2005 to 2014 within the 100 most populated metropolitan areas. They used negative binomial and Poisson regression models to test associations between Uber availability and total traffic fatalities, drunk driving-related fatalities, and weekend- and holiday-specific traffic fatalities.

What was found

The deployment of Uber services had no association with subsequent traffic fatalities across any tested outcome (total fatalities, drunk-driving fatalities, or weekend/holiday fatalities). The abstract reports no numerical risk estimates or confidence intervals.

Why it matters

Despite hypotheses that convenient on-demand rideshare services would reduce alcohol-involved driving deaths, this large metropolitan analysis found no measurable population-level reduction in traffic fatalities during the initial rollout years.

Limits

The abstract reports no numerical effect sizes or precision metrics. The study is ecological at the county/metropolitan level, covers the early 2005-2014 adoption window when rideshare market penetration was lower, and did not evaluate non-fatal crash injuries or specific individual-level ride usage.

Cited by