For the past two years, Mothers Against Drunk Driving (MADD) and Uber have been raising awareness, both on the local level and national level, about the perils of drunk driving. The ride-sharing company proudly points to one its surveys as an accomplishment. More than 80 percent of the riders report the service as helping them avoid drinking and driving.

Uber also proudly mentions a study that indicates a 3.5% to 5.6% decrease in drunk driving deaths in California. However, a newly published study in the American Journal of Epidemiology has another story to tell.

Researchers at Oxford University and the University of Southern California analyzed driving fatalities from 2009 to 2014 in the 100 most populated metropolitan areas. This study used “arguably more complete” set of controls than the California study. The finding during the time frame, when Uber was quickly prospering, was that there was no notable decrease in deaths in general or during peak drinking times, like weekend nights.

There are several reasons as to why the ride-hailing service is not making a difference. One potential factor is drivers who now call Uber are ones that are replacing taxis and public transportation, not their own cars.

Another problem is the sheer numbers. Every year 10,000 people die in the U.S. due to drunk driving accidents, and the number of Uber drivers is too small. Hundreds of thousands of Uber drivers are active each month, but 4.2 million adults are driving impaired.

Not to mention, people that are less rational regard the chances of getting caught as very low, so they drive themselves instead of paying for Uber. Each year, 1.1 million out of the 121 million people involve in drunk driving incidents are arrested.

The study focused on large cities where alternative forms of transportation already exist. It might be better to evaluate smaller cities where taxi and bus service aren’t as prevalent.

David Kirk, co-author of the study, says that there is still hope for ride-hailing companies to have an impact in the future. “It will be important to revisit the findings in a few years, assuming Uber and Lyft have even greater presence in the U.S. and beyond. And if Uber and Lyft can find a way to have greater appeal to drunk drivers (free or reduced priced rides for inebriated passengers), then there is a possibility of bringing fatalities down.”