Efficacy and cost-effectiveness of subsidized ridesharing as a drunk driving intervention in Columbus, OH

•Well-publicized free safe rides to/from hospitality zones reduce drink-driving.•These rides increase alcohol consumption and cost more than they save.•Limiting rides to one-way from the zones does not change these findings.•Return on investment would rise for drink driving crash rates above US rates.•US governments should not devote energy or resources to ridesharing programs. No economic evaluations exist of free or subsidized ridesharing services designed to reduce impaired driving. To evaluate the effects and economics of a 17-weekend program that provided rideshare coupons good for free one-way or round trips to/from the hospitality zones in Columbus, Ohio, coupled with a modest increase in enforcement and a media campaign that used messaging about enforcement to promote usage. Web surveys of riders and intercept surveys of foot traffic in the hospitality zones yielded data on the reduction in driving after drinking and the change in alcohol consumption associated with coupon use. We estimated crash changes from trip data using national studies, then confirmed with an ARIMA analysis of monthly police crash reports. Costs and output data came from program and rideshare company records. 70.8% of 19,649 responding coupon redeemers said coupon use reduced the chance they would drive after drinking. An estimated 1 in 4,310 drink-driving trips results in an alcohol-attributable crash, so the coupons prevented an estimated 3.2 crashes. Consistent with that minimal change, the ARIMA analysis did not detect a drunk-driving crash reduction. Self-reports indicated alcohol consumption rose by an average of 0.4 drinks per coupon redeemer, possibly with an equal rise among people who rode with the redeemer. The program cost almost $650,000 and saved an estimated 1.8 years of healthy life. Across a range of discount rates and values for a year of healthy life, it cost $366,000 to $791,000 per year of healthy life saved. Its estimated benefit-cost ratio was between 0.31 and 0.59, meaning it cost far more than it saved. Ridesharing, coupled with a media campaign and increased enforcement, was not a cost-effective drunk-driving intervention. Although it reduced drink-driving crashes and saved years of healthy life, those savings were modest and expensive. Moreover, the self-reported increase in participant drinking imposed countervailing risks. Even sensitivity analyses that potentially overestimate the benefits and underestimate the costs indicate a significant imbala