The Fallacy of Interpreting Deaths and Driving Distances

Abstract

On December 10, 2012, the National Highway Traffic Safety Administration released annual statistics on traffic fatalities in the United States. The summary report emphasized that the fatality rate had decreased to a historic low of 1.10 deaths per 100 million vehicle-miles traveled. The top graphic vividly showed a 75% decline in traffic risks during the last 4 decades in the United States (Figure 1). In the accompanying press release, then Transportation Secretary Ray LaHood summarized, ‘‘The latest numbers show how the tireless work of our safety agencies and partners, coupled with significant advances in technology and continued public education, can really make a difference on our roadways.’’ These data suggest a large and sustained improvement in road safety, reassuring to motorists, regulators, and expert commentators. The purpose of this editorial is to review misunderstandings that can occur when assessing traffic risks as a ratio of total deaths to total travel distances. At a casual glance, for example, the summary statistic immediately inserts the qualifier ‘‘per million’’ into the expression. Doing so creates an intuitive impression of a miniscule magnitude (verging on comparisons to a 1 per million risk of being killed by lightning each year). The remainder of the expression is sufficiently arcane that people may not notice that traffic risks in the United States are now about 80% higher than the same statistics from the United Kingdom. The large point is that the thousands of deaths each year might merit attention regardless of how they are correlated to driving distances. To be sure, traffic risks increase if motorists travel longer distances. However, the relationship of crash risk to travel distance is complex because longer trips channel vehicles toward controlled protected roads and away from crowded complex intersections. The risks of death with airplane travel, for perspective, are rarely calculated as a ratio of deaths per vehiclemile since the dangers are concentrated around takeoff and landing with much lower risks during intervening distances. Integrating traffic fatalities with community mobility is logical since the risks of a traffic crash would be zero if all persons were stationary. The difficulty arises because no exposure metric is ideal and the summary statistic presumes that all miles were identical. Most other common causes of death in the United States are reported as plain trends despite known cause-and-effect relationships with exposures. From 2000 to 2010, for example, annual deaths from colorectal cancer decreased by 9% (from 57,434 to 52,045) and annual cigarette consumption decreased by 31% (from 435.6 billion to 300.4 billion). Cancer control agencies, however, are unlikely to gauge progress by calculating a metric of ‘‘cancer deaths per million cigarettes.’’ Instead, summary statistics focus on actual counts of deaths, sometimes accompanied by simple adjustments for population size and demographics. The metric of ‘‘deaths per million vehicle miles’’ is thereby irregular and perhaps dehumanizing. Received 3 December 2013 from the Department of Medicine, University of Toronto; Evaluative Clinical Sciences Program, Sunnybrook Research Institute; Institute for Clinical Evaluative Sciences; Division of General Internal Medicine, Sunnybrook Health Sciences Centre; and Center for Leading Injury Prevention Practice Education & Research, Toronto, ON, Canada. This article was supported by a Canada Research Chair in Medical Decision Sciences and the Canadian Institutes of Health Research. The funding organizations had no role in the design or conduct of the study; collection, management, analysis, or interpretation of the data; and preparation, review, or approval of the manuscript. Revision accepted for publication 10 February 2014.