And Then You're Dead

The Autopsy / Case File · Vehicular & Transit

Coded As Something Else

The federal government's official distracted-driving death toll and the federal government's own estimate of the real distracted-driving death toll disagree by a factor of three. Both numbers came out of NHTSA. Only one of them is the one you'll see in a headline.

A fatal crash report has a box for distraction, and an officer can only check it if there's evidence — a surviving witness, a phone record, a driver alive enough to admit it. NHTSA's own 2023 economic-impact study modeled what the official count misses when that evidence doesn't exist, and the resulting number is roughly three times higher than the figure the agency actually publishes as its headline statistic every year.

And Then You're Dead · September 2026 · 8 min read

Join the discussion →
A driver looking down at a phone while operating a vehicle
A public-safety demonstration photo of a driver texting behind the wheel. Whether a real fatal crash gets coded as "distraction-involved" depends entirely on whether anyone survives, or anything survives, to say so.

Every year, the National Highway Traffic Safety Administration publishes a Research Note putting a specific number on distracted-driving deaths — 3,308 in 2022, 3,275 in 2023, figures in the low thousands that get repeated in news coverage, insurance-industry reports, and public-safety campaigns as the authoritative federal count. Every year, that number represents somewhere between 8% and 9% of total traffic fatalities. And every year, that number comes from the Fatality Analysis Reporting System, a database built entirely from what police officers can document at the scene of a crash — which, for a driver who dies before anyone can ask what they were doing in the seconds before impact, is often nothing at all.

The Other NHTSA Number

In February 2023, NHTSA published a different kind of report: "The Economic and Societal Impact of Motor Vehicle Crashes, 2019," an analysis built not from police-reported crash classifications but from a cost-modeling methodology designed to capture the full downstream burden of motor-vehicle crashes — medical costs, lost productivity, legal costs, property damage, and quality-of-life loss. Distraction-involved crashes, under that report's methodology, produced 10,546 fatalities in 2019 alone, along with 1.3 million nonfatal injuries and $98.2 billion in direct economic costs — a fatality count more than three times what that same year's FARS-based Research Note would have reported using the standard police-report classification method.

0
distracted-driving deaths NHTSA's FARS-based Research Note reported for 2019 — the police-report-classified, headline-cited figure
0
distraction-involved fatalities the same agency's 2019 Economic and Societal Impact cost-modeling report counted for the same year

The two figures aren't reconciled anywhere in NHTSA's public communications, and the agency has never retired the lower, FARS-based figure in favor of the higher one. Both numbers remain in active circulation: the Research Note as the citation that appears in press releases and public-safety campaigns, the Economic and Societal Impact figure as the one buried in a 300-plus-page cost-accounting document that safety researchers and insurance actuaries read far more often than the general public does.

What a Crash Report Can't Ask a Dead Driver

The mechanism behind the gap isn't a mystery, and NHTSA's own researchers have documented it directly. FARS classifies a crash as distraction-involved based on evidence available to the investigating officer at the scene: a surviving driver's admission, a passenger's account, a phone company's records subpoenaed after the fact, or physical evidence like a phone found unlocked and mid-message in the wreckage. A single-vehicle fatal crash with no witnesses and a driver who dies at the scene routinely gets coded to a proximate cause — ran off the road, failed to negotiate a curve, overcorrected — with no way to determine, and therefore no box checked, for whatever was happening on the driver's phone thirty seconds earlier.

The National Safety Council has independently documented how large that undercount can run for a single distraction subtype. Reviewing 180 fatal crashes between 2009 and 2011 where investigators found other evidence of driver cell phone use at the time of the crash, NSC found only 52% had actually been coded as cell-phone-involved in the national crash database by 2011 — meaning nearly half of crashes with documented phone-use evidence weren't captured in the official distraction count even when the evidence existed and was found. NSC's broader estimate, drawn from a different data source altogether, put cell-phone involvement in roughly 25% of all crashes — a figure that dwarfs both of NHTSA's own distraction estimates and underscores how much of this mechanism depends on what evidence happens to survive a crash, rather than what actually happened inside it.

A phone screen doesn't survive a windshield the way a blood-alcohol result survives an autopsy. Distraction is the one major killer on American roads that the evidence for often dies in the same crash it caused.

Compare It to a Killer That Does Leave Evidence

The asymmetry becomes clearer set against alcohol-impaired driving, a cause of death FARS captures with far more confidence because it leaves a chemical trace a medical examiner can measure independent of witnesses, admissions, or surviving evidence at the scene. A blood-alcohol test performed during autopsy or hospital treatment produces a number regardless of whether the driver survives to be interviewed, which is part of why alcohol-impaired fatality statistics are treated as comparatively reliable even by researchers who are skeptical of other FARS categories. Distraction has no equivalent forensic marker. There is no blood test for having glanced at a text message, and the digital trail that could substitute for one — a phone's screen-on timestamp, a messaging app's send log — requires a subpoena, a cooperating carrier, and an investigator with the time and cause to pursue it, resources that aren't uniformly available across the thousands of local law-enforcement agencies that feed fatal-crash data into FARS every year.

That structural gap means the FARS-reported distraction share of fatal crashes isn't simply a smaller number than the true share — it's a number that systematically undercounts single-vehicle crashes, crashes with no surviving witnesses, and crashes where investigating agencies lack the resources to pursue phone records, while comparatively over-representing crashes where a passenger or a second driver survived to describe what they saw. That's not a small sampling quirk; it's a bias built into what the classification system is capable of detecting in the first place.

The Number That Actually Drives Policy

The distinction matters beyond bragging rights over which statistic is more alarming. Distracted-driving laws, hands-free mandates, and public-safety-campaign budgets are built and evaluated against the FARS-reported figure, because it's the number published annually, cited consistently, and trackable year over year in a way the periodic and far more methodologically involved Economic and Societal Impact report isn't. A state legislature weighing whether its existing texting-while-driving law is working, or whether a broader phone-restriction law is worth the political cost, is typically working from a data source that its own source agency has separately estimated understates the underlying harm by a factor of roughly three.

NHTSA's continued use of both figures without reconciling them isn't necessarily bad faith — the two reports are built for different purposes, using different, defensible methodologies, and both are the agency's own honest work. But the practical effect is that the number the public actually sees, cites, and legislates around is the smaller of the two, while the more comprehensive estimate sits in a cost-accounting report that safety advocates say deserves far more attention than it gets when the conversation turns to how dangerous distracted driving actually is.

And Then You're Dead

The death certificate says single-vehicle crash, ran off the roadway, no further contributing factors identified. Nobody subpoenaed the phone because nobody survived to be a suspect and nobody was left to ask. Somewhere between 3,142 and 10,546, in the same year, from the same agency, is the actual number of people that happened to — and the honest answer to which figure is right is that the crash report was never built to know.

That's it. That's the whole thing.

Comments

Loading…

Comments are anonymous and moderated.

Sources

Photo: "Distracted driving kills, keep your eyes and mind on the road," by Cpl. Melissa Wenger, U.S. Marine Corps, via Wikimedia Commons (public domain).