There's No Blood Test for Exhaustion. That's Why 5,693 Deaths Disappear Every Year.

Jim Hedlund's analysis for the GHSA found that 17.6 percent of all fatal crashes between 2017 and 2021 involved a drowsy driver.^[1] Apply that rate to 2023's roughly 40,990 total traffic deaths and you land at 7,214. GHSA's published estimate of 6,326 is actually conservative, filtered through additional data-quality screens. Either way, the federal count of 633 sits a full order of magnitude below every independent estimate because FARS codes drowsiness based on whatever the responding officer wrote on the crash report, and officers cannot measure what the human body does not preserve.

For scale: 5,693 uncounted drowsy driving deaths per year is more than the entire annual toll from motorcycle crashes involving alcohol (roughly 1,800) and pedestrian deaths involving distracted drivers (roughly 800) combined.^[2] GHSA calls drowsy driving the "fourth D" after drunk, drugged, and distracted. It kills like the other three but hides in the data like none of them.

Staying awake for 17 to 19 hours produces cognitive impairment equivalent to a blood alcohol concentration of 0.05.^[3] At 24 hours without sleep, you blow a functional 0.10, well past the legal limit in every state. AAA Foundation data shows nearly 20 percent of Americans admit to having driven drowsy in the previous month.^[3] GHSA estimates 1.57 percent of all driving at any given moment is drowsy, meaning roughly three out of every 200 drivers currently sharing the highway with you may be operating at the equivalent of two beers in.

At-risk populations are exactly who you'd guess. Ages 16 to 24. Long-haul truckers. Night-shift nurses. Parents of young children. Seventy percent of parents with kids under 17 report having driven drowsy.^[1] Nobody arrests them for it. Nobody can. Drowsiness is not a crime. It barely qualifies as a data field.

Strongest counterargument: the 17.6 percent estimate carries uncertainty because drowsiness is inferred from crash characteristics (single-vehicle run-off-road crashes at certain hours, no braking evidence) rather than directly measured, which means some portion of those 6,326 may involve other unmeasured factors. Fair. But the same logic applies in reverse: crashes where drowsiness was a contributing factor but a second factor (like speed) was more obvious get coded to the second factor, meaning even the 17.6 percent figure likely underestimates true prevalence in daytime multi-vehicle crashes.^[1]

One limitation worth stating plainly: FARS does not distinguish between microsleep episodes and full loss of consciousness, which means the 633 official deaths capture only the most egregious cases where drowsiness was unmistakable. Every other case vanishes into "driver lost control" or "crossed centerline for unknown reasons."

What you should do: Slept fewer than seven hours last night? You are measurably impaired right now. Pull over if you catch yourself yawning repeatedly or drifting from your lane. A 20-minute nap beats caffeine (which takes 20 minutes to hit anyway). Modern ADAS with driver monitoring (2023+ models from GM, Ford, BMW, Mercedes, Subaru) can detect drowsiness through eye-tracking, but none of them will stop you from ignoring the warning. Only thing that works is sleep. Not coffee. Not the window down. Sleep.