America's Biggest Safety Initiative Just Targeted the Wrong Variable

According to the toxicology reports, and there are roughly half a million of them in the FARS database, impairment rates among drivers in fatal crashes hold at approximately 20% regardless of what vehicle they were driving. Subaru Ascent owners clock in at 8%, Buick Park Avenue owners reach 32%, but most models sit somewhere between 18% and 24%, a band so narrow it might as well be a constant.^[1] Four days ago, NHTSA announced "Pathways to Safer Streets," which it called the most comprehensive traffic safety initiative in its history, and built the entire framework around that constant.

P2SS targets eight strategies: impaired driving, seatbelt compliance, speeding, distraction, law enforcement partnerships, post-crash survival, federal coordination, and data tools. Count the ones that address vehicle engineering, take your time with it, and you will arrive at zero.

I ran the cross-tab because that is what the Toxicology Desk does when somebody announces an $80 million initiative without mentioning the variable that explains the most variance in the dataset. Across 490,736 drivers in FARS fatal crashes from 2014 through 2023, impairment rates cluster around 20% with a standard deviation narrow enough to bore a statistician.^[1] Those equally-impaired drivers were sitting in vehicles that produced death rates ranging from 0.03 per 100 million vehicle miles traveled for the Tesla Model Y to 8.54 for the Hyundai Veloster, a 284-fold spread that P2SS does not acknowledge, does not address, and apparently does not find interesting.

The initiative budgets $80 million for whole-blood transfusion kits in ambulances, citing a 37% mortality reduction in trauma patients who receive blood in the field before reaching a hospital.^[2] Genuinely good policy, and it will save real lives on real highways. But it applies only to people who survive the initial impact long enough for an ambulance to arrive, and whether you survive that impact depends on which 3,500 pounds of metal you chose at the dealership. Honda Accord drivers have accumulated 4,997 excess deaths above the expected rate given their exposure miles, while Ram 1500 drivers show 4,442 fewer deaths than expected.^[1] Same highways, same 20% impairment rate, and yet one vehicle absorbs the crash while the other transmits it through the occupant compartment.

P2SS deserves its strongest counterargument at full volume. Behavioral interventions prevent crashes from occurring in the first place, which is categorically better than building vehicles that survive them. If you eliminate every impaired driver from American roads tomorrow, you prevent roughly 29% of traffic fatalities, somewhere between 10,000 and 11,000 deaths per year based on IIHS 2023 data.^[3] No crash, no structural engineering test, no transfusion kit needed. That is a powerful argument, and the math behind it is real. But the FARS cross-tabulation reveals the uncomfortable follow-up: impaired drivers are distributed essentially uniformly across vehicle models, which means the 284x death rate variance is not caused by differential impairment but by differential engineering. Targeting the 20% who are impaired saves lives within that 20%. Targeting the vehicles those drivers sit in would reduce fatality rates across all 100%.

The limitations are real and worth stating plainly. FARS captures only fatal crashes, so the ~20% impairment constant describes the worst outcomes exclusively and may look different across the 6.7 million annual crashes where nobody dies. Death rate calculations use VMT estimates from the National Household Travel Survey rather than actual odometer data, introducing approximately ±15% uncertainty for low-volume models.^[4] P2SS may contain vehicle engineering provisions not detailed in the public summary. And correlation between vehicle choice and demographic variables like age, income, and geography means we cannot isolate vehicle engineering as the sole cause of the 284x spread without controlling for confounders that FARS does not record.

If you are currently driving a vehicle from the bottom quartile of FARS death rates, P2SS offers you better trauma care after impact and hopes the other driver is sober. If you are shopping for your next car, the single most effective safety decision you can make right now is checking model-specific fatality rates at IIHS and cross-referencing with the FARS query tool at cdan.dot.gov before you sign anything, because the federal government's newest and most comprehensive safety plan just told you it is focused on everything except the machine you are sitting in.