Interesting. The change is almost totally attributable to ESC (electronic stability control), which directly addresses SUV's Achilles' heel. It's also interesting to note the variability in results -- the Audi A6 is one of the safest luxury vehicles (I suspect the Tesla S will change that), and the Honda Accord is remarkably safe for a mid-sized sedan (though the Civic doesn't do as well relative to its class).
It's also interesting to note that pickups can fare quite poorly (the Nissan Titan Crew Cab is among the worst vehicles listed), and even within the same manufacturer, lager models aren't necessarily safer (Mazda 3 at 52 deaths/million miles, Mazda 6 at 60). The error bounds and total miles (exposure) on a number of the larger SUVs are also large and small respectively, suggesting a possible data collection limitation.
That's per million vehicle years not per million miles.
When I read your comment I realized something must be off and went to check the source data. Imagine if the average driver drives 20K miles / year it would take only 50 drivers to get to million miles. If the death rate of cars would even approach 50 deaths per million miles then everybody would stand a good chance of dying in a car accident. Clearly this is not the case.
In 2009 there were 10.5 million accidents, and 35,900,000 motor vehicle fatalities. Your odds of dying in any given accident are 0.3%.
In 2013 projections were for 1,203.6 billion vehicle miles travelled. Travel mileage has actually been declining slightly (about 1%/year) since the 2008 depression.
You could expect to travel 33,526,462 without a death, on average.
I had a slight gash on my head that bled for a little bit, and I have a dislocated clavicle near my breast bone.
I was driving 20 Mph, taking a left turn, the Nissan Titan was doing (by the drivers accounts) 45 - 50 Mph. Practically head-on.
According to the fire department, police and EMT's that arrived on the scene it was a miracle how in-tact the interior was, and that I walked away from it. The insurance adjuster at first wouldn't believe me when I told him to start working up paperwork for totalled while also telling him I had walked away from it. He ended up later calling me while looking at my vehicle, and I quote, "Are you sure you are okay?"
The guy that towed my vehicle from the accident site was surprised to see me after I was released from the hospital looking for my keys (house keys are on the same key ring), he hadn't been told about my status by the cops or anyone, and had assumed I was in crit-condition in the hospital.
Subaru saved my life, and for that I am extremely happy.
With the increasing prevalence of smog checks as a registration renewal requirement, and the inclusion of a highly detailed electronic vehicle record (see for example California's BAR-90 and subsequent records, Washington and several other states have similar records), odometer data is available and is used heavily by insurance companies in assessing risk for a large number of drivers.
It's also interesting to note that pickups can fare quite poorly (the Nissan Titan Crew Cab is among the worst vehicles listed), and even within the same manufacturer, lager models aren't necessarily safer (Mazda 3 at 52 deaths/million miles, Mazda 6 at 60). The error bounds and total miles (exposure) on a number of the larger SUVs are also large and small respectively, suggesting a possible data collection limitation.