I'd say the report is misleading. Fatalities are rare events. They chose cars with 100,000 registered vehicles or more. The categories these vehicles belong to

I'd say the report is misleading. Fatalities are rare events. They chose cars with 100,000 registered vehicles or more. The categories these vehicles belong to experienced around 30 deaths per million cars, so cars that have 100,000 registrations would expect 3 deaths. You could easily get a few models with 0 deaths just as a matter of chance. I don't think these results necessarily show these cars are inherently safer than other cars in the same category.