TC1

As I already explained at least once, very few -- potentially no -- studies have been able to accurately study cycling injury rates, for very well-known ethical reasons.

Even countries that take their cycling very seriously struggle massively to accurately count cycling injuries. So attempting to analyze cycling safety using such numbers, that necessarily have huge error bars, is essentially pointless. This is why, for example, the previously-referenced studies estimate large improvements in cyclist fatality rates could be caused by helmet usage, while our actual real-world statistics have never shown even the slightest such improvement. When you attempt to study a problem using junk data -- which virtually all cycling injury data is, unfortunately -- your results are going to be garbage, due to GIGO.

While not ideal, that means that any accurate safety analysis must be done using fatality statistics -- which are vastly easier to compile accurately.