the biggest risk is sensors. O2 sensors are inherently wonky, and CO2 sensors aren't really available that work. If we had rock solid O2 sensors that didn't drift, get limited, or just sh!t the bed *hopefully the Poseidon sensors prove to fix this*, and we also had CO2 sensors that worked in the environment that we dive in, then you would be able to trust the O2 readings from the sensors, and know if there was breakthrough. If that was the case, the safety would likely get significantly better.
As mentioned above, in OW diving, the difference in say normoxic depths may not be significant because you can carry 2x al80's and get up from anywhere you are with little issue so you are buying some time ability due to gas consumption of OC, but you aren't really getting much of a benefit outside of gas consumption.
In cave diving, the time factor is something that has saved several of my friends who would have died if they were on OC. Carrying enough gas is impractical for a lot of those dives so it saves a lot of logistic concerns.
The big thing about the statistics above are to look at who is doing what kind of dives and whether the dives that are killing CCR divers are still being conducted on OC *they largely aren't*, what the incident rate would be if they were on OC *probably pretty comparable*. It's important to be able to differentiate if it was the CCR that killed them *Wes Skiles*, or they died, and happened to be on CCR but they probably would have died whether they were on CCR or on OC