I am a boating safety professional by trade. Understanding cycling safety has similar challenges. To have decent data, you need to know the number of "exposure hours." It is just not possible to get reliable exposure hour data for cyclists. While you can make some inferences based on the estimated total number of cyclists, you really can't arrive at a point of data reliability doing so. We don't have very good non-fatal accident data at all. We do have decent fatality numbers, but the raw numbers are so small that it's difficult to gain statistical significance in the amount of movement we see either way. The information we have surrounding the circumstances of those fatalities is often unreliable or incomplete as well. I am certainly not saying there's not any worth in the data we have, it's just that we will have to accept that some things will remain a relative mystery.