I'm afraid David Hembrow is just wrong. End of. As Stuart says, you have to break it down in to regions or towns, which is precisely what the CTC has done. Cycling in London is booming, but it's static across almost all of the rest of the UK (if memory serves Glasgow is the other big growth area). And, as cycling in London has increased the number of casualties has remained static. And Hembrow's speculation about fanatics is woeful - he simply hasn't clocked the bombdodger generation perched on floppy saddles and pushing rusty metal
Having said that, you can make a crude statistical link, but it would take a fearsome amount of analysis, probably on a road-by-road basis, to establish beyond doubt that certain circumstances led to a reduction in risk - and most of those circumstances might not be geographical. They might, for example, be related to economic activity or varieties thereof.....
The question, as Lenin put it, is what is to be done? And here we ignore the obvious at our peril. Theorising from national statistics, or even regional statistics doesn't help anybody, and the CTC's Safety in Numbers campaign, while it might convince politicians that the expansion of cycling would not neccessarily increase casualties, it's pretty much devoid of analysis rooted in events. Look at casualties and find common features, or sets of common features. And address those common features.