Fascinating link. Many thanks.
There is a "count point" very close to the recent tragedy (56413). From the download there were, for 2013, the most recent figure 49 pedal cycles with a total for all traffic of nearly 49,000. I am not sure over what period this count was made but at a ratio of 1 push bike to a thousand vehicles this suggests to me what I already suspected, this is a road not often used by cyclists and therefore the total number of serious accidents and fatalities does not give a true picture of the risk of cycling along this road.
I am probably not interpreting this data correctly so if someone who is more familiar with it can do I would be interested.
Given this data is all in a computer somewhere it would hardly be rocket science for the same people who publish it to produce a map showing which roads are the greatest risk. But as a cynic they probably wouldn't want to draw attention this way.
Edit: Just selected several other counting points at random in Devon and looked at the ratios. The problem seems to be the counting points are all main roads so in one sense not much difference between them but the second in the list (6407) gives a bike to vehicle ratio of approaching 1:100 which is more what I might expect for an urban situation but many of the others are around the 1:1000 recorded on the A38.
Not sure where this is leading, only I won't be cycling on the A38 any time soon.