Predicting cycling casualty numbers in London

Continuing the cycling theme, I recently stumbled across an intriguing description of some research by Ben Lewis of Transport for London that was apparently presented to a seminar at the 2010 European Transport Conference.

Entitled 'Assessing the risk to London cyclists from Mayoral policy', it is mainly concerned with proposing a new way to predict the change in cycling casualty rates as overall ridership goes up. Here's the wonkish bit:

Current advice by the Department for Transport is to use an exponent function to estimate accident numbers resulting from an increase in average flow. This paper reviews DfT methodology using London data and seeks to provide policy makers with a more robust methodology using negative binomial regression of site-specific road characteristics: mode share, speed, delay, population density, length of link and number of lanes were tested for significance and a new accident exponent function was found. The results of this study indicate that a negative binomial regression model that uses publicly available data provides a rigorous methodology that can be used to assess the risk that pedal cyclists face on the road network on a link by link basis.

This sounds potentially very useful on its own merits, but there's more:

It was found that the exposure rate for central and inner London is up to five times larger using the bespoke model output rather than aggregated statistics. The model predicts that a quadrupling of cycle activity from 2001 levels without the inclusion of any additional complementary safety measures would result in a 50% increase in accidents in central London and a 30% increase in inner London.

In other words, if the number of cycle trips quadruples without further steps to make roads safer, then the accident rate will come down ('safety in numbers' at work perhaps?) but we will still see significantly more casualties in absolute terms.

You might respond that 50% more casualties is a price worth paying for 300% more cycling, and there's a logic to that, but I think the Mayor and TfL quite rightly aspire to increasing cycling while bringing down the number of casualties. From the summary (I haven't been able to find a copy of the full paper yet), this research seems suggest to suggest that will be a struggle unless the roads get much more cycle-friendly.