Cycling injury risk is an important topic, but few studies explore cycling risk in relation to exposure. This is largely because of a lack of exposure data, in other words how much cycling is done at different locations. This paper helps to fill this gap. It reports a case-control study of cycling injuries in London in 2013-2014, using modelled cyclist flow data alongside datasets covering some characteristics of the London route network. A multilevel binary logistic regression model is used to investigate factors associated with injury risk, comparing injury sites with control sites selected using the modelled flow data. Findings provide support for 'safety in numbers': for each increase of a natural logarithmic unit (2.71828) in cycling flows, an 18% decrease in injury odds was found. Conversely, increased motor traffic volume is associated with higher odds of cycling injury, with one logarithmic unit increase associated with a 31% increase in injury odds. Twenty-mile per hour compared with 30mph speed limits were associated with 21% lower injury odds. Residential streets were associated with reduced injury odds, and junctions with substantially higher injury odds. Bus lanes do not affect injury odds once other factors are controlled for. These data suggest that speed limits of 20 mph may reduce cycling injury risk, as may motor traffic reduction. Further, building cycle routes that generate new cycle trips should generate 'safety in numbers' benefits.
Keywords: Cycling; Injury; Risk; Safety in numbers; motor traffic.
Copyright © 2018 The Authors. Published by Elsevier Ltd.. All rights reserved.