Background In England, six Cycling Demonstration Towns received substantial, town-wide cycling investment from 2005 to 2011, as did 12 Cycling Cities and Towns from 2008 to 2011. The initiatives included a mixture of capital investment (e.g. cycle lanes, cycle parking) and revenue investment (e.g. promotional activities, cycle training). We examined the effects of these investments on the prevalence of cycling, walking and driving to work and whether and how those effects differed by socio-economic deprivation.

Methods This controlled before-after natural experimental study used data on travel to work from successive decennial English censuses up to and including 2011. Our outcomes were the proportion of commuters reporting cycling, walking and driving as their usual, main mode of travel to work (‘commute modal share’). We measured pre-post changes among commuters living in the 18 intervention towns and compared these to changes in three comparison groups (matched towns, unfunded towns and a national comparison group) using the difference-of-differences method. We used the Indices of Multiple Deprivation to measure area deprivation at Lower Super Output Area level and used random-effects meta-analysis to compare effects across towns.

Results Among 1.3 million commuters in the 18 intervention towns, the commute modal share for cycling rose from 5.8% in 2001 (pre-intervention) to 6.6% in 2011 (post-intervention). This represented a significant increase relative to all three comparison groups (e.g. difference-of-differences 0.64 percentage points (95% CI 0.56, 0.73) for intervention vs. matched towns). Walking to work also increased significantly in intervention towns versus comparison towns, while driving to work declined. These changes were observed across all fifths of area deprivation, with larger changes in deprived areas relative to comparison groups or national trends. There was, however, substantial heterogeneity between towns, with the overall positive effects being driven by large increases in a few large towns. The pooled town-level effect estimates were therefore non-significant (e.g. difference-of-differences 0.23 (-0.31, 0.8) for intervention vs. matched towns).

Conclusion The proportion of commuters cycling to work has increased significantly in the intervention towns, while the proportion driving to work has decreased. The distribution of these changes appears relatively equitable, with the observed socio-economic differences being smaller than would otherwise have been expected given unfavourable background trends. The heterogeneity in effects between towns indicates uncertainty regarding the likely impact of comparable investment in future towns. Nevertheless these results support the case for considering and evaluating further town-wide cycling investment.