Table 1. Key assumptions in the health impact modelling.
| Assumption | Sensitivity analyses |
| • Non-linear relationship between physical activity and health outcomes | • Alternative relationships tested |
| • Population physical activity treated as age group and sex specific log normal distributions with minimum threshold of 2.5 MET hours per week1 | • Tested by comparison with HEAT models based on mean time spent walking and mean time spent cycling |
| • Road traffic injuries a non-linear product of distance travelled by each mode, average motor traffic speed, and baseline injuries for each pairwise combination of modes | • Linear model tested; exclusion of speed model tested |
| • Air pollution only modelled PM 2.5; assumed that reduction in emissions from road transport led to equal proportional reduction in primary PM concentrations attributed to road transport | • Assumed proportional reduction in national emissions from all sources led to proportional reduction in total concentrations (including primary and secondary sources) |
PM 2.5 = particulate matter<2.5 nanometres diameter.
1
Median non-travel physical activity was added to travel physical activity for breast cancer, colon cancer, dementia, and depression but not for ischemic heart disease, cerebrovascular disease, or diabetes because walking specific relative risks were used for these three disease groups.