BACKGROUND
Despite strong evidence from observational studies that moderate to high levels of physical activity can have a substantial impact on major non communicable disease, insufficient numbers of adults in the UK are active. Explanations for such low adult levels of physical activity relate to developments in work and home environments, e.g. greater use of cars, declines in walking, and growth in sedentary behaviours. The majority of correlate studies examining the relationships between perceived or objectively measured environmental variables and physical activities have taken place in the USA.

In order to investigate these observations in the UK we developed multidisciplinary research collaboration to undertake a series of studies as part of the Environment and Physical Activity Study (EnPAS) programme. In our first study we analysed data from a large cross sectional study of adult’s perceptions of their environment and walking. We found that women were more likely to walk if they perceived that it was safe in their locality and if there were nearby services, but men were not affected by these considerations. We wanted to test these observations between perceptions of the environment with objectively measured environmental variables and different physical activities.

OBJECTIVE
The objective of EnPAS was to identify which physical environmental variables were associated with specified types of physical activities for adults, after adjusting for possible confounding variables.

METHODS
We conducted a secondary analysis of data from the Norfolk cohort of the European Prospective Investigation of Cancer. We used Geographical Information Systems (GIS) and multivariate models to relate 13,572 participants’ reported levels of physical activity with a range of measures of the local environment. The sample was recruited from local family practices. We calculated the study would have at least 90% power to detect a relative difference of around 40% in the proportion of people who are physically active when comparing those in the top and bottom quartiles of the environmental parameters.

The data for recreational physical activity were collected by the EPIC Physical Activity Questionnaire 2 (EPAQ2), including occupational and travel related physical activity. EPAQ2 has satisfactory test-retest reliability and adequate levels of validity, compared with similar measures. EPIC data provided a range of possible confounding variables, including age, socio-economic status, education, health status and car access.

We created the objectively measured environmental variables using three sources of data (i) GIS derived variables, (ii) secondary data sets, and (iii) environmental audits. We examined:

• The likelihood of reporting any occasions of swimming for recreation in the past month by home proximity in kilometres to nearest public or private swimming pool
• The likelihood of reporting any occasions of cycling for recreation in the past month by quartiles of Road Traffic Volume Index Score (RTVIS)
• The likelihood of reporting any occasions of facility based physical activity in the past month by home proximity in kilometres to nearest public or private sports/exercise facility
• The likelihood of reporting any occasions of walking for recreation in the past month by crime rate per 1000 population per ward
• The likelihood of reporting any occasions of walking for recreation in the past month by home proximity in kilometres to nearest any green space or river walk

RESULTS
We found only one relationship that was statistically significant. Odds for cycling for leisure dropped as local traffic density increased for both genders. For example for women in the highest quartile for RTVIS the likelihood of reporting any cycling for leisure was - OR 0.42, (95% CI 0.32-0.52, p<0.001). Despite finding significant associations between the environmental variables and physical activities in bivariate analysis, access to green space and area levels of crime were not associated with walking for recreation. Distance to facilities had either no or only a small effect on the uptake of different activities.

CONCLUSIONS
EnPAS tested new research questions, pertaining to the relationship between GIS derived measures of crime and road traffic volume and relevant physical activities, as well as examining the importance of access and proximity to places for physical activity. The results of the study are robust because they are drawn on data from a large sample and use detailed behavioural and socio-demographic variables in the analysis. However what remains unknown is if these results are generalisable to other parts of the country and if other unknown environmental factors are associated with physical activity. Further research should use a large representative sample of adults from areas with different variability of environmental and social-demographic variables, and use both objective and subjective measures of physical activity and the environment.