Article Text

Download PDFPDF

The physical environment and physical activity: moving from ecological associations to intervention evidence
  1. Adrian Bauman
  1. Correspondence to:
 Professor A Bauman
 NSW Centre for Physical Activity and Health, Level 2, Medical Foundation Building K25, University of Sydney, 2006, NSW, Australia;

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Can we improve population levels of physical activity? Focusing research efforts on improving the evidence base from (whole community) interventions is a necessary first step.

Recent interest in the physical environment is reminiscent of the pre-individualist era of 19th century public health, when sweeping environmental changes, particularly around sanitation, hygiene, and food supply produced large scale population health effects.1 Current interest in the environment and its role in chronic disease prevention has positively influenced the tobacco control agenda, and more recently, been shown to be associated with population obesity, inappropriate nutrition, and physical inactivity rates.

One driver for environmental research is interest in the causal role of community level variables in health promotion; these include measures of social capital, urban connectedness, social isolation, health literacy, and poverty. This has pervaded recent public health investigation at the community and small area level.2,3

Another reason is scientific curiosity to better understand the determinants of physical activity and obesity. In the 1980s, researchers examined individual cognitions, beliefs, and motivations around diet and exercise. This comprised correlational studies that identified associations between behaviours and a range of theoretical variables derived from social learning theory, the theory of planned behaviour, and motivational readiness to change (self efficacy, behavioural intention, stage of change variables). This resulted in a plethora of cross sectional analytical papers that showed small associations with diet and physical activity, without really “striking gold” in terms of identifying the solve-all correlate(s) that could really improve public health interventions. For example, population physical activity levels in most developed countries were comparatively static or declined during the 1990s, with the exception of Finland and Canada.4 The new epidemiological evidence for health benefits of moderate intensity physical activity5 has stimulated further interest in using social ecology frameworks to describe walking and other “active living” behaviours and their correlates.6,7

A third reason is that this research provides opportunities for new data collection and analytical methods. These include attributes of the environment, using measures of urban residential density, street connectivity, location of shops, schools, and green spaces, and each dataset can be combined in layers to assess community level attributes. The analytical challenges are to use multilevel analytical techniques that permit the inclusion of data at an individual level (from surveys or other individual measurement) and include supra-individual data in the same statistical models.8

The paper by Li and colleagues9 in this issue covers all of these innovative attributes. It shows clear relations between built environmental attributes and walking in a sample of elderly adults. These associations remain strong despite methodological limitations in terms of self report walking measures used, and small numbers of responders in each geographical area. Further investigation of these associations will feed the technophile epidemiologist; physical activity and environment researchers are now preoccupied with improving geographical mapping (GIS measures), objective physical activity measurement (using pedometers or accelerometers) and in combination with global positioning satellite (GPS) devices, so that neighbourhood walking can be objectively quantified. However, I would speculate that future research with better measurement and analytical methods will simply reinforce these associations with walking and physical activity, and perhaps show better evidence of causality through longitudinal studies.

As a technical aside, on the road to estimating the design (clustering) effect in this study, Li reported that 28% of the variability in walking was attributable to between neighbourhood variation9—this is important, because compared with many intraindividual level variables, environmental level variables make a large contribution to explaining physical activity variation.

The paper also contributes to the debate around subjective (perceptions) compared with objective measures of environments—as shown in this study, some individual level perceptions are correlated with walking whether objectively verified or not. Both perceptions and objectively measured environments are separately associated with different types of physical activity.10

Where does this leave us? The scientific imperative demands longitudinal research, better and more sophisticated GIS measures, and better assessment of walking and related behaviours. However, the underpinning rationale and discussion for this kind of study is to “improve physical activity levels, and thereby public health”. This is more challenging, as the need for testing interventions can be delayed for decades while the methods and measures are refined; this occurred with the individual level psychological correlates of physical activity, with many more published exploratory studies than population interventions.

Despite burgeoning correlational and measurement research around environments and physical activity, few opportunistic studies have evaluated the effects of environmental interventions on population physical activity levels,11–13 and the tentative impact of these quasi-experimental designed interventions has been modest. It is timely to prioritise these natural experiments and the opportunistic evaluation of environmental improvements; does manipulation of streetscapes, provision of bike lanes, and interconnected suburban design really change population physical activity and walking rates? Even tentative examples here will provide further evidence on which to progress public health policy decisions.14

Finally, one should sound a note of caution with new areas of public health research. After the enthusiasm produced by “correlational significance”, the subsequent failure to find effective interventions may relegate physical environments to being another unfulfilled “great white hope”. It may be that much broader change is required, with substantive changes to social norms regarding sedentarism, and to increase cultural intolerance towards our toxic food and inactivity environments. With increased community understanding and support, one could imagine mobilising the political will and huge resources to really change the physical environments in importants ways, to more effectively influence physical activity levels in populations.

Can we improve population levels of physical activity? Focusing research efforts on improving the evidence base from (whole community) interventions is a necessary first step.



  • Funding: none.

  • Conflicts of interest: none declared.

Linked Articles