Large scale food retail interventions and diet

BMJ 2005; 330 doi: http://dx.doi.org/10.1136/bmj.330.7493.683 (Published 24 March 2005) Cite this as: BMJ 2005;330:683
  1. Steven Cummins, MRC fellow (s.c.j.cummins{at}qmul.ac.uk),
  2. Mark Petticrew, associate director,
  3. Leigh Sparks, professor,
  4. Anne Findlay, research fellow
  1. Department of Geography, Queen Mary, University of London, London E1 4NS
  2. MRC Social and Public Health Sciences Unit, University of Glasgow, Glasgow G12 8RZ
  3. Institute for Retail Studies, University of Stirling, Stirling FK9 4LA

    Improving retail provision alone may not have a substantial impact on diet

    Ensuring communities have good access to healthy affordable food is one of the government's joined up strategies to improve public health and reduce health inequalities.1 2 Policy solutions for deprived communities without good access—food deserts—have focused on improving provision of food retail as part of a wider suite of recommendations for population dietary change focused around awareness, affordability, and acceptability.3 However, the evidence for the widespread existence of food deserts and their impact on population health has been contested.4 5 This has meant that although retail based policy recommendations to reduce diet related health inequalities now exist,1 2 the evidence to inform how, when, and where to reduce these inequalities is only now emerging.

    Recently completed projects in Newcastle, Leeds, and Glasgow have started to provide us with this evidence.68 The Newcastle study concludes that food deserts exist only for a minority of people who do not or cannot shop outside their immediate locality and for whom the locality suffers from poor retail provision of foods that compose a healthy diet. Key predictors of healthy eating were found to be dietary knowledge, relative affluence, and healthy lifestyle—retail provision was not independently associated with diet.

    The Leeds and Glasgow studies were both prospective evaluations of the impact of large scale food retailing. Utilising an uncontrolled before-after design the Leeds study concluded that access to food improved notably after the intervention. The average distance travelled to the main food store fell to under 1 km, and the percentage of people walking to the main food store tripled to over 30%. Substantial increases in consumption of fruit and vegetables of between 0.25 and 0.5 portions per day were also reported, particularly for respondents who switched to the new provision. In contrast the Glasgow study, a controlled quasi-experimental study, found little evidence for an overall effect of the intervention for fruit and vegetable consumption in portions per day. For those consumers who switched their main food shopping to the new store an improvement in consumption of around 0.35 portions per day was seen though the evidence for this was very weak. A substantial positive improvement in one measure of psychological health (GHQ-12) and a weak positive effect on self reported health was seen in switchers.

    How should this evidence be interpreted? Firstly, the term food desert, although a striking metaphor, has unintentionally led to such polarisation of views by researchers, policy makers, and other interest groups so as to be of limited further use. The authors of the Newcastle study propose that the focus should be on food equity instead.6

    Secondly, ambiguity remains over whether large scale food retail interventions work. Despite the reporting of positive changes in fruit and vegetable consumption in the Leeds study, pre-intervention and post-intervention designs alone rarely provide compelling evidence that an intervention has been successful. Changes in the prevalence of risk factors and outcomes may be observed to change over time in the absence of any intervention.9 Observed changes therefore may not be due to the intervention itself but to an independent secular trend. Additionally, the effects of other ongoing local, regional, or national initiatives may confound the results of evaluations. Without a matched community control, attributing any independent effect of the intervention itself is difficult. Study designs with community comparisons must adequately control for potential confounding factors.

    Overall, retail interventions may have either a small but important effect or no effect on diet and health. Although these studies had similar aims and results, uncertainty over the efficacy of retail led interventions stems from problems of interpretation owing to differences in study design. However, the implications for the future development of dietary interventions are similar. Changes in fruit and vegetable consumption, although small, are consistent with other evidence. Two recent reviews of dietary interventions for cancer risk found an average increase of 0.6 portions of fruit and vegetables per day,10 11 and relatively small increases in fruit and vegetable consumption may have encouraging prospects for the prevention of disease.12 The potential negative impacts of large scale retail interventions need to be understood and accounted for—improved retail provision may also increase the availability of foods associated with poor diet. Activities such as advertising and price promotion that surround store opening may be important mediators of impact and effect. If new retail provision is to have an impact on diet and health, we need a multidimensional approach that also tackles food awareness, affordability, and acceptability in addition to retail change.

    Changing access through improving retail provision alone may not have a substantial impact on diet and health. Changing knowledge without ensuring access seems problematic intuitively. An approach that changes knowledge and access simultaneously may have a better chance of securing improvements in diet and health and a reduction in health inequalities.


    • Competing interests None declared


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    12. 12.