Takeaway food outlet/obesity link: Spun data, faulty policy recommendations in BMJ

fast food outlets As I will review in an upcoming post at PLOS Mind the Brain, the link between the availability of fast food outlets and obesity is weak and probably explained by a variety of other factors, including poverty and restrictions on opportunities to purchase, store, cook, and consume food from supermarkets,  as well as competing demands and preferences

My Mind the Brainslize of  post will discuss some of the reactions that surprised me when I tweeted about this BMJ article.

My expression of skepticism about the ethics or effectiveness of restricting people’s access to fast food outlets had encountered mostly hostile responses. I even lost a few followers. I realized that because it was the early morning and I was in Europe, most of the responses were coming from the UK or continental Europe while Americans were still sleeping.

I later posted something on Facebook, including a link to the article.There the responses were a more sympathetic “take your hands off my slice of pizza”coming from Americans. Thre was much more of a sense that eating is a basic human need, access to food is a basic human right, and restrictions set on that right can become a matter of social justice, particularly when the restrictions occur without the involvement of those who are most affected. Of course, eating is very different than smoking or the misunderstood Second Amendment right “to bear arms.”

I underestimated the cultural differences between the UK and the US in willingness to impose top-down restrictions on people’s freedom of choice. We are, after all, as Winston Churchill observed, two nations separated by a common language.

[The following was uploaded as a Rapid Response at BMJ, awaiting approval March 28, 2014]

There are incentives for researchers to produce findings that support decisions to which policy makers are already committed. Top down decisions to control obesity by restricting access to fast food outlets are a prime example.

Burgoine and colleagues1 previously demonstrated how researchers can serve up illusory ‘obesogenic realities’ to order with arbitrary methodological and statistical decisions and selective reporting. Now Burgoine and other colleagues2 have presented a prime example of this, declaring in their abstract: “Government strategies to promote healthier diets through planning restrictions for takeaway food could be most effective if focused around the workplace.”

This unqualified recommendation is based on a cross-sectional observational survey study of a rural area of England. One needs to read the abstract carefully in conjunction with the actual results to recognize spin being applied to results that even then do not justify this bold recommendation. The abstract selectively reports findings concerning exposure to fast food outlets and consumption of fast food in the work environment. These effects amount to only 5.3 g per day when individuals in lower quartile of exposure are compared to those in the highest. In the results section, we learn that the association for the home environment is more modest and not dose-response. Further, the association for along commuting routes is nonsignificant.

These patterns are examined in multiple linear regression analyses vulnerable to addition or exclusion of arbitrarily selected and poorly measured control variables. For instance, in the results section we learn “In models that omitted supermarket exposure as a covariate, the associations between combined takeaway food outlet exposure, consumption of takeaway food, and body mass index were attenuated towards the null (web figs 5 and 6, upper right panels, respectively). “

The authors are to be faulted for failing to present basic bivariate relationships, reliance on complex multiple linear regression models of cross-sectional data that poorly capture environments or individuals, and an abstract that does not meet results in a way that is accurate or representative.

However, British Medical Journal shares blame for promoting faulty policy recommendations with weak evidence. The journal encourages authors of observational studies to identify clinical and public health implications without noting strength of evidence or the need for intervening randomized trials evaluating interventions between observational studies and policy recommendations. Readers need to ask: would BMJ have accepted this article if the authors had not made policy recommendations unwarranted by their data? If the article were nonetheless accepted, would similar media coverage have been generated by an appropriately modest acknowledgment that within the limits of their data, the authors did not find strong support for limiting fast food outlets as a way of controlling obesity?

Authors of observational studies commonly make recommendations for clinical practice and policy without calling for an RCT.3 BMJ should stop its complicity by reconsidering its policy of calling for clinical and public health implications from observational studies. If the journal policy must persist, then a policy should be implemented of requiring preregistration of design and analysis plans for observational studies,4 just as done with clinical trials. And frank admission of the limitations of cross-sectional observational data should be clearly acknowledged, starting in abstracts.

01. Burgoine T, Alvanides S, Lake AA. Creating ‘obesogenic realities’; do our methodological choices make a difference when measuring the food environment? Int J Health Geogr 2013;12:1-9.
02. Burgoine, T., Forouhi, N. G., Griffin, S. J., Wareham, N. J., & Monsivais, P. Associations between exposure to takeaway food outlets, takeaway food consumption, and body weight in Cambridgeshire, UK: population based, cross sectional study. BMJ: 2014, 348.
03. Prasad, V., Jorgenson, J., Ioannidis, J., & Cifu, A. (2013). Observational studies often make clinical practice recommendations: an empirical evaluation of authors’ attitudes. J Clin Epi, 2013; 66(4), 361-366.
04. Thomas, L., & Peterson, E. D. The value of statistical analysis plans in observational research: defining high-quality research from the start. JAMA, 2012; 308(8), 773-774.