Food policy research: We need better measurement, better study designs, and reasonable and measured actions based on the available evidence


  • Disclosure: The authors declared no conflict of interest.


Lucan and Chambers, in their Letter to the Editor, call for better measurement in order to move food-environment research forward. We wholeheartedly agree. In fact, this issue is not new to the obesity policy agenda (1). However, attaining validated and detailed food environment data in a large-scale (e.g., national) setting is prohibitively costly. Our study included 68,132 women living in 18,186 census tracts across the United States (2). We excluded women who lived in census tracts with a population count of less than 500 and women living outside metropolitan statistical areas because we believed measures of the food environment would not be comparable in urban and rural areas. As we pointed out in our article, most studies to date have analyzed a single type of food outlet (e.g., grocery stores or fast-food outlets) at one time. We examined multiple dimensions of the food environment in a national dataset—and believe that these data and analyses bring the state of the literature forward. We concur that detailed ground observations, such as the ones Lucan and Chambers reference, are ideal. However, these can only be executed in confined geographic settings. Such data would be extremely difficult, if not impossible, to attain on a national scale.

We also agree with Lucan and Chambers' concerns with (1) assumptions that establishments categorized as full-service supermarkets are all comparable; (2) potentially relevant food sources such as farmers' markets and mobile produce stands may not be captured through commercial database listings; and (3) using radial buffers and proximity to a store to capture access without capturing factors such as transportation mode, travel time, and social norms around food purchasing. Such detailed evidence can complement and validate studies based on large national data to ensure that policies are based on a solid scientific foundation.

Importantly, we are working on that. Members of our team are involved with the largest study to date in the United States that is capitalizing on a natural experiment of the elimination of a food desert (1R01CA149105, Does a New Supermarket Improve Dietary Behaviors of Low-income African Americans?). Examination of a natural experiment of this type (i.e., elimination of a food desert) is allowing our team to overcome many of these limitations, from reliance on unvalidated commercial databases (we are conducting food audits to collect price, quality, and availability of food data from all food purchasing venues in residential neighborhoods included in our study as well as the most frequently report venues our enrolled population reports shopping), to a longitudinal quasi-experimental study design with a control or comparison neighborhood, and extensive data on residents' dietary intake, travel mode, time spent in shopping, and experience of food purchasing. We have just completed our baseline data collection and hope that findings from this study when completed will be replicable to other low-income African American neighborhoods across the United States. However, we are indeed confined to one large natural experiment and unlike our published study, will not be analyzing data based on tens of thousands of individuals and census tracts.

We agree that the field faces measurement limitations. Large observational studies, such as the one our paper reported, have imperfect measures. However, given the considerable impact of nutrition on obesity and other health problems, we believe that reasonable and measured actions based on the available evidence need to be considered. Policy makers cannot afford to rely solely on data from detailed studies of a few neighborhoods (one could argue a requirement for grounding approaches) and need to know whether results hold at a national level. Our study does that, and the methods we used are necessary to such a study. Although we agree that the field also needs studies of small areas with rich and detailed measures to complement the national data and help us determine their validity, we still conclude that our findings support restricting the development of fast-food outlets and attracting grocery stores, and are committed to additional research that overcomes the limitations of large studies such as the one we published.

Tamara Dubowitz*, Madhumita (Bonnie) Ghosh-Dastidar†, Rebecca Collins‡, José Escarce§, * RAND Corporation, Pittsburgh, Pa, † RAND Corporation, Arlington, Va, ‡ RAND Corporation, Santa Monica, Ca, § David Geffen School of Medicine, Department of Medicine, UCLA, Los Angeles, Ca. and RAND Corporation, Santa Monica, Ca