Year: 2013
Type: letter
Abstract: Dubowitz et al. provide the latest data linking select aspects of local food environments to select obesity-related health measures ((1)). The authors' lengthy limitations section highlights some of the measurement challenges to conducting this type of research. If future research is to identify the most promising strategies to improve food environments, support healthy eating, and prevent obesity, then the field needs to move beyond some common measurement limitations of early studies. First, the use of unvalidated business-list data is precarious. Although such data were convenient and efficient for establishing foundational associations in early studies, they are less than ideal for the field moving forward. For example, based on detailed ground observations of all food sources from >150 Bronx streets, research in our group has shown the business list used by Dubowitz et al to have a positive predictive value below 50% overall, and a sensitivity for grocery stores of only about 25% (manuscript under preparation). Other researchers have reported similar findings (and the business list having a sensitivity of <10% for fast food) ((2)). More-accurate food-environment data, verified on-the-ground, are needed. Beyond issues of accuracy, the concern with reliance on business lists is two-fold: [1] it inappropriately groups all food sources of a certain type (e.g., as if every supermarket were the same as every other supermarket in terms of selection and quality of products, when empirically this is not the case ((3))), [2] it generally restricts focus to a limited range of food outlets (e.g., “fast-food restaurants,” “grocery stores and supermarkets” ((1))). Studies using business-lists necessarily neglect impermanent sources of food that may be relevant, like street vendors ((4), (5)) and farmers' markets ((6), (7)). They also routinely neglect–more due to investigator decision as opposed to data deficiency a range of other potentially relevant food sources like pharmacies, gas stations, hotels, salons, general merchandisers, clothing outlets, book stores, and other retail selling food and drink ((8)). More-complete availability measures, that include generally neglected food sources, are needed. A separate issue concerns definitions of food-source proximity. While studies like that of Dubowtiz et al commonly use radial “buffers” ((1)), such buffers ignore how people travel in their environments and their barriers to transit. For example, one-way streets and divided highways have different implications for motorists than for pedestrians, and research that fails to consider how people actually access food may miss relevant associations ((9)). An alternative approach to radial buffers, favored by geographers, is to measure accessibility along existing street networks for walking or driving ((10)). More-precise accessibility measures, that consider how people travel to get food, are needed. Dubowitz et al. conclude that their findings support restricting the development of fast-food outlets and attracting grocery stores ((1)). Given measurement limitations, including those above, we do not find strong support for this conclusion. Perhaps greater benefit would come from promoting famers' markets and produce carts … or from restricting the sale of unhealthy foods at salons and pharmacies (but maybe only for those along walking routes and not for those accessed by motor vehicles). To investigate such possibilities and understand different options for intervention, it is time for the field to move beyond the limitations of early research, making use of more accurate, complete, and precise measures.
Source: Obesity
DOI: 10.1002/oby.20015
Authors: Sean C. Lucan, Earle C. Chambers
Institution: Albert Einstein College of Medicine
Language: English
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W2080228704