Do secondary datasets accurately capture the number and types of food outlets near schools?
Public Health Nutrition Editorial Highlight: ‘Assessing the validity of commercial and municipal food environment data sets in Vancouver, Canada‘. Authors: Madeleine IG Daepp and Jennifer Black.
Researchers, public health practitioners and parents have a growing interest in the ways that children are influenced by exposures surrounding schools. Food stores, including fast food outlets, convenience stores, and supermarkets, impact where students spend time, what they eat, and how much food advertising they see. In Canada, the only G8 country without a federal school lunch program, children may be particularly susceptible to the food vendors they encounter near school. But how many and what type of food outlets do children encounter en route to school? Despite a flourishing literature on the consumer food environment surrounding schools , the question is surprisingly difficult to answer because datasets on food outlet locations vary widely in quality [2-3].
In our recent paper in Public Health Nutrition, we tested the accuracy of two commercial and two municipal food outlet datasets commonly used in studies about school and neighborhood access to food stores. We surveyed every street within 800 meters of 26 schools in Vancouver, BC, to identify all fast food restaurants, supermarkets, and convenience stores children might see on their way to and from school, or if they ventured 5-10 minutes down the road during lunch-time. We then compared this “gold-standard” list with data from two commercial data providers, and with municipal health inspections records and business license lists.
We found considerable error rates in all four datasets. At least 20% of the outlets found on the ground were missing from the secondary datasets, and over 25% of the outlets listed the secondary datasets were not observed on the ground. While the business licenses data outperformed commercial data sets, municipal datasets included inaccuracies despite legally mandated reporting requirements—likely because government agencies vary in their ability to maintain and update such listings . Nevertheless, the datasets generally remained representative of the overall food environment. For example, when we ranked schools according to food outlet density—the counts of food outlets within 800m buffers around schools—we found that the rankings were highly correlated between each secondary dataset and the gold standard dataset (Kendall’s Tau ≥ 0.87) .
These findings suggest that that while secondary datasets may not accurately identify every individual food outlet, they can still provide a generally representative picture of the overall food environment surrounding schools. Both commercial and municipal datasets can offer researchers a reasonable sense of the number and types of foods stores surrounding schools for making comparisons between neighbourhoods. However, researchers and policy makers may need to think twice before using secondary datasets to identify specific food outlets or to measure fine grained changes over time. Small-scale, detailed neighbourhood studies will likely still benefit from ground-truthing, but our findings suggest that secondary datasets can play a valuable role in food environment research, and methods proposed in this study can be used for comprehensive assessments of data quality in other jurisdictions.
The paper, ‘Assessing the validity of commercial and municipal food environment data sets in Vancouver, Canada‘ is published in the journal Public Health Nutrition and is freely available until 30 September 2017.
 Williams J, Scarborough P, Matthews A, et al. (2014) A systematic review of the influence of the retail food environment around schools on obesity-related outcomes. Obes Rev 15, 359–74.
 Fleischhacker SE, Evenson KR, Sharkey J, et al. (2013) Validity of secondary retail food outlet data: a systematic review. Am J Prev Med 45, 462–73.
 Lebel A, Daepp MIG, Block JP, et al. (2017) Quantifying the foodscape: A systematic review and meta-analysis of the validity of commercially available business data. PLoS ONE 12, e0174417.