Validation of presence of supermarkets and fast-food outlets in Copenhagen: case study comparison of multiple sources of secondary data

Despite moderately high correspondence between secondary sources of address information and field observation, the findings illustrate that the use of combined sources is recommended.

Abstract

OBJECTIVE: We examined the quality of food outlet addresses provided by secondary sources and determined whether they could be physically located in the field. DESIGN: Addresses of food outlets in fourteen school districts in the northern part of Copenhagen were obtained from multiple business locators. We geocoded 202 addresses using a geographic information system and cross-referenced the sources against each other using a validation grid. Physical presence was determined via street survey. We applied gamma statistics and calculated positive predictive value, sensitivity and percentage agreement to assess the overall correspondence between our test of physical presence and each source of secondary information. SETTING: The study took place within city boundaries of Copenhagen, Denmark. SUBJECTS: Food outlets within fourteen school districts within Copenhagen. RESULTS: Positive predictive value between field results and secondary sources indicated good to excellent correspondence (range: 0·81-0·98), comparable with other studies. Gamma coefficients indicated low to high positive correspondence (range: 0·23-0·98). CONCLUSIONS: Despite moderately high correspondence between secondary sources of address information and field observation, the findings illustrate that the use of combined sources is recommended.