Research Article
Reduced-Item Food Audits Based on the Nutrition Environment Measures Surveys

https://doi.org/10.1016/j.amepre.2015.04.036Get rights and content

Introduction

The community food environment may contribute to obesity by influencing food choice. Store and restaurant audits are increasingly common methods for assessing food environments, but are time consuming and costly. A valid, reliable brief measurement tool is needed. The purpose of this study was to develop and validate reduced-item food environment audit tools for stores and restaurants.

Methods

Nutrition Environment Measures Surveys for stores (NEMS-S) and restaurants (NEMS-R) were completed in 820 stores and 1,795 restaurants in West Virginia, San Diego, and Seattle. Data mining techniques (correlation-based feature selection and linear regression) were used to identify survey items highly correlated to total survey scores and produce reduced-item audit tools that were subsequently validated against full NEMS surveys. Regression coefficients were used as weights that were applied to reduced-item tool items to generate comparable scores to full NEMS surveys. Data were collected and analyzed in 2008–2013.

Results

The reduced-item tools included eight items for grocery, ten for convenience, seven for variety, and five for other stores; and 16 items for sit-down, 14 for fast casual, 19 for fast food, and 13 for specialty restaurants—10% of the full NEMS-S and 25% of the full NEMS-R. There were no significant differences in median scores for varying types of retail food outlets when compared to the full survey scores. Median in-store audit time was reduced 25%–50%.

Conclusions

Reduced-item audit tools can reduce the burden and complexity of large-scale or repeated assessments of the retail food environment without compromising measurement quality.

Introduction

The food environment has been identified as a possible contributor to obesity by influencing consumer behavior, food choice, and ultimately, weight status.1, 2, 3 However, studies to date have shown mixed results when examining the relationship between food environment measures and BMI.4, 5, 6, 7, 8, 9 Although these inconsistencies may reflect real variation in this relationship among different populations, they may also be due to different study designs and to the wide variety of methods used to assess the quality of the food environment. The gold standard assessment would be based on a full inventory of foods available in all retail food outlets in a given area. Because retail food inventory data are not readily available, and would be onerous to compile, other methods have been developed.

Measures of the community food environment3 are widely used. Density of fast food restaurants; distance to the nearest supermarket; direct measures or audits of food stores, restaurants, and other establishments that sell food; and resident perceptions or surveys about the food environment are common in the literature.10, 11 Food establishment density and proximity measures generated from business databases are inexpensive, but do not provide a comprehensive assessment of what consumers can purchase inside a store or restaurant.12, 13, 14 Indeed, categorization of different types of food establishments into healthful versus unhealthful has its challenges and often ignores the mix of healthful and unhealthy foods, such as those found in supermarkets.15, 16 Resident surveys of neighborhood food environments can be costly, depending on length, sample size, and resident perceptions, and reports may be biased by sampling procedures and response rates.17, 18 Direct measures from audits or observations of the consumer food environment within food establishments using trained personnel produce the most objective and detailed assessments but are time consuming and costly.19

A recent review identified 48 instruments that could be used to assess qualities of the food environment in various settings.19 Twenty-one of the instruments reviewed assessed the community or consumer food environments. The measured attributes included access, availability, price, media and advertising, policies, barriers, nutritional quality, nutrition information, use, choice, and affordability. Of the eight that included a store component, two were limited to specific types of communities (low-income, rural), two focused on children, and two audited a limited selection of foods. Testing of the psychometric properties (reliability, validity) had been completed on only one of the remaining tools, the Nutrition Environment Measures Survey for stores (NEMS-S).20 Similarly, of the six tools that were designed to assess restaurants, five were limited in scope, leaving the Nutrition Environment Measures Survey for restaurants (NEMS-R)21 as the only tool that could be used in any restaurant assessing a fairly broad range of food environment characteristics.19 Further, the psychometric properties of the NEMS-R have been tested, with high inter-rater and test–retest reliability and established construct validity.19 Also, to date, more than 50 empirical publications have reported on food environment data from NEMS, to answer a variety of research questions. For these reasons, among the currently available tools, the NEMS measures are considered the gold standard and are widely used.22

The NEMS-S and NEMS-R use direct observational auditing, with NEMS-S examining the availability and cost of healthy options versus less-healthy options over 11 types of foods as well as the availability, cost, and quality of fresh fruits and vegetables.20, 23 The possible score for stores ranges from –9 to 54; the restaurant score ranges from –27 to 63.21, 24 Higher scores relative to lower indicate greater availability and lower cost of healthy options, as well as support for healthy choices. The scoring system assigns negative values for barriers to healthy choices (restaurants) and if the cost of healthier options is greater than the comparable regular options. Positive values are assigned to availability and lower cost of healthier options, attributes that encourage healthy choices (restaurants), and higher quality of fruits and vegetables (stores). Development and evaluation of NEMS-S and NEMS-R as well as the measures, protocols, and scoring systems have been previously published (Appendix Surveys S1 and S2, available online).20, 21, 23, 24

Completion of NEMS assessments requires considerable field time, as well as data cleaning and management time. The length and complexity of these tools make them difficult to use when a large number of outlets require auditing or repeated administration is needed for surveillance. The objective of the research reported here was to develop and validate reduced-item food environment audit tools for stores and restaurants.

Section snippets

Data Collection

Data from the West Virginia (WV) Early Childhood Obesity Prevention Project and the Neighborhood Impact on Kids (NIK) Study25, 26 were used in this analysis. A survey of the food environment was completed in WV from May to November 2011. All retail food outlets in two counties were audited using the NEMS-S and NEMS-R. The survey area was extended beyond county limits using a 5-mile buffer to capture outlets that may be used by participants living near county borders. Outlets were enumerated

Results

The combined data set used to develop the reduced-item audit tool included 820 stores and 1,797 restaurants (Table 1). Median NEMS-S score ranges were 25–37 for grocery and 5–9 and 4–9 for convenience and variety stores, respectively. Median scores for other stores were 11 in WV and 12 in San Diego. Median grocery store component score ranges were 18–25 for availability, 1–4 for cost, and 1–6 for quality. Median component score ranges for convenience stores were 4–7 for availability, 0–2 for

Discussion

Machine learning processes were able to create reduced audit item tools that proved to be valid instruments for estimating respective total NEMS survey scores for stores and restaurants. Field testing of the reduced audit tools, with identified weighting, generated very similar scores to those produced by the full NEMS surveys completed in the same outlet. The WV, Seattle, and San Diego combined database provided a large training data set that included regions of varied geographic and

Acknowledgments

The research reported in this publication was supported by the U.S. Department of Agriculture (USDA) Agriculture and Food Research Initiative Grant number 2011-68001-30049 (Principal Investigator [PI], Partington), the USDA National Institute of Food and Agriculture Grant number 2007-55215-17924 (PI, Glanz), and the National Institute of Environmental Health Sciences of NIH under grant number R01ES014240 (PI, Saelens). The sponsors had no role in study design; collection, analysis, and

References (30)

  • K. Glanz et al.

    Environmental interventions to promote healthy eating: a review of models, programs, and evidence

    Health Educ Q

    (1988)
  • B. Swinburn et al.

    Dissecting obesogenic environments: the development and application of a framework for identifying and prioritizing environmental interventions for obesity

    Prev Med

    (1999)
  • K. Glanz et al.

    Healthy nutrition environments: concepts and measures

    Am J Health Promot

    (2005)
  • S.S. Casagrande et al.

    Healthy food availability and the association with BMI in Baltimore, Maryland

    Public Health Nutr

    (2011)
  • T. Dubowitz et al.

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

    Obesity

    (2013)
  • Cited by (24)

    • Baseline results from the Eat, Play, Live trial: A randomized controlled trial within a natural experiment examining the role of nutrition policy and capacity building in improving food environments in recreation and sport facilities

      2020, Food Policy
      Citation Excerpt :

      Non-packaged items were not audited due to limited availability of nutrition information. An adapted version of the Nutrition Environment Measures Survey–Restaurant reduced item audit (rNEMS-R; fast casual version) was used to assess the overall quality of the concession food environment (Partington et al., 2015). Field testing demonstrated no significant differences between the rNEMS-R and the full valid and reliable NEMS-R (Saelens et al., 2007).

    • Diving into the consumer nutrition environment: A Bayesian spatial factor analysis of neighborhood restaurant environment

      2018, Spatial and Spatio-temporal Epidemiology
      Citation Excerpt :

      Yet spatio-temporal analyses of the NRE require repeated assessment of in-restaurant features, which is costly and time-consuming. Alternative assessment tools, for example, the reduced-item audit tools (Partington et al., 2015) and mobile phone applications (Kanter et al., 2014) could be applied for rapid data collection in future research. Such spatio-temporal analyses could also be computationally challenging, for which fast but approximate inference methods for latent factor models, for example, the Integrated Nested Laplace Approximation approach (Blangiardo et al., 2013; Carroll et al., 2015; Rue and Martino, 2009), might be required.

    View all citing articles on Scopus
    View full text