Data Collection Instruments for Obesogenic Environments in Adults: A Scoping Review

The rise in obesity prevalence has increased research interest in the obesogenic environment and its influence on excess weight. The aim of the present study was to review and map data collection instruments for obesogenic environments in adults in order to provide an overview of the existing evidence and enable comparisons. Through the scoping review method, different databases and webpages were searched between January 1997 and May 2018. Instruments were included if they targeted adults. The documents were categorised as food environment or built environment. In terms of results, 92 instruments were found: 46 instruments measuring the food environment, 42 measuring the built environment, and 4 that characterised both environments. Numerous diverse instruments have been developed to characterise the obesogenic environment, and some of them have been developed based on existing ones; however, most of them have not been validated and there is very little similarity between them, hindering comparison of the results obtained. In addition, most of them were developed and used in the United States and were written in English. In conclusion, there is a need for a robust instrument, improving or combining existing ones, for use within and across countries, and more sophisticated study designs where the environment is contemplated in an interdisciplinary approach.


Introduction
The rising prevalence of obesity and overweight has generated a growing research interest in determining their causes. One of the key factors that has been identified as contributing to the development of obesity and overweight is the obesogenic environment. This has been defined as "the sum of influences exerted by environments, opportunities and life circumstances that promote obesity in individuals or society" [1]. In other words, the obesogenic environment refers to environmental factors that determine consumption and energy expenditure in individuals and influence the development of obesity. The obesogenic environment is a complex concept, and several models have been developed to conceptualise it and explain how it influences the behaviour of individuals [2,3].
When characterising the obesogenic environment through models, differences have been established between the food environment and the built environment [4]. The food environment is defined as the opportunity to obtain food, which includes the availability, accessibility, advertising, and marketing of food [4]. Food can be accessed in various ways from the food environment: in shops (grocery shops, supermarkets, markets), in catering establishments (bars, restaurants, canteens, takeaway outlets), and in the institutions where people spend part of their day (worksites, schools, homes) [5]. On the other hand, the built environment consists of three elements: physical design, land use (residential, commercial, industrial, and other activities), and the transport system. These make opportunities available for physical activity and for healthy and unhealthy food access. Neighbourhoods providing a range of local facilities within an easy active travel (walking and cycling) distance, with good quality infrastructure (such as well-maintained pavements), and which are regarded as safe and pleasant may support physical activity and influence the propensity of an individual to have an active lifestyle. This environment has generally been studied and characterised by the use of questionnaires and geographic information systems (GIS) [6,7].
Although the concept of obesogenic environment has gained widespread recognition over the last decade [8], studies have used different instruments to identify its components. The developed instruments to measure the obesogenic environment assess either characteristics of it related to the home, worksite, schools, shops, supermarkets, and restaurants, or the possibilities for walking or cycling in a given neighbourhood or city [4,8].
It is often stated that the environment exerts an influence on obesity, but further research is required to identify its specific components and how these influence behaviour in order to be able to modify them. Obesity is the result of multiple and complex factors: to identify all of its causes still remains a research goal. However, studies aimed at characterising the environment have utilized various approaches, methods, metrics, and variables; as a result, it is difficult to compare the evidence and scientific characterisation of the environment, as it continues to be unclear and complex [8].
Although the environment impacts all that comes in contact with it, the way in which it influences according to age group is not the same [8]. In addition, there are specific data collection instruments that are classified for each population age group and the characteristics of the studies performed in children are different from those carried out in adults.
The aim of the present study was to review and map data collection instruments for obesogenic environments in adults in order to provide an overview of the existing evidence and enable comparison.

Materials and Methods
A scoping review was performed since the field is heterogeneous and perhaps not suitable for a more precise systematic review because it is necessary to retrieve information from a variety of documentary sources, including research projects, governmental organisations, and scientific articles indexed in databases. This method is used to facilitate a more exhaustive review of all the literature available on the subject and is useful for answering much broader questions [9]. Moreover, the results of the scoping review will give recommendations for measurements and methods for future research in this field [10].
The scoping review method employed was that proposed by Arksey and O'Malley [11], the Joanna Briggs Institute [12], and PRISMA Extension for Scoping Reviews (PRISMA-ScR) [9], formulating the initial research question: What instruments are available for collecting data on obesogenic environments in adults?

Selection Criteria
Instruments were included if they targeted adults aged 18 to 65 or a mixed-age population sample (adolescents and older adults).
Instruments targeting pregnant women, people with a particular pathology or people living in institutions were excluded, as were population surveys, dietary assessment questionnaires, instruments that were not specifically designed to measure the obesogenic environment, and geographic mapping systems. Instruments unrelated to the subject of the initial research question were also excluded. In the event that the same instrument had been used in more than one study, only the study that described the instrument in most detail was included.

Search Strategy
Different databases were searched, including PubMed, Scopus, PsycInfo, Cochrane, and Web of Science, using the following descriptors: "Surveys and Questionnaires", "Environment", "Obesity", and "Adult", varying the search strategy according to the database interrogated (Table 1). Controlled language was used to perform the search, using MeSH (Medical Subject Headings) terms (in the case of PubMed), thesauri, and keywords, depending on each database. A search was also performed using free language in the title and abstract fields for all databases, using the term "obesogenic environment".
In addition, articles cited in previously identified studies that met the inclusion criteria were examined and several webpages was consulted, such as Active Living Research measures, the National Collaborative on Childhood Obesity Research (NCCOR), the National Institutes of Health (NIH), and webpages for existing projects and universities with projects related to the subject and government reports. Documents published worldwide between 1 January 1997 and 31 May 2018 were selected, due to the fact that the concept of "obesogenic environment" appeared for the first time in 1997 [2].

Study Selection
All the identified documents were downloaded in EndNote (EndNote X7 citation management software, Thomson Reuters, Philadelphia, PA, USA). This program was used to remove duplicates and, independently, documents were excluded by title and abstract based on the initial research question as well as inclusion and exclusion criteria. Then, the selected full-text documents were reviewed and those that met the criteria were included.
First, duplicates were eliminated and the documents were assessed by title. The remaining documents were then selected by abstract and full text by two independent researchers. Differences were discussed and resolved by a third reviewer.

Organisation of the Information
Since the obesogenic environment can be measured and classified in different fields, the information was classified according to the data collection instruments and is presented in two tables: the first shows instruments used to analyse the food environment (Table 2), and the second shows those used to analyse the built environment ( Table 3). The papers/instruments in Tables 2 and 3 were ordered by date.
The following information was extracted from each of the documents and entered in a database: author(s)' names, year of study, name of the instrument, city/country where used, type of instrument, the population/sample targeted, the kind of environment analysed, psychometric properties as instrument validity and instrument reliability according to the author's criterion, number of items in the instrument, number of different versions, language, and if the instrument had a cultural adaptation and the country where it was culturally adapted and validated. All this information was collected in Tables 2 and 3.
The built environment (Table 3) was classified according to different aspects of the environment, such as: (1) physical activity environment (places where people are, or can be, physically active); (2) walkability and bikeability in the neighbourhood; (3) worksite physical attributes; (4) neighbourhood design; (5) street-scale features; (6) trail use; and (7) perceived built environment. The places we live, work and so on can either provide or constrain opportunities for physical activity and for healthy and unhealthy food access [8].
In the event that an instrument could be used to measure both food and built environments, it was included in both tables (Tables 2 and 3).
In both tables, types of instruments were classified as checklist or checkbox (a pre-defined list or box of indicator foods which are selected based on predetermined criteria, such as those foods that are identified by the researchers as aligning with current dietary guidance), interview/questionnaire (a pre-determined list of questions that is administered by a trained interviewer or completed by the respondent via self-report), inventory (a form for recording all foods available in a given environment), and market basket (a pre-defined list of foods that represent a range of food choices across a total diet. These foods may be based on foods frequently consumed by the population or may reflect a standardized diet plan), and audit tool (allows systematic observation of the environment, including the presence and qualities of its features) [13,14]. Figure 1 shows a flow diagram of the search strategy. As can be seen, 1500 documents were identified, to which 470 obtained from other sources were added, yielding a total of 1970 documents. After deleting duplicates, we obtained 1474 documents for consideration, which were reduced to 198 following an analysis by title and abstract. The full text of these documents was examined, and finally 91 documents were selected due to their characteristics, in which 92 instruments were described (one of the documents reported two instruments).

General Characteristics of the Instruments
A total of 46 instruments were found for characterising the food environment and 42 for analysing the built environment, and 4 instruments that characterised both environments. Of the identified instruments (n = 92), 79.4% were developed in the United States, 8.7% in European countries, and 7.6% in Australia, whilst 4.3% were from multiple countries.
The majority of the studies reviewed have reported psychometric properties (n = 64), but out of all the studies (n = 92), only 38.0% were reported to be valid and reliable by the authors; 28.3% met some reliability criteria, 3.3% met some criteria of validity, and no mention was made of any criteria in the case of the remaining 30.4%.

General Characteristics of the Instruments
A total of 46 instruments were found for characterising the food environment and 42 for analysing the built environment, and 4 instruments that characterised both environments. Of the  In addition, the instruments used to characterise the built environment were observed to report more psychometric properties than those targeting the food environment, with 88.6% (n = 39) using different types of validity and/or reliability, in contrast to the 54.2% (n = 26) of the food environment instruments.
With regard to language, almost all instruments were written in English (97.6%), and out of this percentage, only 8.3% were also written in another language (French or Spanish); instruments written in languages other than English (Swedish and Arabic) accounted for only 2.4% of instruments. The most widespread data collection method used in the instruments was the checklist (see Tables 2  and 3).
The period 2006 to 2012 witnessed the development of the highest number of instruments. In 2012, two questionnaires were described that characterised the influence of vending machines on the obesogenic environment in public places; one was developed in Australia and the other one in the United States. The instruments contained a median of 33.5 items, with a minimum of 6 and a maximum of 267 items.

Built Environment
A total of 42 instruments were identified that analysed the built environment ( Table 3). Most of them characterised the built environment by studying its physical activity environment (50.0%), followed by neighbourhood design (20.4%), or how these were related to being able to walk or cycle (9.1%). It should be noted that these instruments could collect information on both the population and the built environment (street segments, parks, etc). Only 11 of these 42 instruments considered subjects' perception of the environment [66,76,78,84,85,91,98,106,108,109,113].
Most of the instruments that collected data on the built environment were developed between 2005 and 2006. However, the most widely used instrument, which has the most adapted and validated versions in different languages and cultures, was developed in 2003 [72]. The median number of items in these instruments was 40, with a minimum of 5 and a maximum of 273 items, showing the high heterogeneity of the number of items of the instruments. However, some instruments have also had short versions developed [65,66,72,114].

Instruments That Characterised Both Environments
Only four instruments were identified that considered both the built and food environments to measure the obesogenic environment. The Worksite Environment Measure (WEM) [26] and the Environment Assessment Tool (EAT) [27] focus only on the worksite environment. The EURO-PREVOB questionnaire [58], and the SPOTLIGHT virtual audit tool (S-VAT) [59] include some types of built and food environments. The S-VAT is particularly focused on collecting information about the food store environment, and the built environment was categorized into walking and cycling, public transport, aesthetics, land use-mix, and physical activity facilities.             Tables 2 and 3. NM = Not Mentioned

Discussion
The present review has identified the available evidence on the instruments used to characterise the obesogenic environment, in terms of both the food and built environments. This is the first scoping review on this subject, and the first review of instruments that considers both environments in adults.
Diverse instruments have been developed to characterise the obesogenic environment. Most of them had been developed in the United States and were written in English. Moreover, the majority of the studies reviewed have reported psychometric properties, but out of the all studies only a few were reported to be valid and reliable.
Some studies conducted in this field until 2015 only consider the influence of the food environment [14,116], or only that of the food store environment [117]. The present review includes, as a novelty, the measures developed in adults both at the food environment and the built environment levels, as well as their types. These can be observed in the tables, which show relevant characteristics to facilitate the selection of one or several instruments for carrying out future research. Additionally, this is the first article where the characteristics of the data collection instruments are classified and described in tables in a clear and fast way for consultation.
The importance of determining the factors that constitute the obesogenic environment has led to the development of a large number of data collection instruments. Most of the identified instruments measured the characteristics of the built or food environments, focusing on one particular area of each environment. These include for example physical activity or the characteristics of an area in the case of the built environment, or food shops, restaurants, the workplace, or home in the case of the food environment. Few instruments considered subjects' perception of their environment. However, recent research has shown that perception is a mediator between objectively measured exposure and interaction; consequently, studies that combine both are preferable [8].
Although most of the instruments focus only on one type of environment (built or food), there are four that contemplate both. However, they do not meet all the necessary criteria and different types of environments should be contemplated. First of all, WEM [26] and EAT [27] focus on the worksite environment (measuring food and built environments), but only in the workplace. In addition, in the case of EAT, it covers the physical activity environment, and although it includes the food environment, it does so in a lighter way. On the other hand, the questionnaire EURO-PREVOB [58] could be a good instrument to measure obesogenic environments in Europe. Nevertheless, more work is needed to refine and further test the reliability and validity of this instrument in a range of other environments. Although both types of environment are included, they do not include all the types that exist and that need to improve their psychometric properties. Finally, with the S-VAT [59], it does not contemplate all the types of environments that characterize the food and the built environment. Within the food environment it is particularly focused on collecting information about food store environment, and the built environment was categorized into walking and cycling, public transport, aesthetics, land use-mix, and physical activity facilities. For this reason, it cannot be recommended as a robust and reliable tool to assess both environments.
Both the food and built environment data collection instruments showed a wide disparity in the minimum and maximum number of items, which makes comparison difficult. A high number of items reflected inventories or checklists of foods available in shops or restaurants in the case of the food environment, or a list of the characteristics of a neighbourhood or defined area in the case of the built environment.
Although there has been a significant increase over the last decade in the amount of evidence indicating that the environment exerts an influence, there are still unquestionable gaps in current evidence, as shown in this review. Different measures, definitions, and approaches, and continuing attempts to be novel by creating new instruments to measure the environment have merely generated confusion [8].
However, the results showed that the most widely used instruments to characterise the food environment to date are those developed by Glanz et al. (Nutrition Environment Measures Study in Stores (NEMS-S) [22] and Nutrition Environment Measures Study in Restaurants (NEMS-R) [23]) and Oldenburg et al. (Checklist of Health Promotion Environments at Worksites (CHEW) [15]). With regard to the built environment, the most widely used instruments to identify this environment are the Systematic Pedestrian and Cycling Environmental Scan (SPACES) [64], the Neighborhood Environment Walkability Survey (NEWS) [66], the Analitic/Checklist Audit Tool [71], the Active Neighbourhood Checklist [90], and the International Physical Activity Questionnaire (IPAQ) [74], which are often combined with mapping using geographic information systems (GIS), despite limitations since this assumes that food choices are determined primarily by individuals' proximity to food outlets, without accounting for travel patterns, taste preferences, social norms about where to procure food, or ability to afford foods [13]. They have also served as the basis for various subsequent instruments (Table 4), as well as for studies on the food and built environment.
Few new instruments have been developed for an adult population during the last three years, and this is maybe due to the fact that the majority of studies that were found until May 2018 used the available instruments developed previously, cited in Tables 2 and 3 [118][119][120][121][122][123][124].
Most of the instruments have been developed and used in the United States. This may be due to concern about the high prevalence of overweight and obesity in this country [125]. Furthermore, there was important role played by the NIH, which has the capacity to finance prevention and intervention initiatives in the development of overweight and obesity, as well as in the development of instruments to characterise the environment [126].
Nevertheless, the prevalence of overweight and obesity in Europe is steadily rising and requires more uniform and comprehensive characterisation of the problem and its environments. Although Europe, Australia, and the United States present different food patterns and consumption characteristics, the prevalence of overweight has remained stable or is rising in all three regions [13, [125][126][127][128][129]. Influenced by the obesogenic environment, food habits and patterns of consumption are changing in many developed and developing countries, with a marked move towards low vegetable consumption and high animal protein intake [130]. Hence, obesity rates will continue to rise as long as individuals, society, policy-makers, health professionals, social workers, schools, and prevention campaigns continue to give little importance or priority to the obesogenic capacity of environments [131]. Although public health policies have begun to include measures such as banning marketing of unhealthy foods and taxing unhealthy options [132], there is still a need for vigorous action to prevent and reduce obesity by modifying these environments. It is therefore essential for policy-makers to implement effective interventions that tackle the elements involved in the development of obesity, such as certain sectors of the food industry and food marketing and advertising [133].

Limitations
There is no MESH term for the obesogenic environment. This fact hindered the search for evidence and generated more non-meaningful data or information. Database searches would be easier and more accurate if a new term were created referring to this concept and its types ("food environment" and "built environment").

Conclusions
The present study has provided an overview of the instruments used worldwide to measure the obesogenic environment in adults, identifying the components and characteristics of each tool. Numerous diverse instruments have been developed to characterise the obesogenic environment, and some of them have been developed based on existing ones; however, most have not been validated and there is very little similarity between them, hindering comparison of the results obtained.
Future research should combine validated instruments that characterise the built and food environments and also include subjects' perception of their environment. In addition, validated tools are required in other countries besides the United States, since those that exist are scarce. In conclusion, there is also a need for robust instruments, improving or combining existing ones, for use within and across countries, and more sophisticated study designs where the environment is contemplated in an interdisciplinary approach.
Author Contributions: All authors elaborate the background, discussion and conclusion. A.M.G. reviewed, collected, classified and interpreted the data instrument. All authors read and approved the final manuscript.
Funding: This research received no external funding.