The food environment is often defined as the physical, sociocultural, economic and political environment where one lives, studies and works, and plays an important role in food choices, food availability and, consequently, consumption [1-8].

Given its importance, several measures and instruments related to the food environment have been developed, with the aim of better representing and operationalizing its conceptual aspects [9]. However, despite the progress observed in recent years, the lack of standardization between measures makes it difficult to compare studies and analyze the consistency of associations between the dietary environment and different health outcomes [9-13].

The food environment can be assessed using objective measures, such as systematic observation of the environment or Geographic Information Systems, [2,14] or subjective measures based on people’s perception of the surrounding environment [15-19]. In some studies, in order to improve the complementarity and complexity of the methods, objective measures are used combined with subjective ones [9]. Objective measurements, while reflecting what actually exists, require that data be available, updated, and reflect the food environment. If they are not available, it requires investment in financial terms, human resources and time for data collection, storage, processing and availability [14].

On the other hand, subjective measures, for the most part, are low-cost, and easy to apply and analyze [20]. In addition, perception is an important component of the environment-behavior relationship [21]. How people perceive the environment in which they are inserted seems to be more important than the environment that actually exists [21]. In this sense, understanding this interaction will allow changes to promote improvement in food and food consumption, as well as in people’s health [22,23].

Despite the usefulness and relevance of subjective measures, the use of different instruments, with different characteristics and coming from diverse sociocultural and economic contexts, has considerably limited the comparability between studies [2,9]. Research carried out in Brazil with perceived measures of the food environment used instruments to assess characteristics, such as home, work, school, food store, etc. environments. In general, they are generalist instruments and have not indicated evidence of validity, restricted to a small number of questions that, for the most part, refer to different response scales, which makes comparability between studies impossible [16,18,20,22].

Thus, translating and culturally adapting an instrument for the Brazilian context will expand the availability of instruments for assessing the country’s food environment, stimulating research that uses measures of the perceived environment. This could improve the understanding of how the perception of the environment affects food consumption choices, in addition to allowing the comparison between the results of national and international studies. The aim of this study, therefore, was to cross-culturally adapt the Perceived Nutrition Environment Measures Survey – NEMS-P instrument to assess the food environment perceived in the Brazilian context.

The initial stage of the study aimed to identify instruments for assessing the food environment available in the literature. In view of the volume of research that used measures of food environment; initially, the search for systematic review articles was defined to contemplate original studies with measures of food environment, considering that these reviews should aim for the most relevant studies in terms of using measures of the food environment. To this end, a search was carried out in the Bireme, PubMed and Science Direct databases using the descriptors in English: food environment and review, combined with the Boolean operator “and”. The survey was performed from June to August 2020 and reviews that included studies evaluating the food environment were incorporated, either by qualitative or quantitative measures, and that had been published in the last 10 years. Ten eligible systematic reviews, published between 2011 and 2019, were identified [2,9-12,14,22,24-26].

Then, the original articles included in the eligible reviews were identified. In all, 404 original articles were identified, 344 of which remained on the list after excluding duplicate articles. After reading the titles and abstracts, a total of 311 articles were excluded because they did not include specific studies that used and had instruments for assessing the food environment through perceived measures. Altogether, 33 articles were selected for a full reading. After reading them in full, 24 studies were excluded for not having one or more inclusion criteria (evaluation of the food environment, and/or use of a questionnaire as an evaluation instrument).

Studies retained in the final analysis of findings included self-reported measures of access, availability, variety, and price of food in the neighborhood/local stores and in the household. Eight studies used general measures of the food environment in the form of broad, non-specific questions, and without information about the quality of the measure. The only instrument reported in the studies with information on the quality of the measure was the Perceived Nutrition Environment Measures Survey (NEMS-P). The instrument was previously evaluated in terms of feasibility, internal consistency and test-retest reproducibility [21], in addition to having been used in food consumption surveys in the United States and Spain [27-31]. The NEMS-P consists of 49 questions, mostly objective, divided into six sections: Home Food Environment, Questions about Food Purchase, Questions about Restaurants/Out-of-Home Meals, Your Habits and Thoughts about Food, General Questions About the House and General Questions [21] and has been used to assess the perception of the food environment in different countries such as the United States [29,32] and Mexico [33]. Therefore, NEMS-P was selected to carry out the translation and cultural adaptation for the Brazilian context.

After selecting the instrument, contact was made with the authors of the NEMS-P to obtain authorization for use and translation, which was obtained by email. The translation and cross-cultural adaptation process followed the procedures described in the national and international literature [34,35] and comprised six steps (Figure 1): (I) translation, (II) synthesis, (III) back-translation, (IV) experts committee review, (V) pre-testing the instrument and (VI) obtaining the final version. The translation stage of the original English version into Portuguese began in October 2020 and was carried out independently by two bilingual translators whose native language is Portuguese (T1 & T2). One translator had prior knowledge of the instrument’s objectives, while the other did not. After the translation from English into Portuguese, the translators, together with the researchers, discussed the discrepancies between the two versions (T1 & T2), standardized divergent terms and defined the final version in Portuguese in consent (T12). This (T12) was retranslated independently by bilingual translators; however, their mother tongue was English, following the same procedures as in the translation process. The two back-translated versions (BT1 & BT2) were synthesized into the final back-translated version (BT12) after a meeting with the responsible translators and researchers, in order to standardize divergent terms and define the most relevant translations by consent.

In stage IV, a committee of experts was organized with the purpose of evaluating the translation/back-translation process, and the semantic, linguistic, experimental and conceptual equivalence of the versions, as well as their clarity and objectivity. For the composition of the expert committee, we aimed to obtain a group of 8 to 10 members. In this way, 15 professionals with experience in studies on the food environment and assessment of the environment related to health were invited, all with fluency in English and Portuguese, as indicated in the literature [36]. In possession of the three versions of the instrument: the original version, the final version in Portuguese (T12) and the final back-translated version (BT12), and conceptual information provided by the authors of the instrument, as well as the NEMS-P article, [21] the expert committee, for each question, answered the following questions: (a) “compared to the original version in English, did the question translated into Portuguese maintain conceptual, cultural and semantic equivalence?”; (b) “Is the version translated into Portuguese objective and understandable?”; (c) “compared to the original version in English, is the question back-translated into English altered?”; (d) “What is the relevance of this question in the instrument?”. The answers were presented on a Likert scale adapted for concordance with the questions, according to four possible categories: “no”, “a little”, “quite” and “completely”. The experts also evaluated the relevance of each question, considering the response categories “not relevant”, “a little relevant”, “quite relevant” and “highly relevant”. At the end of each question, a space was made available for possible suggestions and/or questions. The evaluation was conducted remotely using the Qualtrics digital platform. The agreement of the evaluation carried out by the committee of experts was through the answers on Likert scale and a degree of agreement of 70% was determined.

After the changes suggested by the committee, with the intention of evaluating the clarity of the first version of the NEMS-P translated instrument, the pre-test stage V was conducted with a group of 20 people in March 2021. Participants were selected for convenience, and contacted through the researchers’ contact network, considering sufficient distribution for the variables: sex (female vs. male), age/age group (18 to 40, 40 to 59 and ≥60), educational level (elementary/middle and high school vs. undergraduate and graduate), nutritional status, by calculating the body mass index, according to reported weight and height (weight divided by height squared) (eutrophic: 18.5 to 24.9 kg/m², overweight: 25 to 29.9 kg/m² and obesity equal to or greater than 30 kg/m²) [37]. Participation occurred remotely and participants indicated in each item of the questionnaire their answer to the question “how clear is the question?” (Answer options being “not clear”, “somewhat clear”, “fairly clear” and “completely clear”). Eventual comprehension difficulties were indicated, as well as suggestions to improve the clarity of the questions. The suggestions were discussed among the researchers and incorporated or not into the final Portuguese version of the NEMS-P. All steps taken in the survey are illustrated in the flowchart in Figure 1.

All steps and procedures adopted were approved by the Research Ethics Committee of the Pontifícia Universidade Católica do Paraná (Pontifical Catholic University of Paraná (Approval nº 4,458,247/12/11/2020) and all participants signed the Informed Consent Form.

In the first stages, after the two translations were carried out, the translators and researchers met to check for discrepancies in terms and consolidate the words for the synthesis, such as “refrigerator” and “fridge”, “neighborhood where you live” and “neighborhood where you reside”, “if you walk” and “if you go on foot”. The synthesis document was structured with the consent of all.

In the subsequent step, back-translation, the same dynamics occurred for the consolidation of the two versions. In the expert committee stage, among 15 invited researchers, 10 responded to the invitation and agreed to participate. The researchers who accepted had training in the areas of physical education, nursing, physiotherapy, medicine and nutrition. Among the 29 questions evaluated by the expert committee, 22 (76%) were classified as “quite” or “completely” equivalent in conceptual, cultural and semantic terms, and another five questions (17%) were classified as “somewhat” equivalent. Regarding the assessment of objectivity and understanding, 18 questions (62%) were classified as “quite” or “completely” and nine (31%) as “somewhat” adequate. As for the evaluation of changes between the back-translation versions (BT12) and the original, 28 questions were classified as “no” and “a little”, 35% and 62%, respectively. Finally, the majority of the expert committee (66%) considered the questions “quite” or “totally” relevant to the assessment of the food environment in the Brazilian context (Table 1).

Table 1 indicates the result of steps I to IV. The first column displays the items from the original English version, the second the summary of the translations into Portuguese, and the third the summary of the back-translations in English. The last column is the pre-test version in Portuguese after the modifications made by the expert committee.

Some suggestions were made by the committee. In the first question, for example, utensils used by some Brazilian families were added, such as the electric single burner, and cabinet and/or shelf, used for food storage. In the second question, some highly consumed fruits in Brazil were added, such as papaya and orange. In some questions, adaptations were necessary, as in question 4, in which the word “neighborhood”, originally translated as “bairro”, was replaced by “vizinhança”, upon suggestions from the committee. In question 6, at the suggestion of the expert committee, the word “loja”, which is the first translation of “store”, was replaced by “locais”, which means “places”. The changes made to the instrument occurred through discussion between the responsible researchers, maintaining the equivalence of the original instrument, aiming at cross-cultural adaptation to Portuguese and to the reality of Brazil. All modifications suggested by the expert committee are shown in bold.

In stage V of the clarity assessment, most of the participants were women (55%) under 40 years of age (40%), having an undergraduate degree (40%) and without any chronic disease (60%) (Table 2).

The 20 participants rated the 29 questions for clarity and understanding. Only one question (question 4) was evaluated by 80% of the participants as “quite” or “totally” clear and understandable, the other questions having a value equal to or greater than 90%.

This work carried out the cross-cultural adaptation to Portuguese of the NEMS-P instrument, designed to assess the perceived food environment in Brazil. All steps followed methods and procedures recommended, accepted and used in the literature [38,39]. The results indicate that the Brazilian Portuguese version of the NEMS-P has conceptual, cultural and semantic equivalence, understandable for interviewees and comparable to the original English version.

In terms of conceptual, cultural and semantic equivalence, with the exception of the question referring to “time from home to the restaurant you most frequent” (Q-17), whose equivalence was indicated as “quite” or “total” by 70% of the members of the experts’ committee, all other questions obtained values of 90% or 100%. These values are close to those observed in the adaptation of the Caregiver’s Feeding Styles Questionnaire to Portuguese [38]. The evaluation by the committee obtained results with 90% to 100% referring to equivalence. It is likely that the term “sit-down restaurant”, when translated into Portuguese to ensure understanding in the Brazilian context, made the process more difficult. In part, it is possible that this is due to the greater variety of types of restaurants existing in Brazil, such as “a la carte”, “self-service”, “buffet/pay-by-weight”, “all-you-can-eat”, etc., which can lead to confusion about the most appropriate translation of the term. However, for all other questions, the Portuguese version complies with the existing suggestion in the literature, which indicates a minimum 80% positive assessment regarding conceptual, cultural and semantic equivalence [38,39].

With regard to changes in the instrument, in the study that carried out the translation and adaptation of the NEMS-P into Spanish, [40] some types of food were included and excluded and the types of food stores and restaurants were changed, for the instrument to maintain the local reality. Store options were added to the instrument, such as food markets and small specialized stores (fruit and vegetable stores, butchers, bakeries or fish markets). Restaurants include “tapa bars”, which are common in Spain. The adaptation process was conducted by two specialists and subsequently evaluated by a committee composed of four professionals. A pilot test was run with a group of 10 people. The instrument had a content validity index of 0.729 which, according to the authors, is acceptable as it is close to 1 [40]. The name of the instrument was also changed to NEMS-P-MED due to its application being carried out in the context of the Mediterranean.

More than 80% of the experts consulted considered the questions quite or completely objective, understandable and relevant after the translation and cross-cultural adaptation process. In this sense, considering the researchers’ point of view, these data indicate that the Portuguese version of the NEMS-P can be applied in the Brazilian context, ensuring adequate understanding, with questions of relevance to the area of the food environment. This result was confirmed by the participants in the question clarity assessment. Only the question referring to statements related to access, quality and diversity of healthy foods such as fruits, vegetables and dairy products (Q4) was indicated by less than 90% of the participants as a question endowed with enough or total clarity, a value suggested by some authors [41]. According to one of the reports presented, it is possible that some technical terms used in the instrument are due to the subjectivity of each participant; terms such as “easy”, “high quality” and “low fat” may diminish the understanding or clarity of the question. Thus, in general, the results demonstrate that the translated version is sufficiently clear and understandable to be applied in the Brazilian context. However, it is important to highlight that a larger and more diverse sample of respondents may be necessary to identify limitations in terms of clarity and understanding of the questions for some population subgroups, especially considering the continental dimensions of Brazil and its sociocultural variety.

The clarity assessment stage is very important, especially for instruments that can be completed by the respondent alone since the participant will respond to the questionnaire without the assistance of an interviewer. If, when conducting this step, the researcher observes that the participants are having difficulties and/or the evaluation of the items is of low understanding, the method in which several rounds of evaluation are carried out can be used until a consensus is reached that all the words and terms are understandable [42,43].

The study that adapted measures to assess children’s movement behaviors and parenting practices performed the same procedures described in the literature [34,35]. In the clarity assessment stage, applied to parents, some measures were changed and replaced, such as “events” with “activities”, and “sedentary” was replaced with examples of behavior such as “not active” or “sitting for a long time” [44]. This step allows the researchers to identify the interviewees’ difficulties in reading and understanding the questions, and make the changes without compromising the intention and meaning of the original questions, providing the instrument is comprehensible.

As important as the conceptual, cultural and semantic equivalence, in addition to the clarity and comprehensibility of the instrument, is its comparability with the original version. Thus, the expert committee confirmed that the translated and adapted version for Brazil had little change in relation to the original version of the NEMS-P. However, some issues should be analyzed more carefully. In particular, five questions regarding food availability, variety and quality (Q4); price of fruits and vegetables (Q12); distance (Q17) and consumption in restaurants and fast-foods (Q18) and perception of weight loss (Q25) had an evaluation of 70% regarding the changes made in the translation and adaptation process, and the question referring to important factors in the choice of where to buy food (Q10), 60%, indicating no or little change in relation to the original version. These results are close to those of the National College Health Assessment II translation and adaptation study, in which the expert committee evaluated the versions of the steps in the translation process, considering 80% of the questions as unchanged, and the rest as little changed [45]. Despite these values, such changes tend to be more grammatical and interfere little with the meaning and understanding of the question as a whole. Also, to ensure comparability, the final Portuguese version underwent some adjustments, following the suggestions of the expert committee to minimize any impossibility of comparison with the original version, without losing application in the Brazilian context.

It is important to emphasize that the NEMS-P was elaborated following the conceptual model proposed by Glanz et al. [1] and Green and Glanz [21] which addresses individual characteristics, including sociodemographic factors, health status, health behaviors and psychosocial factors. Since the food environment is composed of numerous variables, the instrument is not capable of contemplating the assessment of the food and nutritional environment in its entirety. On the other hand, the instrument allows the evaluation of the food environment through the individual’s perception of the community, consumer and domestic domains [21]. The study of these domains will contribute to a better understanding of the relationship between environment, community and consumption, allowing comparability between different cities, regions and even countries.

On the other hand, it is possible that additional questions will be created and incorporated into the version adapted for the Brazilian context. Different socioeconomic and cultural characteristics lead to very specific aspects in terms of food environment and, consequently, consumption. Thus, although this study presents a standardized evaluation option that is comparable with other studies that have used the NEMS-P it may not be sufficient to capture more or only relevant characteristics for the Brazilian context [29,46,47]. For example, in some Brazilian towns, there are markets, grocery stores and greengrocers with a subsidy for the low-income population or, often, for the entire population [48]. Among these, the administration of Curitiba has 35 units of the Armazém da Família (“Family Grocers”) where the low-income population has access to staple food items at prices on average 30% lower than those found in the region. In this sense, with the aim of better understanding the impact of these policies on eating habits, it is possible that specific questions will have to be incorporated – in particular, given that social media relations, cultural aspects, government policies, and infrastructure, between the country of origin of the instrument and Brazil, are different.

Several studies carried out in Brazil and Latin America have tested associations between food availability in the food environment, accessibility and food consumption in all age groups [18,49,50]. The research results corroborate with each other and with the findings in literature, suggesting that the greater the availability, the greater its consumption [23,51]. However, there is still little evidence regarding the perception of the food environment. Thus, the use of the Brazilian version of the NEMS-P will help and complement these studies with the aim of better understanding these associations.

Although the present study carried out the translation and cross-cultural adaptation of the NEMS-P, the reliability and validity of this version when applied in Brazil are still unknown. Therefore, we recommend that future studies test the reliability of this tool and its validity when compared to other more accurate methods, since it is the first tool to be translated and adapted to assess the perception of the food environment in the Brazilian context.

Some limitations must be considered when interpreting our results. Firstly, the evaluation stage by the expert committee was performed out individually, not allowing feedback on ideas that could emerge, such as what happens during focus groups and group interviews. In addition, during the evaluation by the committee, the relevance of the questions was examined, and not specifically the relevance of the Brazilian context. On the other hand, the remote assessment by experts provided the participation of experts with a good representation of different areas of knowledge and different positions of expertise. Also, since the assessment of clarity in the pre-test stage took place at the time of the SARS-CoV-2 pandemic, this stage could not be carried out in person. Therefore, we were unable to reach the number of evaluators suggested by the literature – from 30 to 40 [39] –, nor the diversity in sociodemographic characteristics that could help to better understand the clarity of the NEMS-P. Since we conducted this step digitally, people with lower education levels and older age, who generally have difficulties accessing and browsing websites and the like, ended up not composing the sample. Thus, there is a certain limitation regarding the number of volunteers who participated, as well as the diversity of their sociodemographic characteristics.

We present a version of the NEMS-P adapted for the Brazilian context, which has adequate conceptual, cultural and semantic equivalence, being objective and comparable to the original version. Future studies should confirm its clarity, reliability and validity.