Anthropometric and Dietary Indicators Applied in Population-based Surveys: a Systematic Review.

Background: Anthropometric and food consumption indicators are important in dietary, nutritional, and health status conditions assessment and monitoring of the population, as a mechanism to identify changes or trends in consumption and to understand the relationship between dietary exposure and varied health outcomes. The aim was to identify population-based health and nutrition surveys, conducted with adults and the elderly, and performed in the Americas, Europe, and Oceania, in order to investigate the more common anthropometric and food consumption methods used, their applicability, and their limitations. Methods: Electronic databases (LILACS, PubMed, and SCOPUS) were systematically searched for studies published between 1997 and 2017 in Portuguese, English or Spanish. Forty-ve studies (49% carried out in the Americas) met the eligibility criteria and were included in the review. The data were analysed in 2018. Results: The methodological quality of most of the studies (64.4%) was classied as moderate, according to the Agency for Healthcare Research and Quality checklist for cross-sectional studies and the Newcastle-Ottawa scale for cohort studies. Forty percent of the articles evaluated only food consumption, 31% just anthropometry, while 29% evaluated food consumption and anthropometric measurements. The most used food survey methods were food record (31% of studies) and the 24-hour dietary recall (22% of studies). Body mass index (BMI) was the most used indicator for anthropometric nutritional status assessment. Although most of the studies used the World Health Organization classication criteria, these studies did not adopt the different cut-off points for BMI classication for adults and the elderly. Conclusion: BMI and methods that record current consumption, such as the food record and the 24-hour dietary recall, were the main methods of assessing nutritional status, taking into consideration the easy application, low cost, and good reproducibility. terms was used: (Inquérito OR pesquisa) AND (saúde OR nutrição OR dieta OR ‘consumo alimentar’ OR ‘estado nutricional’ OR antropometria OR “composição corporal”) AND (nacional OR populacional). The searches were performed from April to July 2017 and conned to articles published between January 1997 and July 2017, in Portuguese, English, or Spanish. Data were analysed in 2018.

The reliability in the application of the method, the standardization of measures, and the possibility of international comparisons in uence the reliability of the data and the reproducibility of the studies. Thus, this study aimed to perform a systematic review of population-based health and nutrition surveys, conducted with adults and the elderly in the Americas, Europe, and Oceania, to identify which anthropometric and food consumption methods were most frequently used, along with their applicability and limitations. These continents were selected because the dietary habits of Western populations are similar, in that there is a growing inclusion in the diet of foods high in calories, fat, and re ned sugars, while they have low consumption of fruits and vegetables and ber, in contrast to the diets in Eastern (Arab and Asian) and African countries.

Design
This was a systematic review of original studies, prepared according to the recommendations of Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2009 [14], designed with the objective of guiding the dissemination of systematic reviews and meta-analyses in health eld. The review was registered in the international prospective register of systematic reviews (PROSPERO) under the number 2017: CRD42017071392. LILACS the following combination of terms was used: (Inquérito OR pesquisa) AND (saúde OR nutrição OR dieta OR 'consumo alimentar' OR 'estado nutricional' OR antropometria OR "composição corporal") AND (nacional OR populacional). The searches were performed from April to July 2017 and con ned to articles published between January 1997 and July 2017, in Portuguese, English, or Spanish. Data were analysed in 2018.

Inclusion and exclusion criteria
The papers were considered eligible when they met the following inclusion criteria: 1) Observational studies; 2) Performed in adult and/or elderly populations; (3) Surveys carried out over the last 20 years (1997 to 2017); 4) Conducted in Europe, Americas, and/or Oceania; and 5) Available in Portuguese, English, or Spanish. Systematic reviews, meta-analyses and studies involving pregnant women were excluded from the study.

Synthesis and comparison of results
Initially, two evaluators independently reviewed the titles and abstracts to verify if those met he proposed eligibility criteria. Then, the complete texts of the papers were read. Subsequently, each evaluator completed, independently, a data mining spreadsheet, including, in addition to the evaluation of the methodological quality, the following items: 1) Characteristics of the studies: authors, year of publication, place of study (country and continent), and the evaluation of the methodological quality score. The population of each study was described according to the number and age of participants. 2) Characteristics of surveys: name, year, variables, method, frequency of application of the nutritional survey, and evaluation criteria of the anthropometric nutritional status and the respective cut-off points. Discrepancies in the evaluation were resolved through discussion between the evaluators, and, in case of doubt, a third reviewer was consulted.
Evaluation of the methodological quality of the studies The methodological quality of the selected papers was evaluated and scored according to the recommendations of the Newcastle-Ottawa quality assessment scale (NOS) for cohort studies, and the Agency for Healthcare Research and Quality (AHRQ) checklist for cross-sectional studies. The NOS evaluation consists of eight questions, which include items such as: selection of participants, comparability between the subjects and veri cation of exposure. The papers are scored as 'Good', 'Adequate', or 'Poor', in accordance with the score received on each item of the scale (items are identi ed with one or no stars). The sum of these items (stars) classi es the article. The AHRQ consists of 11 items, with the options 'Yes', 'No', or 'Unclear'. A score '0' is attributed to items evaluated with 'No' or 'Unclear', and score '1' for those evaluated with 'Yes' [15].
To better present the results, the score evaluated by NOS was converted into quality categories, based on the document from the AHRQ [16]. Based on these results, three categories for evaluation were established: 0-3, 4-7, and 8-11, indicating low, moderate, and high quality, respectively.
Criteria for evaluation of methodological quality of primary studies The following indicators of anthropometric evaluation and assessment of habitual consumption were considered as standard in the assessment of the methodological quality of the primary studies: 1) BMI: two or more measurements; 2) food surveys: at least two applications (for 24-hour dietary recall and food record), as recommended in the literature [17]. To check the quality of the anthropometric data collected, we investigated if the papers mentioned calibration of anthropometric instruments and training of the interviewers.

Results
The search recovered 615 documents. In addition, another 22 articles were added by manual search. 52 documents of the 637 were duplicated and for this reason were excluded, resulting in 585 abstracts. After reading the title and abstract, 496 articles were excluded. Thereafter, after considering the inclusion and exclusion criteria of 89 articles read in full, 45 studies were included in the review (Fig. 1).
The evaluation of habitual consumption, dietary pattern, ingestion of foods or food groups, energy, and macro and/or micronutrients, were the purpose of the application of the R24h in nine studies [18,19,34,52,55,58,60,62,65]. The R24h was used to evaluate consumption of alcohol and breakfast (being a consumer or not) in two other papers [31,53].
With regard to the quality of the measures for composition of BMI, 11 papers reported that interviewers/researchers were trained to perform the measurements [20,25,53,27,28,30,31,33,35,43,45], four studies reported the calibration of equipment [25,27,30,37], and only two articles [25,26] reported performing two repetitions of each measurement of weight and height, as standardized by the WHO for collection of these data. Only the study of Meller et al. [25] reported all the information about the survey (training of the interviewers, calibration of equipment, and replication of anthropometric measures). Table 1. Characteristics of the studies and population-based surveys on health and nutrition, performed in the Americas, Europe, and Oceania, from 1997 to 2017. Food consumption surveys are used to collect information about the preparation and consumption of food, through observations by skilled personnel [10]. In the present analysis, the food record was the most frequently used nutritional survey to assess food consumption in large population surveys, followed by the R24H. Some countries have used this type of dietary inquiry, including Brazil [21][22][23], as well as developed countries such as Spain [40,41,61], England [42,44,49,69], France [46,47], the United Kingdom [48], and Switzerland [56].
Although allowing a detailed assessment of food consumption, the food record dispenses with the interviewer and does not feature respondent bias. However, it does present some limitations such as the requirement for literacy and high motivation [10]. It is possible that preference for its use in European countries is due to the easier application since an interviewer is not necessary. In this sense, it reduces the costs of the research. It should be emphasized that use of this method has greater reproducibility in populations with a high level of education.
Wide variations in the type of equipment used for data collection were observed in these studies, such as the use of tablets, digital cameras, photographic recordings, telephone interviews, and printed forms, which require a certain degree of skill for their use. It is also interesting that most papers that used the food record applied it for four or more days [21, 42-44, 46-49, 56]. Accordingly, the choice of this method over a long period of time requires some skills from the interwire, such as: good collaborative ability, motivation, and an understanding of the importance of the study. These avoid possible biases resulting from delay in the implementation of the method, such as underreporting or overestimation, as well as withdrawal of the interviewer [70].
The R24h also provides details of food intake and offers greater convenience to the respondent than food record. However, there is a possibility of both respondent and interviewer bias, and the method requires trained interviewers. In this review, only four articles described the training of the interviewers [18,31,52,58]. As all the papers were based on national surveys, generally using secondary data from databases, it is assumed there was a lack of information on interviewer training. Furthermore, the information was not available in the database, but was described in the survey methodology.
Both the food record and the R24h require multiple days of evaluation to estimate habitual intake [10]. Recording of several days of consumption when using short-term records, such as the R24h and the food record, is necessary to remove intrapersonal variability, reducing the random error inherent in usual food consumption [71].
In relation to anthropometric measurements, most of the studies used the BMI as an indicator to assess anthropometric nutritional status. The evaluation of body composition may be relevant as it evaluates the role of body components to health, as well as its relations with the emergence of NCDs. There are various methods for the assessment of body composition that are considered accurate and sophisticated, such as hydrostatic weighing and dual-energy x-ray absorptiometry (DEXA). However, their use in epidemiological studies is impractical because of the high cost. Therefore, BMI and waist circumference are widely used for anthropometric indicators in population studies due to their practicality and low cost [72].
In practice, the anthropometric data are compared for the interpretation of anthropometric nutritional status to reference values, which were obtained from speci c populations [73]. Thus, it is necessary to consider different cut-off points for the interpretation of distinct life cycles, such as aging.
The WHO [74] de nes an aged or 'older person' as an 60 years old individual or older. This de nition, according to the WHO, is appropriate for developing countries; however, in some contexts, especially in developed countries, considering the signi cant increase in life expectancy in recent years, the classi cation of 65 years or more for an elderly person may be more appropriate.
However, it's considers that chronological age is not necessarily a precise marker in the monitoring of changes related to aging, because there are important differences related to health, participation, and levels of interdependence among people of the same age. Therefore, it is worth emphasizing that for the purpose of formulating public policies (one of the purposes of national surveys), these variations among older people should be taken into account [75].
Despite its wide dissemination in epidemiological studies, the use of BMI for assessment of anthropometric nutritional status of the elderly has been questioned, owing to the changes in body composition resulting from the aging process [76]. Thus, the employment of the same criteria for the classi cation of BMI in the general adult population and elderly patients in particular is controversial, considering the reduced stature, accumulation of adipose tissue, and reduction of lean body mass and water in the body that occurs in the elderly individual [77].
The criteria of the WHO (< 18,5 kg/m² -Underweight; 18,5 to 24,9 kg/m² -Normal weight ; 25,0 to 29,9 kg/m² -Pre-obesity; >30,0 kg/m² -obesity) [ This review contains some limitations that must be considered, such as the possible non-inclusion of all health and nutrition surveys conducted on the continents surveyed, owing to the fact that not all databases were consulted. However, to circumvent this limitation, a manual search was performed to obtain a greater number of studies for inclusion.
It is worth mentioning this study included articles of high methodological quality; thus, the results presented here depict the more common methodologies used and accepted in the scienti c community for assessment of food intake and anthropometric nutritional status, allowing, however, their critical analysis, to improve the reproducibility of health and nutrition studies.

Conclusion
This review identi ed the food record and BMI as the most common indicators of the evaluation of food consumption and anthropometry in population surveys conducted with adults and the elderly. The food record and the 24-hour dietary recall were the preferred methods for studies in developed countries, where high education levels and motivation of residents facilitate the application of these methods. BMI, owing to the ease of obtaining its component measurements and the wide application of its use, was the method of choice of the studies that performed anthropometric evaluation. Nevertheless, most of the studies did not use different classi cations in the BMI assessment of adults and the elderly, disregarding the speci cities of the changes in body composition and the physiological process of aging widely reported in the literature. The lack of clarity or omission in relation to information regarding the quality of dietary or anthropometric surveys, such as interviewer training, calibration of equipment, or replication of anthropometric measurements, were also issues observed in this review that should be reported.
Hence, it should be emphasized that, in population studies that assess health and nutrition, it is important to devise a proper study design, by selecting the method of food survey that best ts the objectives, the use of anthropometric indicators feasible in epidemiology, and the use of cut-off points appropriate for the population studied, with a view toward reducing the biases and providing valid data. Population surveys can thus provide reliable guidance for the formulation of public policies consistent with the epidemiological pro le identi ed in these studies.