Evaluation of the international standardized 24-h dietary recall methodology (GloboDiet) for potential application in research and surveillance within African settings

The global nutrition transition associated with an increasing prevalence of overweight and obesity is a real health and societal challenge affecting particularly low- and middle-income countries (LMICs) worldwide, including those in Africa [1–3]. This transition is characterized by rapid changes in diets shifting from a traditional and prudent high fibre, low-fat diet to a diet rich in saturated fat, animal source foods, salt, processed foods, added sugar, lower intake of fruits and vegetables, fibre and complex carbohydrates and increased consumption of meals eaten out-of-home, especially from fast foods chains [4–6]. In addition, physical activity has substantially decreased, especially in urban areas with increased sedentary behaviour in the secondary and tertiary economic sectors activities, as compared to primary sector activities which are more prevalent in rural areas [7]. This tangled situation has resulted in the steady increase in non-communicable diseases (NCDs) such as cardiovascular diseases, respiratory diseases, diabetes, and certain types of cancer in the LMICs [8, 9]. Concernedly, micronutrient deficiency is still tremendously prevalent within these populations, leading to a double burden of malnutrition difficult to be subdued by the fragile economy of the concerned countries [10, 11].

Accurate assessment and monitoring of malnutrition in LMICs, especially early detection of changes in the nutritional status of the populations, might require the collection of high-quality dietary data because individual food consumption is intimately related to the nutritional status. Dietary data can be collected for several purposes including setting of nutritional recommendations, exposure and safety assessment as well as overall nutritional surveillance.

We have previously illustrated from an inventory conducted in 18 African countries that various dietary assessment methodologies are used across the continent, predominantly for surveillance and research purposes [12]. Amongst these methodologies are the 24-h dietary recalls, food frequency questionnaires (FFQ), dietary records and dietary history. The 24-h dietary recall is an open-ended method intended to report detailed information about all foods and beverages consumed by a respondent, in the preceding 24 h or over the previous day [13, 14]. The interview can be paper-and-pencil-based, computer-assisted or rarely self-administrated. The method relies on short memories and allows the quantification of all the foods and beverages consumed over the period concerned.

Although only few of the used methodologies were reported to have been validated, their reported scientific outcomes have shaped the understanding of the food consumption landscape in Africa. However, in practice, there is a need to appraise the validity of the methodologies used and strengthen individual food consumption data collection and comparability throughout Africa. The optimal way to achieve this goal is to build on pre-existing experience and accelerate transfer of knowledge and capacity building among researchers and countries, considering the specific scientific, social, cultural and ethical African context. Diverse initiatives have already been launched on the continent to cover this gap. Although these initiatives do not strictly embed the same objectives, they act as fundamental components to cover the food consumption collection field in Africa. For more than six decades, dietary intakes activities in Africa have been pioneered by diverse institutions including the Food and Agriculture Organization of the United Nations (FAO), Harvard University, Wageningen University and the French National Research Institute for Sustainable Development (IRD) through the development of country- or region-specific food composition tables (http://www.fao.org/infoods/infoods/tables-and-databases/africa/en/), the training of local research teams, as well as the designing of studies on dietary assessment methods use in epidemiologic studies [15, 16]. The International Network of Food Data Systems (INFOODS, http://www.fao.org/infoods/infoods/en/) benefitted from these experiences, and through its regional data centre AFROFOODS published the West-African food composition table ⁽[17]⁾and developed relevant courses [15]. Besides this, the FAO/WHO Global Individual Food consumption data Tool (FAO/WHO GIFT, http://www.fao.org/nutrition/assessment/food-consumption-database/en/) aimed at increasing the harmonization and availability of existing data. The International Dietary Data Expansion (INDDEX, www.inddex.nutrition.tufts.edu) project led by Tufts University was designed to increase the availability, access, and use of household and individual food consumption data in low-income countries by developing open source and open access tools for dietary data collection, processing, analysis, and policy application.

The International Agency for Research on Cancer (IARC), the affiliated specialised institute of the WHO on cancer research, has developed a 24-h dietary recall software for the collection of standardised dietary data for epidemiological, surveillance and monitoring purposes [18]. A series of quality control measures are utilized from the preparation of the databases to data collection and management [19]. The software based on the GloboDiet (previously EPIC-Soft) methodology, has already been used in 19 European countries for both research and surveillance purposes. In addition, the GloboDiet methodology has successfully been adapted in Latin America (Brazil) and in the Republic of Korea. The specific adaptations needed for this first Asian country were recently published [20].

The implementation of GloboDiet in Africa will require adaptation based on prior in-depth understanding of the specific context of the African continent, with its diverse populations, languages, food sources and habits. Through a consultation panel consisting of both African and international experts with known and recognised experience in the field of dietary assessment, this paper reports on the evaluation findings of GloboDiet’s interview steps and other various aspects towards a feasible platform to be used for surveillance, monitoring and research purposes in diverse African settings.

The GloboDiet methodology designed for nutritional research and surveillance based on the 24-h dietary recall is a tool that has, initially, been used to collect dietary consumption data in the EPIC study. The cumulative experience that went into the development, testing, validation and implementation of this international software enables the establishment of strong science-based evidence on its reliability, detailed information, robustness and flexibility that made it possible to be adapted amongst other countries and contexts [23, 27]. Likewise, the main feature of the GloboDiet 24-h dietary recall software is its design that ensures a high level of standardisation of data collection, management and analyses within and between countries. Therefore, the consultation of the panel of African and other experts was an efficient way to share information on the methodology and receive feedback on the potential of the existing version and additional adaptations needed for Africa.

Within the GloboDiet software, the general information on the interviewee (e.g. age, sex, height and weight) is important for final controls on the reported total energy and macronutrient intakes vs. recommendations, as well as for individual identification information (e.g. interviewee’s identification code). It should be recommended that in the context of low-literacy, anthropometric measures be entered from a participant sheet to avoid self-declared information, which is a plausible source of errors [28, 29]. The “special days” within the general information might also concern local market days in rural Africa, because foods available in households and consumed on that day are more diverse, and do not reflect routine consumption [30]. In addition, these foods available and then consumed that day fluctuate with the purchasing power of the households.

To fully describe a consumed item, a combination of two food description pathways is used in GloboDiet. The implicit pathway uses the name of the foods as self-standing (e.g. injera uniquely could refer to the Ethiopian sourdough bread made of teff), whereas the explicit description requires facets and descriptors (e.g. injera may require at least the facets: food production/purchase location, salt content). The use of facets and descriptors privileged in GloboDiet is also used in food description worldwide by organisations such as the European Food Safety Authority [31]. Through the consultation of the experts, it was confirmed that most of the facets/descriptors available in GloboDiet European versions are also suitable for the description of foods and recipes consumed in urban areas in Africa. However, as in any preparation of a GloboDiet version, typical African foods will need to be included in GloboDiet databases and combined implicit and explicit descriptions will be applied to each food and recipe during data collection. Additional descriptive pieces of information that may be deemed necessary to collect are sources of drinking water [32, 33] and water treatment methods [34, 35] because drinking water is still a vector of microbial infections and chemical hazards in Africa. Other examples of description consideration details that might be interesting to consider are application of treatments to decrease cyanogenic agents in cassava [36], or the covering of street-vended foods. It is also important to consider food fortification status in the description of the foods available in Africa since several foods and staples are now subjects to mandatory or voluntary food fortification programs in Africa [37]. Food fortification should be included in African databases if interviewees are likely to recall the food items and/or if fortified foods could be recalled during the interview and dealt a priori.

The major challenge in food quantification that was raised unanimously by the experts is the individual consumption quantification from a shared plate, considered as a notable issue in studies using the 24-h recall [38, 39]. To address this issue, some researchers reported the distribution of HHMs beforehand by field enumerators, and asked the participants to serve individual portions [40, 41]. However, Kigutha [42] hypothesized that this might affect eating habits of the subjects, and therefore introduce biases in individual intake estimation. Asobayire [43] quantified individual intake by weighing participants before and after eating and adjusting for beverages volume consumed in separate bowls. This approach, besides weight measurement constraints and errors, is only valid to estimate the consumption of a single food and recipe, without accurate quantification of diverse item consumed. In the Gambia, an algorithm has been applied to approximate individual portion consumed from a shared bowl considering body weight, sex and age [44]. The formula used in this algorithm has not been tested in other settings, suggesting that, if an algorithm is the chosen approach to solve the shared plate issue, there is a need to find a systematic and reliable algorithm to estimate individual intake from shared-plate.

Besides the consumption from a shared plate, food quantification methods usually applied in GloboDiet have been acknowledged by the panel of experts as valid for quantification of African foods and recipes. Pictures have been extensively used for quantification in Africa [45] and validation studies have been carried out [46, 47]. Picture atlas in GloboDiet is a series of a minimum of four high-quality unpixellated coloured photos at incremental food sizes. Adaptation of the existing GloboDiet photos to the African context will require updating of the foods and recipes to be photographed and the deletion of the cutlery not in use in African settings. Food grids developed and validated by GAIN for the Food Assessment Coverage Toolkit (FACT) could be utilized as a primary tool [48] and used a complement to validated picture of foods for the quantification of individual intake. Likewise, previous studies suggest that HHMs including cups, mugs, spoons and plates are extensively used for quantification in Africa [49–53]. Adaptation of the HHMs booklet will include photos of local HHMs, but also having these onsite during the interviews to ease quantification. However, in contrast to foods pictures and HHMs, shape/volume quantification appears to be laborious because misrepresentation of the actual size of the foods may occur in relation to the visual perception of the subjects. Shapes are predominantly used in GloboDiet to quantify breads and wedge-shaped portions and have not been extensively used in Africa, which imply a thorough analysis of applicability and adaptability. Overall, there is a necessity to validate the combination of methods of quantification proposed by Willett [13] comprising household measures, photographs, geometric shapes, standard units, and other innovative methods likely to be developed in the very context of LMICs. In addition to individual portion size estimation, memory lapses have been reported as substantial sources of errors in food intake assessment in Africa [54]. Therefore probing and various controls are important to capture all the foods consumed, their description and quantification in order to prevent misreporting. GloboDiet contains a series of probing and control steps from database preparation, data collection, calculation and data cleaning in order to decrease these errors [19]. Apart from usual probing such as sugar for tea and topping for breads, African-specific probing can be included according to specific behaviours in some African regions. For example, condiments such as the stock cubes which are ubiquitously found in African countries can be candidate for probing. Another important issue in Africa would be the need to extensively train field enumerators for the collection of individual food consumption data, considering the local environment.

The contributions of the experts were constructive and will serve as the foundation for the possible future implementation of the GloboDiet methodology in Africa. It can be summarized that successful and standardised assessment of individual food consumption across Africa is possible but would need the inclusion and combination of description and quantification methods available in GloboDiet, as well as the specific propositions drawn from this consultation. Furthermore, particular attention should be given to the shared plate consumption, to improve quantification of individual intake. Also, a clear definition of dietary supplements will be necessary to handle the African specific situation.

The possible implementation of the GloboDiet methodology in the African context will be an opportunity for public health, nutrition, food safety and cancer researchers to collect and analyse high-standard quality food consumption data. Preliminary work to assess the needs and gaps to adapt the GloboDiet methodology is a prerequisite before any roadmap for implementation. Africa is a heterogeneous continent constituted of more than 50 countries with various dietary cultures and the most efficient approach will be to develop several versions of GloboDiet-Africa either by country or by region, with few pilot countries as starting point. Further steps will therefore include, but not exclusively: collection of foods and recipes from African surveys, detailing of appropriate facets and descriptors, exploration of suitable quantification pathways and their validation in pilot studies.

Africa needs to collect regular data on nutrition and its determinants, as a starting point of ingrained nutrition surveillance programs capable to support the continent in the process of fulfilling the Sustainable Development Goals (SDGs). Overall, concerted actions will be needed and collaboration between and with African networks will be the key to successful implementation.