Food Composition Table

- Congo Basin Countries -

Background

Food Composition Table for Congo Basin Countries

Food composition tables (FCTs) are important tools to convert food amounts into nutrient intake which are fundamental to many activities related to nutrition. Ideally, they should be country-specific due to several factors that can affect the composition of foods among countries.

Comprehensive FCTs are not available for most countries in the Congo Basin region and to try to help felling this gap, the International Center for Research in Agroforestry (ICRAF), also known as World Agroforestry Centre put together a FCT including 517 food entries for use in the region. The initiative was part of the GML project, developed by Erasmus Tang.

The present consultancy aims to review and assess the quality of the compiled FCT and to suggest appropriate changes to improve the FCT.

This website was created to share all relevant files and videos to document the comprehensive review of the compiled FCT.

Timing and objectives of the consultancy

The consultancy ran from March 1st to December 21st 2021 for a total os 60 working days with the following objectives:

Play Video

Introduction

1. Initial Assessment - User Database

The “User Database” corresponds to the information that will be made available for users. A systematic checking was was performed in this level of the database was performed according to the FAO/INFOODS Guidelines for Checking Food Composition Data prior to Publication of a User Table/Database before starting a detailed data checking of the metadata

The main findings are listed below:

   List of strengths

  • Good coverage of food and components
  • Cooked foods and recipes were also included
  • Food description looks complete in most of the entries (but a detailed review is needed to compare the description in the original sources)

 

   Issues to check and/or review

  • Sum of proximates: for about 150 foods, out of the 516 in total, the sum of proximates is outside the acceptable range (95-105) proposed by INFOODS. For another 90 foods it was not possible to calculate the sum of proximates since some components (mainly moisture) were missing
  • Missing values: for some components there are many missing values e.g. dietary fiber (missing for 20% of the foods), selenium (missing for 82% of the foods) and pantothenic acid (missing for 69% of the foods). Missing values should be either completed or the component should be excluded from the FCT
  • Implausible values: very high/low values were identified in the user database (e.g. very low moisture value for boiled amaranth leaves or very high iron value for plantain)
  • Inconsistent nutrient definitions: vitamin A expressed in both RE and RAE show inconsistent values
  • A precise documentation is missing for the components presented in the excel file

 

The excel file where the assessment was performed and the explanatory video are linked below (please see “related videos and files”).

Additionally, to access the quality User Database the screening questions proposed by the ‘FAO/INFOODS Evaluation framework to assess the quality of food composition tables and databases’ were used. This tool was developed to assess the data quality in a standardized manner (not published yet, additional information is available in the FAO/INFOODS webinar in data quality of food composition tables). The screening questions contain eight questions to evaluate the minimum requirements for FCTs regarding data presentation, food and component description and coverage, documentation and publication format. After answering to the questions, the FCT receives a score (maximum 120 points) and if the score is below 100 the FCT us considered of low-quality meaning that it needs substantial improvements.

The questions and answers for the FCT-CBC are available on the video ‘Quality of the FCT using FAO/INFOODS Evaluation Framework’. The final score was 75 (considering that the final database will be disseminated also in excel format) indicating that an improvement is needed. The main problems were the mixture of different denominators (e.g. data for some foods are presented on dry weight) and the inconsistency of nutrient definitions and lack of a precise identification. The review and an improved documentation will increase the total of points for the FCT-CBC to an acceptable score.

Related videos and files

Play Video

Systematic checking of the ‘User Database’: main findings explained

Play Video

Quality of the FCT using FAO/INFOODS Evaluation Framework

2. Archival Database: assessing the main sources of data

2.1 Food Composition Tables

Mainly four FCTs were used during the compilation of the FCT-CBC. In order to access the quality of these datasets, the screening questions proposed by the ‘FAO/INFOODS Evaluation framework to assess the quality food composition tables and databases’ were also applied to each FCT.

The total score for each FCT included in the FCT-CBC are summarized on Table 1 and additional comments are also provided below.

Table 1. Main food composition tables used as source of data and score obtained in the FAO/INFOODS Evaluation Framework

Note: Maximum score is 120 and the FAO/INFOODS Evaluation Framework considers that if the score for a FCT is below 100, it should be considered of low quality and the FCT needs substantial improvements.
The five FCTs were compiled in a unique excel file to facilitate the data checking. Since the component’s documentation are not available yet, it was necessary to check their definitions in the original sources. To facilitate the process, each  component presented in the FCTs was linked to the corresponding INFOODS component identifier (tagname) which are of great importance for a precise identification of the data.
 
If different tagnames for one nutrient exist, it means that depending on the analytical method, mode of expression or definition used it will result in different nutrient values (Klensin et al., 1989) (to learn more about the importance of using the tagnames you can watch the video ‘Importance of component identification’).
 
The excel file including all data available in the five FCTs linked to the respective tag name are available for download at the end of this topic.

Tanzanian Food Composition Tables (2008)

Represents the main dataset used in the FCT-CBC (reported as source of data for 91 foods).

Data presented in the Tanzanian FCT were derived from old FCTs, mainly from the International MiniList (1988-1992) and Kenya food database (1993). Some recipes were calculated but they were not performed to the INFOODS’ recommendations (nutrient retention factors were not considered).

The main limitation for using this FCT is that the components are not well described in terms of analytical or calculation methods and expression adding many uncertainties to the values (they may be based on obsolete methods since they were derived from old FCTs). The lack of documentation is important especially for fat, fiber, vitamin D, E, niacin and folate. Additionally, data for vitamin E data seems not correct (it is originally expressed in µg in the original source and when converted to mg the values are very low).
 
Therefore, my recommendation is that this FCT should be excluded (or at least taken as the last option). It this FCT is used to attribute a nutrient profile I recommend that all values should be checked for plausibility against data from another high-quality FCT.

Nigerian Food Composition Table (2017)

Contains analytical data for foods consumed in Nigeria (compiled from literature or from unpublished results available from the authors).

Similar to the Tanzanian FCT, the main limitation for using this FCT is that the components are not well described in terms of analytical or calculation methods and expression adding many uncertainties to the values. The lack of documentation is important especially for fat, fiber, vitamin D, E, C, niacin and folate. Additionally, many gaps on the nutrient profiles (missing values) may be identified.
 

Thus, using this FCT as a source of data will also require carefully checking of the data against another high-quality FCT. Many components may also need complementary data from other FCTs to allow for a complete nutrient profile.

West African Food Composition Tables (2012)

Total score = 120

Published by FAO/INFOODS this FCT presents a very detailed documentation. A new version was released in 2019 which was expanded in terms of foods and components, including many recipes. It is advisable to use the 2019 version instead of the previous 2012 version.

USDA National Nutrient Database for Standard Reference (2019)

Total score = 115

 Published by the U.S. Department of Agriculture, this is a very comprehensive FCT both in terms of foods and components. This FCT may be used, especially to fill the missing values from the other sources of data.

Kenya Food Composition Table (2018) – suggested additional source

Total score = 120

The Kenya FCT is also an African FCT of sound quality, but it was not used during the compilation of the FCT-CBC. I suggest that it should be included as potential source of data if needed during the review.

Related videos and files

Play Video

Main sources of data - Food Composition Tables

Play Video

Complementary video: 'Importance of component identification'

2.2 Scientific Articles

Data from more than 130 scientific articles were compiled. This is a good approach since it reflects the composition of local data, even though it is more time-consuming (for compiling and checking). Given the relevance of such data it is of great importance to check their appropriateness for inclusion in the FCT-CBC.

Although the different levels of data management are identified in the database, the documentation is not enough to allow the identification of the data source for all values presented in the database (you can find additional information about this topic in the video ‘Compiling data for FCT/FCDBs’ linked below). Thus, to facilitate the data checking some changes in the structure of the database were performed:

  • INFOODS tagnames were added to allow for a precise component description;
  • all foods compiled from the same reference were put together;
  • codes were added to identify each scientific article (BiblioID) and compiled food (record number).

 

The main issues identified in the compiled articles were:

  • the food description should be compiled as complete as possible to allow for a good food matching in the reference database;
  • data expressed on dry matter basis without moisture information (should not be used);
  • in some articles the sum of proximates is outside the acceptable range (should not be used);
  • the conversion of data originally presented on dry matter basis to fresh weight basis was incorrect (I have re-calculated it where necessary);
  • in some cases, the compiled data was not available on the original source without any documentation indicating from where the values came from (in these cases it was not possible to check the information and the data is highlighted in the reviewed spreadsheet);
  • review articles were compiled – it is more appropriate to compile from the original sources where we can find additional information.

Related videos and files

Play Video

Complementary video: 'Compiling data for food composition tables or databases'

3. Reference Database - detailed assessment

The data documentation to allow the assessment of the reference database was not available in the excel file, therefore, before starting the data checking a new ‘reference database’ was compiled to allow a detailed comparison of the original data against the foods included in the user database. The review was documented for each food (in columns T, U and V) and specific comments and recommendations were provided in the excel file (link to download the excel file is available at the end of this topic).

A summary of the checks and changes performed on the reference database are listed below:

Data documentation

  • no documentation (or user guide) was provided for checking. It should include a complete documentation about the criteria adopted, data sources and component description (nomenclature, modes of expression, tagnames and conversion factors used)

Components

  • all values for each component were reported under the same column without any further documentation; now they were checked and are reported under the appropriate INFOODS tagname to allow for a precise component description;
  • in some cases, the compiled data was not available on the original source without any documentation indicating from where the values came from (in these cases it was not possible to check the information and the data is highlighted in the reviewed spreadsheet);
  • there are many missing values, especially for some specific foods or components; my recommendation is that after the review: (1) if a component has many missing values it should not be presented in the FCT-CBC and (2) if a food has many gaps that could not be filled by appropriate values, it should also be excluded from the FCT-CBC;
  • some values are different from the original sources; in some cases, I assume that it is due to rounding, however the rounding procedures and number of decimal places used are not consistent within the database, so they should be standardized for the user database;
  • for some components, including dietary fiber, vitamin A, D, E, folate and niacin the different tagnames SHOULD NOT be mixed since the values are not equivalent.
  • energy and available carbohydrates should be recalculated and not directly compiled from other sources. FAO/INFOODS guidelines recommend that these components should be calculated as follows:
Energy (kJ/100 g EP) = total protein (g/100 g EP) x 17 + total fat (g/100 g EP) x 37 + available carbohydrates (g/100 g EP) x 17 + dietary fiber (g/100 g EP) x 8 + alcohol (g/100 g EP) x 29

 

Energy (kcal/100 g EP) = total protein (g/100 g EP) x 4 + total fat (g/100 g EP) x 9 + available carbohydrates (g/100 g EP) x 4 + dietary fiber (g/100 g EP) x 2 + alcohol (g/100 g EP) x 7

 

Carbohydrates, available (calculated by difference) (g/100 g EP) = 100 – (total protein + total fat + dietary + alcohol + ash) (g/100 g EP)

Food matching and description

  • food description was divided in different columns (and incomplete in some cases), for an easier identification of the food the details were presented as complete as possible in a single cell (column H) for each food;
  • FAO/INFOODS published a guideline on how to link foods from different datasets. I strongly recommend that you follow this guideline to match all foods included in the FCT. The guideline is available in the INFOODS website and your can also see the complementary video suggested below.
  • in some cases, the food description is different between the source of the nutrient data and the food presented in the FCT-CBC; it can be misleading for users (it is indicated in the individual comments, some examples are available in the lines 5, 11, 17 and 538); if similar foods are not available for a given food description (e.g. chinchin) they should not be included in the FCT-CBC;
  • some raw foods were matched to cooked foods and adjusted using nutrient retention factors (according to the original notes); however this is not the recommended procedure since both yield and nutrient retention factors should be taken into consideration (more details are available in the video ‘Principles of recipe calculation’ linked below); in this case you should either make the appropriate calculation or use data for the cooked foods available in other FCTs; note that (1) for the calculation the moisture content in the raw food is required and that (2) for some processing methods (e.g. fermentation) there are no nutrient retention factors available so it is not possible to estimate the nutrient profile in this case;
  • the “source” is not indicated in a standardized manner, I recommend that they should be standardized indicating not only the data source but also the food code in the original source (for data from other FCTs).

Related videos and files

Play Video

Complementary video: 'How to link foods from different datasets?'

Data aggregation

Play Video

Example: How to borrow data from FCTs?

Play Video

Complementary video: 'Types of data included in Food Composition Tables or Databases'

Play Video

Complementary video: 'Principles of recipe calculation'

Created by Fernanda Grande (2021). The content of this website is continuously reviewed and updated.

Last updated 08.10.2021