 Good morning, good afternoon and good evening to all participants and to the presenters to this very important FAO in foods webinar on food quality or food composition data. I'm very much looking forward to interesting presentations that we will have today on the different aspects of food quality not only from our food composition perspective but also from the user perspective. And as you can imagine food the quality of the data is very important not only for the food composition tables but also for the users. So I'm looking very much forward to the presentation. So I will present the FAO Infoods Evaluation Framework. Propasi Puvastin and Konchit Yuk Branson will present the Asian System for Quality Evaluation of a Food Composition Table and Databases. Marine Oseh Dukes will present the Eurofear Quality Index for Analytical Data and Sandra Crispin will say us from the user perspective how our quality in food composition used and what is missing. So with this one I will say that all the participants they can participate not by speaking but there you have two buttons. One is the Q&A and one is chat. So you can put your comments and questions into these two chats and at the end we will have half an hour where we will be able to respond all to your questions. Having said so I would like to start with the presentation of the FAO Infoods Evaluation Framework to assess the quality of food composition tables and databases which we call very shortly the Evaluation Framework and it is a product from many different people so from me from Barbara Stademeier, Anna Winsen, Marine who is here, Super, Propasi who is also here, Konchit who is here, David Haydovitz and Patricia Knane-Niarella. So what is the background? We know that evaluation systems are very important to systematically assess the quality of the data and or the product to make decisions of inclusions or exclusions for usage. For food composition we already have some evaluation frameworks available but they are on analytical data and they are from the USDA and from Europe but we did not have one which is global for the published food composition tables and databases and as we all know food composition is really the basis for almost everything that we do in nutrition and increasingly also in agriculture. So if this basis is of a bad quality everything else which is built on it will be of a bad quality as well or on the country if it is a good quality then we have better chances that everything else is on a good quality as well. So what was the aim? We wanted to develop this evaluation framework that will assist compilers as well as users to assess the quality of a food composition table in a standardized measure and this is very important so that everybody is using the same framework to determine what is of good and what is of less good quality and this will help the users to select the most appropriate food composition table that they have and to demonstrate and now for the compilers to demonstrate areas where they could do some improvements and we hope by this one as well that to motivate decision makers in making better decisions on the investment of in high quality of food composition tables and databases. So the main target groups are the compilers and the users. How did we do that? We started in December 2014 with a call through the listserv and we built a working group. Then Barbara, Stalemeyer and me we were then preparing the first draft which was circulated in many rounds and once we have had the more or less the frame then we started to put these themes that we had into and build up some questions on how we can be evaluated. So and this took quite some time to make a proposition and then fine tune and improve it in several rounds within the working group and we also tested it to see if what we are proposing makes sense and then it was really finalized the last year the text and the Excel file and at that stage I really would like to say also a great thank you to an undergrander who added all the macros to the to the Excel file and now I hope that we would have published it already but it is still in the editing and layout process and will be published very soon but in any case in 2021. So how is it looking like? We have three parts. The first part is the screening questions which are eight questions and if you get through it then you are invited to do the full evaluation which is part two. So we have seven categories and in each category as I said before we have a lot of questions to which you can answer and then you get a score and then at the end of this full evaluation we will have a quality score per category and this quality score per category is then used to do the final scoring of the whole database which is in part three. So which are the principles? The principle is that first of all we define the categories and then in all the categories we created questions and sub-questions and to each of them the criteria and the sub-criterion we scored them if they are highly important, important, medium important or low and depending on that one this category, this criteria got a score which is either 120, 90, 60 or 30 and then all the questions which are normally answered by yes or no or not applicable or sometimes in rare cases you needed to say if it is applicable for all the foods in your database or if it is applicable to the archiving the requirements of the user database. And based on these answers and the points which were attributed to them we calculated a score for each of the categories and as I said before we have two main target groups, the compilers and the users and the users they cannot assess that much the quality of the data so these questions would then only be asked to the compilers and then at the end as I said for the entire database we had then a score to say the quality of the entire food composition database. So here are the eight screening questions so our foods presented per 100 gram edible portion then are the denominators and the SI units used so and they are mentioned are the tax names of inputs used and then we have a list of what we consider a minimum list of components that should be included in each of the food composition tables and then are the food and food names and their description well specified are certain foods covered not only raw foods but we also wanted to have food as consumed then the source of data are they provided and not for the whole database but on the food level or the value level and then the format is a format that is user-friendly meaning it is Excel, CVS similar or access and you can see the priority it's either one or two and then for this one you get either 10 or 20 points. The final score in the maximum is 120 and if you have less than 100 meaning that one of the high priority questions or two of the less priority questions you failed then we consider it is not worthwhile to go into the food evaluation. So the food evaluation we have seven categories one is the documentation, the second is the food coverage, third is the food identification, four is component and value coverage as well as the expression, five is the quality of the data and this is only for the providers because normally the users will not be able to know and then the seventh category is the year of publication and the frequency of output and here on the right hand side you can see the number of questions that we have in each of the categories so it ranges really from only few to quite a number 66 this is then for the compiler or 65 for the users and as I said before the quality of the data has no question for users owned for the compiler and then based on the questions and the score which is obtained then we have a quality which is ABCD which is either very high above average average or below average depending on the score achieved the percentage of the of the maximum points and here are some some it's one example only so for example in the category of three was the food identification our scientific names provided and then it is for all relevant foods for certain or not at all or is the edible part there is or the preparation and processing stage so and again for all for non officer and depending you can see that you can get a higher score and if it is none then it is zero and then based on the categories so we have seven categories and for each category the maximum point is four points so for ABCD and based on this one we have then the overall score for the food composition table which is then AA, BB, CC, DD and applying the same kind of scoring so and as the maximum score will be different for compilers and users the points to be achieved are different and then the interpretation is so if you arrive above 85 it's a high quality the second category is above average average and below average and I would also like to acknowledge people who have contributed at the beginning but dropped out for certain reasons during the process so and they are mentioned here and I would really like to thank them as well and the conclusion is that this is the first attempt to evaluate not data but the whole food composition database from an international perspective and for users as well as for compilers and we are sure it can be improved in the future and we hope that this result will be helpful to improve existing and future food composition tables and bed databases as well as to publish them on the inputs website so and the next one I would like to ask Kutchit and Papasi to present the one from Anshanset. First of all greetings again to all from Thailand and Papasi Puyasitian and my colleague from the Institute of Nutrition Mahidun University. We would like to acknowledge FAO in Food and Dr. Root Sharon Deere for inviting us to share our experience on the development of the Anshanset food system for quality evaluation of how to publish food composition table and database since some of the participants may not be the food composition database generator may I first review the step of to establish food composition database in good quality food composition database the first step we have to set the the objective and then select and prioritize food and nutrient to be analyzed which can be based on the national food consumption survey and the nutritional problem in each country okay and then prepare sampling plan and collect a representative sample then we prepare the sample and we prepare the sample and to be homogenized and then analysis after that we prepare the after the analysis we should use the standard method and instrument and we should have quality control system internal and external after that we get the analyzed data and we prepare archival database reference database and user database you can see that along the along the process we have the first system you refer and USDA guideline for evaluation data quality of the quality of the analyzed data and we have also the second system FAO in Food guideline for checking the data prior to publication we have to check all detail and all information the data and all information before be published and and now we also need the evaluation system for assessing the quality of the published food composition database so these are the steps to establishing good quality food composition database this slide show the filmcham table which were available in 2015 the oldest version was published in 1997 and the newest one is was published in 2015 before that different food commission tables present different detail information of foods and nutrients with different data quality some use different food commission data format at this time some country were in the process of updating the national food composition database in 2015 you see the international life side institute southeast ASEAN ASEAN region survey the status of the national food commission table in ASEAN and they provide financial support to organize the first round table ASEAN food you will see discussion on the development of quality food mission table and database in ASEAN and also the development of an evaluation system for assessing the quality of published national food commission table in ASEAN country fortunately at the beginning of the 2015 Dr. Ruth distributed the first draft of FAO in food evaluation framework and criteria on the quality of the published food commission table and database via the food composition discussion group so we prepare a draft version of ASEAN food evaluation system based on the FAO in food first draft and used for round table discussion the round table discussion led to harmonization and agreement of the idea and the first version of the quality evolution system was developed and used for evaluation of the old version of national food composition table in ASEAN after the trial the final ASEAN food system for quality evaluation was developed in March 2016 it composed a set of 12 criteria with several sub criteria together with awaiting and schooling systems the final ASEAN food quality evaluation system will be presented to you now by Dr. Kanjit Prasong. Okay for the developing ASEAN quality assessment of the published food conversion database the round table discussion is emphasized on the main component of the quality system they are composed of the criteria sub criteria waiting score for the criteria ranking score for the sub criteria and also the scoring system to indicate the quality of the evaluated food conversion database. The all the references that we used during the discussion round table discussion so starting from the the framework as we mentioned several times and the guidelines from the euro for from the FAO and also from the infill circuit book I think the community of food composition we know about this the total criteria of the ASEAN food quality system is we decided to have 12 criteria they are the first in the year of quality evaluation we also have a concert on the waiting score of east criteria for the first of the year quality evaluation we provide only one the second criteria is the percentage number of food items in the updated version food provision table compared to the previous system the waiting score is four the maximum the maximum score in east criteria is five so we also discuss a lot on on the score on east criteria so this is the content first form the ASEAN member the third criteria is the document of the information to the user and also the score is four the four is the food description again four the fifth person contribution of the Naray data from the laboratory within the county we have the maximum score of five the number six is number of individual data sets where in more than or equal more than three for preparation of the user database the maximum waiting score is five number seven on the food conversion derived from the compute data of borrow data including the recipe calculation the score provided only three the number eight is copper newtons which is very important part of the database so the content set is five score number nine missing newtons data the score is three the quality content from the laboratory this is the part that for the user we cannot know about the laboratory part so we use the cell assessment by the evaluator but we emphasize on this the criteria we also provide the maximum score of five number 11 is compilation to use for the establishing and updating food conversion database the waiting score is four and the access of the food conversion table food consumption database the maximum score is five so according to the table of criteria you can see that the the maximum score is number five number six number eight and number ten from this when we we we have each sub criteria what i will present to you rather but now the the when we see the maximum score in this column and in this slide i want to show you only the major contribution of the east titillia which is i already put the yellow color in four parts of east titillia okay the the quality assessment system we also evaluated by the user using the excel templates but i think uh it is the same of the input framework but we have much easier than than the the framework before we go into detail may i introduce to you to the waiting score of for the titillia and ranking score for the titillia five is sufficient as the most important or most combined with the titillia or sub titillia and uh descending order into one less important is mean if not meet the titillia and the same as the waiting score for the sub titillia five is the percent of the most combined with the titillia and one is not combined with the titillia for the first for the first titillia we emphasize on the year of publication and again the waiting score only one the over channel of the pub beach you can see if we publish really over channel we will got only one for ranking score but if the food commission service publish very recently we will got the maximum score five and the titillia number two the percentage of number of food item in the updated version if the database has has the number of the food item less than 10 percent of the total so they will got only one of the ranking score whereas when they have more than 50 number of food item that update in the compared to the previous version they will got maximum score five in this webinar we use the Thai food convention database as a example for the evaluation the evaluator in each country will feel only the the the color in the in the column if only so the score that they give when they will put into this column and then the excel will calculate the total score for each titillia and also for the final score the titillia number three about the documentation of the information to the user they also have several sub titillia here something from food sampling sample preparation and copper nutrient unit of explanation I think our update information are really important so we we will provide the score about four for each sub titillia the number four about the food description we divided into three parts of the food description for the first group draw and cook single food if they provide local name English name scientific name and also part of food maturity physical state cooking method so they will got the maximum score of four in each sub titillia and for the homemade dish or food so in the restaurant or supermarket they also need to have some sub titillia here and also the same ad for the p-package food they also have some sub titillia number five about the person contribution of the analyze data from the laboratory if the more data that are right in the rapidly is mean that the data is come from the analyze data the maximum score of five will be given whereas the really low number of the analyze data they will have only low ranking score the number six about the number of individual data set where in more than three for the preparation of user database and again the more of the individual data set which is mean this is more quality they will get they will cut the maximum score fine and number seven for the food composition that are derived from the compute data or borrow data which is consists of several sub parameter sub titillia here so they also have the to fill in each sub titillia the major important part is the copper nutrients we also discussed a lot on the priority of each parameter here starting from the main nutrients dietary fiber sugar stars major major element trace elements and also also the water soluble vitamin fat soluble vitamin 30 as equal to slow and also et cetera here so they also have a lot of sub titillia which we are emphasizing based on the nutrition problem in each country and we have the content set that i think may be different from country to country but this is the content set of the asian group member the titillian number nine is the missing newton data in the food transmission database if we have less missing data it means high quality so this is we will get got the maximum score fine and number 10 is the quality control for the lab analysis as i mentioned to you that this is part this titillia we need to use the cell assessment because the user for all the evaluator may not know the the the quality in the laboratory for example the laboratory already accredited ISO 17 0 25 or not using standard analytical method or not using internal quality control or not so this is the important part for the analysis the kind illia number irrelevant is the compilation to use for establishing and updating food conversion database as we know that okay we have the for example input compilation to for the uh use for the compilation and we we also provide some cell develop compilation to our commercial you can see the skull view uh uh just being ordered here the last titillia of the evaluation uh system is the assessment of the food conversion database we also provide as a pin set pin thread version which is have sub titillia only three score but when we provide in the pdf the score is going up and in the excel this is the maximum maximum score for the searchable we also make the decision that we will have only four for each score okay all the 12 titillia and sub titillia you can see they have a lot of number to be viewing for the thailand food conversion database we we we end up with the 12 titillia with the maximum score is around 200 and when we put the score for evaluation and this is example for the thailand that we got around 90 percent of quality score and the same as loose uh presented if we classify as a confidential cost uh as a grade a for the high quality it means it's maximum uh uh more than 85 percent so and this is the same titillia okay the result that we got from the asian member country sorry that we cannot uh put the name of each country but we can we can see that okay in the asian country we need the improvement for the one that they have low low score and according to the law table discussion we can see that they have many uh sub titillia and titillia to be emphasized when you publish the food competition database so i think this is our example of exercise in this case so may i move to uh uh to summarize these are the curriculum version of food composition uh table and database in indonesia in malaysia and in philippines and they they have also the uh online food composition database okay and this is the current version of food conversion table in thailand we have a two source of two data source the uh in 2016 we developed online uh version of titillian database and in in 2018 the titillian food conversion table published in by municipal public health also uh developed both database online and uh during the quality evaluation result that obtained in 2015 have been considered and used by the asian food member countries in preparation for preparation the update version of the national food composition table and database for improving the quality of the the new one so we expect to see uh much improved quality of the food composition table and uh database in asian however the quality evaluation of the update version have not yet been uh finished so uh it is on process now so we expect to see the much improvement in terms of the quality of the database in asian yeah that's our our appreciation thank you so much uh kunjit and prapasi for this presentation and you can see that many things are very similar but uh very different uh so the um the waiting is quite different between the two concepts so uh and to be seen how useful these two concepts in parallel will be um may i remind the participants that you have the possibility to write down in the chat or in the q and a your questions and um and uh and comments please do so so may i invite now marine to present the order of your quality index for analysed data thank you russ yes i'm marine ozerzoek from the french adjet c for food environmental and occupational health and safety um my presentation will be about the ufia quality index for analytical data so you refer quality index analytical data there are some words to clarify um on this topic the first word is your fur um i will give you an insight of what urofia is the second concept is data quality once the meaning of data quality is defined we will need a method and a tool to measure it so we need a system and urofia has created such a system will make another view of it this system consists of a set of questions grouped under different topics and the scouring methodology is associated with these questions it leads to it leads to the production of a score which reflects the overall result of the quality assessment and we call this score the quality index after presenting this data quality assessment system um i will share with you what i've learned from this work let's start with information on urofia urofia means european food information resource um initially it was a five-year project funded by the european commission and its objectives were essentially to strengthen scientific and technological excellence in food data bank system this for the benefit of european food and nutrition research to reach these objectives a urofia quality framework was built and then the urofia quality index was created as part of this framework urofia now is a network and also a non-profit association of 150 members mainly research institute government departments and the universities urofia participates in various projects from the european union and from the european food safety agency and gets funding from this urofia also operates the input regional data center in europe its current mission remains very close to its initial mission it still advocates for food information but in larger geographic geographical area as initially you can have more information on urofia.org the circle here is representing the urofia data quality framework um it's indeed covering many different aspects of quality and food composition databases for example a flowchart of the compilation process identifying critical control points um was was produced you also have practical recommendation concerning recycling calculation component identification well even an official standard which is stood by the european committee for standardization was created based on urofia work and so urofia worked on quality evaluation of values the question to be answered is slightly different from the one um that was presented by ruse and by kunchit and paprasi the question is is the data to be evaluated appropriate for inclusion in national food composition database so data quality is defined this way today we will focus on the quality of any analytical data for any food and for any nutrients which method did we use to produce this urofia data quality system a task group has been set up um with urofia members and we made a review first of existing data qualities uh data quality assessment system we had the basis system which is another urofia work um it's dedicated to bioactive substances foods uh we have also four national system from the u.s from france from italy and from german the difficulty with data quality assessment is that it's not pure science i mean some decisions have to be taken based on compiler experience therefore some national data quality system were never published in scientific literature and one benefit of being in a project was that we were able to have access to systems from members that were never published and we could be inspired by them and we tried to make a compromise between all these different systems the task group produced a draft and other urofia partners have given their inputs on this draft and the system was tested by a compiler the quality assessment for analytical data we've made consist of seven categories that you can read here food description component identification sampling plan number of analytical samples sample handling analytical method and analytical performance defining these categories was more or less easy because there was a good consensus between the different systems on these categories and today we will focus only on two categories food description and analytical method um in each of these seven categories we have identified a list of critical issues each issue was considered in a question and possible answers to this question were either yes no or not applicable we have this not applicable answer when the question is not relevant for the pair food and nutrient considered in the end the number of questions per category ranges from two to sixteen and to assess the quality of the value with the urofia system you have to answer up to 33 questions in total here is the list of questions in each category this is of course too small for you to read but i show it to you so that you can form an opinion on the complexity of the system you will understand that i won't be able to present all the questions now and therefore i will focus on on two categories only in the food description category we have indeed three sub categories of questions the first applies to all types of food the second applies to manufactured goods and the last applies to composite dishes made at home or in restaurants for example here on these slides we have questions related to all types of food you can see that you have already 12 questions there in each question the urofia quality assessment system questions the availability of satisfying details concerning specific issues for example food group cooking method preservation method the first question is for example is the food group now this question is asked because if we have a data on mustard for example it can be wanted to know whether we are talking about the seed about the seasoning another question is on the edible parts if relevant is it clear if the food was analyzed with or without the inedible part you can notice in this question that we have the words if relevant we have it because for some foods the question is not relevant example are salt juices and applesauce we do not want to ask questions and count the positive answer i mean yes as in favor of good data quality whereas the answer is obvious therefore we have this not applicable answer it's up to compiler to decide whether a question is relevant or not for the given food and nutrient the urofia data quality assessment gives sometimes advice and examples for example we have this question in the system if relevant is the months or season of productions indicated among the tips accompanying this question you have that it is especially relevant for fresh foods and vegetables for example well despite the number of questions in this category it may be a category not too complicated to assess for compiler because they are familiar with all these aspects that are indeed to be considered when using food description coding system such as long wall of food x2 that compilers know very well so as a preliminary conclusion on this category for assessment i would say that it includes a high number of questions but number is not necessarily associated with complexity nevertheless the more question you have the more time you spend to perform the assessment and care was taken to limit as far as possible the number of questions indeed in this category for assessment complexity lies in decision and relevance of the question for the given couple food and nutrient so till now we have explored some aspects of the category food description and since we have a quality assessment system focusing on analytical data let's focus now on the category analytical method this category consists of two questions only is the analytical method appropriate and are the key method states appropriate one major input of urethra is to provide the support in the assessment of analytical method this input is the urethra guidelines for assessments of method of analysis it's also named gamma in short it's freely available on the web you will find a wiki analytical method for vitamins um created originally by urethra and supplemented by the project tds exposure which is a a european project financed by the european commission in gamma for each vitamin you have the current golden standard for analysis its principle its key steps you also have the list of precautions to be taken for sample preparation for example for vitamin b2 its high sensitivity to light is mentioned together with its sensitive sensitivity to alkaline condition you have also a list of set certified reference materials and standard reference material in gamma on the list of proficiency testing schemes other methods available are also considered so with all the information provided in gamma it's then possible to answer the two questions that you can see about above but also questions for other category um the category analytical quality control um uh questions and sample handling so the creation of gamma was a very important step forward because compilers assessing data quality have reported during urethra meetings lacking knowledge regarding analytical techniques in a sense um gamma is a foundation supporting ability to answer many questions of the urethra data quality assessment system so after considering two categories of the urethra data quality assessment system i will say a few words on scarring strategies used in this system scores per category uh range from one to five one corresponding to the lowest quality and five to the highest quality of course we have here two examples of different strategies for scarring the first example is for food description to determine a score for this category it was decided to multiply by five the number of yes answers and to divide it by the number of 11 questions for another category the category component identification and related terms the scarring strategy is totally different this category consists of the three following question if the component described unambiguously is the unit unequivocal is the matrix unit unequivocal to score it um you have to give the maximum score which is five when you have three yes any other combination of answers would lead to score one you'll then understand that scarring is not straightforward and that it's necessary to have an automatic determination of scores based on answers given um this automation was implemented in food case the food data management system that is promoted by urethra but it could of course be implemented in any other food data management system um urethra advises to store not only uh scores obtained for each category but also all answers that were given the idea is not only to keep traceability but also to allow future checking under maybe updates once you have a score per category you can determine the total score for the data that was assessed simply by summing the seven scores per category the value obtained is the quality index it ranges from seven to 35 it then can be recalculated to a maximum of 100 to be more easy to handle after this presentation of the urethra quality index for analytical data i mean only for two categories um i would like to share with you some lessons learned thanks to this work the first one is that making such a system requires to make compromises if you define two precisely questions and if you have too many questions you may have a good robustness but performing a data quality assessment would be extremely time consuming on the other hand if you have excessive simplicity uh on the very limited number of questions well misinterpretation will be possible we would obtain a quality score but if its repeatability and reproducibility is very bad then it would be pointless another challenge is updates and extension today gamma applies only to vitamins well ideally it should be extended to other nutrients than vitamins and gamma should also be updated another kind of update that could be planned is to propose a data quality assessment system for non-analytical data that wasn't done uh unpublished till now or at least not not finalized a third challenge is to promote the use of this system among compilers it's indeed rarely used we should explore the reasons why should we emphasize the importance of data quality assessment is the system too time consuming what are the difficulties encountered by users it would be an interesting options to collect more feedbacks from compilers by organizing round robin tests for example i conclude with the successes obtained from this work well this system is the first attempt to make a common data quality assessment system for analytical data in europe before that we had different countries having a different system and we also had a majority of countries having no system of a vibrato the second success is that uh well this system is associated with gamma and thanks to it it was possible to overcome the difficulty of assessing aspects related to laboratory analysis and to conclude i will emphasize the benefit of this system that we had not planned in fact the list of questions uh we have made is a very useful checklist when designing a food composition study uh when preparing a report a scientific paper describing new data so thanks to AFIRO for its invitation to attend this for listening thanks also to all contributors of the work presented today you can see the name of the various articles of uofia publication and this world cloud and the uofia quality for quality assessment system for analytical data can be found on the uofia website here is a long link thank you hello thank you so much marina for this presentation and may i invite sundra to to give her perspective from a user perspective which i think it's very interesting because we as compilers we are always so uh bringing so much importance to the to the quality of the data that we are presenting and sundra will walk us through some of her perspective of how this would be used or if it is thanks i need permission to to share my screen actually perhaps marina second yes i have difficulties to close the window i'm afraid i should quit and and be back so as to let you um so i so as to let you present sundra sorry um it's on the on the on the top it is when you have two screens it can be on the other screens away yes but uh i just um uh have lost the window um the zoom window so i'm back sorry no yes can you as an administrator uh change that you yeah now i think i've managed yes okay sorry so let's see um i should do the other screen this did not happen when we tried can you see my screen now hello can you hear me because i cannot see you everything is perfect okay great so uh thanks a lot wrote for the invitation uh it's my pleasure to share this presentation with you this morning for i use a perspective how our quality food composition you use it and what's missing i will be talking uh in reference to the evaluation systems that have been presented but also i will give a bit of perspective on the quality of food composition data overall use it uh as a user before i proceed with the topic itself i would like to define who is the user i'm going to be talking about it in this presentation uh as you may know food composition data attend a variety of courses including different user groups and this may include different backgrounds such as nutritionist dietitians but also economists journalists and many other specialists these groups may have common needs but also specific needs we may be interested in estimating nutrient contents of foods but also for the labeling nutritional claims any other purposes as such i'd like to define here with you that my view my user perspective is for nutritionist dietitians who performs dietary assessment especially nutritional service i will also be talking a bit about individual level assessments but i cannot extrapolate much more than that because this is my experience i also delight to analogy the following with you uh as you may know dietary assessments or overall assessments on on diet uh they have an step of food matching linking for consumption with composition data and this varies from practice to practice how this is done in this variation we'll have implications on data quality i think this is very important especially with systems with software or assessments that we use software so we pre-defined matching pre-incorporated matching between consumption composition data because this really often implies less quality of the data and this is very common at individual level assessment around the world other assessments we do the posteriori matching which allows to improve the data quality it's also more challengeable but it's common at the population level in nutritional service furthermore i would like to acknowledge that as a user of food composition data in dietary assessment very often this food matching this linkage is not done by food composition cables specialists and in practice i rarely see evaluation systems being considered in terms of evaluation systems you have seen in this uh this morning or afternoon depending on where you are uh evaluation systems uh that usually are focusing on analytical data uh possibilities exist but in practice lia is done to recognize the quality of food composition data in the final dietary information for nutritional service there is a long way uh between identifying the table the data that you want to use and reporting the final values and i'd like to invite you to reflect on this because in fact i can't find a good table to be used but in practice in practice uh in dietary assessment we are mixing food composition tables and no evaluation systems exist to judge the user calculated and estimated values so i would say that there is a gap here and uh perhaps in foods evaluation system is considering uh the user perception perspective and uh is taking into account some of these needs but i also think that before uh we decide which food composition papers available in the report what's the real number at the end there is a long way of aspects to be considered and then uh i'd like just to highlight some fundamental questions that usually we should ask or we ask ourselves while um performing these assessments neither for the matching option with predefined values already measured or doing after we collect the data these questions they need to be asked the difference is that with predefined matching little can be done actually to improve the product of the data and when you are doing this afterwards you allow yourself to make more conscious choices and more can be done to improve the product of the data so usually we will ask ourselves if there is a national representative table available and then a question that usually is hard to answer and perhaps now with the evaluation system will be easier it is if this table is of good quality uh because it's not so easy to evaluate that so we would like then probably to prefer to use uh one with an evaluation system in place and another question that i think is important for us as users uh to ask ourselves if the food composition table is of no use because sometimes it's not easy to define this is a good table but it's easier to actually identify bad compilations or bad food composition tables that will not add good quality of the data for us and here i'd just like to give an example that also i think should could help us identify the needs of users is that in brazil as you may know we have two more than one brazilian national table these two taco and tbca are considered of good quality a lot of nutritionists and other backgrounds the professionals they do use these two tables but one is providing data more in the analytical level uh values uh and then if i take the example price i have rice type one cooked and that's it and tbca is a compilation where we'll consider uh preparations and the rice will have the addition of oil onion garlic and salt and as such uh contents of lipid and sodium of course will be higher in the tbca table and as a user i will prefer to use for dice assessment the tbca values and this is not always understood it's not always done in the country uh but there is something we need to identify uh in order to to proceed with the analysis another question we would like to ask considering that more than one food composition table is usually used in our assessment is if there is an order of preference and the answer is yes and of course we would like to go with more quality first less quality after this is not always is to to to do it uh perhaps now with evaluation systems in place uh we could evaluate that and make better choices i also need to ask myself when using food composition tables for that assessment is uh if i can use how can i use the values from different tables and a lot of uh assessments may be using uh information interchangeably uh without consideration the need for armonization and standardization this is the problem that i usually see in practice and i think we should avoid we need more understanding that nutrient definitions for fortification practices are different from country to country modes of expression they need to be uh also standardized and in all other aspects that are highlighted in these evaluation systems you have seen presented this morning i will not talk much about about that more about that uh but we have the food concourse and the it file inputs uh platform that uh as you may know more than i than i know uh covers very well this topic and the last question i think we should ask ourselves or maybe not the last one but an important one to be asked is uh depending of the nutrient of the interest well how is the coverage of the table i'm in working with uh do i have missing values uh should i borrow and is um or estimate it a nutrient values for all foods this is a task that can be done and uh should be reported more often here i bring you the example of Caribbean survey we are performing at the moment this is preliminary and uh preliminary report but uh so far from the foods we have been uh assessing uh most of the foods most of the nutrients they have a good coverage above 90 80 percent of the nutrients or the dietary components and there are components such as diet um carotenoids uh which will not have a good cover and therefore values will be underestimated and this will have implication in the quality of the data um here's a bit um cut it but uh because i was asked to talk about uh how data quality is used i also like to stress that although we don't have uh evaluation systems in place to assess uh what we are using there are some guidelines uh specifically from strobenut released in 2016 which guide us would give us some guidance on what we should minimally report when presenting our survey results specifically related to food composition data we have four guidance the first one is a very interesting one which uh suggests us to describe and justify food composition data usage explain procedure to match food composition with consumption data and describe the use of conversion factors even uh mention uh that we should also have clear the number of missing values in food composition and how these were treated there are three other guidances uh one to describe any statistical method used to combine dietary or nutritional data if applicable is the one i find very interesting important that limitations in food composition data should be described as well uh as you may may be aware most of the dietary assessments uh that use food composition data that report nutrient intake do not report much on the food composition data usage we barely see the name of the food composition table use it and more can be said about the whole exercise in order to give us some uh more understanding of the quality of the data unless they also suggest to provide data collection tools and data as online material and this would be the case for the food composition data itself so going towards the end of my presentation um i also do like to mention two complex aspects that i think we should consider in the assessment of the data quality as far as we are combining data from consumption and composition uh diverse descriptor systems are used between the service and this can hamper the nutrient values achieved in dietary assessments you have seen that all evaluation systems presented this morning they had some evaluations on food description and while we can guarantee that from the composition part we cannot um guarantee that from the consumption part we have a lot of reporting that will come in a very simple way i have seen people only reporting for instance fish without mode of preparation uh but you also should have the type of the fish you can also give more explanation on how was cooked if it was fried and what type of oil was used type of fat and uh this kind of differences in food description uh certainly will imply um different quality of data and then just to reflect on uh if a bad value is presented will this be due to the composition or to the lack of information provided in another aspect is the recipe desegregation dilemma food nutrient values can be hampered by the lack of these details if more recipes preparations are needed after the survey how can you guarantee the data quality if i as a user need to do more recipes because this was reported in my survey uh how can i guarantee that this was done as the other food composition tables did i think we need guidelines for that we also need cooperation with compilers and there is also the need for compilators to share what's used behind the calculations and to finalize reflecting what's missing i think that is uh it is missing transparency and used procedures in food matching especially from pre-existing matching to enable us to assess the data quality i also think we need more transparency on the reporting of the food composition data used in the assessments and the need for evaluation system to assess final nutrient values in nutritional service because i believe there is a long way before you define the table or when you define the table to find the reported values with complex aspects to be considered uh that we will not guarantee the quality of the data at the end if you've not considered that thank you very much thank you so much i would really like to thank all the presenters they have done a tremendously good job in presenting what we as compilers think and how we assess the quality either of the analytical data of the whole database and then from the user perspective of how this is used and and probably yes we need to go further but this is then probably not for the food composition community to do but it is more for other communities to do like the dietary assessment community so and so really thanks to all the presenters and i also would like to thank Sol Ruiz who is our IT specialist and who is making this webinar possible so thanks a lot to her as well and i would like now to go to the part of the question and answers so we have some questions the first one is from the asian quality assessment is it publicly available and where you are unmute uh you have to unmute Kuntit okay now at present we need not put into the online system but if the one that needs to have the the template so we can send to to the link to the mail and in the future i think we encourage to to use the input FAO framework rather than the asian system so um also another question for Kuntit and Papasi um you presented there were some asian food composition tables which did not score that well where would they need improvements okay according to the the the score that the food consumption table that has low score we found that the number of samples to be analyzed is one of the of the important factor and and also the copper nutrients also give the low score and the full description and the percent of analyzed data if it is less the they will get low score and also the document to the user lacking of document to the user and detailed information of food composition database so these are the causes of the low score the low score of quality score so if you can if you consider improve those cost factor the score will be increased okay and and i think if you look at the criteria for evaluation you can find a way to increase quality in your food composition table or database okay thank you so much Kuntit and Papasi um there was a question if it is possible to join the evaluation process in in Israel and i think the evaluation frameworks they are done so you can use them to improve your own food composition table in any country so it is done in a way that it is universally applicable if it is the analytical data evaluation or if it is the evaluation of the whole food composition table so but these processes are for the moment they are closed and they are well the oil refuel one is published and the the other ones are going to be published then there is a question to marine uh where to find gamma is it published so if it is published can you in the chat probably put the link yes i'm doing it right now so you'd have to answer yes and then there is another question what is the difference between euro fear in the inputs guidelines probably marine you would like to answer that one okay um the euro fear guidelines um at the level of a unique analytical value where the FAO guidelines uh i mean the FAO quality framework presented by ruse is at the level of a food composition database as a whole so it's a very different level uh one is very close to the data the other one is at a higher level so it is very different so you know if a food has let's say 30 nutrients you have to do this evaluation for each of the 30 values of this one food and then you go to the next one and this is why it is so extensive and then there was another question so the euro fear system is interesting but it is very time consuming so it could help to to simplify it and probably also that it will always be done by two people so what is your opinion marine um yes i understand that it's um that simplification is needed but we we have had difficulties finding the right balance between simplification and possibility of misunderstandings that's the difficulty of making a compromise but i'm sure that it could be it could be improved based on your on your recommendation and concerning your other suggestion with which was to um ask two different compilers to perform the assessment yes i think it's a it's a super idea it's a it's more or less the round robin one robin test with only two compilers but it's true that it will increase the time spent on the evaluation so it's it will increase the robustness of the of the evaluation but it was also it will also increase time spent so we should find the compromise and maybe keep on working on the subject and probably you can share how many of the existing food composition tables and databases in europe have used this system for all the data and published it in europe i think it's less than five so it's not um and some use it in in an adapted way so it's still rarely used because it's too time-consuming so yeah okay thank you um then there is a question uh for sundrum so um it is valuable to develop an ongoing automatic data quality assessment including assessing beyond the food food composition part so would you like to make some comments on that yes i do think it's valuable i do think we need to bridge the two fields and or at least speaking from the consumption work when we receive this data and we are using different manners or mixing tables uh i do think uh is harder any practice is not done to assess the quality of these these sources of information so on an automated uh well not necessarily an automated but a system that will go beyond uh the analytical data that i think a bit of as input has been proposing i think it's already a good step forward but also perhaps considering other aspects that will be uh important for our choices when because we are usually linking our matching information at the food level uh using different tables as i said so i think assessing more uh this step would help us to provide more quality of the data because in practice a lot of mixing of information is happening i know that those who are at the national level doing uh already trained by eurofear or many other frameworks have a good knowledge on that but a lot of assessments across the world in against speak for brazil uh are hampered by the lack of understanding of these of how this data is prepared what do i do with this data that was told to me that it was of good quality but now i'm mixing with other tables that is often not good quality uh so i think that is a gap there that needs to be filled that somehow not sure for by uh if by food composition uh but the food composition field but certainly communicating with the food composition field yeah so meaning that we need in the future probably much more collaboration between the dietary assessment and the food composition expert to come up with some uh some ideas on how to uh to evaluate like we you said you know we did in the uh in the evaluation framework that we proposed for published food composition data and analysis and databases to see how what where are the important steps and you mentioned some of the food matching and this was also we have had a very nice seminar last week of not last week on the 30th of march on food matching because food matching is really a highly important step and i think in the quality assessment of the nutrient uh uh intakes it's not that uh that well considered and i think there are other steps which we need to consider so for the moment we have some guidelines which are separated and and separated people are looking at the set and this at the things differently so probably this is one of the things that would be needed to be done in the future having said so sundry you have some more questions so dr sundry about the name displayed for food which includes the preparation of food should it also consider the yield factor and or the retention factor used to calculate the nutrient items thank you so if the question is not only that you do a good matching on the food level but you check that the data which is behind is also of a good quality this is i think what the person would like to ask yeah if i understood correctly uh i guess um defining this or identifying this need for including these factors depends of what kind of information you are linking or you are looking to use usually you want to look into those aspects because you need to consider the amounts that will be linked and how was reported in your survey and retention factors are important but i think it all depends how the data are presented in the table if that has already been considered in the food composition part you don't need to do uh that again in the consumption part or i mean depending on how you are reporting your data and that varies so much from system to system the the dietary assessment should be able to handle that and the food composition data should already be should should have already taken that into account when reporting information on 100 grams per consumed food for instance uh but we usually find useful that food composition tables provide that information because you maybe use it for further calculations you know not all food composition tables are presenting those values yeah so if i understand correctly you wanted to say that you you would choose a food composition table who has demonstrated that they are applying the yield and nutrient retention factors correctly um compared to a food composition table who did not demonstrate that or which did not yield exactly so then we have another question very often when aiming for exact match for food matching researchers need to do new recipe calculation the additional challenge is the lack of good databases of yield factors and retention factors for reliable recipe calculation so how do we check for quality in recipe calculation and this is probably a question for everybody of us so probably kunted and papasi would you like to to start um you're still mute no you don't want most of our data is some analyzed data we haven't borrowed from anyone so we have no experience on food matching yes so what the kunted uh what kunted but papasi is just saying is that there are some tables who are just analyzing the data they are not calculating so most of the the values that you will find in these tables they are only analyzed therefore this would not be applicable other tables are much more um um adding some calculated values like for cooked foods of recipes so marine what would you say on on this issue well in the french food composition table uh table the number of data obtained from recipe quest uh quality uh calculation is very very low i think we have less than maybe 20 foods but what i would say concerning recipe calculation is that yes we know that we need factors for i mean yield factor and nutrient retention factors but i suppose we should also pay attention to the way the recipe was made i i mean uh sometimes we have um uh we know that someone has um has consumed food that is a recipe but we do not have the associated recipe so we have to search for a representative recipe to match with the with the name of the recipe we have because we have some only the name of the recipe and i think um um um maybe regarding the number of source considered um regarding the type of source considered of obtaining list of ingredients and uh cooking methods um there's something to well uh to quality assess also in this um in these domains not only in calculation mm-hmm yeah thank you and i would like to add that for example in the uh best african food form food composition table that was published last year we uh all the recipes uh which are used in the um in the food composition table they are with the ingredients they are all published there uh with the ingredients with the amount with the yield factors and we have a whole list of more than 400 uh foods with their uh yield factors and we also publish the retention factors which was used so and this is some of the things that we would like to see in every published food composition table because it helps a lot to understand the quality of the food composition table and and gives more information to the users of which yield factor was used if it is appropriate for them uh which retention factors which ingredients of each recipe and i think that would be helpful and probably really in the future we should think about a a nice database for yield factors which would be available all for for everybody we already have some which is from bogner and from leona verkstrom but it is not very accessible so probably this is one of the things uh some people can think about that one in the future and i guess it would be really extremely important to have that um but uh again you know look at the food composition tables that you want to use and see if this is part of the documentation that they are publishing and this is also part of the things in the evaluation framework proposed by uh power and inputs that this is always be given so this is one of the important aspects of the documentation and it should really be there sundry you have a thought yeah i just would like to stress i have said that my presentation but how important for a user is to have access to that information to be able to do new recipe calculations that are always needed we cannot assume that all recipes are going to be there for your use and if you think about modes of preparation different modes of preparation that will happen so often and we need that we need to change one ingredient and you are limited to to that information if there is no transparency about how how was done and once this is displayed the factors and even the way which recipe method was chosen to calculate those recipes it will help us to have complete assessments with new recipes that are always needed and just to reflect a bit uh on the caribbean example we are having now with root we have noticed that a lot uh the need for an evaluation system for recipes quality control uh root has some ideas on checking uh water content of recipes that are always a problem uh there is room there for improvements uh it's a room for improvements from both sides composition and consumption fields thank you so uh there is another question is there a data quality evaluation guideline system for food data taken from scientific publications or journals so the answer very short is yes you can use the or of your standard uh then there is another question for or of your data quality assessment at what range of total quality index should a paper meet for it to be included in the food composition data in case if the value is below 33 because you say the uh quality ends at 33 the question is uh on the limit um that allow access to to to to publication well in fact we can also define other limits i other limits i mean um there could also be a limit for inclusion within a data bank so um we can we we have not yet decided uh we have not yet made a decision on what um on what is what the limit should be for inclusion in a scientific paper because we think um it's up to the reviewer also to to make an assessment what i would say is that um the list of uh criteria we have in the ufia quality system is a checklist then depending on the use you want to make uh on the specificity of the data suppose you have totally new data on the food that was never analyzed maybe it it can have a bad score on a category but the data can still have an interest if you do not have better data so we haven't decided about a limit and i think it's not up to ufia to to decide uh that and it's the same for inclusion in uh in a national database if you have nothing better than what you have collected in some cases you may still want to use this data because um suppose you don't have any data people who are uh doing um assessment studies they still need a value and they and you will have to produce a value so um maybe it's better to use what you have instead of um while letting users alone and using maybe a zero or the the or the mean or the median of the this nutrients in this food category so i would say that defining limits um well first is it's difficult and maybe it's not always appropriate so that is the eternal difficulty so we aim for the highest quality and sometimes we oblige to include data of less quality and therefore it is so nice that if we would have the possibility to indicate that this is a data point that is included for because there is nothing better available um yeah because we need to make progress but we also need to make progress in the highest possible um uh quality um we are almost there and there is um um there are some comments and a lot of comments they say you know really thank you so much very informative i truly enjoyed very refreshing so a lot of people really enjoy um the the presentation that we have seen today and one person says yes please i did not see the the link to gamma yet so if you can do that before we we close um so again you know really thank you to all the presenters for really excellent presentations and to the audience for staying with us for such a long time we have had 111 person joining us in in this webinar so thank you so much for for all of you and i hope that you found it interesting and quality of data is really always important so this is why this seminar is really very important and i would like to ask everybody to uh to have some last comments before we close couldn't it and Papasi would you like to start okay i think that we we can learn from the webinar that i think the quality of the system is very important part for the good quality database for the user so it means we need to emphasize on the quality in all the steps of the making the the food composition database Papasi you would like to add something if you want to have a good quality food composition database just study the criteria for evaluation and follow that study the the highest marks the highest marks for the evaluation so follow that so you will get a good mark and become like a really helpful food quality food content database yeah marine well i think the way to quality is a long way it's a time-consuming way but as as shown by Ruth in her presentation it's the basis of very important conclusions conclusions so i would like to motivate you to invest in this topic for for the benefits of the results that will be produced based on the food composition values Sandra i completely agree with marine quality it's a long way to good quality data um i invite you i invite everyone to actually look into that from the beginning to the end of the assessments especially those who are going to use the data afterwards uh so for digital assessment or any other purposes there are more things to be looking looking to act and we are not always doing but we should do it and let's make use of these evaluation systems because i think they are a great start of understanding what we are receiving what we are getting yeah in terms of quality yeah so thank you so much and i can see that we have now the gamma web link in the chat so those who want to use it can download it from from the chat and i would also like to say that quality of data is important so if everybody who is working on some part of anything is really making sure that their part is of highest quality and if everybody would do the same so i think we would really move towards the direction where we want things to be done so that every part of a dietary assessment every part of a food composition table starting from the sampling the the generation of the food the the compilation and the presentation um and the publishing i think we would really make a lot of of progress and there is also a question where the evaluation framework will be downloadable so it will be soon available on the inputs website so i think it will take another two or three months and it will be available on the inputs website so you will find it there and it will be like everything else that we do in in our inputs it will be free of charge and having said so i would really like again to thank all the presenters and also to thank Sol for her assistant and for all your great questions and and comments and i really hope that in the future everybody will contribute to a higher quality and that we will and i'm looking forward to that one so thank you so much and with this one i would like to to close our very interesting webinar and again thanking everybody for contributing