HAL Id hal-01539470httpshalarchives-ouvertesfrhal-01539470

Submitted on 26 May 2020

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents whether they are pub-lished or not The documents may come fromteaching and research institutions in France orabroad or from public or private research centers

Lrsquoarchive ouverte pluridisciplinaire HAL estdestineacutee au deacutepocirct et agrave la diffusion de documentsscientifiques de niveau recherche publieacutes ou noneacutemanant des eacutetablissements drsquoenseignement et derecherche franccedilais ou eacutetrangers des laboratoirespublics ou priveacutes

Distributed under a Creative Commons Attribution - NonCommercial - NoDerivatives| 40International License

From the SAINLIM system to the SENS algorithm areview of a French approach of nutrient profiling

Marion Tharrey Matthieu Maillot Veacuteronique Azaiumls-Braesco Nicole Darmon

To cite this versionMarion Tharrey Matthieu Maillot Veacuteronique Azaiumls-Braesco Nicole Darmon From the SAINLIMsystem to the SENS algorithm a review of a French approach of nutrient profiling The Proceedings ofthe Nutrition Society 2017 76 (3) pp1-10 (First view) 101017S0029665117000817 hal-01539470

Nutrition Society Summer Meeting 2016 held at University College Dublin on 11ndash14 July 2016

Conference on lsquoNew technology in nutrition research and practicersquoNutrient profiling as a tool to respond to public health needs

From the SAINLIM system to the SENS algorithm a review of a Frenchapproach of nutrient profiling

Marion Tharrey1 Matthieu Maillot2 Veacuteronique Azaiumls-Braesco3 and Nicole Darmon141UMR NORT (Uniteacute Mixte de Recherche ndash Nutrition Obesity and Risk of Thrombosis) Aix-Marseille Universiteacute

INSERM INRA 1260 Marseille France2MS-Nutrition Faculteacute de meacutedecine La Timone AMU Marseille France

3VAB-nutrition 1 rue Claude Danziger Clermont-Ferrand France4INRA 1110 CIRAD SupAgro CIHEAM-IAMM MOISA (Markets Organizations Institutions and Strategies of

Actors) Montpellier France

Nutrient profiling aims to classify or rank foods according to their nutritional compositionto assist policies aimed at improving the nutritional quality of foods and diets The presentpaper reviews a French approach of nutrient profiling by describing the SAINLIM systemand its evolution from its early draft to the simplified nutrition labelling system (SENS) algo-rithm Considered in 2010 by WHO as the lsquoFrench modelrsquo of nutrient profiling SAINLIMclassifies foods into four classes based on two scores a nutrient density score (NDS) calledSAIN and a score of nutrients to limit called LIM and one threshold on each score Thesystem was first developed by the French Food Standard Agency in 2008 in response tothe European regulation on nutrition and health claims (European Commission (EC)19242006) to determine foods that may be eligible for bearing claims Recently theEuropean regulation (EC 11692011) on the provision of food information to consumersallowed simplified nutrition labelling to facilitate consumer information and help themmake fully informed choices In that context the SAINLIM was adapted to obtain theSENS algorithm a system able to rank foods for simplified nutrition labelling The imple-mentation of the algorithm followed a step-by-step systematic transparent and logical pro-cess where shortcomings of the SAINLIM were addressed by integrating specificities offood categories in the SENS reducing the number of nutrients ordering the four classesand introducing European reference intakes Through the French example this reviewshows how an existing nutrient profiling system can be specifically adapted to support publichealth nutrition policies

Food Nutrient France Labelling Nutrition information

This review illustrates how an existing nutrient profilingsystem can be specifically adapted to support publichealth nutrition policies with the example of a Frenchapproach of nutrient profiling The present paper isdivided into four sections The first section introducesthe preliminary indicators to estimate the nutritionalquality of foods which laid the foundations for the

French SAINLIM (score of nutritional adequacy ofindividual foods (SAIN) score of nutrients to be limited(LIM)) system The second is devoted to the presentationof the SAINLIM system including the context in whichit was generated the calculation principle and itsimprovements The third explains how a nutrientprofile originally designed to help regulating health and

Corresponding author N Darmon email nicoledarmoninrafr

Abbreviations AFSSA French Food Standard Agency DRV daily recommended value EC European Commission ED energy density FampVfruits and vegetables LIM the score of nutrients to be limited MRV maximal recommended values NDS nutrient density score SAIN score ofnutritional adequacy of individual foods SENS simplified nutrition labelling system

Proceedings of the Nutrition Society Page 1 of 10 doi101017S0029665117000817copy The Authors 2017

Proceedings

oftheNutritionSo

ciety

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0029665117000817Downloaded from httpswwwcambridgeorgcore INRA - Versailles-Grignon on 13 Jun 2017 at 140250 subject to the Cambridge Core terms of use available at

nutrition claims was transformed to an operational sys-tem for simplified nutrition labelling in the context ofboth European and French legislation on the provisionof food information to consumers The fourth sectionsituates the simplified nutrition labelling system (SENS)in the international context of food labelling by empha-sising its specificities in relation to other nutrient profilingsystems underlying simplified nutrition labels launchedworldwide

Emergence of indicators to estimate the nutritionalquality of foods the nutrient density score and the score

of nutrients to be limited

In 2005 the Dietary Guidelines for Americans recom-mended that consumers give priority to nutrient-densefoods defined as those with a high nutrients-to-energyratio and then consume discretionary energy dependingon their energy needs(1) Although the concept of nutrient-dense foods was well understood as this time no uniformstandard of nutrient density existed In this context a jointFrancendashUS research team developed a scoring system toestimate the nutritional adequacy of foods Precursor ofthe SAIN the nutrient adequacy score was defined asthe mean of percent daily values for sixteen nutrientsexpressed per 100 g of food but also per 418middot4 kJ(100 kcal) as a nutrient density score (NDS) or per unitcost (nutrient-to-price ratio)(2) The idea that food pricesand diet costs may be one factor limiting the adoptionof healthier diets especially by the low-income consumerand may contribute to the observed socioeconomic dispar-ities in health(3) was reinforced by this innovativeapproach that put forward the relationship between nutri-ent content energy density (ED) and food costs The newconcept of nutrient profiling appeared therefore as a rele-vant tool to rank foodstuffs according to their contribu-tion to a balanced diet and better consider the issue offood costs and nutrient-to-price ratios(4) A scoring systembuilt on two indicators was developed (1) the NDS basedon twenty-three qualifying nutrients (ie positive nutri-ents) (2) the LIM based on disqualifying nutrients

To identify the optimal base of calculation indicatorswere expressed by 100 g or 418middot4 kJ of food or per stand-ard serving sizes Results showed that indicators basedon serving sizes and on 418middot4 kJ were preferable forpositive scores(5) Since negative scores incorporatedenergy-providing nutrients (ie macronutrients) it waspreferable to express them per 100 g of food in orderto not penalise foods with a low ED (such as fruit andvegetables (FampV))(5) In the absence of standardisedEuropean serving sizes the team therefore decided toexpress the NDS per 418middot4 kJ while the LIM wasexpressed per 100 g of food

The NDS summarises the positive aspects of the foodby assessing the percentage of adequacy with the recom-mendations for essential nutrients It was defined as anunweighted arithmetic mean of the percent adequacyfor the qualifying twenty-three nutrients present in thefood composition tables and for which a daily

recommended value (DRV) existed(6) The generic equa-tion is

NDS =sum23

p=1 NutpDRVp

23times 100

where Nutp is the quantity (in g mg or μg) of nutrient pprovided by 418middot4 kJ (100 kcal) of the food and DRVpthe French corresponding DRV (mean for men andwomen) expressed in the same unit as Nutp

( 7)The LIM score summarises the unfavourable aspects

of the food It was defined as the mean percentage ofthe maximal recommended values (MRV) for three dis-qualifying nutrients that should be limited in a healthydiet sodium simple added sugars and SFA(6) The gen-eric equation is

LIM =sum3

p=1 LipMRVp

3times 100

where Lip is the content of limiting nutrient p in 100 g ofthe food and MRVp is the maximum recommended valuefor nutrient p expressed in the same unit as Lip TheMRV for SFA and added sugars correspond to 10 ofthe recommended energy intake of 8368 kJ (2000 kcal)The MRV for sodium is 3153 mg corresponding to adaily intake of 8 g NaCl

NDS and LIM were found to accurately characterisethe nutritional quality of individual foods(6) Thusmedian NDSLIM values of foods selected in diets mod-elled by linear programming increased with increasinglystringent nutritional constraints included in the modelsie with increasing nutritional quality level of the mod-elled diets(6)

Additional consideration of energy cost (ie the cost ofenergy) disclosed that foods(6) and food groups(4) differwidely in terms of nutritional quality and cost Com-bined together NDS LIM and energy cost were usedto identify foods with good nutritional quality for theirprice revealing that those foods must be preferentiallyselected to obtain healthy diets at a low cost(6) Theseresults put forward strong evidence that effective dietaryguidance must take into account both the nutrient profileof foods and their nutrient and energy costs to allow con-sumers to identify and select optimal foods at an afford-able cost

Two-score nutrient profiling system the SAINLIMsystem

The notion of nutrient profiling appeared in the early2000s fostered by a widespread consumer interest fornutrition information More aware of the link betweendiet and health they strongly expressed the desire toget informed about the nutritional properties of thefood they eat Improvements in nutrition labellingand consumer communications policies on health claimsappeared as a solution to make the existing point-of-purchase environment more conducive for healthychoices To avoid misleading consumers about theoverall nutritional quality of foods the European

M Tharrey et al2

Proceedings

oftheNutritionSo

ciety

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0029665117000817Downloaded from httpswwwcambridgeorgcore INRA - Versailles-Grignon on 13 Jun 2017 at 140250 subject to the Cambridge Core terms of use available at

Fig 1 From the SAINLIM (score of nutritional adequacy of individual foods (SAIN) score of nutrients to be limited (LIM)) system to the simplified nutrition labelling system (SENS) algorithm

AFrench

nutrientpro

filingsystem

3

Proceedings of the Nutrition Society

httpsww

wcam

bridgeorgcoreterms httpsdoiorg101017S0029665117000817

Dow

nloaded from httpsw

ww

cambridgeorgcore IN

RA - Versailles-Grignon on 13 Jun 2017 at 140250 subject to the Cam

bridge Core terms of use available at

Commission (EC) decided to define the eligibility tonutrition or health claims according to the nutrientprofile of each food(10) Considering the overall nutri-tional status of a food the main issue was to identifythe limiting nutrients for which dietary intake shouldbe reduced and positive nutrients for which increaseddietary intake is desired in terms of nutritional goalsof public health policies In this context the FrenchFood Standard Agency (AFSSA) initiated a workinggroup which ended proposing an original nutrient profil-ing scheme the SAINLIM system (Fig 1)(89) ThisFrench system was derived from the two afore-mentionedindicators the NDS and the LIM considering that posi-tive and negative nutrients are both needed to define foodhealthiness Primary threshold value was allocated toeach indicator hence defining four nutrient profileclasses Threshold for good nutrient density was set at5 given that a SAIN gt 5 is equivalent to 5 of nutri-tional adequacy for 418middot4 kJ corresponding to 100 ofnutritional adequacy for 8368 kJd (reference dailyenergy intake) For the LIM a high content of limitednutrients was defined by a LIMgt 7middot5 ie 7middot5 for100 g corresponding to 0 excess in disqualifying nutri-ents for 1337 gd (mean daily food intake observed in theFrench population)

The SAINLIM system calculation principle

The twenty-three nutrients considered in the previousNDS were included in the first version of the SAINdeveloped by the AFSSA SAIN23 Nevertheless froma practical perspective it would have been challengingto require this level of information for both administra-tors and economic operators Several expressions weretested for the SAIN score and the final one was limitedto five basic nutrients proteins fibre vitamin C calciumand iron and one optional nutrient vitamin D(SAIN5opt) Namely when the vitamin D ratio(VitD(μg418middot4 kJ)DRVvitD) was higher than the lowestratio among the five basic nutrients this lowest ratiowas replaced by the vitamin D ratio in the equationTo ensure the relevance of this selection differentSAIN were generated from 613 foods of a French data-base (INCA1) by randomly selecting five seven nineeleven or thirteen of the twenty-three initial nutrientsSpearman correlations were calculated between the simu-lated scores and the contents of each of the twenty-threenutrients in the table Results showed that there was nobenefit to include more than five nutrients in the SAINconsidering that the number of strong correlations waslittle affected by the number of selected nutrients(8)The choice of nutrients reflects a balance between theneed to include nutrients of importance to public health(inadequate intakes of iron calcium and vitamin C existin the French population) nutrients markers of key foodcategories that are subject to nutritional recommenda-tions (iron for meats calcium for dairy products fibreand vitamin C for FampV) and nutrients markers ofother essential nutrients (proteins in foods are usuallycorrelated with that of other essential nutrients such asvitamins B2 B3 B5 B12 iodine selenium or zinc)(8)

As well as NDS SAIN is a marker of the nutrient densityand is expressed per 418middot4 kJ of food The LIM is thesame as the one previously developed The score is calcu-lated per 100 g of cooked or rehydrated food Because anegative score based on 100 g is more lenient towardsdrinks due to their low ED(5) the LIM score was multi-plied by 2middot5 for soft drinks The weighting factor of 2middot5was chosen considering that a portion of 250 ml is rele-vant for liquids

Based on the SAIN and LIM values and on the thresh-old defined for each score each food was allocated to oneof four possible SAINLIM classes Foods with the mostfavourable nutrient profiles are in Class-1 and foodswith the least favourable nutrient profiles are inClass-4 In response to the European regulation on nutri-tion and health claims (EC 19242006) AFSSA proposedthat only foods present in Class-1 would be eligible tohealth claims and foods in Class-1 or Class-2 to nutri-tion claims(8)

Four years after the AFSSA report an Americanresearch team proposed a novel method to selectingand weighting nutrients for nutrient profiling of foodsand diets(11) Their statistical approach was based on cor-relations between nutrients intakes and an overall indexof dietary quality the Healthy Eating Index scoreApplied to the National Health and Nutrition Examin-ation Survey survey population the novel method ledto a model containing five qualifying nutrients (proteinfibre calcium unsaturated fat and vitamin C) andthree disqualifying nutrients (SFA sodium and addedsugar) Except for iron (not selected with the HealthyEating Index-correlation study) these nutrients turnedout to be the same as the nutrients included in the SAINLIM system despite using different approaches and popu-lations These results further strengthened the choice of thenutrients in the SAINLIM system and raised the possibil-ity of extending it beyond France and Europe

Improvement of the SAINLIM system

Vitamin D was first chosen by AFSSA as the optionalnutrient of the SAIN5opt score to correct misclassifica-tions for oily fish Subsequently changes were introducedto improve the nutritional relevance of the system Asdescribed by the AFSSA report analyses were conductedon the French food database (INCA1) to identify whichessential nutrients were or were not correlated with thenutrients included in the SAIN5opt score The underlyingassumption was that only nutrients strongly correlated(R2 gt 0middot5) with one or several of the six nutrients includedin the score would be represented (though indirectly) bythe score It turned out that all water-soluble nutrientsnot included in the SAIN5opt score were in fact stronglycorrelated with at least one nutrient of the score(Table 1)(8) In contrast none of the fat-soluble nutrientswere strongly correlated with nutrients of the SAIN5optexcept DHA which was strongly correlated with vitaminD Thus a more refined version was developed depend-ing on the lipid contents of the food SAIN5optDndashLip97For foods providing more than 97 of their energy aslipids four fat-soluble nutrients namely vitamin E

M Tharrey et al4

Proceedings

oftheNutritionSo

ciety

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0029665117000817Downloaded from httpswwwcambridgeorgcore INRA - Versailles-Grignon on 13 Jun 2017 at 140250 subject to the Cambridge Core terms of use available at

vitamin A α-linolenic acid and MUFA were used asoptional nutrients and up to two replacements wereallowed between optional and basic nutrients Forother foods vitamin D was kept as the only optionalnutrient(12) The next version of the SAIN namedSAINcat was proposed as the part of the Optimed pro-gramme(9) designed to help small food companies in theMediterranean areas for the promotion and nutritionaloptimisation of their products The aim was to derivefrom the SAINLIM system a more nutrient-sensitivesystem that could be used as a tool to predict the evolu-tion of the nutritional quality of a given food productunder the impact of a reformulation or a food processAs a special feature the SAINcat algorithm integratednutritional specificities of food categories with the intro-duction of the new notion that the optional nutrientshould be a category-specific nutrient(9) Based on an epi-demiological analysis of the contribution of the differentfood categories to the intake of nutrients in the diets ofFrench adults the following category-specific nutrientswere identified folates (for fruits vegetables andlegumes) magnesium (for cereals) riboflavin (for dairyproducts) niacin (for meat and eggs products) and vita-min D (for fish)

Practical use of the SAINLIM system in interventionand research studies

By classifying individual foods based on their overallnutritional quality the SAINLIM system is able to iden-tify foods that should be encouraged to promote healthyeating(1213) This is especially relevant when translating

the nutrient profiles of individual foods into concreteand quantified advice to promote overall nutritional bal-ance Therefore the scope of the SAINLIM system waslargely extended leading to multiple public health impli-cations It provided new insights into how diet costaffects consumer choice and diet quality and should becounted among the key socioeconomic determinants ofhealth(14) Among other things it was used to identifyfoods with a good nutritional quality for price(6) toexplore the impact of food price policies (subsidies andor taxes) on the expenditures and nutritional quality ofthe diet(15) or to evaluate the impact of food transforma-tions on final nutritional quality of food products(16) TheSAINLIM system was also used as a tool to support theOpticourses intervention that aimed to reduce socialinequalities in nutrition and health by improving thenutritional quality of food purchases in populationswith budgetary constraints Through participatory work-shops in socioeconomically deprived neighbourhoods ofMarseille (France) an educational tool named theGood Price Booklet was constructed Based on theSAINLIM ratio to define nutritional quality andnational food prices this booklet lists over 100 goodnutritional quality foods and their good price definedas the price below which a food of good nutritional qual-ity can be considered as relatively inexpensive (httpwwwopticoursesfr) In addition a 6-month interven-tion in retail shops was performed to increase the visibil-ity and attractiveness of those inexpensive foods withgood nutrition through shelf labelling and marketingstrategies(17) This social marketing intervention wasfound to improve food purchasing behaviours in

Table 1 French Food Standard Agency assessment of the nutritional relevance of the SAIN using Spearman correlations between the sixnutrients composing the SAIN5opt and eighteen other nutrients(8)

SAIN5opt components

Water-soluble nutrients Fat-soluble nutrients

Protein Fibres Vitamin C Ca Fe Vitamin D

Water-soluble nutrientsVitamin B1 0middot45 0middot49 0middot54 0middot24 0middot54 minus0middot13Vitamin B2 0middot64 0middot10 0middot34 0middot50 0middot49 0middot09Vitamin B3 0middot63 0middot25 0middot34 minus0middot08 0middot58 0middot03Vitamin B5 0middot53 0middot29 0middot49 0middot33 0middot51 0middot08Vitamin B6 0middot46 0middot44 0middot59 0middot25 0middot60 minus0middot05Vitamin B9 0middot19 0middot69 0middot72 0middot47 0middot58 minus0middot09Vitamin B12 0middot58 minus0middot45 minus0middot20 minus0middot07 0middot21 0middot42K 0middot40 0middot61 0middot71 0middot44 0middot68 minus0middot19Mg 0middot46 0middot59 0middot57 0middot54 0middot70 minus0middot20I 0middot53 minus0middot29 minus0middot12 0middot42 0middot02 0middot38Cu 0middot27 0middot57 0middot50 0middot18 0middot73 minus0middot13Se 0middot70 minus0middot23 minus0middot10 0middot04 0middot31 0middot34Zn 0middot73 0middot10 0middot16 0middot40 0middot54 0middot02

Fat-soluble nutrientsDHA 0middot46 minus0middot39 minus0middot24 minus0middot22 0middot06 0middot61ALA 0middot09 0middot13 0middot16 0middot15 0middot14 0middot09Vitamin E 0middot02 0middot30 0middot33 0middot04 0middot20 0middot06Vitamin A 0middot05 0middot27 0middot40 0middot41 0middot23 0middot23MUFA 0middot03 minus0middot50 minus0middot50 minus0middot27 minus0middot31 0middot30

Nutrients are expressed in percent of their daily recommended value (per 418middot4 kJ) after log transformation of the variables Correlation above 0middot5

A French nutrient profiling system 5

Proceedings

oftheNutritionSo

ciety

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0029665117000817Downloaded from httpswwwcambridgeorgcore INRA - Versailles-Grignon on 13 Jun 2017 at 140250 subject to the Cambridge Core terms of use available at

disadvantaged neighbourhoods In the context of foodlabels an experimental study on consumersrsquo beliefsabout the tastiness and healthiness of food items (esti-mated from the SAINLIM system) revealed that simplenutrient information influences consumersrsquo grocerychoices(18) More recently the SAINLIM system hascontributed to a better understanding of the relationshipamong the three dimensions of sustainability in foodsenvironmental impact nutritional quality and price(1920)

A new nutrient profiling system operational for foodlabelling in Europe the SENS system

In Europe the European regulation (EC 11692011) onthe provision of food information to consumers conveysEU rules on general food labelling and nutrition label-ling(21) Individuals are influenced by various contextualfactors that influence their food purchasing behaviourwhich include nutrition labelling on food products(22)Interpretive labels have been proposed as a potentiallyeffective approach to assist consumers in choosinghealthier products(23) This approach requires an effectiveand suitable nutrient profiling system able to rank foodsaccording to their nutritional composition(24) Becausethe European regulation has not stipulated which nutri-ent profiling system should be used to implement label-ling several nutrient profiling systems have beenproposed Among them a group of experts and membersof French food retailers and industries initiated the devel-opment of the simplified nutrition labelling system(SENS Fig 1) in line with the European regulation(EC 11692011) fostering multi-stakeholders and collab-orative interventions(21) The SENS algorithm isdescribed (N Darmon J Sondey V Braesco et alunpublished results) and in a report of the FrenchAgency for Food Environmental and OccupationalHealth amp Safety(25) Adaptations from the SAINLIMto the SENS were introduced progressively followingan iterative process described in detail in a dedicatedreport(26) The implementation of the algorithm followeda step-by-step systematic transparent and logical pro-cess as encouraged by WHO in the guiding principlesmanual and methodological framework for developingnutrient profiling(24) The main steps were as follows

(1) Define the purpose of the nutrient profiling modelThe SENS aims to classify food according to theiroverall nutritional quality both between and withinfood groups as well as between similar foodproducts

(2) Decide whether to use an existing model or to developa new model It was decided to derive the SENSfrom across the broad SAINLIM system in viewof its many advantages Hence the SAINLIMwas developed by independent experts under theaegis of the AFSSA It was formerly recognisedby the French authorities as a relevant system inthe context of the European regulation on nutritionand health claims (EC 19242006) In addition theSAINLIM system was initially developed to be

adaptable and improvable as described in theAFSSA report(8)

(3) Decide on the scope of and exemptions to the modelIn accordance with the European regulation (EC11692011) the SENS algorithm mainly concernspackaged foods but it can be calculated on allfoods including non-packaged ones (fruits fishmeat etc) Alcoholic beverages are not in thescope of the system

(4) Decide the number of food categories that will beused in the model (if any) Category-specific nutri-ent profiling systems including the specificities offood categories were previously found to be moreeffective in promoting an achievable healthy dietprovided that the number of categories is moderateand an across-the-board-approach is maintained(27)Accordingly a limited number of relevant food cat-egories are taken into account in the SENS systemThe algorithm distinguishes three main groupsadded fats beverages and solid foods The latter issubdivided in six categories cereals cheese otherdairy products eggs fish and other solid foodsThis categorisation differs slightly from theSAINcat to be closer to that considered in nutrientprofiling systems proposed by the EC(28) and theWHO Regional Office for Europe(29)

(5) Decide which nutrients and other food componentsshould be involved As the SAINLIM system theSENS is derived from two indicators a NDS forqualifying nutrients (SAINSENS) and a limited nutri-ent score for disqualifying nutrients (LIMSENS) Inthe SAINSENS for solids foods proteins and fibrewere kept as basic nutrients vitamin C was replacedby the percentage of FampV and iron was removedFood category-specific nutriments were introducedfibre (for cereal-based products) calcium (for dairyproducts) and proteins (for eggs and fish) withdependent weighting factors to give greater promin-ence to these nutrients in the calculation whereneeded Other specific nutrients were selected in theSAINSENS for beverages and added fats respectivelyvitamin C and FampV for beverages and α-linolenicacid and MUFA for added fats In the LIMSENSadded sugars were replaced by free sugars (addedsugars plus sugars naturally present in honey fruitjuices and concentrates) in accordance with recentWHO recommendations on the need to reduce freesugar consumption(30) Besides including free sugarsin the calculation will provide an incentive for themanufacturers to develop strategies to reformulateproducts high in sugar by fostering nutrient-denseingredients containing naturally present sugar ratherthan ingredients with lower nutritional value TheSENS algorithm could encourage food companiesto reformulate their products by improving the nutri-ent density while avoiding replacement of nutrient-dense ingredients by empty ingredients

(6) Decide the reference amount for the model As in theSAINLIM system the SAINSENS represents anutrient density and is expressed per 418middot4 kJwhile the LIMSENS is expressed per 100 g

M Tharrey et al6

Proceedings

oftheNutritionSo

ciety

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0029665117000817Downloaded from httpswwwcambridgeorgcore INRA - Versailles-Grignon on 13 Jun 2017 at 140250 subject to the Cambridge Core terms of use available at

(7) Decide whether to use scoring or thresholds (orboth) The previous SAINLIM classification wassensitive to small changes in the nutrient compos-ition of a food due to a strong edge effect Thiscould lead to a leap over classes even directlyfrom Class-1 to Class-4 To overcome this issue sec-ondary thresholds were added to the primary ones(5 for the SAIN and 7middot5 for the LIM) These thresh-old values of 10 15 35 40 for LIMSENS and 2 3middot57middot5 10 15 for the SAINSENS were selected pragmat-ically to reach a better ordered classification and toallow food with a similar LIMSENS to be discrimi-nated against their SAINSENS and vice versa

(8) Decide which numbers should be used to determinethe thresholds or score The European regulation(EC 11692011) states that labelling should bebased either on the harmonised reference intakesin Europe set out in Annex XIII or in their absenceon generally accepted scientific advice on intakes forenergy or nutrients(21) In the SENS algorithmexisting reference intakes in Europe replacedFrench DRV for SFA sodium calcium and pro-teins The 2015 WHO recommendation was usedfor free sugars(30) FAO and EFSA (EuropeanFood Safety Authority) recommendations wereused for MUFA and α-linolenic acid(31) In theabsence of reference intakes in Europe for fibres inAnnex XIII of EC 11692011(21) a reference intakeof 20 g was chosen as an intermediate between theEFSA reference values(32) for children (14 gd) andfor adults (25 gd) It was lower than in the originalSAIN (ie 25 g) to better value foods rich in fibres(as a lower reference value increases the ratioFibres(g418middot4 kJ)DRVfibres thereby increasing theSAINSENS)

(32) FampV were divided per 10 to avoida disproportionate importance of this ratio relativeto other ratio of the equation

Implementation of the algorithm requires three mainsteps The first is to collect information on ingredientsto categorise a selected food in one of the eight food cat-egories described earlier Due to its specific compositionmilk is considered as a dairy product rather than as abeverage In accordance with the French nutrition andhealth programme added fat is defined as fats addedby oneself such as oil butter cream mayonnaise or vin-aigrette(33) As suggested by the EC to avoid a food beingclassified in more than one category the selected foodmust have more than 50 of the sub-category ingredientfor 100 g to belong to that sub-category (except forcheese requiring an amount gt70 in accordance withthe French decree no 2013-1010) Once the sub-categoryis identified the second step requires the collection ofspecific nutritional information to calculate the adequateSAINSENS and LIMSENS Among nutrients required forthe calculation of the SENS algorithm only energy pro-teins SFA and sodium are subject to the mandatorynutrition declaration However manufacturers shouldbe able to provide information about other nutrients orcomponents In the calculation FampV include all freshand processed fruits (tubers nuts seeds dried fruits

and legumes excluded) In the third step the selectedfood will be positioned on a map with the SAINSENSand LIMSENS as axes to display their allocation in oneof the four SENS classes Water is automatically clas-sified in Class-1 (LIMSENS = 0 and SAINSENS is infinitegiven that water is energy-free) Other beverages thatmay be present in Class-1 when the algorithm is appliedare systematically downgraded to Class-2 to deliberatelyhaving water as the only beverage represented in Class-1Other exceptions concern energy-dense foods thosefoods with EDgt 1673middot6 kJ (400 kcal)100 g that may beallocated to Class-1 or Class-2 by the algorithm are sys-tematically downgraded to Class-2 and Class-3 respect-ively Finally foods with EDgt 2092 kJ (500 kcal)100 gand Na gt 200 mg100 g that may be allocated toClass-3 are systematically downgraded to Class-4

As reported by WHO validation is a key step in thedevelopment of a nutrient profiling system Althoughno gold standard exists for defining a healthy food vari-ous validation methods have been proposed to ensure themodel classifies foods properly These include assessmentof construct validity that is testing the nutrient profilemodel ability to complement and support food-baseddietary guidelines in the regions in which it is appliedThe construct validity of the SAINLIM system was per-formed using diet modelling with linear programming fordesigning healthy and unhealthy diets Although basedonly on few key nutrients the system was able to predictthe ability of a given food to facilitate or to impair thefulfillment of a large number of nutrient recommenda-tions(1234) The SENS algorithm was validated by differ-ent approaches Its implementation on the CIQUALFrench database showed that SENS classification wasconsistent with nutrition principles and with food‐baseddietary guidelines (N Darmon J Sondey V Braescoet al unpublished results) Results also showed theSENS classification was positively associated with ED(N Darmon J Sondey V Braesco et al unpublishedresults) which is considered as one validation elementof the nutrient profiling model(35) Nevertheless this clas-sification was not ascribed solely to the ED and itsstrength lies in its capacity to also differentiate foodsaccording to their nutrient density Other analysesshowed that the SENS algorithm leads to a hierarchicalclassification of foods considering their contribution tonutritionally adequate diets suggesting that healthychoices can be advocated based on the relative contribu-tion of foods from each SENS class(26)

The SENS in the international context of food labelling

Initially driven by the regulation EC 19242006 to con-trol access to nutrition and health claims in Europethe scope of nutrient profiling systems has been extendedto various applications including marketing of foods tochildren regulation of fortification fiscal food imple-mentation the use of economic tools to orient food con-sumption or product labelling logos or symbolsRegarding the latter the rise of overweight and obesityhas focused policymakersrsquo attention on the improvement

A French nutrient profiling system 7

Proceedings

oftheNutritionSo

ciety

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0029665117000817Downloaded from httpswwwcambridgeorgcore INRA - Versailles-Grignon on 13 Jun 2017 at 140250 subject to the Cambridge Core terms of use available at

of nutrition information to consumers Nutrition declar-ation tables may be difficult to understand for the aver-age consumer and simplified nutrition labelling suchas front-of-pack logos or symbols might help consumersto spot a synthesised nutrition information when pur-chasing foods(36) In that connection different formatsof front-of-pack labels have been launched worldwideSummary indicators were first introduced by non-profitorganisations and government agencies such as theGreen Keyhole in the late 1980s(37) Used in SwedenNorway and Denmark this voluntary positive category-specific label is only placed on products that meet specificrequirements relating to fibres salt sugar fat and satu-rated fat set for a given food category Following theEuropean regulation (EC 11692011) the UK has optedfor a nutrient-specific Traffic Light system developed bytheUKFoodStandardsAgency combining colour andper-centage of DRV for energy fat saturated fat sugars andsodium(38) In the USA Facts Up Front was launched in2011 as a joint initiative of the Grocery ManufacturersAssociation and Food Marketing Institute in the absenceof government-endorsed scheme(39) This label showsenergy per serving and information on three nutrients tolimit in the diet saturated fat sodium and sugar and pos-sibly up to two positive nutrients if providing more than10 of the DRV per serving In 2014 Australian andNewZealand governments have developed in collaborationwith industry public health and consumer groupsthe Health Star Rating system ranking foods from half tofive stars(40) The algorithmdriving the calculation accountsfor energy saturated fats total sugars sodium andcases-specific points allocated to positive nutrients proteinfibres and the amount of fruit nuts vegetables and legumesIn France in addition to the simplified nutrition labellingbased on the SENS system other nutrient-specific labelshave been proposed and are presently tested in real-life con-ditions the Traffic Light system as applied in theUK(38) animproved Guideline Daily Amounts (percentage ofGuidelineDailyAmounts) label(41) andafive-colour systemderived from the UKOfcom nutrient profiling model(42)

Derived from the SAINLIM system the SENS pre-sents the unique advantage of not combining the twoindicators that account for qualifying and disqualifyingnutrients respectively This allows a separate evaluationof the positive and negative aspects of each food consid-ered individually Modifications were made from theSAINLIM system to make it more operational for sim-plified labelling in Europe Missing data in the algorithmcomputation could however hinder its implementationNot all components used in the SENS are subject tothe mandatory nutrition declaration such as fibres freesugars and the proportion of FampV Information aboutcontent of vitamin C calcium MUFA and α-linolenicacid may also be required according to the food categorywhich the product belongs to However it seems legitim-ate to ask for this information as adding these nutrientsbetter reflects the nutrient density of specific foodsBesides manufacturers are expected to have a good com-mand of their recipes and should be able to provide infor-mation about nutritional specificity of their products

Conclusion

The SENS system is an adapted version of the SAINLIM system while addressing its shortcomings by inte-grating specificities of food categories reducing the num-ber of nutrients ordering the four classes and introducingEuropean reference intakes Through the evolution fromthe French SAINLIM system to the SENS algorithmthis review shows how a nutrient profiling system canbe adapted according to scientific knowledge publichealth orientation and legislation Food labellingschemes designed to help steer consumersrsquo choicetowards healthier products are derived from variousnutrient profiling systems differing in terms of nutrientsselection grouping of food products type of referenceamounts and cut-off values Despite WHO guiding prin-ciples manual for developing nutrient profiling designinga relevant tool remains challenging and no system is per-fect making comparisons and decision difficultNevertheless present nutrient profiling systems alignwith basic nutrition principles making them a valuabletool to support public health nutrition policies

Financial Support

This review received no specific grant from any fundingagency commercial or not-for-profit sectors M M isemployee of MS-Nutrition MS-Nutrition has receivedfees from the several food companies and sectors thathave financially contributed to the development of theSENS These companies had no role in the design andconduct of this review

Conflicts of Interest

None

Authorship

N D V A-B and M M developed the concepts pre-sented in the manuscript M T wrote the first draft ofthe manuscript N D and M M contributed to writingof the manuscript and proposed critical comments Allauthors approved the manuscript for publication N Dhas primary responsibility for final content

References

1 US Department of Health and Human Services amp USDepartment of Agriculture (2005) Dietary Guidelines forAmericans 6th ed Washington DC USA

2 Darmon N Darmon M Maillot M et al (2005) A nutrientdensity standard for vegetables and fruits nutrients per cal-orie and nutrients per unit cost J Am Diet Assoc 1051881ndash1887

M Tharrey et al8

Proceedings

oftheNutritionSo

ciety

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0029665117000817Downloaded from httpswwwcambridgeorgcore INRA - Versailles-Grignon on 13 Jun 2017 at 140250 subject to the Cambridge Core terms of use available at

3 James WP Nelson M Ralph A et al (1997)Socioeconomic determinants of health Thecontribution of nutrition to inequalities in health BMJ314 1545ndash1549

4 Maillot M Darmon N Darmon M et al (2007)Nutrient-dense food groups have high energy costs aneconometric approach to nutrient profiling J Nutr 1371815ndash1820

5 Drewnowski A Maillot M amp Darmon N (2009) Shouldnutrient profiles be based on 100 g 100 kcal or servingsize Eur J Clin Nutr 63 898ndash904

6 Maillot M Ferguson EL Drewnowski A et al (2008)Nutrient profiling can help identify foods of good nutri-tional quality for their price a validation study with linearprogramming J Nutr 138 1107ndash1113

7 Martin A (2001) The lsquoapports nutritionnels conseilleacutes(ANC)rsquo for the French population Reprod Nutr Dev 41119ndash128

8 AFSSA (2008) Setting of nutrient profiles for accessingnutrition and health claims proposals and argumentshttpswwwansesfrfrsystemfilesNUT-Ra-ProfilsENpdf(accessed October 2016)

9 Rouveyrol C Lesturgeon A Georgeacute S et al (2014) GUIDEPRATIQUE OPTIMED Une deacutemarche drsquooptimisation etde valorisation nutritionnelles baseacutee sur deux outils origi-naux Guide ACTIA [in French]

10 The European Parliament and the Council of the EuropeanUnion (2006) Regulation (EC) No19242006 of theEuropean Parliament and of the Council of 20 December2006 on nutrition and health claims made on foods OffJ Eur Union L 404 9ndash25

11 Arsenault JE Fulgoni VL Hersey JC et al (2012) A novelapproach to selecting and weighting nutrients for nutrientprofiling of foods and diets J Acad Nutr Diet 112 1968ndash1975

12 Darmon N Vieux F Maillot M et al (2009) Nutrientprofiles discriminate between foods according to their con-tribution to nutritionally adequate diets a validation studyusing linear programming and the SAINLIM system AmJ Clin Nutr 89 1227ndash1236

13 Maillot M Drewnowski A Vieux F et al (2011)Quantifying the contribution of foods with unfavourablenutrient profiles to nutritionally adequate diets Br J Nutr105 1133ndash1137

14 Darmon N amp Drewnowski A (2015) Contribution of foodprices and diet cost to socioeconomic disparities in dietquality and health a systematic review and analysis NutrRev 73 643ndash660

15 Darmon N Lacroix A Muller L et al (2014) Food pricepolicies improve diet quality while increasing socio-economic inequalities in nutrition Int J Behav Nutr PhysAct 11 66

16 Achir N Kindossi J Bohuon P et al (2010) Ability of somefood preservation processes to modify the overall nutri-tional value of food J Food Eng 100 613ndash621

17 Gamburzew A Darcel N Gazan R et al (2016) In-storemarketing of inexpensive foods with good nutritional qual-ity in disadvantaged neighborhoods increased awarenessunderstanding and purchasing Int J Behav Nutr PhysAct 13 104

18 Jo J Lusk JL Muller L et al (2016) Value of parsimoniousnutritional information in a framed field experiment FoodPolicy 63 124ndash133

19 Masset G Vieux F amp Darmon N (2015) Which functionalunit to identify sustainable foods Public Health Nutr 182488ndash2497

20 Perignon M Masset G Ferrari G et al (2016) How lowcan dietary greenhouse gas emissions be reduced withoutimpairing nutritional adequacy affordability and accept-ability of the diet A modelling study to guide sustainablefood choices Public Health Nutr 19 2662ndash2674

21 The European Parliament and the Council of the EuropeanUnion (2011) Regulation (EU) No 11692011 of theEuropean Parliament and of the Council of 25 October2011 on the provision of food information to consumersamending Regulations (EC) No 19242006 and (EC) No19252006 of the European Parliament and of the Counciland repealing Commission Directive 87250EEC CouncilDirective 90496EEC Commission Directive 199910ECDirective 200013EC of the European Parliament and ofthe Council Commission Directives 200267EC and20085EC and Commission Regulation (EC) No 6082004 Off J Eur Union 304 18ndash63

22 Cohen DA amp Babey SH (2012) Contextual influences oneating behaviors heuristic processing and dietary choicesObes Rev Off J Int Assoc Study Obes 13 766ndash779

23 Cecchini M amp Warin L (2016) Impact of food labelling sys-tems on food choices and eating behaviours a systematicreview and meta-analysis of randomized studies ObesRev Off J Int Assoc Study Obes 17 201ndash210

24 World Health Organization (2010) Nutrient ProfilingReport of a WHOIASO Technical Meeting GenevaSwitzerland WHO

25 ANSES (2016) Faisabiliteacute de la classification des alimentsselon lrsquoalgorithme proposeacute par la FCD Comparaison desreacutesultats obtenus agrave ceux du systegraveme 5-C inteacutegrant les ajuste-ments du HCSP httpswwwansesfrfrsystemfilesAUTRE2015SA0253pdf [in French] (accessed October2016)

26 Darmon N Maillot M Braesco V et al (2015)LrsquoAlgorithme du Systegraveme drsquoEtiquetage NutritionnelSimplifieacute (SENS) Deacuteveloppement description et valid-ation Rapport remis le 23 deacutecembre 2015 agrave lrsquoAgence natio-nale de seacutecuriteacute sanitaire de lrsquoalimentation delrsquoenvironnement et du travail (ANSES) [in French]

27 Scarborough A Arambepola C Kaur A et al (2010)Should nutrient profile models be lsquocategory specificrsquo orlsquoacross-the-boardrsquo A comparison of the two systemsusing diets of British adults Eur J Clin Nutr 64 553ndash560

28 European Commission (2009) Working document on thesetting of nutrient profiles Brussels Belgium httpswwwsenatfreuropetextes_europeensa0006pdf (accessedOctober 2016)

29 World Health Organization Regional Office for Europe(2015) Nutrient Profile Model Copenhagen DenmarkWHO Regional Office for Europe

30 World Health Organization (2015) Guideline Sugars intakefor adults and children [Internet] Geneva SwitzerlandWorld Health Organization

31 EFSA Panel on Dietetic Products Nutrition and Allergies(2010) Scientific opinion on dietary reference values forfats including saturated fatty acids polyunsaturated fattyacids monounsaturated fatty acids trans fatty acids andcholesterol EFSA J 8 1462

32 EFSA Panel on Dietetic Products Nutritionand Allergies (2010) Scientific Opinion on Dietary ReferenceValues for carbohydrates and dietary fibre EFSA J 8 1462

33 Hercberg S Chat-Yung S amp Chaulia M (2008) The FrenchNational Nutrition and Health Program 2001-2006-2010Int J Public Health 53 68ndash77

34 Clerfeuille E Vieux F Lluch A et al (2013) Assessing theconstruct validity of five nutrient profiling systems using

A French nutrient profiling system 9

Proceedings

oftheNutritionSo

ciety

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0029665117000817Downloaded from httpswwwcambridgeorgcore INRA - Versailles-Grignon on 13 Jun 2017 at 140250 subject to the Cambridge Core terms of use available at

diet modeling with linear programming Eur J Clin Nutr67 1003ndash1005

35 Drewnowski A Maillot M amp Darmon N (2009) Testingnutrient profile models in relation to energy density andenergy cost Eur J Clin Nutr 63 674ndash683

36 Hersey JC Wohlgenant KC Arsenault JE et al (2013)Effects of front-of-package and shelf nutrition labeling sys-tems on consumers Nutr Rev 71 1ndash14

37 TheNordicCouncil ofMinisters (2010)TheKeyholeHealthychoices made easy httpnordendiva-portalorgsmashgetdiva2700822FULLTEXT01pdf (accessedDecember 2016)

38 Department of Health the Food Standards Agency (2016)Guide to creating a front of pack (FoP) nutrition label forpre-packed products sold through retail outlets httpswwwgovukgovernmentuploadssystemuploads

attachment_datafile3008862902158_FoP_Nutrition_2014pdf (accessed November 2016)

39 Grocery Manufacturers Association and the FoodMarketing Institute (2017) Facts Up Front httpwwwfactsupfrontorg (accessed November 2016)

40 Australian Government Department of Health and Ageing(2014) Health Star Rating httphealthstarratinggovau(accessed November 2016)

41 Food and Drink Federation (2016) GDAs explainedGuideline Daily Amounts httpwwwfoodlabelorguklabelgda_valuesaspx (accessed November 2016)

42 Ducrot P Julia C Meacutejean C et al (2016) Impact of differ-ent front-of-pack nutrition labels on consumer purchasingintentions a randomized controlled trial Am J Prev Med50 627ndash636

M Tharrey et al10

Proceedings

oftheNutritionSo

ciety

httpswwwcambridgeorgcoreterms httpsdoiorg101017S0029665117000817Downloaded from httpswwwcambridgeorgcore INRA - Versailles-Grignon on 13 Jun 2017 at 140250 subject to the Cambridge Core terms of use available at