Using reference nutrient density goals with food balance sheet data to identify likely micronutrient deficits for fortification in countries in the Western Pacific Region Rosalind S Gibson, Research Professor, Department of Human Nutrition, University of Otago, Dunedin New Zealand Tommaso Cavalli-Sforza, Nutrition & Food Safety, WHO Regional Office for the Western Pacific, Manila, Philippines
Possible sources of food consumption data for fortification FAO Food Balance Sheet data National level (180 countries) data on food supply available for consumption per capita (~95 food commodities) Household Income & Expenditure Surveys data on amount food purchased = proxy for apparent consumption at household level National food consumption surveys Individual level data: limited data sets in LICs Prevalence of inadequate & excess intakes with an appropriate design Each has limitations and errors
Methods to assess likely micronutrient deficits using national data Compare nutrients per capita with populationweighted mean requirement value based on EAR takes into account different nutrient requirements of individuals by weighting: uses national population distributions by age & sex for each country (UN, 1994) no guarantee that intakes distributed among individuals in population will satisfy their own nutrient requirements 2. Uses nutrient density values (per 1000 kcal) per capita for comparison with Reference Nutrient densities
Advantages of using nutrient density per capita for comparison with Reference Nutrient Density Errors arising from differences in quality of FBS data across countries are less marked Density values much less influenced by age & sex in a population Enables a single reference nutrient density (per nutrient) to be applied which does not depend on weighting based on national population distribution data
Study objectives 1. To calculate micronutrients (MN) densities of daily food supply per capita from food balance sheet data for countries in WPR 2. To compare MN density estimates by country with reference MN density goals 3. To identify MNs likely to be inadequate in national diets in WPR 4. To identify potential staple food vehicles for fortification
Methods: Step 1 Calculating nutrient densities of daily food supply from FBS data 1. Access FAO FBS data for 95 standardized food commodities for each country 2. Calculate nutrient content of each commodity from daily per capita energy value given in FAO FBS (i.e., mg nutrient/kcal) takes into account waste during food preparation & extraction rate of cereals (Wuehler et al. 2005) 3. Calculate daily per capita amounts of energy & nutrients by country 4. Calculate daily per capita nutrient densities by country
Examples: Ave daily per capita amounts of energy & MN available in food supply County Energy (kcal) B-1 (mg) B-2 (mg) Fe (mg) Zn (mg) Ca (mg) Cambodia 1995 0.58 0.56 4.8 6.4 113 Philippines 2355 1.05 0.99 6.8 6.7 260 Laos 2152 0.82 0.75 7.3 7.3 183 N Zealand 3150 1.44 1.86 11.8 10.3 630 Nutrient values based on WorldFood Minilist
Examples: Average daily per capita MN densities available in food supply Nutrient per 1000 kcal Cambodia Laos Philippines New Zealand B-1 (mg) 0.3 0.4 0.4 0.5 B-2 (mg) 0.3 0.3 0.4 0.6 Ca (mg) 57 85 111 200 Fe (mg) 2.4 3.4 2.9 3.5 Zn (mg) 3.2 3.4 2.9 3.3
Methods: Step 2 Compare MN density per capita with the Reference Micronutrient Density Goals This comparison reveals likely micronutrient shortfalls in the available food supply per capita
Methods: Step 2. What are Reference Micronutrient Density Goals? These have been designed such that a diet with average micronutrient densities equal to the Reference Goal should be adequate to meet the functional needs and generate some modest body reserves for at least 95% of individuals in a population* Based on: Micronutrient EARs & Energy requirements Assume distribution of nutrient density is normal with a CV of 10% for the mean *Details given in Beaton (1995): Fortification of foods for Refugee Feeding. Technical Background Report : Derivations and Analyses. Prepared for CIDA
Step 2: How to calculate Reference Nutrient Density Goals? 1. Ave necessary nutrient densities calculated for 8 lifestage groups for coverage of all but ~ 5% of individuals 2. Highest density for each life-stage group selected as Reference Nutrient Density Goal 3. For this life-stage group expected coverage is 95% 4. For all other groups coverage will be higher For other approaches for calculating reference nutrient density goals see Backstrand (2003) Pub Hlth Nutr 66:829-837 See: IOM (2003): DRI Applications in Dietary Planning
Age-sex-physiologic groups used to calculate necessary micronutrient densities* Groups 2.5 y M & F 7.0 y M & F 12 y M & F 15 y M & F 17 y M & F 25 y M & F Pregnant Lactation Requirements used by Beaton (1995): Energy: FAO/WHO/UNU (1985) Nutrients: FAO/WHO reports; WHO (1996) Note for Fe & folic acid: pregnant women are excluded*. *From Beaton (1995)
How to derive necessary group mean nutrient densities for 8 age-sex-physiologic groups? Methods based on probability approach. All need: EAR for each nutrient & energy requirement per age-sex group Assume intakes & nutrient requirements are independent & low correlation between intakes of MNs & energy Assume a normal distribution for nutrient intake & densities from available food supply Assume CV of mean intake is 10% - set at daily mean nutrient density per capita Assume 95% coverage (probability of adequacy = 0.95) For alternative methods e.g. Monte Carlo simulation: see IOM (2003)
Estimating group mean nutrient density to achieve 95% coverage Methods: IOM Mean conventional group intake method per Mean group requirement per 1000 kcal XXX 1000kcal to achieve 95% coverage for the group 5.0% Nutrient density Factor assumes CV 10% & coverage of 95% 1. EARs for B-1,B-2 or niacin expressed as mean requirement per 1000 kcal for males aged 25 y 2. Assume CV = 10% for mean nutrient density per capita 4. Multiply mean group requirement/1000 kcal by adjustment factor of 1.19* 5. Yields necessary mean group intake/1000 kcal for 95% coverage 19% > EAR
Examples: Necessary mean MN densities by MN & class of subjects (for 95% coverage) Age class B-1 B-2 Fe* 2.5y M 0.35 0.48 0.55 7.0y M 0.35 0.48 0.85 12 y M 0.35 0.48 1.06 15 y M 0.35 0.48 1.03 15y F 0.35 0.48 1.35 17 y M 0.35 0.48 0.48 25 y M 0.35 0.48 0.49 25 y F 0.44 0.60 1.32 Pregnancy 0.35 0.48 - Lactation 0.34 0.48 0.56 *Amount utilizable iron needed
Examples: Reference Micronutrient Density Goals (expressed as units/1000 kcal) to cover normative needs* Micronutrient Reference and unit MN density Vit B-12 (ug) 0.8 B-1 (mg) 0.44 B-2 (mg) 0.60 Niacin (NE) 6.5 Vit B-6 (mg) 0.41 Fe (mg) 18 L; 9M; (1.35)* Zn (mg) 10 L; 2.9 M Ca (mg) 550 L= Low; M=moderate bioavailability Value in ( ) is an estimate of amount of utilizable iron that should be provided *Excludes pregnant women as supplements needed Covers all groups unless stated otherwise. * From Beaton (1995)
Estimated MN density per capita of food supply compared to Reference Nutrient Density Goals* Nutrient per Cambodia Laos Philippinnes New Goal 1000 kcal Zealand B-1 (mg) 0.3 0.4 0.4 0.5 0.40 B-2 (mg) 0.3 0.3 0.4 0.6 0.60 Vit B-12 0.9 0.9 (μg) 1.9 2.9 0.8 Ca (mg) 57 85 111 200 550 Fe (mg) 2.4 3.4 2.9 3.5 18(L) 9 (M) Zn (mg) 3.2 3.4 2.9 3.3? 10 (L) 2.9 (M) From Beaton (1995) L= low; M = moderate bioavailability
Step 3: Likely micronutrient deficits in available food supply per capita in countries in WPR Country# Cambodia (8) Laos (7) Viet Nam (7) Kiribati (6) Philippines (5) Malaysia (4) New Zealand (2 or 3) Likely micronutrient deficits B-1,B-2,Ca,Fe, Zn, Cu, folate, Vit A B-2, Ca, Fe, Zn, Cu, folate, Vit A B-2, Ca, Fe, Zn, Cu, folate, Vit A B-2, Ca, Fe, Zn, folate, Vit A B-2, Ca, Fe, Zn, Cu Ca, Fe, Zn, Cu Ca, Fe, Zn? Note that Vits D, C, & I & Se were not included
4. Identifying potential food vehicles from FBS data FBS data provide estimates of daily per capita amounts of ~95 commodities by: grams/day; kcal/day; % kcal from commodity This can be used to identify potential food vehicles BUT: Provides no information on coverage of food vehicle within country* Provide no information on purchasers of food vehicle* Provide no information on availability of food vehicle for those most at need* Provided at household level by HIES
Identifying food vehicles from FBS data County Wheat per capita g/d Kcal/ d % kcal Rice per capita g/d Kcal/ d % kcal Cambodia 6 18 1 453 1527 71 Mongolia 306 841 30 8 30 1 Laos 2 6 0 470 1516 69 New 191 569 19 18 67 2 Zealand In Cambodia & Laos: low availability of wheat for consumption vs. in Mongolia & New Zealand: low availability for rice These staples not suitable food vehicles for fortification in these countries
Potential food vehicles and suggested fortificants for countries in WPR Country Food vehicles Micronutrient fortificants Cambodia Rice Condiments? B-1*,B-2,folate, Fe*, Zn*, Ca, Cu Mongolia Wheat flour B-1, Fe*, Zn*, Ca, Cu Laos Rice Condiments B-2, folate, Fe, Ca, Zn, Cu New Zealand Wheat flour Ca, Fe*, Zn*? * Biomarker data from national surveys and/or small studies confirms deficiencies of +these micronutrients
Conclusions Comparison of nutrient densities per capita from FBS with Reference Nutrient Density Goals can identify likely micronutrient deficits Countries can be ranked according to number of likely deficits FBS provide NO information on distribution of nutrients available for consumption at household or individual level FBS data can identify potential food vehicles for fortification but does not provide data on coverage, most needy, or purchasers of food vehicles Reference nutrient density goal approach can also be applied to HIES data to yield information on likely MN deficits at HH level as well as coverage, purchasers of food vehicle, & most needy households.
1. Estimating group mean nutrient density to achieve 95% coverage Methods: IOM Mean conventional group intake method Mean group Requirement (EAR) XXX to achieve 95% coverage for the group 5.0% Micronutrient amount per capita *Factor assumes CV 10% & coverage of 95% 1. Select EAR for nutrient for males aged 25 y 2. Assume CV= 10% for mean daily amount per capita 4. Multiply EAR by adjustment factor of 1.19* to yield a necessary group mean intake for 95% coverage ( 19% above EAR) 6. Divide the necessary group mean by energy requirement for males aged 25 y to yield necessary mean nutrient density
Identifying food vehicles & coverage for fortification from HIES data Food item Nationwide No. HH % Rural No. % Urban No. % Pop. Quintile Expenditure Quintile 6,434,534 100 5045,528 78 1389,006 22 1 % 5 % Wheat flour 1112,852 17 675,967 13 436,885 31 76.613 8 515778 29 Bread 1263,497 20 541,425 11 722072 52 35,538 4 678834 39 Sugar 4679,455 73 3388,525 67 1290,929 93 435330 45 1578085 90 From Fiedler et al. (2008)
How to set fortificant levels in the food vehicles Select potential staple food vehicle for fortification: e.g. wheat flour Identify likely micronutrient deficits per capita (FBS) or per HH (HIES) Change micronutrient values in FCT for chosen food vehicle to reflect probable micronutrient levels required to meet deficits Recalculate data: iterative process until simulated MN densities per capita (or per HH) meet reference nutrient density goals Once these levels achieved, then coverage per capita for the chosen micronutrient will be at least 95% Express amount of MNs required per 100 g food vehicle
Identifying food vehicles & coverage for fortification from HIES data Food item Nationwide No. HH % Rural No. % Urban No. % Pop. Quintile Expenditure Quintile 6,434,534 100 5045,528 78 1389,006 22 1 % 5 % Wheat flour 1112,852 17 675,967 13 436,885 31 76.613 8 515778 29 Bread 1263,497 20 541,425 11 722072 52 35,538 4 678834 39 Sugar 4679,455 73 3388,525 67 1290,929 93 435330 45 1578085 90 From Fiedler et al. (2008)
Example of an excerpt of a standardized Food Balance Sheet Country Year Population.(thousand) Food supply per caput Commodity Kg per year Grams per day Kcal per day Protein per day (grams) Fat per day (grams) 95 food commodities in 176 countries
Comparison of Food Balance Sheet vs. Household Income & Expenditure Surveys Attribute Food Balance Sheets HIES data Availability FBS data for ~176 countries; variable quality Fewer HIES; Variable quality Level of data National level only Proportion w. inadequate /excess intakes No data on distribution at individual or HH level by age/sex No. Crude estimate if CV of inter- person intakes known. Cannot identify sub-groups at risk Household (HH) level Data on distribution across HHs but not within HHS No. Crude estimate if data converted to man values. No data on whether vulnerable persons are reached
Sources of nutrient composition values WorldFood Minilist : 53 nutrients (1800 foods) from 6 countries Egypt, Kenya, Mexico, Senegal, India, Indonesia. Available at: http://www.fao.org/ifoods/ International Network of Food Data Systems (INFOODS) Regional organizations maintain regional databases: e.g LATINFOODS; ASEAN-FOODS. Directory at: http://www.fao.org/infoods/ USDA Database: http://www.nal.usda.gov/fnic/foodcomp/ Concise South African Food Composition Table: 36 nutrients 1457 items: http://www.mrc.ac.za/foodcompositionadv/start.jsp
Assumptions by Beaton (1995) for derivation of Reference MN Density Goals Assume all individuals in the household share same mix of foods so nutrient density of household nutrient density of diets of individuals in the group (except children < 2 y) Assume individuals in household share available foods in proportion to their relative energy needs
Identifying food vehicles with FBS & HIES data Other considerations Methods of storage & food processing & preparation: to identify potential losses of food fortificants Whether food vehicle is centrally processed permits controlled fortification Stability of food vehicle to varying temperatures Technology fortification issues Presence of absorption inhibitors in food vehicle: phytate; polyphenols
Limitations with Food Balance Sheet Method Limitations in calculating food supply data per capita Quality of data for food diverted for nonhuman food uses (eg: animal feed, seed, manufacture) often limited Do not include waste on farm, during distribution and processing, and at household or food service level Accuracy of population data varies across countries: adjustments for refugees and migrants/ tourist or seasonal workers not always made No information on actual distribution of foods available for consumption by: region; urban/rural; SES; season; within households; age/gender/lifestage groups Limitations with per capita nutrient content of food supply
Limitations of using Food Balance Sheet data for identifying potential food vehicles for fortification No information on distribution of potential food vehicle within country; by region (urban/rural); by SES; within HHs No information on whether potential food vehicle is purchased by HHs & thus more accessible to fortification No data on whether the commodity is available for consumption by those in need No information on consumer preferences Only provides information on one food at a time Does not account for political sensitivities