How do methods matter Yan Bai PhD Candidate Friedman School of Nutrition Science amp Policy Tufts University Seminar at International Food Policy Research Institute Washington DC November 13 2019 ID: 921166
Download Presentation The PPT/PDF document "Modelling the Cost of Nutrient Adequacy ..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Modelling the Cost of Nutrient Adequacy over the Life Cycle: How do methods matter?
Yan Bai, Ph.D. CandidateFriedman School of Nutrition Science & PolicyTufts UniversitySeminar at International Food Policy Research InstituteWashington DC, November 13, 2019
Slide2Introduction
Methodology and Data
Modelling for Different Age-Sex Groups
Conclusions and The Future
2
Photo:
Stevier Kaiyatsa
METHODOLOGY AND DATA
INTRODUCTION
MODELING
FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
OUTLINE
Slide3introduction
3
Slide4WHY MODEL DIETS?
Identify foods or whole diets to meet nutrient gaps in the population
For advocacy, nutrition education or behavioral change
Diet recommendations for key populations (e.g. pregnant women or infants)
Measuring diet costs/affordability for targeting & designing social transfers
Measuring poverty (Allen 2017 AER)
Identifying critical nutrient gaps and deficienciesEvaluating impacts of (bio)fortification, reformulation and supplementsEvaluating impacts of agriculture, economic, and nutrition policiesEvaluating food systems
How efficient is a given food system for delivering nutrients?4
METHODOLOGY AND DATA
INTRODUCTION
MODELING
FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide5THE COST OF NUTRIENT ADEQUACY
CANDASA developed Cost of Nutrient Adequacy (
CoNA
):
Use least-cost way to meet nutrient needs for
an adult woman of 19-30 y/o
35 upper & lower bounds of energy + 21 nutrients for a balanced dietCan be measured in any time and location in the world But concerns about internal & external validity of CoNA
:We know requirements vary by age and sexCoNA produce diets heavy in plant-based foods, but major bioavailability concernsInfants/toddlers have additional constraints: small stomachs, chewable, palatable, ….In short, methodological assumptions could matter a lot
5
METHODOLOGY AND DATA
INTRODUCTION
MODELING
FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide6Methodology and data
6
Slide7Methodology AND DATA
Optimization: linear programming
Widely applied in both Economics and Nutrition since Stigler (1945)
Measuring poverty, tracking and comparing nutritious diets, making recommendations
The Cost of Nutrient Adequacy (
CoNA
) :CoNA: min. Ck =
Σipiqi subject to relevant dietary reference intake requirements
Food price data collected for other purposes:Food retail prices collected by national statistical agenciesChoice of items aims to represent all good & services consumed
Underlying data are typically confidential7
METHODOLOGY AND DATA
INTRODUCTION
MODELING
FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide8Dietary Reference Intakes (DRIs)
Estimated Average Requirements (EARs)
meet needs for 50% of a healthy population
Recommended Dietary Allowances (RDAs)
and
Adequate Intake (AI): meet needs for 97.5% of a healthy populationUpper Levels: eliminating risk for toxicity of over intakes
DIETARY REFERENCE INTAKES
Incorporated DRIs for long-term health
Acceptable Macronutrient Distribution Ranges (
AMDRs
) and Chronic Disease Risk Reduction Intakes (
CDRR)
METHODOLOGY AND DATA
INTRODUCTION
MODELING
FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide9DIETARY REFERENCE INTAKES (CONT’D)
The DRIs specify min. & max. levels for energy and up to 21 nutrients
9
Source: Dietary Reference Intakes (DRI) values for the US & Canada, last revised 2011. Available online at https://www.nal.usda.gov/fnic/dietary-reference-intakes
Energy
balance
1 calorie constraint
Macronutrient
ranges
For
protein
, fat and carbohydrates
3 lower bounds
3
upper bounds
2 average requirements
Micronutrient requirements
For 8 minerals (calcium, copper, iron, magnesium, phosphorous, selenium, zinc and sodium) and also for 10 vitamins (niacin, riboflavin, thiamin, folate, and Vitamin A, retinol, B6, B12, C, E)
16 average requirements
10 upper limits for toxicity
METHODOLOGY AND DATA
INTRODUCTION
MODELING
FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide10Dri
will vary across age-sex groups
Nutrient requirements vary by age, sex over life courses
METHODOLOGY AND DATA
INTRODUCTION
MODELING
FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide11DRI will vary by individual status
Nutrient bioavailability may alter nutrient requirements and lead to the need for adjustments
Multiple factors may affect
WHO and FAO provide DRIs (RDA equivalent) of Zinc and Iron for all gender-age groups assuming lower bioavailability
METHODOLOGY AND DATA
INTRODUCTION
MODELING
FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide12MODELING FOR DIFFERENT SEX-AGE GROUPS
12
Slide13Nutrition cases
Four methodological variations considered:
EAR only – meet needs of 1/2 population, recommended for population assessments;
The traditional way of making the least-cost diet constraints;
EAR plus upper limits (UL) and Acceptable Macronutrient Distribution Range (AMDR)
“default case” for the current
CoNA applications
Using Recommended Dietary Allowance (RDA) or Adequate Intake (AI) to replace EARTargeting almost entire healthy population’s needs;RDAs for Indispensable Amino-Acids – “protein quality” impacts the least cost?
And for Adult Groups: check if dietary zinc and iron sufficient given predicted bioavailability
METHODOLOGY AND DATA
INTRODUCTION
MODELING
FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide14CONA Cost ($) by age & sex
Lactating women and adolescent men are facing the highest CoNARDA substantially raises the cost, with CoNA in most age-sex groups greater than the national poverty line
METHODOLOGY AND DATA
INTRODUCTION
MODELING FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide15CONA Cost ($) per calorie by age & sex
Per 1,000kcal, females are consistently facing higher CoNA than male groups, and the CoNA density is increasing as people are aging
METHODOLOGY AND DATA
INTRODUCTION
MODELING FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide16Change in
cona from % change in CALCIUM EAR/RDACalcium has most expensive for the adolescent boys and girls
METHODOLOGY AND DATA
INTRODUCTION
MODELING FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide17Cona
food composition in adult groupsHigh contribution of plant foods due to missing bioavailability assumptions?
METHODOLOGY AND DATA
INTRODUCTION
MODELING FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide18Zinc/iron absorption at lower levels
Dietary zinc in
CoNA
food selections always above WHO standard:
Almost no zinc deficiencyDietary iron is far below the WHO DRIs: 5% absorption of many food selections,
Lots of deficiency
METHODOLOGY AND DATA
INTRODUCTION
MODELING
FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide19Using higher requirements for iron
Malabsorption of Iron leads to a much higher CoNA in adult women
Implausible pulse/nut/seed diet: iron fortification for women?
METHODOLOGY AND DATA
INTRODUCTION
MODELING
FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide20Modeling
for infants and toddlers
Current approach underestimates costs for young children?
Food selection and preparation
Is the food item suitable for infant/toddler feeding?
Do cooking methods relate to nutrient retention and food volumes?
Maximum daily feeding volumes? (small stomach problem)Need to consider additional nutrients and estimated energy requirements Breast feeding assumptionsBioavailability and health challenges like persistent diarrhea, infections, etc.
At least make adjustment for protein digestibility?
METHODOLOGY AND DATA
INTRODUCTION
MODELING
FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide21PRELIMINARY Analysis strategy
We are using the ICP price data for 25 low-income countries
Model for infants from 7-12 months, and toddlers from 13-36 months
First step focuses on food selection and preparation assumptions:
Certain foods are excluded for the infant group, such as red/white meat?
Food preparation assumptions from the West African Food Composition Table
Nutrient loss during food preparation (nutrient retention rates)Food volumes will change during preparation (food yield factors)Make different yield factors of cereals for infant and toddler groups (~7 vs ~2.5)
Assume functional gastric capacity: 30 g/kg reference BW and ~4 meals for infants and ~5 meals for toddlers (WHO, 2003) Reference BW: average of the median BW at each month
METHODOLOGY AND DATA
INTRODUCTION
MODELING FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide22PRELIMINARY Results
Average
CoNA
for the infant group is USD0.52 for male, and USD0.75 for female;
For the toddler group, it is USD1.12 for male and USD1.13 for female
METHODOLOGY AND DATA
INTRODUCTION
MODELING FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide23Top 5 foods being selected - NEPAL
Infant MaleInfant FemaleToddler Male
Toddler Female
METHODOLOGY AND DATA
INTRODUCTION
MODELING
FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide24Top 5 foods being selected - Malawi
METHODOLOGY AND DATAINTRODUCTION
MODELING
FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTUREInfant Male
Infant Female
Toddler Male
Toddler Female
Slide25CONCLUSIONS AND THE FUTURE
25
Slide26CONCLUSIONS AND THE FUTURE
Rigorous & realistic costing of nutritious diets has many applications
However, relatively little previous work explores how much methods matter
Our future goals in this project are:
Establish how methods matter (methodological study)
Create a user-friendly dietary modelling tool (open access)
Apply these methods to different types of data & different populations
METHODOLOGY AND DATA
INTRODUCTION
MODELING
FOR DIFFERENT AGE-SEX GROUPS
CONCLUSIONS AND THE FUTURE
Slide27Thank you! Q&A
27