A presentation on my PhD work in Guatemala over the past two years, which highlights new unique tools to understand a 'wicked' problem. Not all results have been peer-reviewed.
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
The complexities of child stunting in guatemala
1. The Complexities of Child Linear
Growth in the Western Highlands of
Guatemala: Aflatoxins and Enteric
Dysfunction
By
Lee Voth-Gaeddert, M.S., EIT
Environmental Systems Engineering PhD Candidate
Missouri University of Science and Technology
Adviser: Dr. Daniel Oerther, DoS-GFS
Financially Supported by:
US Peace Corps, USAID Nexos Locales, USAID Feed the Future (Texas A&M)
2. Outline
• The Problem
• Aims and Methods
• Results
• Child Linear Growth
• Aflatoxins
• Environmental Enteric Dysfunction
• Recommendations
3.
4. Aims and Methods
Aims:
• Analyze critical factors to linear growth of
children (child stunting)
• Analyze factors which influence transmission
of aflatoxin/fungus
• Analyze factors which influence transmission
of pathogens
8. Linear Growth – Network
Analysis
All Children
Severe Stunting
(< -3 SD)
Mild Stunting
(-3 to -2 SD)
No Stunting
(> -2 SD)
9. Linear Growth – Cluster
Analysis
If HAZ & Diarrhea are key “clusters” what other
variables are also key “clusters”?
1. Poverty
2. Food – Oil, Butter, Margarine
3. Practices Soil Conservation
4. Bred Animals Last Year
5. Knows Warning Signs for Problems in Pregnancy
6. Presence of Soap at Handwashing Station
7. Knows Warning Signs for Problems w Sick Child
8. Food – Other Fruits and Veggies
9. Food – Sweets & Chocolates
10.Spent $$ on Electricity
10. Field Work…
October 2016 and February 2017
Held health assemblies
Conducted household visits for observations and maize samples
Fecal samples pending….
11. Linear Growth – Aflatoxins & Enteric Dysfunction
Test Model of Linear Growth, AFB measure,
& Enteric Dysfunction 2016 n = 320
Test Model of Linear Growth, AFB measure,
& Enteric Dysfunction Change n = 120
12. Linear Growth – Aflatoxins
Confirming relation between AFB measure
and AFB symptoms 2016; n = 320
Dynamic Relationships Between AFB
measure & AFB symptoms; n = 120
14. Linear Growth – Full Model
October 2016 February 2017
n = 372 n = 300
15. Take Aways
• Increased putative aflatoxin exposure was negatively correlated with
child linear growth
• Increased putative aflatoxin exposure was negatively correlated with
enteric dysfunction
• Increased putative aflatoxin exposure was positively correlated with an
increase of symptoms of aflatoxin exposure
• Increased child play time was negatively correlated with child linear
growth
• Increased prenatal health (vitamins and controls) was positively
correlated with child linear growth
21. Take Aways
• 1 month before harvest improved maize storage and post-harvest
practices were negatively correlated with more AFB symptoms
• 1 month before harvest an increase in observations of fungus in the
household maize supply was negatively correlated with an increase in
AFB symptoms
• 1 month before harvest the improved purchase habits of the mother at
market was negatively correlated with an increase in AFB symptoms
• 2-3 months after harvest only delayed sale at the market and
remoteness of market were correlated with AFB symptoms
• Potential for market based intervention….
28. Enteric Dysfunction - Departments vs
Environment – Model Validation
• Utilized data from previous study in northern Quiche (2011)
• Held department constant (Quiche), varied environment (mountains v tropical)
Result: Structure of model validated, but sanitation became important in northern
Quiche suggesting the department was consistent in quality (good or bad) and
environment (tropical) potentially influenced sanitary transmission pathway.
• Utilized data from my site (San Vicente, Momos, Toto)
• Varied department (Toto v San Marcos), held environment (mountains)
Result: Structure of model validated, suggesting no detectable difference in
department influence, potentially environment was more important
32. Take Aways
• Trans-regional models show consistency in the floors and flies pathways of
disease transmission
• Regional groupings also appear among Huehue - San Marcos and
Quetzaltenango - Totonicapán
• Between 2012 and 2013 San Marcos appears to improve the most
• Having soap at the handwashing station and improved sanitation was
correlated more often with chronic health indicators (EED/Stunting)
• Data suggest the environment may influence pathogen transmission
pathways more than departments
• In Momostenango, water treatment was correlated with EED cross
sectionally, while an improvement in water source was correlated with an
improvement in EED
• CBA suggests with potentially high discount rate multi-barrier interventions
can decrease cost of interventions and improve long term cost recovery
33. Recommendations
• Economic improvement and food diversification were shown to be consistent
in impacting child linear growth
• Aflatoxins, prenatal health and ‘floors’ were important for child linear growth
• Env. Enteric Dysfunction is either consistently wide spread or not a problem
• Two ways to attack aflatoxins, demand side or supply side; data suggests both
have benefits
• Increasing “value” of fungus free maize (or decreasing “value” of fungus laden
maize) at market level could influence both?
• Multi-barrier intervention approaches with decisions on topic made at
department or municipality level (user-centered design approach)
35. Variable Explanation Scale
ZHAZ Measure of height to age of child and standardized
based on World Health Organization growth charts
Given in standard deviations and normalized to [0, 1]
EED Latent [reflective] variable, created from the
manifestations of Diarrhea, previous day use of oral
rehydration therapy (ORTuse), and total number of
children in the household (TotChild)
Diarrhea; 0 = yes, 1 = no
ORTuse; 0 = didn’t use yesterday, 1 = used yesterday
TotChild; real value given
Diarrhea Has the child had a bout of diarrhea within the past two
weeks?
0 = yes, 1 = no
KnowPregProbs Does the mother know warning signs of problems in
pregnancy?
0 = yes, 1 = no
MomHeight Does the household treat their water in anyway? Real value normalized [0, 1]
CMycoB Latent [reflective] variable, created from the
manifestations of having problems with fungus in maize
crop (CornDisease) and/or problems with the cultivation
of the maize (ProbsCornCulti)
CornDisease; 0 = yes, 1 = no
ProbsCornCulti; 0 = yes, 1 = no
NoFood30Days Has the household had no food in the home at least one
time in the last 30 days?
0 = yes, 1 = no
ChildDietD Child diet diversity based on 24 hour food recall; food
groups add up to find total (label CDDS); scale
incorporates breastfeeding practices and age of child
into score
4 = If < 7 months & exclusively breastfed
CDDS - 1 = If < 7 months & not exclusively breastfed
2 = If >/=7 months & exclusively breastfed
CDDS = If >/= 7 months & exclusively breastfed
CDDS – 1 = If not breastfed
36. Figure 3 – SEMs on factors associated with child stunting in
Guatemala
All data model; n = 1943 Huehuetenango model;
San Marcos model; Quiche model;
37. Figure 3 – SEMs on factors associated with child stunting in
Guatemala
Northern Totonicapán model; Quetzaltenango + Southern Totonicapán model;
Structural equation modeling is our primary tool. Boxes are observed variables, circles are latent variables, arrows are hypotheses, U=unstandardized parameter estimate, S=standardized parameter estimate. Models to be read as regression analysis.