Connecting the dots between human malaria, livestock, and under-nutrition: An unexpected pathway

Connecting the dots between human malaria, livestock and under-nutrition:

An unexpected pathway

Jeffrey K. Griffiths, MD MPH&TMWith data from Nassul Kabunga, PhD

NK: International Food Policy Research Institute, UgandaJKG: Tufts University School of Medicine, TMC GeoMed ID, and

Schools of Nutrition, Engineering, Vet Med March 25, 2015

What do we do?• The mission of the Nutrition Innovation Lab is to discover how

integrated interventions of agriculture, nutrition and health can achieve large-scale improvements in maternal and child nutrition in Asia and Africa and enhance institutional and human research capacity around agriculture, health and nutrition in Africa and Asia …. Blah blah blah

• Seek real, biological linkages: agriculture → → nutrition, health

• Today’s story is all about why cows are good for malnourished children and bad for them because of malaria. Summary: Cow Bednets

Simple Idea – Not Enough FoodLeads to Malnutrition; solution= Food.

• Stunting• Wasting• Small for

Gestational Age/Low Birth Weight

• Micronutrient Deficiency (Fe, Zn, vitamin A, Iodine)

The Intestine

Not Enough Nutrients

• 800,000 neonatal / 3.1 million childhood deaths per year. 165 million stunted children.

• If top 10 nutrition interventions targeted to 34 countries with 90% of childhood deaths …

• Reduce deaths by 15%, stunting by 20%, acute wasting by 61%. (For < $10 billion per year).

Lancet June 2013

Bad News: Lancet review (2013) of how much “food would fix” – not much (20%).

ADOLESCENT, PRECONCEPTION, GESTATIONAL, AND MATERNAL NUTRITIONADEQUATE CALORIES (PROTEINS, FATS, CARBOS) IN ALL LIFE STAGESDIVERSITY OF MICRONUTRIENTS, VITAMINS, HIGH QUALITY PROTEINSOPTIMAL BREASTFEEDING, RESPONSIVE FEEDING PRACTICES, STIMULATIONGOOD COMPLEMENTARY FEEDING 6-23 MONTHS, DIETARY DIVERSITYWEALTH, EDUCATION – [BE SURE TO CHOOSE YOUR PARENTS WELL] Others…..

PREGNANCY EARLY CHILDHOOD

All areKnownRevalentNutritionActions

MYCOTOXINS IN FOOD

HUMAN AND ANIMAL

PATHOGENS

UNHEALTHYINTESTINAL MICROBIOME

MICRO- AND MACRO-

NUTRIENTS

PERMEABLE (“LEAKY”)AND INFLAMMED GUT

Griffiths Innovation Lab for Nutrition 7

MYCOTOXINS IN FOOD

HUMAN AND ANIMAL

PATHOGENS

HEALTHY INTESTINAL MICROBIOME

MICRO- AND MACRO-

NUTRIENTS

NORMAL GUT – NOT PERMEABLE

Griffiths 8

• What are the benefits of having improved cattle? – poverty, income, nutrition

• Malaria in Uganda drives nutrition outcomes such as anemia and stunting

• We find livestock ownership is linked to ↑ malaria in women & children, which makes biological sense; several potential reasons.

• Implications for programming – health and nutrition facets to agriculture and vice-versa

Improved Dairy Cows, Milk Yield, Household Welfare and Child Nutrition Outcomes in Uganda:

Evidence and Impact Pathways

Nassul Kabunga, PhDIFPRI-Kampala Office &

Nutrition Innovation Lab for Africa

Presentation at USAID, Washington D.C.March 9, 2015

FEED THE FUTURE INNOVATION LAB FOR COLLABORATIVE RESEARCH ON NUTRITION - AFRICA

• Milk is a very important dense animal source food, rich in essential micronutrients to complement plant-based local diets

• Households in developing countries face constraints in accessing milk due to limited availability and high cost

• There are several hypothesized pathways for impact due to adoption of improved dairy cows:1. Adoption improves milk yield and supply2. Improved milk yields reduce cost and increase availability improve milk

consumption; also stimulate the integration into modern markets3. Adopter households would improve farm incomes, which in turn can be used

to buy food of better quality (nutritious) /quantity4. Broader effects of improved incomes could be demonstrated through

reduction in food insecurity and poverty outcomes5. Women empowerment if dairy cows are owned/controlled by women

Context: Milk and Improved Dairy Cows in Uganda

Nutrition Innovation Laboratory for Africa

• There is still a gap in understanding the impact (and impact pathways) of improved cow breeds on milk yield, milk sales and broader welfare, as well as nutrition outcomes

• A few studies exist elsewhere:– Nicholson et al. (2004) showed that ownership of dairy cows

increased household-level intakes of dairy products as well as cash incomes (184 households in coastal Kenya)

– Nicholson et al. (2003) and Rawlins et al. (2014) show a possible positive association between child linear growth and ownership of dairy cows by the foster households in Kenya and Rwanda

• But, existing literature presupposes that the benefits of adoption accrue uniformly across small and large farms, ignoring marginal benefits and costs due to scale

Improved Cattle More milk, income ? Nutrition


• We use Uganda National Panel Survey (UNPS) for 2009/2010– compiled by the Uganda Bureau of Statistics (UBoS) as part of a

nationally representative household survey– comprises 2,975 households selected from the sample of

households surveyed in Uganda National Household Surveys—UNHS 2005/2006

• A two-stage, stratified, random-sampling design was followed to generate the UNPS sample– in the 1st stage, enumeration areas were randomly selected from

the 4 geographical regions based on probability proportional to size

– in the 2nd stage, 10 households that had been randomly selected in UNHS 2005/2006 were re-interviewed, except in cases where the respondent could not be traced

Methods - Data


• We aim to measure the average treatment effects on the treated (ATT)the difference in outcomes of adopters with and without the technology

• We assume there is selection bias-employ propensity score matching (PSM) techniques to isolate adoption effects– we implement nearest neighbor matching (NNM) and test the

robustness by kernel based matching (KBM)– we check for sensitivity of unobservable effects by doing

Rosenbaum bounding tests and covariate matching methods– we check covariate balancing tests and impose common support

conditions– (If you are unfamiliar with PSM, this is a way to control for

confounding, since households are matched to similar households)

Methods – Empirical strategies


Results – Descriptive and Econometric ($$$)

Results –PSM causal effects on enterprise-level indicators

Mean Observations

Outcome: AdoptersNon-

adoptersAverage

effect: ATT AdoptersNon-

adopters

Milk yield NNM 232 64 169*** 149 700

KBM 232 69 164*** 149 700

Milk consumption

NNM 66 27 39** 149 700

KBM 66 28 38** 149 700

Milk sales NNM 29 11 18*** 149 700

KBM 29 10 19*** 149 700

References:Improved dairy cows in East Africa can yield up 1500l/cow/yearWHO recommended milk consumption = 200 l/capita/year


Results –PSM causal effects on household-level welfare indicators

Meals NNM 2.77 2.56 0.20** 147 698KBM 2.77 2.55 0.22*** 147 698

Food expenditure (ln)

NNM 10.17 10.02 0.15** 134 700KBM 10.17 10.03 0.14** 149 700

Nonfood expenditure (ln)

NNM 9.60 9.47 0.14 131 668KBM 9.71 9.58 0.13 145 668

Food poverty headcount

NNM 0.14 0.27 –0.13** 134 700KBM 0.14 0.26 –0.12*** 149 700

Nonfood poverty headcount

NNM 0.30 0.44 –0.14** 134 700KBM 0.28 0.43 –0.15*** 149 700

Mean Observations


adoptersAverage

effect: ATT AdoptersNon-

adopters


Descriptive statistics: Household level outcomes- Enterprise level outcomes seem to influence household level outcomes1US $=UGX 2,014

Food poverty Nonfood poverty

0%

10%

20%

30%

40%

50%

60%

13%

27%32%

51%

Adopters Nonadopters DifferenceMeals 2.76 2.46 0.30***


-

5,000

10,000

15,000

20,000

25,000

30,000 29,760

27,960 25,260

15,840

Adopters (N = 159) Nonadopters (N = 747)

Results –PSM causal effects on child nutritional indicators

Mean Observations


adopters ATT AdoptersNon-

adopters

HAZ NNM –0.95 –1.43 0.48* 108 572

KBM –0.95 –1.44 0.49** 108 572WAZ NNM –0.55 –0.70 0.15 108 569

KBM –0.55 –0.70 0.15 109 569


Descriptive statistics –child nutrition outcomes- Nutrition indicators for children are generally poor; several standard

deviations below the reference mean (0)• Children in adopter households are better nourished

HAZ WAZ

-0.93

-0.51

-1.48

-0.83

Adopters (N=120)Nonadopters (N=625) WAZ

HAZ


Results – Heterogeneous effects


Heterogeneous effects by scale –enterprise level

Scale by herd size Scale by farm acreageATT S.E. ATT S.E.

Milk yield: Large 119** 47 312*** 108Small 186* 112 25 33

Milk consumption: Large 65** 31 89*** 34 Small -0.29 7.6 -5.2 9.6Milk sales: Large 18.7*** 5.4 28.7*** 5.2 Small 10.9** 3.9 6.7 3.8


Heterogeneous effects by scale –household level Scale by herd size Scale by farm acreage

ATT S.E. ATT S.E.Meals:

Large 0.13 0.09 0.09 0.08Small 0.30*** 0.08 0.32*** 0.08

Food expenditure: Large 0.09 0.09 0.08 0.09 Small 0.20*** 0.08 0.21*** 0.06Non-food expenditure: Large 0.12 0.11 0.14 0.13 Small 0.11 0.11 0.08 0.10


Heterogeneous effects by scale –household poverty and child nutrition

Scale by herd size Scale by farm acreageATT S.E. ATT S.E.

Food poverty: Large –0.11** 0.05 –0.07 0.05Small –0.13** 0.05 –0.16*** 0.05

Non-food poverty: Large –0.13** 0.06 –0.11 0.08 Small –0.19*** 0.07 –0.20*** 0.06HAZ: Large 0.86*** 0.32 0.80** 0.32 Small 0.22 0.32 0.28 0.29


• We find positive effects on own-milk consumption and milk sales improved dairy cows may help farmers increase milk intakes and integrate into high value markets

• Adoption increases the number of meals consumed and spend 15-21% more on food item adopters are seemingly more food secure than non-adopters with traditional cattle breeds.

• Consequently, adopters reduce chances of falling below the food and non-food poverty line by 10-15%

• We find no effects of adoption on WAZ (underweight) but on HAZ (stunting) children in adopter households are taller!

• By scale effects, we observe milk yields increases are for both small and large farms: but the yield effect is 70% more on farms with small herds. Yields are also higher with large acreage

Summary: Adoption of Improved Cattle



Repeated Panel Survey Data 6 Districts (4 North, 2 South)• 3,630 households Oct-Dec ‘12• Revisited 3,330 in Oct-Dec ‘14• Examines farm production,

income, practices, nutrition, health in Districts where Uganda Community Connector active.

• Obtain hemoglobin (anemia) and rapid detection test (RDT) malaria prevalence

• Focused on children 0-24 months & women of child-bearing age

39

Lira, Uganda. Child collecting water protected springUniversity Illinois/Nutrition Innovation Lab 2013


41


42

MALARIA in 2012MALARIA in 2012


43

Hgbs in children with and out malaria by age 2012

WITHOUT MALARIA

WITH MALARIA


Malaria Prevalence and Hemoglobins in Children 0-24 Months.

• In 2,008 index children aged 0 to 24 months, malaria prevalence increased from < 3% to 50% with age in 2012. Overall mean Hgb values differed significantly in young children with (n=483) and without (n=1450) malaria (9.94 + 1.57 versus 11.20 + 1.56 g/dL, p< 0.001) as well as by month of age.

• Update 2014: similar difference in Hgbs in this age range. 9.85 vs. 11.31, n=216 vs 1,013.

44


45

Green – no falciparum malariaRed – with f. malaria*Kisoro – no index children had malaria

*All differences between children with and without malaria in a given district are significant

DISTRICT SPECIFIC DATA


46

Crude malaria rate in children 0-24 m

LIRA AND DOKOLOHIGHEST RATES

AGAGO AND KOLEMUCH LOWER


47

LIRA & DOKOLOHAD NO INSECT RESIDUALSPRAYING

AGAO & KOLEHAD INSECTRESIDUAL SPRAYING

Effect of IRS on Malaria and Anemia ….Am. J. Trop. Med. Hyg., 88(5), 2013, pp. 855–861


48

2012 vs. 2014Children under 24 months Pos Neg % PositiveDokolo 2012 no IRS 158 132 290 (54.5%)Dokolo 2014 no IRS 92 116 208 (44.2%)p=0.0291 Lira 2012 no IRS 185 162 347 (53.3%)Lira 2014 no IRS 74 97 171 (43.3%)p=0.0397 Kole 2012 sprayed 58 294 362 (16.0%)Kole 2014 35 192 227 (15.4%)p=0.8168, NS Agago 2012 sprayed 50 266 316 (15.8%)Agago 2014 20 254 274 (7.3%)p=0.0014


49

2012 vs. 2014 Children <= 24 months Pos. Negative % Positive

Kamwenge 2012 51 338 381 (13.4%)

Kamwenge 2014 10 166 176 (5.7%)

P=0.0081

Altitude > 1500 meters

Kisoro 2012 0 323 323

Kisoro 2014 0 239 239

NS (no malaria)

Malaria and Stunting•Two recent reports from Uganda risk of stunting with malaria. •Asymptomatic malaria at 1 year of age was significantly related to a worse length-for-age score (-1.40 vs -0.79, p=0.014). •When clinical episodes of malaria in the first year of life counted, the trends for both reduced length for age and for stunting were very highly significant (p < 0.001). Mean LAZ with 0 episodes was -0.73; with 1 episode -0.95; and with 2 or more -1.23. •The risk of stunting for children with 2 or more episodes of malaria was 2.12 (1.38-3.27) times than for children without a clinical episode. (prospective study: Muhangi et al 2013)

•Arinaitwe et al 2012 report that moderate to severe stunting was 24% more likely in children with malaria (p=0.02) after adjustment for HIV status, breastfeeding, chemoprophylaxis, and urban/rural residence (cross-sectional, so )

50

Nutrition Innovation Laboratory for AfricaLivestock and Malaria 2014 data

51

Livestock ownership Caregiver (female 18-49 years)

Index child (< 24 months)

Children aged 24-59 months

Household owns any livestock (incl. poultry)

0.049**(0.016)

0.113***(0.000)

0.172***(0.000)

Household owns livestock (excl. poultry)

-0.005(0.800)

0.064**(0.041)

0.123***(0.000)

Household owns cows 0.045**(0.026)

0.170***(0.000)

0.223***(0.000)

N 2,445 1,029 1,966

***, ** and * denote that the p-value is less than 0.01, 0.05 and 0.1, respectively.Analysis excludes Kisoro because there is no malaria there. Kabunga et al, unpublished.

We don’t know why this relationship between livestock and malaria is present, but it is seen in both the 2012 and 2014 panel sets.


Why should owning livestock be related to malaria in humans?

• Perhaps the mosquito transmitting malaria can also feed on cattle, increasing their numbers near the house (in Uganda, people tend to keep their cattle nearby to discourage thieves).

• Perhaps milking cattle early in the day occurs at a time when the vector mosquitoes are more active – a timing issue.

• Troughs for cattle can be mosquito breeding sites. Cattle footprints may act as tiny pools where mosquitoes can breed.

• Each of these possibilities can be assessed in partnership with entomologists, livestock experts, and malaria specialists.

• Once the reasons are clear, then interventions can be designed to reduce the malaria risk.

52

IRS – only household• Spray households – kill off An. gambiae that

rest inside the house.• An. arabiensis, which also feeds on cattle,

predominates, and since it does not rest inside the house, IRS does little to decrease vector. It bites outside the house.

• Maintains malaria cycle.

53


Anopheles gambiaeThis species complex consists of:• Anopheles arabiensis (zoophilic)• Anopheles bwambae• Anopheles merus• Anopheles melas• Anopheles quadriannulatus (zoophilic)• Anopheles gambiae sensu stricto – evolving into 2

new species from the Mopti and Savannah strains• Anopheles gambiae cryptic subgroup ‘Goundry’

54

Fanello et al, Med & Vet Enteomology 2002

55


56From Kiszewksi et al., 2004. American Journal of Tropical Medicine and Hygiene 70(5):486-498.


Zoophilic mosquitoes bite humans and cattle. Insecticide treatment of cattle or their enclosuresis more protective of humans with zoophilic vectorssuch as Anopheles arabienesis in Africa than the “distractive” effect which putatively has cattle drawing mosquitoes away from human (Franco et al PLoS One 2014).

57

For example: if the relationship is because of zoophilic mosquitoes….

Nutrition Innovation Laboratory for AfricaFrom Dr. David Hole, CDC/CGH entomologist

• VECTOR: “in IRS-sprayed Apac District (6 years), An. arabiensis predominates. In those areas [which are] arabiensis-heavy, peak biting occurs before midnight (Apac) and more often outdoors. In Lira, peak biting occurs after midnight and often indoors, showing a difference in biting activity between these species in these 2 districts which are next to each other in Northern Uganda….. folks are fairly active in the evening. If arabiensis is truly biting earlier as the data suggest, it will occur at a time when fewer people are sleeping under bed nets.

• BREEDING SITES: I have seen some cattle up there, but not in large numbers. I can’t say I know if watering troughs would even breed these mosquitoes, but rain-filled earthen footprints certainly will and if cattle are sleeping next to huts then I suspect breeding sites will be close to huts as well.”

• TIMING: “Hi Jeff, I forgot to add that I don’t know at what hours milking occurs, but if it is early in the morning, shortly after sunrise, I suspect that there is not a lot of anophline biting activity by that time.”

58

‘Cattle Corridor’ in Uganda

Graph comparing districts in and outside the cattle corridor

Data source: Uganda HMIS Districts considered in the cattle corridor: Abim District,Alebtong District,Amolatar District,Amudat District,Amuria District,Apac District,Buhweju District,Bukedea District,Dokolo District,Gomba District,Ibanda District,Isingiro District,Kaabong District,Kabarole District,Kaberamaido District,Katakwi District,Kiboga District,Kiruhura District,Kiryandongo District,Kole District,Kumi District,Kyankwanzi District,Kyegegwa District,Kyenjojo District,Lira District,Luwero District,Lyantonde District,Masindi District,Mbarara District,Mitooma District,Mityana District,Moroto District,Mubende District,Nakapiripirit District,Nakaseke District,Nakasongola District,Napak District,Ngora District,Ntoroko District,Ntungamo District,Otuke District,Oyam District,Pallisa District,Rakai District,Rubirizi District,Rukungiri District,Sembabule District,Serere District,Sheema District,Soroti District

0

10

20

30

40

50

60

Jan

Feb

Mar Apr

May Jun Jul

Aug

Sept Oct

Nov Dec

Jan

Feb

Mar Apr

May Jun Jul

Aug

Sept Oct

Nov Dec

2013 2014

Mal

aria

inci

denc

e (c

ases

per

1,0

00)

Month of year

Inside cattle corridor Outside cattle corridor


Net fencing around cattle“In the absence of cattle nearly 3 times more An. gambiae (p =0.0001) landed on humans. The deltamethrin-treated net significantly reduced (nearly three-fold, p= 0.0001) culicine landings inside enclosures. The sporozoite rate of the zoophilic An. ziemanni, known to be a secondary malaria vector, was as high as that of the most competent vector An. gambiae; raising the potential of zoophilic species as secondary malaria vectors. After deployment of the ITNs a deltamethrin persistence of 9 months was observed despite exposure to African weather conditions. The outdoor use of ITNs resulted in a significant reduction of host-seeking culicines inside enclosures.” 61

Maia et al. The Effect of Deltamethrin-treated Net Fencing aroundCattle Enclosures on Outdoor-biting mosquitoes in Kumasi, Ghana.PLoS ONE 7(9):e45794

Enclosure hasDeltamethrin-impregnated netsaround it


For example: if the relationship is because of zoophilic mosquitoes…. Consider residual spaying in housesAND spraying cattle enclosures with Deltamethrin. Suggest: promote livestock for nutrition benefits ANDby understanding pathway, malaria transmission potential while promoting cattle health

62

X X

X

X

X

Thesis:• Livestock ownership can improve human nutrition.• Residual insecticide household treatment clearly

decreases human malaria burden (death, morbidity, anemia, possibly stunting).

• Owning cattle can increase likelihood of humans having malaria in Uganda. (Unclear why yet).

• Integrated programs ‘should’ perhaps promote improved livestock ownership AND promote insecticide treated enclosures just as IRS is promoted, or remove breeding sites, etc. Tactic will depend upon the reason for the relationship.

• New agriculture → nutrition & health pathway63

Just some of the actors!• Edgar Agaba• Joyce Kikafunda• Bernard Bashaasha• Flo Turyashemererwa• Annet Kawuma• Maura Mack (USAID)• Nassul Kabunga• Eileen Kennedy• …. Many others

• Shibani Ghosh• Liz Marino-Costello• Luke Ascolillo• Patrick Webb• Robin Shrestha• Chris Duggan• Wafaie Fawzi• Nilupa Gunaratna• Gerald Shively• William Masters

64

Thanks!

Questions: Jeffrey.Griffiths @ tufts.edu

Photo: JK Griffiths Tanzania 2008 65

Documents

Connecting the dots between human malaria, livestock, and under-nutrition: An unexpected pathway