1

Patrick Webb

Kathmandu, August 2013

What the Lancet Nutrition Series of 2013 tells us (and what it doesnt)

2

Dozens of contributors 120+ pages of content (with yet more online) Complex analyses

3

Coverage rates for [many] interventions are either poor or non-existent.

Factoids from Lancet Series 2013

The cost of scaling up needed interventions to 90% coverage is US$96 billion per year

20% of stunting by 24 months attributed to being born SGA

165 million children under five are stunted.

Undernutrition is responsible for 45% of all under five child deaths

10 interventions at scale prevents 1 million child deaths and averts 20% of all stunting

4

Evidence for intervention impacts updated 34 countries = 90% burden of stunting [36 in 2008] Nutrition actions in emergencies discussed. Moderate wasting included [not in 2008] Impact on mortality and stunting [vz DALYs] Obesity discussed [no evidence of interventions] 10 interventions modeled to assess impact and cost

of scaling up [90 vz 99% in 2008]

What was different in Lancet 2013 (from 2008)

5

6

Mortality in children younger than 5 years by 15% (range 9-19%)

Stunting by at least 20% (range 11-29%)

Severe wasting by 61% (range 36-72%)

Current evidence and modeling of impacts 10 interventions @ 90% coverage can reduce:

But, thats only 20%. That leaves 80% of the stunting problem to be solved!!! Enter

nutrition-sensitive programming, multi-sector integration, and enabling policy environments.

Source: UNICEF (2013) Improving child nutrition

Nepal today

20% cut

0.1

.2.3

.4.5

dens

ity

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6height-for-age z-score (stunting)

Mountains Hills Terai

Source: DHS 2006

Children below 5 years (n=5,237) by agroecological zone (from left to right, means = -2.27, -2.02, -1.89)

Height-for-age Z-scores (stunting) Nepal, 2006

9 Source: Shively (2013)

PresenterPresentation NotesLets make sure everyone is comfortable with Z-scores

One consequence of low levels of development is widespread malnutrition.

Z-scores are standard indicators of nutritional outcomes in children.

HAZ is a measure of long-term nutritional adequacy (low HAZ = stunting).

WHZ is a measure of short-term nutritional adequacy (low WHZ = wasting).

Z-scores are constructed by comparing a childs growth for a given age or weight to the median in an adequately nourished reference population. The Z-score is calculated and scaled to represent standard deviations.

A natural question to ask is what combination of observable features explains these nutritional patterns?

10 Source: Black et al. (2013) Lancet Series 2 Paper 1

11 Source: Ruelet al. (2013) Lancet series 2 Paper 3

Nepal

13

14

15

A doubling of per capita income from agriculture is associated with 15-21% point decline in stunting.

1. Productivity 2. Diet Quality 3. Empowerment

4. Food system safety 5. Delivery platforms

Agriculture to Nutrition Pathways

16

17

18

Research priorities in 2013 Lancet Series Paper 1: What role of dietary factors in stunting? Paper 2: Rigorously test feasibility of integration of multisector interventions versus co-location?

Paper 3: Can health workers effectively assess and interpret height-for-age measures?

Paper 4: What strategies are effective at enabling multisectoral coordination and coherence for nutrition?

More than 60 research priorities defined in the 4 main papers.

19

We know more than ever what kinds of things work; lets get on with it! But

generate rigorous evidence in doing so.

What is needed, feasible and cost-effective is context- and need-specific. Tailoring!

Its never either/or! We need targeted treatments, universal prevention, and nutrition-sensitive

actions. The latter requires much careful research.

Please bear in mind: a) High bar for admissible evidence!

[debate on legitimate sources of evidence]

b) Other interventions possible!

c) Not all actions needed (in every setting)!

d) Evidence base still evolving!

Limitations should be : A large proportion of the evidence on interventions is still derived from efficacy trials as opposed to effectiveness studies and hence variations exist in estimates of effect size for various interventions. Few robust assessments in programme settings and available data from observational studies do not permit ready assessment of intervention effectiveness.

Nutrition effects resulting from agricultural and other food system policies and programmes are very difficult to assess with RCTs, partly because treatments cannot be randomised and because the effect pathway is long. Yet the most promising opportunities for improvement of health and nutrition are undoubtedly found in such policies, and not in home gardens and other minor projects which are amenable to study within the framework of randomised trials.

Pinstrup-Andersen (2013) Commentary on Lancet Series 2

23

Lives Saved Tool (LiST)

LiST is a linear, mathematical model that describes fixed associations between inputs and outputs that will produce same outputs each time model is run.

Outputs are changes in population level of risk factors (such as wasting or stunting rates, or birth outcomes such as prematurity or size at birth) and cause-specific mortality.

Model assumes that changes in distal variables, such as increase in income per person or mothers education, will affect mortality by increasing coverage of interventions or reducing risk factors.

24 Source: Black et al. (2013) Lancet series II Source: Black et al. (2013) Lancet series II

Slide Number 1Slide Number 2Slide Number 3Slide Number 4Slide Number 5Slide Number 6Slide Number 7Slide Number 8Slide Number 9Slide Number 10Slide Number 11Slide Number 12Slide Number 13Slide Number 14Slide Number 15Slide Number 16Slide Number 17Slide Number 18Slide Number 19Slide Number 20Slide Number 21Slide Number 22Slide Number 23Slide Number 24