Which wealth groups experience the greatest reductions in stunting? New multi-country paper

When we report stunting rates we tend to focus on the entire population.  

But we know that wealthier households tend to have lower rates of stunting.  This is because higher household wealth can buy more and better food, clean water, improved sanitation and higher quality healthcare.  It is not a perfect relationship by any means, but it is regular. 

But what is happening to this disparity in stunting rates across income groups as overall stunting rates improve?  And the lowest wealth groups reducing stunting more quickly than the highest wealth groups or is the opposite true? A new paper using MICS and DHS datasets and authored by Caryn Bredenkamp and colleagues, published in the International Journal of Epidemiology, answers these questions.

The answers tell us something about the pro-poor sensitivity of nutrition actions.  Countries that do better at reducing stunting in the wealthier rather than the poorer groups have nutrition plans that are geared to addressing the easier cases.  Those that are better at reducing stunting among the lower socio-economic status groups have figured out a way of reaching the poorest children in an effective way.  

Which way is better?  It seems obvious to conclude that the second way is preferable.  And work by Carlos Carrera and his UNICEF team suggests that the benefit cost ratios are higher if we focus on the poorer groups.  But is it so straightforward?  First, every child has a right to avoid stunting.  Second, generating results in the higher wealth groups may generate greater political support for stunting reduction and result in more sustained funding to reach those in the bottom wealth groups.

Anyway, what did the authors find?  Using a concentration index to summarise stunting disparities across different wealth groups, they conclude:

* Out of 53 countries, 31 show no statistical change over the past 10 years or so in the concentration index.  11 countries show a widening in inequality of stunting outcomes (many from South Asia) and 11 countries show a narrowing of inequality of stunting outcomes (the Europe and Central Asia region and the Middle East and North Africa region). 

* Most sub-Saharan African countries show a constant level of stunting inequality

      But the concentration index is a summary statistic.  It can improve for a variety of reasons, for example because the very poor are catching up with the rest, or the middle level wealth groups are catching the wealthier but leaving the poor behind.  

      When we examine the data by wealth quintile the paper shows that for the countries where inequality is increasing, it is because the wealthiest group is doing much better than the rest in reducing stunting (India and Ethiopia are particularly vulnerable in this regard).  

      For countries where inequality is reducing, there are fewer consistent patterns: sometimes it is because the bottom quintile is improving, sometimes the bottom two quintiles and sometimes because the upper quintile is showing a worsening level of stunting.

      What are the policy implications of these findings?  The paper does not draw out any, and that is probably because it is difficult, for the reasons given above.  Ideally we want all groups improving, with public policy geared more to the poorest groups because (in theory) children in the wealthier groups have more private resources to draw on.

      But for me this paper raises the question, yet again, of why we prioritise a stunting rate.  

      The stunting rate simply counts whether a child is above or below the -2 threshold for standardized height for age (i.e. if ZHA <-2) The equivalent in poverty measurement is P0, the poverty rate.  But as any poverty economist will tell you the poverty gap (P1, or the average gap between income and the poverty line) and the poverty gap squared (P2 or the average of the squared poverty gap—even more sensitive to large gaps between income levels and the poverty line)--are more meaningful measures of the poverty sensitivity of policy and programmes.  

      I don’t know why we do not calculate the average ZHA gap (S1) and the average ZHA gap squared (S2) as well as the rate of stunting (S0).  The mortality, morbidity and economic consequences of being at -2.5 ZHA are surely more serious than being at -2.1 ZHA.  We care if a child moves from -2.5 to -2.1, but it would still be classified as stunted.

If we calculated S1 and S2 we could understand more about progress and its distribution than we can with our current stunting measure (S0). I have seen it tried once, by Saul Morris (an excellent addition to the rapidly strengthening nutrition team at the Children's Investment Fund Foundation) but nowhere else.

      Why not?