18 December 2015

Catching up with the debate on catch up growth and development for young children

Below are two posts from important researchers working on infant and young child growth.  They both agree that the first 1000 day period after conception is a key intervention opportunity because growth velocity is so high.  They also agree that we should not restrict interventions to this period—we should take whatever opportunity in the life cycle that presents itself to invest in better nutrition.

Where the 2 sets of authors part company is on the nature of the evidence that supports their arguments for investments post the first 1000 day period.  Both authors think such investment is essential (we should never give up on a child!), but the Young Lives authors’ bolster their argument with the fact that for the children in their cohort studies, their standardised height for age scores (HAZ) improve between ages 5 and 8. 

Jef Leroy argues that the movements of HAZ do not reflect improvements in linear growth but are simply due to bigger standard deviations of HAZ as we move up the age groups.  Leroy says a better marker of any catch up growth would be the absolute difference between measured height and the median age- and sex-specific height obtained from the growth standards (HAD).  In addition, Leroy says while height is important for human development, the main story is in the things height is a marker for such as cognitive development and attainment.

If I were the Young Lives research team I would see if I can replicate my HAZ results with HAD.  I would also look at potential catch up in other markers of development such as cognitive attainment scores. 

Either way the policy implications are not very different for the 1000 Day period, but they could be for the post 1000 Day period: interventions to stimulate catch up in growth may be different from those to stimulate catch up in, say, cognitive attainment.  So it is important to air contrasting views.

Thanks to both blog authors and let the debate continue! 

Early is best… but it’s never too late to help stunted children

Paul Dornan and Andreas Georgiadis, Young Lives

The scale of the problem is immense: WHO global estimates suggest 162 million children under five are stunted (that is, too short for their age). Stunting is an indicator of chronic under-nutrition resulting from inadequate food intake, poor health, and poor child care. It is hard to underestimate the consequences of under-nutrition for child survival, health, and development, as it contributes to the deaths of 3.1 million children per year. Improving children’s nutrition should be at the heart of development policy.

There is consensus that good early nutrition is the foundation of children’s survival, growth, and learning. The Sustainable Development Goals reiterate earlier commitments to reduce the number of stunted children under-5 by 40% by 2025. How would it be possible to achieve and surpass this target? Considerable recent attention has been given to nutrition and care during the first 1000 days after conception. Although this period of time is critical, there is growing evidence that improvements in children’s nutrition beyond that early period could yield additional benefits in terms of child development and well-being. That’s encouraging, as we need to make use of every opportunity to improve children’s nutrition. The evidence for this is set out in a new policy brief from Young Lives.

The good news is the attention nutrition has received in recent years. The London Olympics of 2012 was used by the British and Brazilian Governments, along with the Children’s Investment Fund Foundation to initiate the Global Nutrition for Growth Compact, with the stated aim of putting nutrition at the centre of development policy, which has now been signed by 26 Governments. Rio 2016 marks the point at which governments have committed to check in on the progress and to gather more supporters. The same initiative also saw the launch of the Global Nutrition Report, an authoritative advocacy tool providing strong evidence that emphasises the importance of action to improve nutrition.

Powerful arguments have emerged suggesting that experiences during the first 1000 days after conception (up to 2 years old) are critical for children’s later development. Some go further, arguing that beyond the first 1000 days, stunting and its implications for development are irreversible. The prioritisation of a focus on nutrition during infancy has important merits: children are most vulnerable and more likely to die as a result of poor nutrition and diseases during this period. And for those who survive, there are other important reasons to increase the focus on the very early years – with James Heckman, for example, identifying this period as the most efficient point to intervene to secure long-term benefits.

But while prevention is better than cure, there are increasing reasons to think that alone it isn’t enough. The evidence produced from analysis of the Young Lives cohort data, summarised in the policy brief, suggests that – first, while child growth is more plastic in infancy and early childhood, there is still potential for recovery of growth deficits and there is a risk for growth faltering  after the first 1000 days.

Therefore, interventions implemented during the first 1000 days should be sustained in order to promote growth recovery and prevent further faltering. Among the factors found to be associated with changes in growth after the first 1000 days were parental schooling, household poverty, maternal height, and community health infrastructure and these are some of the factors to be prioritised by policy interventions aiming to promote child growth and nutrition.

Alongside Young Lives, other studies have also produced evidence of post infancy growth changes. Analysis of the Cebu cohort study in the Philippines identified change in stunting and growth between age 2 and age 8 years and a recent multi-country analysis published by UNICEF’s Office of Research also emphasised the potential for catch-up growth. Evidence from Young Lives indicates that post-infancy physical recovery is associated with school attainment, suggesting that the gains may extend to other domains of children’s lives. While prevention of under-nutrition during the early days should be prioritised, there is still potential for interventions in later periods to tackle early nutrition and growth deficits.
Headlines from the YL policy brief

So what does this evidence mean for policy and what should change?

1.     Nothing should undermine the central importance of the first 1000 days for child nutrition. But, the focus on the first 1000 days should not inhibit efforts to improve child nutrition in later periods.
2.     While it’s a traditional complaint that researchers always call for more research, in this case not enough is known about interventions which might bring about post-infancy recovery. Improving the knowledge base by trialling or testing post-infancy nutrition interventions may help to open up new opportunities to remedy under-nutrition and some of its consequences.
3.     One stand-out conclusion is that there is more potential in the role of the school for improving children’s nutrition, and that doing so can support better school attainment. Young Lives evidence suggests that school feeding programmes may promote growth recovery, particularly for children who were severely under-nourished in infancy. There is also evidence that nutritional supplementation in school-aged children can yield benefits in terms of children’s learning and performance in school. In short – let us have more thinking about the role of schools as a platform for interventions to support better nutrition.

Our new policy brief summarises the latest findings from an ongoing programme of work. We hope this work will open up new opportunities for sustaining and supporting children’s healthy development, showing that opportunities exist for a longer time window than sometimes thought. For the central message we end where we began: early is best, but it’s never too late.

Counterpoint

Jef Leroy, IFPRI

I read the Dornan and Georgiadis post with great interest.  I strongly agree with their bottom-line conclusions (1) we should continue to focus on the first 1,000 Days (2) but there is a need to also study children’s potential to benefit from interventions after 2 years of age (and pre-pregnancy).  This is the conclusion my coauthors and I reached after analyzing growth of around 300,000 children from 51 countries (see Journal of Nutrition).  

However, I don’t think that the evidence cited by Dornan and Georgiadis really supports this conclusion.

Here is my thinking:

* Height-for-age Z-scores (HAZ) are a useful tool to (1) describe groups of children at a specific point in time (e.g., “mean HAZ in Burundi in 2010 was -2.2”), (2) compare groups of children (e.g., “mean HAZ in 2010 in Burundi was higher than in Rwanda”), and (3) assess changes in groups of children over time (e.g. “mean HAZ in Rwanda decreased from -1.8 in 2010 to -1.6 in 2014/5”).

* But, as we demonstrated in two recent publications (Journal of Nutrition and BMC Pediatrics), HAZ is not the right tool to assess changes in children’s height as they age.  Why?  First, the standard deviations (SDs) used to construct HAZ (they are in the denominator) are cross-sectional in nature (they reflect the variance in height for children of specific age and sex). As a consequence, they are inappropriate to measure changes in linear growth over time (i.e. as children age).  Second, the SDs of HAZ increase as we move up age groups, so we do not know whether changes in HAZ as children grow older are due to changes in the numerator (the difference in the child’s height for age from the median age- and sex- specific height for age of a healthy reference population) or the denominator (SD).  Changes in linear growth as children age should be analyzed using height-for-age differences (HAD), the absolute difference between measured height and the median age- and sex-specific height obtained from the growth standards.

* As far as I can tell, none of the articles referred to in the blogpost (Young Lives analyses, CEBU work, UNICEF study) look at child growth. They assess changes in HAZ as children grow older. The problem with the analyses of these “HAZ trajectories” is that there is no evidence showing that growing children track along specific HAZ-trajectories.

* As there is no “expected HAZ trajectory”, I do not see how one can formulate meaningful hypotheses to be tested or how one can interpret deviations from HAZ trajectories.  If there is literature on HAZ trajectories, the authors should refer to it.

* None of the articles cited define catch-up growth properly. My view is that the only meaningful definition of catch-up growth is (partial) recovery from a linear growth deficit accumulated in the past (see BMC Pediatrics). Catch-up growth in height is only possible when children grow faster than the expected velocity (for their age and sex) so they can make up for the lost growth in height (see this video for a visual demonstration with Duplo blocks). The only way then to assess catch-up growth is in absolute terms (that is, cm), as expected velocity can only be defined in absolute terms (there is no “expected HAZ trajectory”). As a consequence, none of the articles referred to in the post provide evidence of catch-up growth.
Jef Leroy explaining how lost growth can be made up

* Given that HAZ cannot be used to establish catch-up growth, the YL findings boil down to the following. First, children who have a taller mother, whose parents have gone to school longer, who live in less poor households based in communities with better health infrastructure are taller. Second, taller children do better in school. These findings are interesting and relevant, and they confirm what we have known for long time. The studies do not provide evidence, however, of catch-up growth, they do not identify the determinants of catch-up growth, nor do they establish that catch-up growth has long-term consequences in other domains.

The recent focus on catch-up growth potentially distracts from the most important issues. We care about growth faltering as a development outcome per se because short stature increases the odds of obstructed labor; stunted mothers also have smaller babies. That is it. Linear growth is not a causal determinant of poor development, schooling, productivity, etc.  It is a marker.  So if we care about these last outcomes (we do), we should focus on them directly, and not get sidetracked with doing more and more catch-up in linear growth analyses.

3 comments:

Andreas Georgiadis said...


It is true that change in HAZ may partly reflect changes in the denominator but, as the response suggests, change in HAZ may also reflect changes in the numerator that expresses differences between the height gain (changes in linear growth over time) of the child and the reference child during a given period. Therefore, if a child grows faster in a given period relative to the reference child then this is going to be reflected in the change in HAZ. In fact, the only difference between HAZ and HAD, that is argued in the response as the appropriate measure for changes in linear growth, is that the former does not express changes in linear growth in absolute terms (height deficits relative to the reference child are expressed as negative numbers) and that it expresses linear growth in standard deviations of the height distribution of children of the same gender and monthly age, obtained by the growth standards, instead of cm. It is not also clear whether cm or SD is the best measure to express differences, given that the objective is not necessarily to measure height differences per se but any underlying nutritional deficits. This is because, we do not know if a constant relative height deficit in cm as children age reflects a constant nutritional deficit or whether an improvement in the position in the reference height distribution as children age, even when height deficit in cm remains constant, does not reflect any relative improvement in nutrition.


The studies by Outes-Leon and Porter (2012) (add Link) and Singh et al. (2014) (add link), who use Young Lives data, produce evidence that the change in HAZ between age 1 and 5 years was larger among children with lower HAZ at age 1 year. These studies, use various methods to control for the possibility that this change in HAZ is not due to an increase in the standard deviation of the reference distribution as children age. This further implies that the change in HAZ was the result of a change only in the numerator of HAZ. Therefore, the interpretation of these findings is that children with lower initial height compared to the reference child, experienced faster height gain (growth) than the reference child over this period. Based on the preferred definition of catch-up growth provided in the response, as height gain of the reference child of the same age and gender provides an indication of the child’s expected trajectory this is evidence of catch-up growth.

As discussed above the studies by Outes-Leon and Porter (2012) and Singh et al. (2014) provide evidence of catch-up growth. The studies cited using Young Lives (Schott et al. 2013) or other data sets such as Cebu do not identify determinants of height but of growth relative to that of the reference child. Moreover, because of the method used, the determinants identified by these studies do not vary depending on initial height, i.e. they are the same among those with initial height deficit and those with initial height surplus. This is why, these studies identify determinants of catch-up growth. Finally, a number of studies cited in our blog establish an association between higher growth after early childhood relative to the reference child and cognitive achievement and school attainment. The latter evidence is suggestive but not conclusive that faster height gain later in childhood, that reflect improvements in nutrition, leads to gains in learning and other developmental outcomes. Nevertheless, another study cited in our blog (Grantham-McGregor, 2010), that reviews the evidence of the impact of nutritional (iron) supplementation interventions, based on randomized trials, highlights that supplementation has led to improvements in cognitive function among school-aged children.

This is one view. Another view is that linear growth is an indicator of chronic malnutrition that in turn has been found to have causal effects in the short-run in terms of child health, development, and schooling (Maluccio et al., 2009) but also in the long-run on economic productivity and poverty (Hoddinott et al., 2013).

Unknown said...

Dear All,

This is an interesting debate. Having read Jeff Leroy's recent papers and watched the short video, I found the explanation of what should be "catch-up growth" using Duplo blocks quite convincing.
However, I must say that I still haven't made my mind about what the HAD metric represents and what it brings really in our appraisal of changes in population-level growth over time. It seems to me that one flaw in the way HAD works is that it compares the average growth of a population of children to the median growth of the WHO standards, without taking into account their growth status at start.
For example, I'm puzzled by the fact that an imaginary cohort of children who would be born at -1 SD of the WHO 2006 standards (i.e. 48.0 cm for boys; while the median is 49.9 cm) and who would strictly follow the -1 SD growth trajectory within the WHO standards, would exhibit a HAD of 4.7 cm at the age of 5 years (they would be 105.3 cm tall on average, while the median is 110.0 cm). The HAD would then have increased by 2.8 cm. Does that mean that the growth of those children would have worsen?? I'm not sure.
My understanding is that catch-up growth should be defined by a growth velocity higher than the one which is expected given the growth status at the time the follow-up starts, rather than higher than the median growth velocity of the WHO standards.

Fortunately enough, we all agree on the utmost importance of improving children nutrition status during the first thousand days and, as Lawrence wrote, there are no really different policy implications of the way we measure growth deficit or change in deficit for that period.
Looking at determinants of catch-up growth (using one metric or the other) is good; but further research would be welcome to inform us on how children who have exhibited (or not) some degree of catch-up growth perform during adolescence or even adulthood in terms of physical and intellectual development and economic achievement.

Best regards

Yves Martin-Prével
IRD (Institute of Research for Development) - Montpellier, France
Nutripass Research Unit (Food and Nutrition Research in the Global South)





Unknown said...

Dear All,

This is an interesting debate. Having read Jeff Leroy's recent papers and watched the short video, I found the explanation of what should be "catch-up growth" using Duplo blocks quite convincing.
However, I must say that I still haven't made my mind about what the HAD metric represents and what it brings really in our appraisal of changes in population-level growth over time. It seems to me that one flaw in the way HAD works is that it compares the average growth of a population of children to the median growth of the WHO standards, without taking into account their growth status at start.
For example, I'm puzzled by the fact that an imaginary cohort of children who would be born at -1 SD of the WHO 2006 standards (i.e. 48.0 cm for boys; while the median is 49.9 cm) and who would strictly follow the -1 SD growth trajectory within the WHO standards, would exhibit a HAD of 4.7 cm at the age of 5 years (they would be 105.3 cm tall on average, while the median is 110.0 cm). The HAD would then have increased by 2.8 cm. Does that mean that the growth of those children would have worsen?? I'm not sure.
My understanding is that catch-up growth should be defined by a growth velocity higher than the one which is expected given the growth status at the time the follow-up starts, rather than higher than the median growth velocity of the WHO standards.

Fortunately enough, we all agree on the utmost importance of improving children nutrition status during the first thousand days and, as Lawrence wrote, there are no really different policy implications of the way we measure growth deficit or change in deficit for that period.
Looking at determinants of catch-up growth (using one metric or the other) is good; but further research would be welcome to inform us on how children who have exhibited (or not) some degree of catch-up growth perform during adolescence or even adulthood in terms of physical and intellectual development and economic achievement.

Best regards

Yves Martin-Prével
IRD (Institute of Research for Development) - Montpellier, France
Nutripass Research Unit (Food and Nutrition Research in the Global South)