A live conversation between Jan-Maarten Wit (Leiden), Andrea Merker (Stockholm ) and Robert M Malina (Austin), moderated by Ivo Arnhold (Sao Paulo, Brazil)
Contributors: Abiola Oduwole, Ze’ev Hochberg, Mark Sperling, Martin Ritzen, Alexander Jorge
Transcribed from the live conversation by Uri Klempner
Ivo Arnhold: We start with 1) the theoretical considerations on girls and boys growth references versus standards by Robert Malina, followed by 2) WHO growth standards by Andrea Merker, followed by 3) theoretical reasons why at least at the present time the global standard is not likely to represent all healthy people on the globe by Jan-Maarten Wit, followed by 4) the pros and cons of using WHO standards in countries that have their own national growth references and brought by Andrea Merker as pro and Jan-Maarten as con, and then 5) the pros and cons of using WHO standards in countries that do not have their own national growth references by Bob Malina as pro and Jan-Maarten against. We want to make sure that that at the end we have enough time for the general discussion and suggestions for future communication between scientists and the lay public on this topic. So to start I will pass it on to Bob.
1) Theoretical considerations on girl´s and boy´s growth references versus standards
Robert Malina: I’ll talk a little bit from experience because back in the late 60s early 70s I trained the technicians for several of the American health surveys. I did the anthropometry field training for them and so on. And these in many ways were the precursors for the US growth charts first in the late 70s, the first version by Peter Hamill et al. in 1977 (Vital Health Stat 11.1977;(165)i-iv, 1-74) and then a 2000 version (Kuczmarski et al, Vital Health Stat 11.2002;(246):1-190.). In the context of growth charts, the charts are a reference for comparison.
Very often people talk about growth standards. The word standard is a problem because a standard implies this is what you should strive for. It is a value judgment. What is the standard? What’s the standard of beauty? and so on. So to me, growth charts are a point of comparison, and any time you come to the perspective of a point of comparison it is a choice. Why am I comparing? Am I comparing a single child? Or am I comparing a sample of children? and so on.
This is part of the issue and in the development of the US charts; by way of background they are based on five or six consecutive surveys and there was no secular change across those surveys in the US. The samples initially were selected to represent the American White and American Black population and then eventually they oversampled Blacks, and then Hispanics to fill out the Hispanic part of the American population (the Hispanic Health Survey). And so the unique feature of the U.S. reference data is that it is based on an ethnically heterogeneous group, which is probably suitable for the U.S. population, seeing as evidence for genotypic differences in height are not that marked among the different ethnic groups. One advantage of the U.S. reference values is they are based on adequate sample sizes; you are talking five or six hundred children per age group in each sex. It is thus a good probability that the 95th and the 97th percentiles are real children. On the flip side, if you have proportionally small sample sizes, the upper percentiles are projections which may or may not be real. So that is a unique thing for the U.S. dataset. And of course it is also a heterogeneous population and so on.
But what many people do not realize about the American growth charts: the situation for body weight is altogether different. In the last survey before 2000, this was perhaps the beginning of the obesity overweight epidemic in the US. So when they updated the US reference charts in 2000, they did not use the body weights of the last survey. Why? They made a public health decision that this (i.e., increase in body weight relative to height) was not healthy; unfortunately, many people have not read the background of the US charts, which is often overlooked in discussions of the US growth charts. The WHO charts are a little bit different because part of the charts are based on the original 1977 US charts.
Of course, growth charts are based on cross-sectional data. It is a cross-section of the population. Now, growth charts based on longitudinal data for individual youngsters are altogether different. Many of you use Tanner’s modifications of the American charts (Tanner and Daviers, J Pediatr 1985;107:317-29). The charts were based on the Harpenden data (UK) shifted to the left for early maturation to accommodate US children and youth. It shows basically the British data shifted to the left to adjust for potential population differences in maturation. And of course, any time you use a growth chart during adolescence somehow or other you must consider inter-individual differences in pubertal timing and the growth spurt. You are more aware of that than I, because you work with individual cases. I am working with samples, which is a very different perspective.
Another concern is placing growth charts in the context of secular change, the tendency for larger size and early maturation. In the U.S., we have had no secular change in height, on average, through the 2006 and 2008 surveys. Medians are pretty much superimposable over time; the differences are relatively small for height. Weight is a different story. I am focusing largely on height because it is what most of us work with.
Secular changes of course begin in childhood; they are not just limited to adolescence. This is relevant to the most recent version of Tanner-Whitehouse 3 skeletal age protocol (TW3 SAs). The rationale for changing the protocol was to accommodate secular trends to larger size and earlier maturation. Yet, the secular change in maturation in TW3 SA is only apparent in adolescence not before. But, data from Belgium and also the Netherlands suggested a secular increase in height but no change in the timing of puberty and the growth spurt (see Malina et al., Sports Med. 2017 Nov 21. doi: 10.1007/s40279-017-0827-7). Moreover, in many places the trend has largely stopped. Japan is a good example and for many years. The US is the same. So, as far as I can see, secular changes are real and are taking place in many populations, but in the traditional populations considered (largely populations of European ancestry), the secular trends have slowed considerably or stopped. So, with this, I will stop.
Jan-Maarten Wit: With regard to the situation in US where you have no secular change: is this because of the influx of other ethnicities? If that is so, I would expect a larger standard deviation over time. So, have you seen a larger SD in the last 50-60 years in the U.S?
Malina: Standard deviations or the percentiles in different surveys pretty much overlap consistently. And of course, most of these were done prior to the big influx, because the most recent influx occurred around 2000 and thereafter. We are a nation of immigrants so keep in mind this is always a potential confounder. On the other hand, it is an indicator of real human variability. So in many ways the percentiles of the U.S. charts I think can accommodate a good deal of it. We are going to miss a few but that is to be expected.
Ze’ev Hochberg: In a recent report on the secular trend, in 1896 Americans were the among the tallest in the world, whereas in 1996 they were around number 50, mostly due to the minimal secular trend among Americans (Elife 2016;5:e1341).
2) WHO growth standards
Andrea Merker: I am a Ph.D. student from Karolinska Institutet in Stockholm. My Ph.D. project focuses on growth curves and growth in extreme short stature.
To give a short background on the WHO curves, this project started in the 1990s. A group of experts considered that the growth pattern of healthy breast-fed children was significantly different from the NCHS growth reference from the 1970s. This expert group concluded that growth during infancy and early childhood ages is globally quite similar, and that growth during these ages is driven by the environment including nutrition.
This was the starting point and most probably also a reason for why the WHO multinational growth reference study (MGRS) focuses on ages 0 to 5 years. The aim was thus to produce a global “standard”, i.e. a reference showing normative growth irrespective of background population. Another motive for developing a growth standard might have been merely political, meaning to promote breast feeding as growth of the breast-fed child should be the biological norm.
Also, I think we have to distinguish between the (so-called) standard for 0 to 5 years on one hand and a reference for 5 to 19 years on the other hand. Only the first part of the charts, meaning ages 0 to 5 years, is a product of a prospective multinational study, based on 8500 children from Brazil, Ghana, India, Norway, Oman and the US, with a longitudinal part 0 to 2 years and cross-sectional data covering ages 2 to 5 years. The second part is based on the data from the (US) NCHS77 reference, as Bob has already mentioned. The NCHS77 growth reference is based on US data from the 60´s and 70´s. The full NCHS77 reference was previously used as the official WHO growth curve. A smooth connection between the standard and reference part was created by mixing data points from both data sets during overlapping ages.
Regarding weight for age, the WHO presents data only up to 10 years of age and recommends following BMI for ages thereafter.
The WHO standard has, according to WHO, been endorsed by more than 140 countries; often by national breast-feeding centers but also by the International Pediatric Association and the International Union of Nutritional Science. To what extent the WHO charts are used in these countries is uncertain. In the US, for example, the WHO curves are recommended only between 0 and 2 years of age, for older ages CDC2000 growth reference are recommended. In UK, the WHO charts are recommended between 2 weeks of age and 4 years followed by the UK90 growth reference. In Canada, the WHO charts are recommended between 0 and 19 years.
So then of course we can discuss differences between references. Studies have shown that about 60 percent of children in Bangladesh aged 0 to 5 years have a height below minus 2 standard deviations of the WHO standard. Dutch children, on the other hand, are at five years about 0.7 standard deviations taller. And I think that at adult height Dutch people are at least one standard deviation taller.
There are always differences in methods between references including how the data was collected, inclusion criteria, and data presentation (e.g. SD or centiles), data exclusion rules or smoothing techniques. One should also discuss what differences are clinically meaningful although data might be statistically different.
Lastly, it can sometimes be difficult to compare differences due to differences in growth tempo.
Wit: The difference between the Dutch nation-wide reference and the WHO charts is even more than 0.7 SD: young adults are on average 8 centimeters taller than the mean adult height derived from WHO charts.
Merker: A systematic review (Natale & Rajagopalan, BMJ Open. 2014;4:e003735) comparing the means of the WHO child growth study to other studies with similar methodological background showed generally small, maybe negligible, differences for height. Weight varied more than height. Development of head circumference showed most differences.
3) Theoretical reasons why at least at the present time the global standard is not likely to represent all healthy people on the globe
Wit: I will talk about some theoretical arguments. I believe that WHO is actually not suitable for the whole world. The dogma of WHO preparation is that the actual environment at present is the only factor important for stature. And that ethnicity, DNA difference and so on do not play any role at all. This has been the dogma also in their papers. I would like to challenge this dogma.
It’s also worthwhile to mention before that there are also bio-statisticians who say that there was some selection bias in the WHO study, because if breast feeding was not continued long enough, the babies were excluded from the study. The children who were tallest stayed in the study longer than the ones who were in the lower half of the population. So you have already a sort of selection there and that has been demonstrated by the Dutch statistician Stef van Buuren (Nestle Nutr Workshop Ser Pediatr Program. 2010;65:167, Discussion 175).
Now, about the four arguments that I would like to propose that are speaking against the WHO dogma. In the first place, if you look at data of heights in the whole world, the difference between the shortest countries or ethnicities and tallest ones is about 20 centimeters. So that’s the whole range. It is about three standard deviations, if you think in terms of adult height. But 20 cm is the real measured difference actually between two countries. That’s a lot. That’s of course the starting point.
What about the secular change? The whole concept of secular change is not mentioned in any of the WHO papers. They do as if it doesn’t exist. Now, it is certain that there is a secular change. I think everybody here in this room knows that. There is more secular change in the high income countries than in the low income countries. In many countries, like indeed in the US, there is no or hardly any secular change. As far as I know the US is the only high income country where there is no secular change. Most other high income countries still show a secular change, although in some (like Sweden and the Netherlands) height appears to have reached the maximum for the given genetic background. In some low income countries a secular change is seen over the last decades, but not in all. Secular change is probably not only positive. If one looks back at long term history there is a sort of wave that can go up and down. In European countries the change has been positive in the last 150 years, but before that time it has not always been positive.
Hochberg: In the comparison of 1896 to 1996, the secular trend was negative in some countries in Africa.
Wit: Is it? In the paper that I saw, secular change in African countries was actually not negative and not positive. Mean height has remained stable over time. Of course one has to look at the quality of the data in the studies, because it’s fairly difficult to do a good study in such countries. The methodology is often of unknown quality.
So what about secular change? How fast is it? In general, the secular change is very slow. Even in high income countries it is about 1.2 cm per decade. That is not much. So if you look at the initial difference between the countries of 20 cm, it will take multiple generations for a country that is at the low side in this distribution to get close to the average of all countries. Secular change is not only very slow, but it is also not continuing forever. In Sweden, I think in the 1980s the secular change has stopped. In the Netherlands it stopped in 2000. Average height in the 4th nation-wide growth study in 1997 and the 5th growth study in 2010 are completely equal (Schonbeck et al, Pediatr Res. 2013;73:371). So we believe at the present time that the secular change in Holland has now stopped, maybe because we have reached our optimum for our DNA. But there are also people who think that there are negative environmental changes, for example that people do not drink enough milk anymore.
What would the secular change be if a low- or middle income country would be able to rapidly guarantee optimal living conditions, including optimal food? Could one expect that in 2030, as is the aim of WHO, stunting would be out of the world? How would it ever be possible, even if there would be an optimal secular change, to have that aim actually accomplished? So this is the first thing about secular change that I would like to mention.
The other thing is that secular change is something that happens in the whole population. So also the healthy well-fed rich children in the 1850s were short. Actually the healthy well-fed German children in a Hamburg school in that time in 1860 were as short as healthy well-fed Indonesian boys at the present time (Hermanussen and Wit, Horm Res Paediatr. 2017;88:38). So the position of the whole distribution was changed. So, apparently it’s not the actual feeding pattern of a person, but there’s something changing very slowly about the whole distribution of a population.
It is also interesting that the secular change does not respond very quickly to circumstances. You have the example of the last World War: German males who were born in 1920-1925 had some hold up of the secular change but very short and only neutral. It was not going down, it just stabilized, and then it went off again in the later birth cohorts. And we saw the same thing in Japanese students. So there is some effect of the environment on the secular change, but it’s actually small and only temporary.
There is the assumption that it is more the social economic conditions that have to do with the secular change. However, there may be other factors. A recent study showed that tall men had a relatively high number of children. So, this may suggest that secular trend may be associated with selection for gene variants that are associated with tall stature. However, my colleague Michael Hermanussen commented that this could be an exception, because normally people who have more money have less children than people who have low income.
I still think that there might also be genetic differences. As I mentioned, WHO said that doesn’t play any role, and that it’s only the environment. But there are some indirect data that genetic differences exist. It would be strange if with more than 400 genes that are associated with height, ethnicities would not differ in the distribution of height-associated gene variants, while there are so many other physical characteristics that differ between ethnicities.
For example, South Asian children living in the Netherlands for more than two generations are still much shorter than the ethnic Dutch, while they have excellent economic conditions. So it appears likely that the difference in height is not related to the environment or the food they eat; it’s more likely that there are genetic differences associated with body stature between children of South Asian and northern European origin. In line with this argument, in England and in the Netherlands babies of South Asian origin are about 400 g lighter than the average of all other ethnicities. So the growth pattern seems to be rather special in South Asian children, which has probably little to do with the environment. (Visser et al, Early Hum Dev 2009;85:737).
Besides differences in stature, there are also differences in body shape. Japanese boys are almost at WHO level half way through adolescence, but they mature earlier and their adult height is about 5-6 cm shorter than WHO. Although they have already had a very long period of positive secular change (that has now almost stopped actually), they are still shorter than WHO in adult height. So that is also in my opinion an indication that there are genetic differences that have something to do with height. So the idea of WHO that only the current environment is the determining factor of body stature is incorrect, in my opinion.
4) Pros and cons of using WHO standards in countries that have their own national growth references
Pros
Merker: So, about the pros of the WHO curve. We decided to differentiate between countries that have a long-standing growth charts’ tradition like in the Netherlands and other Northern European countries, and countries that don’t have own national curves. In Sweden, for instance, we have a long tradition of our own growth curves with a special curve layout.
First, I’d like to highlight that the WHO growth standard is an extensive work using a very sound method. And yes, a lot of children have not been included in the data analysis because the mothers did not comply with e.g. the breast feeding or non-smoking criteria. For the longitudinal part (0-2 years), data from only 13% of the recruited children was used for the reference construction. And for the cross-sectional part it was only 31% of the data. Also, extreme data were excluded by removing the upper 3% from the cross-sectional sample before the final curves were modeled. But still I think it has been an extensive and well-structured work, and that these curves serve as a good “matrix” for following an individual’s growth as well as for comparing different populations towards a fixed standard.
Secondly, we should have in mind what are the curves used for. Are they for clinical use and for following an individual child? Or are they used for public health services or epidemiological surveys with the intention to compare populations for instance. I think if we would like to compare populations with each other, then the WHO curve is a very good benchmark. If 60% of children in Bangladesh grow below -2 SDS, this also gives a picture of growth and health in Bangladesh compared to other countries in the world, and might suggest that public health initiatives are still needed.
Also, with the WHO standard tables and charts that are published one can easily calculate standard deviation scores. One can easily express and illustrate the growth pattern of a certain child, which makes it easier to discuss the growth pattern of an individual with any colleague around the world.
For clinical practice, the child’s growth pattern in relation to the growth pattern in the family might be more important than the absolute position in any growth chart or a potential position above or below a certain cut-off line. In that sense, any growth curve might be seen as a “matrix” for following the individual’s growth. It might therefore be an advantage to stay with a single matrix, for example the WHO matrix, instead of shifting between different references.
In short, I think that the WHO growth standard/ reference is quite good for comparison purposes and for research.
Cons
Wit: I would like to focus on the situation of the countries where height is considerably taller than the WHO standard. I am speaking about the Netherlands, Denmark, Sweden, Norway, Germany, etc. Adult height in the Netherlands is 8 cm taller the WHO standard, and young adults in Germany and the Nordic countries are probably 6 cm taller. I also focus on the clinical use of the charts. If one would use WHO standards instead of the national reference diagrams, one would lose sensitivity to pick up short stature syndromes. This has been documented for Turner’s syndrome quite nicely in Norway and Finland. In the Finnish study, by the age of 2 years 72% of girls with Turner syndrome have height measurement below the 3rd percentile using national reference charts, in contrast to only 36% using the WHO standard (Saari et al, JAMA Pediatr. 2013;167:194). At the opposite part of the spectrum, if for a tall child one would use the WHO standard instead of national reference charts the specificity would be lower to pick up pathological causes of tall stature. For North-European countries, it would be unwise to use the WHO standard. Just because clinicians would like to use growth charts in the clinic and pick up pathology.
Abiola Oduwole: We’ve also been worried because we’ve found out that the WHO chart is good for us. Using the WHO chart it is now obvious that stunting is not as common as previously observed. Environment do affect growth, as studies comparing rural/urban or high and low social class show more stunting in the rural and lower social economic class. In fact, children from the upper economic class as shown by our studies have similar curves with CDC chart. This has also been observed as previously stated in the US. Despite these observations we’re worried that we need to have a growth chart specifically for ourselves, that we may be missing out short stature, tall stature children just because we’re using the standard. We are now using our own standard in the clinic so to pick up those kind of children. The chart needs a review and more participants and should be longitudinal to pick out secular trend and influence of the environment and culture. Nigeria also has one of the highest prevalences of children with heamoglobinopathy. Most of these children are classified as stunted or short stature. Developing a growth chart just as it was done for Turner syndrome and Downs Syndrome may be worthwhile. So, I agree that each country should actually have their own growth charts, which they could use for their own citizens. Then, maybe the WHO charts should be for comparing countries; that is OK, but to observe the children of a country, to be fair and not miss out the two extremes, personalized growth is advocated.
5) Pros and cons of using WHO standards in countries that do not have their own national growth references
Pros
Malina: I will speak largely in the context of Mexico. I began working in Mexico in 1968 studying indigenous children in the south of Mexico Oaxaca, which felt the last earthquake severely. And I worked there most of the 70s and I was back in the field in 2000 and was back just a few years ago. But I will share with you data from one community that I studied in 1968, 1978, and again in 2000. And these are schoolchildren of 6 to 17. I chose to use the NCHS (the U.S. reference) as my point of comparison.
I will use the boys 6-11 years as an example. The data are age adjusted. In ‘68 and ‘78 their heights were well below the fifth percentile of the US reference. For example, the average for boys in ‘68 was 117 cm, and in ‘78 it was 117.5 cm; the U.S. fifth percentile is 123.8 cm. This gives you some idea how short they were. But their body weights were at the U.S. fifth percentile – 21.3 kg in ’68 and 22.2 kg in ’78; the U.S. fifth percentile was 22.7 kg. As a result, the BMI in 1968 was at the U.S. 25th percentile. And in 2000, it was at the U.S. median. The same trend was apparent in girls and for adolescents. So, I am working with a very short population that has weights that are almost too much for their heights; this is characteristic of indigenous children and of indigenous populations in Mexico, especially in the south. The state of Oaxaca has 16 different indigenous (linguistic) groups.
And historically in Mexico, the heights of indigenous adults decrease from north to south. This was first shown in the 1890s, and has been subsequently replicated. It is still true of contemporary indigenous children. Using data from a special school program for indigenous children in 2012, those in the north of Mexico are the tallest; heights decrease somewhat into the central part of the country, but once you get to the south and south-southeast (Yucatan peninsula) they are, on average, the shortest. Mexico has no growth charts. They have one but it is based on the work of Ramos Galvan which is based on children from his private practice in Mexico City and do not apply to indigenous children.
I want to raise the question regarding the overweight issue. What do population differences have to do with the definition of overweight? In the group that I work with, they also have proportionately short legs. And what is the impact of proportionally short legs (or a longer trunk) and short stature on the significance or meaning of the BMI? I do not know the answer but that is something we need to consider.
Oduwole: Could that be genetic?
Malina: I would say probably a combination. I give you a case in point in this single village. The youngsters made a significant gain in height from ‘78 to 2000; 6 cm gained over 20 years. But what happened in the village at that time? Well in the mid-80s, the village went through the epidemiological transition, meaning preschool mortality was finally lower than adult mortality. So something happened in their lives. All these youngsters were born after the epidemiological transition. So conditions are improving. What they are specifically? We can hypothesize diet, they now have a doctor coming to the community every day, and have a nurse there seven days a week. So things are changing but ever so slowly; it is going to take another generation. Hopefully, I will be around to do another follow up!
Wit: 6 cm in 20 years; that’s not so slow.
Malina: No. That is major, it is a major gain, and in the younger children it is mostly in the legs, in the adolescent boys just the legs, and in girls it is still ambivalent. So there may be a sex difference. I do not know the answer but I am just telling you what I see in the field and in the data. There is a slight secular change in adult height. But that began probably among those born in the 50s and 60s. So it is changing but we do not know the answer.
Mark Sperling: So, you chose South Mexico with an indigenous population which would be different from Mexico City with its Spanish and French and now other European admixture. But all over the world now my impression is there’s a lot more admixture of genes with population shifts than there was even 50 years ago. And even in Eastern European countries, which I just visited recently, I see faces that are clearly Middle Eastern or African that I’m sure didn’t exist until the 80s 90s maybe 2000. And so, has there been a change in the admixture of the genes that you know in Oaxaca rather than just environment? Jan-Maarten said that WHO claims it is environment rather than a change in the genetic admixture.
Hochberg: On a similar topic with regard to national growth charts: countries don’t have really any biological significance. A country is a political entity, and the fact that we group its children together into a reference doesn’t mean much. I was convinced by Jan-Maarten today that for countries like northern Europe or the Scandinavians it makes sense to use the country specific nation reference. But this is a small enclave with regard to the rest of the world. If you think about the boy in Bangladesh who is growing very steadily on the 25th percentile of his national reference. I don’t know if they have a reference for growth. He’s stunted, this boy. So the information of stunting of that boy who apparently grows normally would be missed if they use a national reference.
Using the standard makes a lot of sense to me, and I actually like it the way that WHO has done it. It’s OK to exclude the Netherlands and Sweden but for the rest of the world, it makes a lot of sense to me.
I think that the fact that WHO strives to eliminate stunting by 2030 – we don’t need to hold our breath to 2030. It’s OK if it takes another generation or two. And here you tell us about 6 cm in a period of 20 years. I am impressed.
In the study of Barry Bogin of Maya children who moved to America, they show that within a single generation the Maya children grew as well as American children. So this is not genes admixture. This is rather the environment that is changing. It was just mentioned here that leg length grew more than anything else. The same Bogin advocates for many years now that the leg length is a better measure for the effect of stunting by the environment. When you see change in leg length, this is the effect of the environment.
Malina: I will just comment about the admixture question. The issue in most Indigenous communities is not admixture, it is out-migration. Something is missing here. This is apparent in the work of Barry Bogin dealing with Guatemalan children of Mayan ancestry in Florida and in Guatemala. They are getting progressively taller, and the gain was in leg length. My Oaxaca children are midway between the Mayans in Guatemala and in the U.S. There is a community in New Jersey that I recently heard of that is populated largely by people from Oaxaca, specifically from the community I have worked in. The difficulty right now is that they do not identify themselves given to the idiot we have in the white house. We have to be very careful; it’s a very sensitive issue. They are interested but to identify them, will create political problems for them.
Sperling: I think these are all valid points, what’s happening in Holland, or the populations you described. I now live in New York so I’m amazed at the tall Chinese that I see; they appear to be really quite tall, based perhaps on my bias, evolutionary bias. But I do think that we have an opportunity in today’s world to try and distinguish genetics, which is clearly an important factor, from environment. And you have these populations, large segments of population, that have moved to cities like New York or Antwerp or Mexico City that would give an opportunity to look at the difference between environment versus genetics. And that doesn’t apply to the child who is growing up in Bangladesh, because there, that genetic admixture does not exist. It’s primarily environmental.
Malina: We have data to some extent on that issue. In Mexico you have what they call Colonias where largely poor people live; the children from Colonias in Oaxaca are the exact same size as children in Colonias in Mexico City.
Your comment about the Chinese, I am assuming you mean the Han Chinese. I am not aware of studies of indigenous children of different ethnic groups in China. There are a lot more interesting groups in China that need to be studied.
Cons
Wit: Let me first start with the arguments against the use of the WHO charts in low-and middle income countries. Last year Michael Hermanussen and I were asked to advise the Indonesian Pediatric Association in the debate that was going on in that society to continue using WHO standards or to develop their own national reference charts. The clinical pediatricians had observed that a high percentage of healthy well-fed Indonesian children are way below the 3rd percentile of the WHO standard, and thus would be considered stunted. This was confirmed by an Indonesian study that was done on multiple islands published by Batubara et al, and it was again confirmed by nationwide data. At 18 years, height in males and females was approximately 12 cm and 9 cm shorter than WHO standards, respectively. The form of the 50th percentile stature in males suggests that adult height maybe a few centimeters taller than the mean stature at 18 years, so there’s a sort of delayed puberty also there. So that difference with WHO charts for adult height might be a little less than it looks now. I think the main problem caused by using WHO standards is that completely healthy well-fed children in such country are called stunted. This is also a problem at the clinical level, in determining which child needs to be further investigated for short stature. If one would perform a diagnostic work-up for every child with a height <-2 SDS for the WHO standard chart, the diagnostic yield would be low.
It's also a problem at the policy level. I mentioned that WHO has set targets that before 2030 the frequency of stunting should be brought down to normal. There is a difference between the slow pace of secular trend in a country with an increasing standard of living and the fast increase in height observed in children who move from one country to another, for example to the US. Such fast change was also seen after the Wall fell in Germany; the Eastern German kids were much shorter when the Wall fell, but they showed an enormous increase in height for about five years until they were equally tall as West-German children. East-German children were in a good nutritional state, so there are apparently other factors that have an effect on growth.
It has also been shown that food supplementation programs aiming at diminishing the incidence of stunting have no or only very little effect on body stature.
So I think that if WHO standards are used for assessing pathological stunting in many low and mid-income countries, this would lead to a waste of resources. It would aim at the wrong targets. The aim should be to improve the socio-economic conditions in which children live, social justice, hygiene, nutrition, education, etc. Furthermore, we need significantly more research to better understand the reasons for these differences in heights.
Not only in Indonesia, but also in most other South Asian children height is considerably below the WHO standard. Even in social strata that are well nourished, wealthy, healthy and well educated. Using standards that are too high for the well-fed local population does not only lead to overestimation of stunting, thus loss of specificity, but also low sensitivity for detecting pathological causes of tall stature.
I'd like to add that we did also a study on the children of Moroccan and Turkish origin in the Netherlands. In 1997 there was still a 5 cm difference in children and 10 cm in adult height between children of Moroccan and Turkish origin and the individuals of Dutch origin. Thirteen years later these minority groups had accomplished a large secular change of about 3 cm, while the height of children of Dutch origin had stabilized (Schonbeck et al, PLoS One. 2015;10:e0124686). So, one can predict that in another 10, 15 or maybe 25 years, individuals belonging to these minority groups may have a similar height.
With regard to the comment of Bob Malina, I can confirm that indeed short legs have an influence on BMI. When an individual has short legs, there is relatively more trunk. The trunk represents more weight than legs, so BMI is higher in an individual with relatively short legs.
Malina: Have you followed the data from the INCAP (Institute of Nutrition of Central America and Panama) supplementation study in Guatemala? The program lasted for 5 or 6 years. I believe the evidence is suggesting that the youngsters who were on supplementation beginning in the late 60s were as young adults taller than their parents. And the children born to those who were supplemented up to age 3 are taller than those before. So there simply may be a generational effect on the effects of the INCAP supplementation program. That is a study that should be considered in this context.
General discussion
Arnhold: I would like to ask Andrea to add information on the Colombian experience. There the WHO’s standard is used, but still Colombians are considering to change to their own reference charts, apparently. Can you comment on advantages and the reasons?
Merker: Yes, I’m working together with pediatric endocrinologists from Colombia on constructing national curves for Colombia. This project was initiated quite long time ago. Meanwhile, the WHO came also to Colombia with their trucks to roll out its standard while heavily marketing them.
In this Colombian growth project, I’m responsible for the data analysis and the clinicians have asked me for arguments against the WHO charts. They also wondered what curves to use. I think that an advantage of our Colombian curves is that they include also body proportions (sitting height and leg length) and waist circumference references. These are an important complement to any clinical evaluation. For instance, when one is screening for Turner syndrome, a typical increased sitting height relative to total height can easily be appreciated.
For Colombia, we have been saying that it’s of utmost importance to follow the child´s growth, not just having one measurement that should decide whether the child is sick or not.
Another comment regarding weight, the official WHO curves only cover 0 to 10 years. A Canadian group of pediatric endocrinologists have subsequently modeled weight all through 19 years of age (Rodd, Metzger and Sharma, the Canadian Pediatric Endocrine Group (CPEG) Working Committee for National Growth Charts, BMC Pediatr. 2014;14:32). So this extension should of course be implemented in those countries that use the WHO charts.
Also, can I comment on the construction of own (national, regional or ethnic specific) curves? Some people want “own curves”, but constructing representative references or curves takes resources. It takes for instance resources to get a sample that is representative or big enough to construct references. And a sufficient sample size is needed to get reliable cut-off levels and outer lines beyond 2 standard deviations. Creating national growth charts demands money and other resources that some countries might instead wish to invest in other health care programs.
Oduwole: We, the pediatric endocrinologists in Nigeria, have a group chat. What we found is that those who are in private schools had growth curves that are similar to the American growth chart. And those who were in public school had growth curves that was quite similar to the WHO. So I think it depends on nutrition. Accessibility to good health programs has a lot of influence on growth. Because those that are in private schools have access to hospitals, access to good food, access to supplements, whereas those in public schools don’t have half of that. So obviously the ones who are in the private school with a lot of things which kids in the American school do have, are similar in growth. Whereas those who don’t have, they’re more likely to have an issue; we have a lot of stunted children there, whereas 30% of those in private schools are obese and overweight. So it depends. I agree with you. Something is going on.
Wit: I think it’s very important to look at the various parts of the world. I think in Africa probably WHO charts are okay. From the data I have seen I have the impression that the South Asian children will probably need many decades to come close to the WHO standard. I saw data from Pakistan, also from private schools, on well fed children, whose height was far below the WHO standard, similarly to the situation in Indonesia. So it is dependent on the area, I think.
Alexander Jorge: For the first 5 years of life, where the genetic influence is lower, an international growth curve should be appropriate to assess growth and to identify children that are malnourished or in a stressful environment. After the second to the fifth year of life the genetic influence starts to increase and then one may have two different sorts of populations. First, in populations in genetic isolation a national growth chart is very important, as well in populations that are considerably taller, possibly due to improvement in nutrition in subsequent generations. In contrast, populations like Brazil have a huge admixture, so that it’s really difficult to construct a growth reference curve that is specific for our country. It changes every day because we have different population every day in our country.
With regard to sitting height, there are of course some conditions with an increased sitting height/height ratio, like SHOX haploinsufficiency. But I saw a lot of adult patients in whom their sitting height increased only because they gained weight, while their height did not change. I think it’s impossible to give a general rule to use a universal growth standard or a national reference. It depends on the age of children that we are analyzing as well as the changing population. Like in Holland, the Turkish and Moroccan individuals start admixture with locals, which changes the picture completely. What growth chart do we need for mixed ethnicity children?
Malina: You raise very good points. It goes back to what question you are asking. Are you just surveying a population? Any survey is relative. For Brazil, I just read a paper where they have new growth charts based on predicted time before peak height velocity (just published in Nutrition and Metabolism). If you are four years before peak height velocity here’s your chart. Unreal. However, the equations to predict time before peak height velocity were developed on youth of European ancestry and also have major limitations.
Martin Ritzén: I agree with Jan Maarten that the WHO figures have underestimated the genetic influence. It’s a mixture of populations from different parts of the world with different backgrounds and different growth potential and whatever. But when we sit with a patient with short stature, we look at the parents, that’s the first question. How tall are the parents, how tall are siblings? And we make a family tree with heights. So it’s obvious to all of us that genetic influence is very important. So I wonder if the WHO charts were constructed for a political purpose in order to even out all differences.
Malina: I sent you that 1974 paper by Jean Pierre Habicht et al. (The Lancet 303,611-615). When I was at INCAP we compared upper and lower class children from birth to 7 years from several parts of the world. There were no differences among the upper class children. The major difference was between social classes. When that paper came out, Jim Tanner criticized that paper because we were saying if the children’s environment is healthy then the growth of the children is pretty uniform. We caught a lot of flak for that paper, back in 1974 before the growth charts.
Arnhold: If the WHO concentrated on breastfeeding and not smoking, these charts were already looking more at the standard reference. This may be a political way of increasing the duration of breastfeeding and avoiding smoking.
Wit: WHO has started from the dogma that everybody is equal, so heights should be equal too. Then the conclusion is that it’s only environment that has an influence on growth. I think the data show otherwise.
Merker: The WHO growth child standard is a nutritionally driven initiative with the focus to improve infant and child health, thus focusing on growth at ages 0 to 5 years only.
Comment by Gabrriele Haeusler Associate Professor, Medical University of Vienna
1.To my opinion, the terms growth standard, growth reference, and growth chart were best explained by Tim Cole in his 2012 review in Ann Hum Biol 39: 382.
I guess, these definitions would make this discussion easier to follow:
A growth reference is a statistical summary of anthropometry in a reference group of children, usually presented as the frequency distribution at different ages. The reference group is often representative of some geographic region at a particular time, e.g. Great Britain in 1990 (Freeman et al.1995). The statistical summary involves the mean and standard deviation or alternatively the median and selected centiles, conditioned (usually) on age and sex. Growth references describe how children grow and the references can be applied to other children to establish whether or not their measurements are typical of the reference group.
A growth standard is essentially the same as a growth reference except that the underlying reference sample is selected on health grounds. Thus, it represents a healthy pattern of growth and the standard shows how children ought to grow rather than how they do grow.
A growth chart is a growth reference or standard presented as a visual display for clinical use and in this sense it is a graphic design. (Note that the term ‘growth reference’ or ‘growth standard’ is sometimes applied to the growth chart that displays the reference or standard, but strictly speaking this usage is wrong.)…”
2. I would like to introduce two papers discussing the issue which standards/references to use: Ziegler-EE, Nelson SE: The WHO growth standards: Strengths and limitations. Curr Opin Clin Nutr Metab Care 2012; 15: 298-302 and WHO 2006 Child Growth Standards and 2007 Growth Reference Charts: A Discussion paper by the Committee on Nutrition of the European Society for Pediatric Gastroenterology, Hepatology and Nutrition (2013). J Pediatr Gastroenterol Nutr; 57: 258.
They suggest that the WHO growth standards should be used for age 0-2(5) years ( for the good reasons explained in detail by Andrea Merker above) followed by national references . The authors recommend that NATIONAL bodies should decide, whether the WHO 5-19 reference should be used (a decision paying attention to the genetic population background).
We have done so in Austria and combined the WHO 2006 growth STANDARS with 4-19 cross sectional 2013 national growth REFERENCES ( Gleiss-A…Haeusler-G, Ann Hum Biol 2013, see attachment ) in our growth CHARTS and for calculation of SDS.
3.In my view, the use of national growth references gives the background for evidence-based diagnostic work-up for short stature as proposed in the Dutch paper by Grote et al 2008 in Arch Dis Child: The deviation from the reference by a defined SD score, together with information on parental height and percentile crossing sets the ground for a diagnostic workup.
The case of the mentioned boy growing on the 25th percentile (Ze’ev’s example) would not be suspicious for GHD or SHOX deficiency- but at the same time is one of many good reasons to improve health care and other important issues in underpriviledged countries in general.
4. Body proportions are very important, both with respect to diagnostic workup (mild skeletal dysplasias like SHOX deficiency do not have to wait until GH testing has shown a normal result) and observations on population catch-up growth (the paper by Bogin has been mentioned already). Note that catch-up growth in China has been predominantly established by longer legs!
In our 2013 Ann Hum Biol paper (Gleiss-A et al) we provide the data for sitting height, leg length and ratio SH/LL. As our references for height are very similar to the latest German and Belgian data, we think that our references could easily be used at least in these countries! After the Zurich Longitudinal Study by Prader et al we do not have valid reference data on body proportions.
BMI: In our 2015 paper (Mayer-M et al) we propose to use Equi-BMI- a term introduced/method developed by Cole in order to account for the obesity trend. The Austrian Equi-BMI data are likely suitable for other European countries, too!
COMMENTS TO THIS CONVERSATION ARE TO BE MAILED TO IVO ARNHOLD [iarnhold@usp.br]