Worldwide Timing of Growth Faltering: Implications for

Nutritional Interventions

Roger Shrimpton, MSc, PhD*; Cesar G. Victora, MD, PhD‡; Mercedes de Onis, MD, PhD§;
Rosângela Costa Lima, MD, MSc‡; Monika Blössner, Dipl oec troph§; and Graeme Clugston, MD, PhD§

ABSTRACT. Objective. It is widely assumed that almost all of this burden, with 70% of all early child
growth faltering starts at around 3 months of age, but mortality and malnutrition concentrated in sub-
there has been no systematic assessment of its timing Saharan Africa and South Asia. Despite setting a goal
using representative national datasets from a variety of of reducing malnutrition by 50% at the World Sum-
countries.
mit for Children,3 few countries in these 2 regions
Methodology. The World Health Organization Global
Database on Child Growth and Malnutrition includes the will have been successful in achieving this goal by
results of 39 nationally representative datasets from re- the end of the decade. It is our contention that one of
cent surveys in developing countries. Based on these the reasons for failure is the lack of clear definition
data, mean z scores of weight for age, length/height for and common understanding of what the problem is.
age, and weight for length/height were compared with The purpose of this article is to try to bring more
the National Center for Health Statistics and Cambridge clarity to these issues.
growth references, for children younger than 60 months. Waterlow4 proposed a functional classification for
Results. Mean weights start to falter at about 3 child malnutrition that separated children who had
months of age and decline rapidly until about 12 months,
with a markedly slower decline until about 18 to 19
acute malnutrition from those with chronic malnu-
months and a catch-up pattern after that. Growth falter- trition. The acutely malnourished children were
ing in weight for length/height is restricted to the first 15 those with adequate height for age but inadequate
months of life, followed by rapid improvement. For weight for height (wasted). The chronically malnour-
length/height for age, the global mean is surprisingly ished children were those that had inadequate height
close to National Center for Health Statistics and Cam- for age (stunted). Chronically malnourished children
bridge references at birth, but faltering starts immedi- could also be acutely malnourished, in which case
ately afterward, lasting well into the third year. they would be both stunted and wasted. The Water-
Conclusions. These findings highlight the need for
low Classification was called a functional classifica-
prenatal and early life interventions to prevent growth
failure. Pediatrics 2001;107(5). URL: http://www. tion, because it helped decide what interventions
pediatrics.org/cgi/content/full/107/5/e75; growth, body were needed by the children being surveyed.
height, body weight, infant nutrition disorders, child nu- Stunted children need little attention, because stunt-
trition disorders. ing cannot be reversed and they are in no immediate
danger of dying. On the other hand, wasted children
ABBREVIATIONS. UNICEF, United Nations Children’s Fund;
need urgent medical attention to prevent death.
WHO, World Health Organization; NCHS, National Center for The Waterlow classification was extremely useful
Health Statistics; SD, standard deviation; NHANES, National for guiding curative interventions, but it did not
Health and Nutrition Examination Survey; IMCI, Integrated Man- indicate how the processes of becoming either
agement of Childhood Illnesses.
stunted and/or wasted could be prevented, because
it does not indicate when, what, and how the stunt-

S
tunting affects 182 million (33%) and being un- ing and/or wasting happens.5 Neither does it facili-
derweight affects 150 million (27%) of the tate understanding at what age actions might be
world’s children1; these are associated with taken to prevent child malnutrition from occurring.
over half of the 10 million annual deaths of children The Child Survival and Development Revolution,
under 5 years.2 Developing countries account for promoted by the United Nations Children’s Fund
(UNICEF) during the 1980s, contributed to the mon-
From the *Nutrition Unit, UNICEF, New York, New York; the ‡Postgrad-
itoring of children’s growth becoming an almost uni-
uate Program in Epidemiology, Universidade Federal de Pelotas, Pelotas, versal practice,6,7 The objective of growth monitoring
RS, Brazil; and the §Department of Nutrition for Health and Development, is the early detection of growth failure, to allow
World Health Organization, Geneva, Switzerland. timely remedial interventions and prevention of fur-
Dr Shrimpton’s current affiliation is with the Centre of International Child
Health, Institute of Child Health, University College London, London,
ther growth failure. Several authors have suggested
United Kingdom. that growth faltering starts at about 3 months of life
Received for publication Aug 30, 2000; accepted Nov 21, 2000. in developing-country settings,5,8,9,10 However, most
Reprint requests to (C.G.V.) Postgraduate Program in Epidemiology, Uni- such publications have looked only at weight
versidade Federal de Pelotas, CP 464, 96001-970, Pelotas, RS, Brazil. E-mail:
[email protected]
growth, not height growth, and were not population-
PEDIATRICS (ISSN 0031 4005). Copyright © 2001 by the American Acad- based studies of national scale. The present paper is
emy of Pediatrics. the first attempt to investigate globally the patterns

http://www.pediatrics.org/cgi/content/full/107/5/e75 PEDIATRICS Vol. 107 No. 5 May 2001 1 of 7

and timing of growth faltering and to discuss impli- RESULTS
cations for developing interventions for prevention The initial analysis consisted of plotting the mean
of child malnutrition worldwide. z scores by age for each country dataset, against the
NCHS growth reference. In general, the national
curves had remarkably similar shapes within each of
METHODOLOGY
the regions of the world, although their relative po-
The World Health Organization (WHO) Global Database on Child sitions showed some variation (data available on
Growth and Malnutrition11 collates data from cross-sectional an-
thropometric surveys worldwide; 39 national samples from less-
request).
developed countries, each with ⬎500 children aged 0 to 23 Figure 1 shows the global mean z scores, calcu-
months, were reanalyzed to provide information on mean z scores lated from the 39 national survey population sam-
by month of age. Table 1 shows the datasets included in the ples. At birth, the average weight for age, length for
analyses with the corresponding numbers of children, and the age age, and weight for length are quite close to the
ranges for which the data were available. All but 3 surveys pro- NCHS reference. Growth faltering occurs thereafter,
vided data from birth onward; 13 provided data until age 35 or 36
months, 2 until 47 months, and the remaining 24 until 60 months.
so that by 18 months the mean values of weight for
All surveys were conducted between 1987 and 1997. age and length for age are between 1 and 2 standard
For every age group in each national sample, the mean weight deviations (SD) below the reference median value,
for age, length/height for age, and weight for length/height z while the mean weight for length is about ⫺0.6 SD.
scores were calculated using the National Center for Health Sta- The way the 3 indices behave during the first 3
tistics (NCHS) reference.12 Mean weight for age and length/height months of life is different. Although weight for age
for age z scores were also calculated relative to the Cambridge
reference,13 resulting from criticisms of the NCHS growth curves.5
stays horizontal and parallel to the reference for the
For each of 3 regions (Africa, the Americas, and Asia), the first 3 months, length for age decreases. Conse-
average value of the national means was calculated. The global quently, the weight for length z score increases in
average of all studies was also calculated. this period. This observation raises the question of

TABLE 1. Nationally Representative Samples Included in the Analyses*

Region Countries Dates of Age Range Number of
Survey (Months) Children
Latin America and Caribbean Bolivia 1993–94 0–35 2860
Brazil 1996 0–59 3815
Paraguay 1990 0–59 3389
Peru 1996 0–59 13431
Colombia 1995 0–59 4408
Guatemala 1995 0–59 7768
Haiti 1994–95 0–59 2794
Dominican Republic 1991 0–59 2883

Trinidad and Tobago 1987 3–36 840

Asia Pakistan 1990–91 0–59 4056
Nepal 1996 0–35 3705
Sri Lanka 1987 3–36 1993
India 1992–93 0–47 24421
Bangladesh 1996 0–59 4787
Turkey 1993 0–59 3152
Thailand 1987 3–36 1857

Africa Kenya 1993 0–59 4763

Rwanda 1992 0–59 4386
Tanzania 1996 0–59 5344
Uganda 1995 0–47 4775
Zambia 1996–97 0–59 5443
Zimbabwe 1994 0–35 2014
Comoros 1995–96 0–35 921
Madagascar 1992 0–59 4240
Malawi 1992 0–59 3233
Namibia 1992 0–59 2430
Benin 1996 0–35 2273
Burkina Faso 1992–93 0–59 4278
Côte d’Ivoire 1994 0–35 3341
Niger 1992 0–59 4052
Nigeria 1990 0–59 5606
Mali 1995–96 0–35 4678
Ghana 1993–94 0–35 1818
Senegal 1992–93 0–59 3865
Central African Republic 1994–95 0–35 2310
Cameroon 1991 0–59 2380
Egypt 1995–96 0–59 9766
Morocco 1992 0–59 4532
Tunisia 1988 3–36 1996
* Source: WHO Global Database on Child Growth and Malnutrition.

2 of 7 WORLDWIDE TIMING OF GROWTH FALTERING

 Fig 1. Mean anthropometric z scores by age for all 39 studies, relative to the NCHS reference (0 –59 months).

why height growth is faltering when weight growth in relation to the reference. The lines are less steady
is apparently normal. after 36 months because they are based on fewer data
After the first 3 months, the falls in weight for age points as some studies did not include 3- and 4-year-
and length for age are similar until around 12 olds (Table 1). At 3 months, the Latin American and
months of age, when weight for age stops decreas- Caribbean mean z score is about 0.25 SD above the
ing, stabilizing at around ⫺1.4 SD. Length for age African mean, which is about 0.5 SD above the Asian
continues to fall through until 24 months of age. The mean. All means drop sharply until about 10 months,
increase at 24 months is artificial and caused by the but run roughly parallel to the reference or even
disjunction of the 2 datasets used for the NCHS catch up slightly after 12 months. The differences
reference (Fels and National Health and Nutrition between the 3 regions that were present at birth
Examination Survey [NHANES]).5 Ignoring the dis- remain throughout the first 5 years of life.
junction, mean length/height for age continues to In Fig 3, the length/height-for-age profiles of the
decrease until around 40 months. This finding was three regions are compared. Unlike weight, the mean
confirmed when the data were plotted against the length z scores at birth are very similar in all 3
Cambridge reference (see below). The weight-for- regions and the same as the growth reference. In all
length profile declines after 3 months to about ⫺0.7 z 3 regions, the mean z score of length for age falls
scores at about 18 months. Thereafter, it slowly in- sharply from birth to about 24 months. The magni-
creases to about ⫺0.25 SD around 40 months of age, tude of the drop in Latin America and the Caribbean
when it remains parallel to the reference profile. Data is about ⫺1.25 SD, whereas in Africa and Asia the
for 4- and 5-year-old children should be interpreted drop is ⫺2 SD. Again, interpretation of the profiles is
with caution because some of the datasets were re- affected by the 24-month disjunction in the NCHS
stricted to children under 3 or 4 years. reference, but after the 24th month the process of
The behaviors of these 3 growth profiles shows stunting seems to continue, albeit at a much slower
that the processes of becoming stunted and of be- rate.
coming wasted are independent of each other. The The process of wasting in the 3 regions is shown in
process of stunting begins at birth and continues Fig 4. At birth, Asian children are on average 0.4 SD
during the first 3 years of life. The process of wasting below African children, who are in turn 0.4 SD below
is concentrated in the period of 3 to 15 months. Latin American children. In each region the mean z
Thereafter, there is a process of gradual improve- score rises about 0.5 SD, reflecting the maintenance
ment in the wasting situation, so that by about 40 of weight growth (Fig 2) in the face of length falter-
months there is little and the global mean is just 0.25 ing (Fig 3). From about 3 months of age, the mean z
SD below the reference. However, the stunting that scores for weight for length fall in all regions—about
occurs in the first 3 years is not recovered thereafter. ⫺1 SD— until about 9 to 15 months of age. At some
In Fig 2, the weight-for-age means for Africa, Asia, time during the second year, mean length for age
and the Americas are compared. The 3 profiles have starts to improve in all regions, but regional differ-
similar shapes but start and finish at different levels ences persist. By 36 months of age the African child

http://www.pediatrics.org/cgi/content/full/107/5/e75 3 of 7
 Fig 2. Mean weight-for-age z scores by age, relative to the NCHS reference, by region (0 –59 months).

Fig 3. Mean length/height-for-age z scores by age, relative to the NCHS reference, by region (0 –59 months).

is about 0.25 SD thinner, and the Asian child about All analyses were repeated using the Cambridge
0.75 SD thinner than the average Latin American and reference (graphs available on request), but the con-
Caribbean child. A possible tendency toward over- clusions were not affected.
weight children is observed in Latin America, where Differences in the faltering patterns for height for
mean weight for height tends to be above the NCHS age and weight for age were highly consistent across
median after 39 months of age. countries. For the 35 countries with data from birth

4 of 7 WORLDWIDE TIMING OF GROWTH FALTERING

 Fig 4. Mean weight-for-length/height z scores by age, relative to the NCHS reference, by region (0 –59 months).

onwards (Table 1), 32 had negative z score changes in the timing of growth faltering. The findings also
mean length for age from birth to 3 months. The contradict common knowledge that faltering relative
average reduction was 0.37 z score (SD ⫽ 0.36). For to international references starts at only about 3
weight for age, the average reduction was 0.07 z months of age,5 because faltering in length starts
score (SD ⫽ 0.36); 17 of the 35 countries had reduc- immediately after birth. This conclusion is supported
tions, and 18 had no change or increases. by studies of the growth of individual children,
showing that most of those malnourished at ages 3 to
DISCUSSION 5 years already presented anthropometric deficits at
This is the first attempt to investigate the timing of the end of the first year of life,16,17 Additional re-
growth faltering on a global scale. The first striking search is needed to elucidate the reasons for the
finding was how similar the patterns of growth fal- differences in the timing of height and weight falter-
tering were in different developing countries, not ing.
only within a region but also globally, despite the The importance of intrauterine growth retarda-
different instruments and measuring techniques tion18 on weight for age is evident. Because the
used in the various surveys. A possible limitation is curves for the 3 regions have similar shapes but
that although only nationally representative samples markedly different starting points, improvements in
were included, data were not available for every intrauterine growth per se can be expected to reduce
country in each region, and thus the regional means the prevalence of underweight. The sharp faltering
are not necessarily representative. It was felt that in the first year of life confirms the importance of
weighting the results according to the country’s pop- concentrating preventive interventions at this age or
ulations was not warranted, for the same reason, and earlier. Curative interventions, on the other hand, are
also because similar shapes of curves were obtained still important for older children, who will make up
from the different surveys. the majority of those diagnosed with malnutrition in
The NCHS reference has been criticized for several health facilities.
reasons,5,14 including its inappropriateness to reflect The fact that in the 3 regions mean birth lengths
the growth of breastfed infants and the 24-month were very close and similar to the NCHS reference
disjunction. It is reassuring, however, that the anal- median was unexpected and merits additional inves-
yses with both the NCHS and Cambridge references tigation. Accurate measurement of length is com-
produced results that were very similar, with the plex, but it is unlikely that errors would be sys-
exception of the artifacts resulting from the 24-month tematic in all studies. Some type of intrauterine
disjunction in the NCHS data.5 programming of birth length seems to be present, so
The use of data from 39 countries revealed inter- that adverse fetal conditions affect weight but not
esting findings not apparent from previous analyses. length. In at least 1 study prenatal interventions have
The findings confirm the notion that, at the popula- been shown to affect postnatal growth19 despite hav-
tion level, wasting and stunting show different be- ing no effect on birth weight.20 Therefore, prenatal
haviors,15 a fact not widely appreciated in relation to interventions may still help prevent length faltering.

http://www.pediatrics.org/cgi/content/full/107/5/e75 5 of 7
The main priority should be the development of and reliability of length measurement that would
effective interventions to stop the remarkable falter- limit its wide adoption. Whatever the measurement
ing— upwards of 2 SD in Asia—that occurs from being taken, growth monitoring constitutes primar-
birth to 18 months. Interventions should be sustained ily an individual-based, largely curative intervention
until the third year of life because faltering continues in that it happens after the fact. Although it certainly
until this age. has a role to play, the present results suggest that a
Faltering in weight for length or height shows a preventive, population-based approach—aimed at
very different pattern. Starting at about 3 months, it improving the nutritional situation of affected pop-
stops around 12 months and is followed by a marked ulations rather than individuals24—would be more
recovery, particularly in Africa. In Latin America and desirable, with efforts directed primarily at mothers
the Caribbean, the recovery phase leads to mean and young infants. Much more attention needs to be
values above the NCHS reference at around 40 given to the prevention of low birth weight.
months, indicating a higher prevalence of over- Our findings have important policy implications.
weight. This is particularly worrisome because cor- The data show that interventions during the earliest
onary heart disease seems to be particularly common periods of life—prenatally and during infancy and
among men who were thin at birth but whose weight early childhood—are likely to have the greatest im-
caught up so that they had an average or above pact in preventing child malnutrition. Types of inter-
average body mass from age 7 years.21 ventions include culturally adaptable nutrition edu-
It is unlikely that survival bias could explain the cation on diet and feeding practices (with food
apparent catch-up in weight for length. First, preva- supplementation if necessary), together with coun-
lences of wasting were low in many countries that seling on improved parental caring skills.25 Given
showed this pattern—for example, in the Latin the complexity of the underlying and basic causes of
American countries wasting is typically between 3 the problem, new local specific efforts must be made
and 5%,15 and mortality between ages 1 and 4 years to understand the specific economic, behavioral, di-
is quite low. Second, it has been shown5 that weight- etary, and other factors affecting child growth and
for-length curves are remarkably similar in shape development. A number of community-based pro-
across a wide range of wasting prevalences. The grams have had a substantial effect on declining
whole weight-for-length curve seems to shift and child malnutrition rates.26,27 Wider application of
there is no evidence that its shape would change such programs is needed. Special effort should be
resulting from selective mortality of wasted chil- made to improve the situation of women as primary
dren. child caregivers with particular attention to their
Community-based growth monitoring was the health and nutrition throughout the life cycle. Simi-
central plank of the child survival and development larly, a strong focus on complementary feeding and
revolution promoted by UNICEF during the 1980s.6,7 continued attention to the protection and promotion
The aim of growth monitoring is to weigh children of breastfeeding remain key components for tackling
and plot the weight on a growth chart to detect the problem.
growth faltering early, allowing community workers
to advise mothers on how to improve the growth of ACKNOWLEDGMENTS
the faltering child. An evaluation of community- This study was financed by the Nutrition Unit of UNICEF and
based growth monitoring programs in 6 countries in by the Department of Nutrition for Health and Development of
the World Health Organization.
the early 1990s came to the conclusion that the We are grateful to Dr Allen Shoemaker for his assistance in the
weighing of children was often a mechanical act, reanalysis of the national datasets.
which led to little action.22 Insufficient effort went
into training community workers. Growth monitor- REFERENCES
ing has also been criticized for targeting all children 1. United Nations Administrative Committee on Coordination/Sub-
under 5 years rather than being aimed at the first Committee on Nutrition. Fourth Report on the World Nutrition Situation:
couple of years of life, when nutritional status is Nutrition Throughout the Life Cycle. Geneva, Switzerland: United Nations
Administrative Committee on Coordination/Sub-Committee on
being determined.15 Nutrition; 2000
A somewhat different approach was adopted by 2. Pelletier DL, Frongillo EA, Habicht JP. Epidemiologic evidence for a
the Integrated Management of Childhood Illness potentiating effect of malnutrition on child mortality. Am J Public Health.
(IMCI) initiative. Promoted jointly by WHO and 1993;83:1130 –1133
UNICEF, this is a broad strategy to reduce child 3. UNICEF. First Call for Children. Summit Declaration and Child Convention.
New York, NY: UNICEF; 1990
mortality and morbidity in developing countries.23 4. Waterlow JC. Note on the assessment and classification of protein-
IMCI recommends that all children be weighed and energy malnutrition in children. Lancet. 1973;2:87– 89
that their weight be plotted on the weight-for-age 5. World Health Organization. Expert Committee on Nutrition. Physical
chart. This preference for assessing weight for age Status: Uses and Interpretation of Anthropometry. Geneva, Switzerland:
World Health Organization; 1995. WHO Technical Report Series, Report
rather than assessing adequacy of growth velocity is No. 854
based on the recognition that often the sick child who 6. Cash R, Keusch GT, Lamstein J. Child Health and Survival: the UNICEF
comes to a health facility has no previous weight GOBI-FFF Program. Beckenham, England: Croom Helm; 1987
with which to compare current status. 7. Morley D, Woodland M. See How They Grow: Monitoring Child Growth for
Based on our findings, monitoring length or height Appropriate Health Care in Developing Countries. London, United
Kingdom: MacMillan Press; 1979
would be more relevant than weight monitoring for 8. Waterlow JC, Thomson AM. Observations on the adequacy of breast-
the assessment of nutritional status. However, there feeding. Lancet. 1979;2:238 –242
are obviously many issues related to the feasibility 9. Rogers IS, Emmett PM, Golding J. The growth and nutritional status of

6 of 7 WORLDWIDE TIMING OF GROWTH FALTERING

 the breast-fed infant. Early Hum Dev. 1997;49(suppl):S157–S174 tation during pregnancy and postnatal growth. Lancet. 1992;340:623– 626
10. Beaton GH. Small but healthy: are we asking the right question? Eur 20. Kardjati S, Kusin JA, De With C. Energy supplementation in the last
J Clin Nutr. 1989;43:863– 875 trimester of pregnancy in East Java: I. Effect on birthweight. Br J Obstet
11. de Onis M, Blössner M. WHO Global Database on Child Growth and Gynaecol. 1988;95:783–794
Malnutrition. Geneva, Switzerland: World Health Organization; 1997. 21. Eriksson JG, Forsen T, Tuomilehto J, Winter PD, Osmond C, Barker DJ.
Doc WHO/NUT/97.4 Catch-up growth in childhood and death from coronary heart disease:
12. Hamill PVV, Hamill PVV, Drizid TA, et al. NCHS Growth Curves for longitudinal study. Br Med. 1999;318:427– 431
Children Birth–18 Years. Vital and Health Statistics Series 11 No 165. 22. Pearson R. Thematic Evaluation of UNICEF Support to Growth Monitoring.
Hyattsville, MD: USHEW-PHS; 1977. DHEW Publ. No. (PHS) 78-1650 Working Paper Series No. 2. New York, NY: UNICEF; 1995
13. Cole TJ, Freeman JV, Preece MA. British 1990 growth reference centiles 23. WHO. Division of Child Health and Development. The Management of
Childhood Illness in Developing Countries: Rationale for an Integrated Strat-
for weight, height, body mass index and head circumference fitted by
egy. Geneva, Switzerland: WHO (IMCI Information); 1997
maximum penalized likelihood. Stat Med. 1998;17:407– 429
24. Rose G. Sick individuals and sick populations. Int J Epidemiol. 1985;14:
14. Victora CG, Morris SS, Barros FC, de Onis M, Yip R. The NCHS
32–38
reference and the growth of breast and bottle-fed infants. J Nutr. 1998;
25. WHO. A Critical Link. Interventions for Physical Growth and Psychological
128:1134 –1138
Development. A Review. Geneva, Switzerland: World Health
15. Victora CG. The association between wasting and stunting: an interna-
Organization; 1999. Doc WHO/CHS/CAH/99.3
tional perspective. J Nutr. 1992;122:1105–1110 26. United Nations Administrative Committee on Coordination/Sub-
16. Huttly S, Victora CG, Barros FC, Vaughan JP. The timing of nutritional Committee on Nutrition (ACC/SCN). Managing Successful Nutrition
status determination: implications for intervention and growth moni- Programmes. Geneva, Switzerland: United Nations Administrative Com-
toring. Eur J Clin Nutr. 1991;45:85–95 mittee on Coordination/Sub-Committee on Nutrition; 1991. ACC/SCN
17. Rivera J, Ruel MT. Growth retardation starts in the first three months of State-of-the-Art Series Nutrition Policy Discussion Paper No. 8
life among rural Guatemalan children. Eur J Clin Nutr. 1997;51:92–96 27. United Nations Administrative Committee on Coordination/Sub-
18. de Onis M, Blössner M, Villar J. Levels and patterns of intrauterine Committee on Nutrition (ACC/SCN). How Nutrition Improves. Geneva,
growth retardation in developing countries. Eur J Clin Nutr. 1998; Switzerland: United Nations Administrative Committee on Coordina-
52(suppl 1):S5–S15 tion/Sub-Committee on Nutrition; 1996. ACC/SCN State-of-the-Art
19. Kusin JA, Kardjati S, Houtkooper JM, Renqvist UH. Energy supplemen- Series Nutrition Policy Discussion Paper No. 15

http://www.pediatrics.org/cgi/content/full/107/5/e75 7 of 7