Nutritionall status following malaria control in a Vietnamese
ethnicc minority commune.
LeLe Q. Hung u; Peter J. De Vries '; Tran Q. Birth2;
PhanPhan T. Giao u; Nguyen V. Nam 3; Piet A. Kagerl.
1)) Division of Infectious Diseases, Tropical Medicine & AIDS, Academic Medical
Centerr F4-217, PO Box 22700,1100 DE Amsterdam, the Netherlands.
2)) Department of Tropical Diseases, Cho Ray hospital, 201 B Nguyen Chi Thanh,
Districtt 5, Ho Chi Minh City, Vietnam.
3)) Center for Malaria & Goitre Control, Phan Thiet, Binh Thuan Province, Viet Nam.
NutritionalNutritional status after malaria control
Objective:: To study whether control of malaria leads to catch up growth or an
increasee of obesity in a marginally nourished population
Setting:: A Vietnamese ethnic minority commune in southern Vietnam.
Design:: Repeated annual anthropometric surveys were performed from 1995 to
2000.. Z-scores for height, weight and BMI for age and weight-for-height were
determinedd by using NCHS 1978 and CDC-2000 reference tables and by the LMS
Intervention:: Active malaria control which reduced the parasite carrier rate
fromm 50% in 1994 to practically nil in 1998.
Results:: Inhabitants were generally of short stature and very thin. Using the US
referencee tables, the prevalence of moderate/severe stunting among children was 53%
// 24% and of wasting 27% / 9% in the first survey in 1995. Physical condition and
normall daily activities of most inhabitants were normal. The repeated LMS-Z-scores
uncoveredd a significant recovery of stunting, extending into preadolescence, including
thee development of a pubertal growth spurt for girls and enhancement of pubertal
growthh in boys, after control of malaria. The mean (95% CI) annual increase of Z-
height-for-agee was 0.11 (0.09 - 0.12) for boys and 0.14 (0.13 - 0.15) for girls (p <
0.001).. As a consequence, weight-for-age and BMI Z-scores decreased without
indicationn of developing obesity.
Conclusion:: Control of malaria induces catch up growth extending into
preadolescentt age in a Vietnamese ethnic minority population with a chronic state of
loww food intake, without indication of developing obesity.
Thee rapid socioeconomic changes in Asia are paralleled by the so called
epidemiologicall and nutrition transitions.(1;2) Traditionally, public health programs
focussedd on communicable diseases and malnutrition because these were the most
importantt health hazards. The interaction between infection and nutrition, however, is
nott clear, and it difficult to define which of these problems should be addressed
first.(3)first.(3) To date the situation is even more complex due to the epidemiological
transitionn to non-communicable diseases, many of which are associated with un-
balancedd diet and overweight.
Thee association between malnutrition in early life and overweight at later
stages,, has been addressed before.(4) Stunted growth during gestation or early infancy
predisposess for overweight and obesity when nutritional intake increases. Overweight
predisposess for cardiovascular disease and the metabolic syndrome.(5) It is the
negativee effect of a rapid nutrition transition. An important question is if control of
infectiouss diseases has the same detrimental effect on body dimensions.
Thiss matter is very complex because the epidemiological and nutrition
transitionss are unequally spread among the population of many middle and low
incomee countries. Especially children of low income groups, suffer from nutrient
deficienciess and communicable diseases, leading to retarded growth whereas the
transitionn to high energy food intake with non-communicable diseases is affecting the
moree affluent strata of societies.(6)
Thee socioeconomic changes in Viet Nam over the last decade came at a
variablee pace for different population groups.(7) In the early 1990s Viet Nam was one
off the poorest countries of the world. During the last decade it transformed rapidly
intoo a middle income country and living conditions changed accordingly. The most
vulnerablee populations were and still are the ethnic minorities. They share much less
inn the economic progress than the economically more active ethnic Vietnamese
(Kinh)) and they endure hard living conditions with malnutrition, endemic malaria and
otherr infectious diseases.
Inn 1994 a general health programme was initiated in Phan Tien village, an
ethnicc minority community located in the forested mountains of Binh Thuan
province,, in southern Viet Nam. Malaria was hyperendemic with parasite carrier rates
off up to 50%. Inhabitants were thin and of short stature but without overt
malnutrition,, performing normal daily activities including cultivating the land and
workingg in the forest. The hyperendemic malaria situation was brought to almost
completee eradication within four years.(8) This was complemented by eradication of
hookwormm infections (unpublished
improvementss seemed to lack behind but solid anthropometric reference tables of the
minorityy populations in Viet Nam and criteria of optimal values were lacking. Growth
chartss and tables of the National Center of Health Statistics/World Health
data). Nutritional and socioeconomic
NutritionalNutritional status after malaria control
Organizationn (NCHS/WHO) were based on a US reference population and criteria for
optimall growth and body dimensions were not clear.
Thee first aim of this study was to establish the rate of malnutrition in this
populationn and to see whether control of the infectious diseases, mainly malaria, had a
positivee effect on child growth and body dimensions. In this study we also laid down
aa methodology to construct growth charts with an accepted method, the LMS method,
andd a sensitive method to follow secular trends over a relatively short time span.
Thee secondary objective of this study was inspired by the difficulties in the
interpretationn of anthropometric data in this population. It was not clear if the use of
internationall reference tables was justified. A marginally nourished population is
muchh different from an industrialised country's reference population. Using the US
criteriaa generally overrates the rate of malnutrition in developing countries.(9) This
mayy frustrate statistical analysis by crowding of data in the tail of the distribution
curvess as well as the mathematical conversion of cut off points to define desirable
bodyy dimensions. There are also physiological arguments to object to extrapolating
thee definition of optimal body dimensions to other populations, especially the ethnic
minorityy groups in Viet Nam, whose members are generally shorter and leaner than
thee Kinh majority.(10-12)
Therefore,, in order to improve the interpretation of the anthropometric
indicatorss in the ethnic minority population of Viet Nam, the secondary objective of
thiss study was to establish the correlations between body fat, the body mass index
(BMI)) or its Z-score, (Z)BMI, and the Z-score of weight for height, Z-WH.
SUBJECTSS AND METHODS
Populationn and study site
Thee study population consisted of all inhabitants of Phan Tien who participated
inn the annual malariometric surveys from 1995 to 2000. The population of Phan Tien
iss composed of nine ethnic groups. Two thirds of the population belong to the Rac
Layy and Ta Lop ethnic minorities. Remaining groups are ethnic Vietnamese (Kinh),
Nung,, Nop, K'Ho, Tay, Hoa and Cham. In 1994, when the village had just been
established,, the population was 716; it increased to 1088 in 2000. People lived family
wisee with 5 to 6 persons in one house, with clay walls and a thatched roof. The local
economyy was simple, based on small-scale slash and burn subsistence farming and on
whatt the forest offered as food. During the dry season it took several hours to reach
thee village from the main road. During the raining season the village could not be
reachedd by car. Governmental support included a program of rice supplementation for
vulnerablee groups, such as pregnant women and children, which was usually shared
withinn families. Before July 1994, there was no health care facility and water supply
wass from a small river near the village. Mid 1994, three wells were drilled with the
aidd of UNICEF, but due to their limited supply of safe water most people adhered to
usingg river water. Safe drinking water was ensured at the end of 1997 by establishing
eightt functioning wells. Electricity was introduced at the end of 2000.
Despitee the general socio-economic improvements in Viet Nam, which started
att the end of the nineteen-eighties, the living conditions in Phan Tien remained very
poor,, largely similar to other ethnic minority communes in the remote areas of Viet
Nam.. Clinically significant malnutrition was not observed throughout the study
period.. A few children had slightly hypo-pigmented hair and approximately 15% of
childrenn younger than 16 years old had clinical signs of anemia (pallor). Hb was
measuredd from 1996 onwards and this showed severe anaemia (Hb < 7g %) of
maximallyy 2.1%, in 1996. Diarrhoea was noted infrequently while taking the interim
historyy at the surveys, and this was confirmed by findings in four faeces examination
surveyss in 1997 (2x), 1998 and 1999 (rate of diarrhoea always less than 0.5%).
Designn and procedures
Inn 1994, a full census was done, houses were numbered and all individuals
weree registered. A record was completed for each individual with name, unique
identifier,, age, sex, household and ethnic group.(8) Most of the inhabitants of Phan
Tienn do not usually record or remember their date of birth, only their age in years or
thee year of birth. In the database the date of birth was fixed to the middle of the solar
calendar,, i.e. 1st July, so that age could be recalculated to months. This introduces an
errorr of maximally 6 months in age for every individual, but the group average is
accurate,, except for children under 1 year old. This small group, of maximally 35,
wass excluded from the analysis; their results appeared in the repeated measures
analysiss after they had passed the age of one year. For subgroup analysis, three age
groupss were recognised: children from 1 to 10 years, teenagers/adolescents from 10
throughh 23 years and adults older than 24 years.(13)
Malariaa surveys weree done annually at the end of the rainy season from 1994 to
2000.. Demographic data were updated at every survey, registering newborns, deaths,
newcomerss and those who left. In 1997 an additional survey was performed at the end
off the dry season in school children only.
Fromm 1995 onwards, measurement of height and weight was added to the
Physicall examination was performed of all subjects by the same physician and
aa short interim medical history was obtained. Waterlow's criteria were used to
classifyy the degree of malnutrition. The criteria differentiate between primary signs,
i.e.. growth retardation, oedema, mental unhappiness and hepatomegaly; and
secondaryy signs, i.e. hypo-pigmentation of the hair or skin, anaemia and
Bodyy weight and height were measured of all children through the primary
schooll and all adults who were present in the village at the time of the annual surveys
andd did not refuse to participate in the survey. Height was measured in standing
NutritionalNutritional status after malaria control
positionn on bare feet to the nearest 1cm by using a portable stadiometer, which was
placedd on the flat hard floor and against a bare wall. Body weight was measured to
thee nearest 0.1 kg with a digital electronic scale. The weight of children between one
andd two years was measured by weighing the mother with and without the child in her
Measurementt of skin folds and middle upper arm circumference was only done
inn 1995. Thickness of skin folds was measured to the nearest 1mm with a standard
Harpendenn skin fold calliper at four sites: overlying arm biceps and triceps,
subscapularr and supra-iliac regions on the right side of the body, in standing
positional6)) Mid-upper arm circumference (MUAC) was measured at the mid-point
betweenn the tip of the acromial process and the tip of the olecranon process.(13)
Dataa processing and analysis
Usingg the NCHS/WHO reference tables of 1978 and the Center for Disease
Controll (CDC) tables of 2000, Z-scores were calculated for children's height-for-age
(Z-HA),, weight-for-age (Z-WA), weight-for-height (Z-WH) with the programme
NutStat,, a part of EPI-info 2000 (Centers for Disease Control and Prevention, Atlanta,
GA,, USA). This calculates the Z-HA and Z-WA for individuals from birth up to 18
yearss (NCHS/WHO) or 20 years (CDC 2000) of age. Z-WH is calculated for boys up
too 11.5 years of age and shorter than 145 cm, and for girls up to 10 years of age and
lesss than 137 cm with the NCHS/WHO tables. With the CDC 2000 tables ZHW is
calculatedd up to the age of 36 months and for children from 77 to 121 centimeters in
Withh both reference tables, the interpretation of results was similar: stunting:
Z-- HA < -2; underweight: Z- WA < -2; wasting: Z- WH < -2, severe malnutrition: any
off these Z-scores < -3. The BMI was calculated as weight (kg)/height (m2). Z-scores
weree again calculated with the two US reference tables and with the LMS method, for
alll subjects, including adults. The interpretation of the BMI is different for the two
USS reference tables and is shown in Table 1. BMI and WH Z-scores were compared
too skin fold thickness, a height and weight independent measure of body fat.
Tablee 1: The interpretation of US reference values for BMI for age.
Cutt off values for
<< 5th percentile
Z-scoress < -2
>> 85th and < 95th percentile
>> 95th percentile
>> +2 Z-scores
>> +3 Z-scores
Att risk for overweight
Growthh charts were constructed by gender for HA, WA, WH and BMI, using
thee LMS method on the first measurements available in the dataset of children and
adolescents.. The LMS method generates smooth centile curves of data sets that do not
necessarilyy show a normal distribution^ 17; 18) For this purpose dedicated software
wass used: LMS-Pro (Cole and Green, Institute of Child Health, Aberdeen, UK).
Calculationn of centiles is based on establishing smooth curves for skewness
(L),, the median (M) and the coefficient of variation (S). We aimed at smooth curves
forr LMS starting with the minimum degrees of freedom for the L, M and S (19;20).
Thee goodness of fit with different degrees of freedom for L, M or S, was assessed
withh the Q-test, provided in the Pro-version of the LMS program. LMS also generates
valuess of L, M and S for age specific intervals, which can be used to calculate the Z-
scoree of every individual given age, height and weight, with the following formula:
ZZ = [(y/M)L - 1] / LS when L ^ 0
ZZ - log(y/M) / S when L = 0
wheree "y" is weight or height for the respective age intervals or, weight for height
specificc intervals. (21) By this method, the Z-scores of the repeated measurements
weree calculated, using the set of all first measurements as the reference table. Ages up
too 24 years were included for constructing the HA and WA charts, ensuring that the
dataa also included all individuals with recent cessation of growth. WH charts were
constructedd using the same age groups in order to increase the available number of
dataa points for fitting. It is not conventional to calculate WH up to the age of 24 years
butt there is no objection to apply the LMS method to a larger age group and it
assistedd us in the interpretation of BMI. BMI is now the preferred indicator of
overweightt in all age groups, but it is unknown if that also applies to underweight.
Longitudinall data were analysed to detect trends over the 5 years of the study
period.. The repeated post-rainy-season Z-scores of all individuals were analysed as a
functionn of time in a linear mixed effects model. The linear regression coefficients
weree generated by restricted maximum likelihood methods (REML). The Z-scores of
thee single survey done at the end of the dry season was compared to the other surveys
usingg generalised linear models ("repeated measures ANOVA").
Inn order to find whether the anthropometric indicators were different between
thee ethnic groups, subjects were re-classified into 3 main groups: Rac Lay, Ta Lop
andd remaining inhabitants. The latter also included subjects with parents of different
ethnicc groups. Z-scores of population sub-groups were compared using ANOVA.
Alll routine statistics were done by using SPSS 11.5.1 (SPSS Inc., Chicago, II,
USA)) Mixed effects modelling was done with S-plus 2000 (MathSoft Inc., Seattle,
WA,, USA). Statistical significance was accepted when p<0.05.
Overr the complete study period, 91% of the population of Phan Tien was
examinedd at least once with a median of three times (male/female: 53%/47%;
NutritionalNutritional status after malaria control
children/teenagers/adults:: 35%/35%/30%). Children who were over 1 year in 1995
weree examined every year.
Basedd on the Z-scores derived from the NCHS/WHO 1978 and the CDC 2000
referencee populations, the overall prevalence of stunting during the first survey in
19955 was 53% with both tables. Severe stunting occurred in 24% and 22% of the
overalll population, respectively. The prevalence of wasting was 13% and 27%
respectively,, including 1.4% and 9% severe wasting. Less than 0.6% of the overall
populationn was at risk for overweight. Among teenagers and adolescents stunting and
wastingg were particularly common: 60% stunting, over 12% wasting and no obesity,
ass compared to both US reference tables.
II I I I I I I I I I I I I I I I I I I I I I I I
33 5 7 9 11 13 1517192123 25
544 72 90 108 126 144 162 171
33 5 7
ii i i i i
99 11 13151719212325
i i i i i i i i i i i i
C Q Q
II I I I I I I I I I I I I I I I I I I I I I I I I
1 33 5 7 9 1113151719212325
Figuree 1: Growth charts of the population of Phan Tien, a Vietnamese ethnic
minoritiess commune. Charts were generated with the LMS method, based on the first
ofof the measurements between 1995 and 2000. In all panels the 5', 50' and 95'
centilescentiles are shown. Solid lines represent males and broken lines represent females.
Chartss for HA, WA, WH and BMI were established with LMS. Figure 1 shows
thee median and the 5th and 95th centiles. The annual growth rate was maximal in
boyss at the age of 11; for girls a growth spurt at pubertal age could not be discerned.
AA strong linear correlation between the LMS Z-scores and the Z-scores by US
referencee populations was observed but with a large intercept of >-2 to -1 SD (Figure
2).Longitudinall analysis showed a significant change of Z-scores among children and
adolescentss younger than 23 in 2000, over the years (Figure 3). In a linear mixed
effectss model, a significant increase of Z-HA and Z-WA of children of both sexes
(p<0.0011 for all) was observed. The mean (95% CI) annual increase of Z-HA was
0.111 (0.09 - 0.12) for boys and 0.14 (0.13 - 0.15) for girls (p < 0.001). HA increased
fasterr than WA, particularly in girls. As a consequence, WH and BMI Z-scores
- 1 00
1 2 3 4
- 2 - 11 0 1
Figuree 2: LMS and US Z-scores in Phan Tien, a Vietnamese ethnic minority
commune.. Open circles: NCHS 1978, crosses: CDC 2000.
19955 1996 1997 1998 1999 2000
Year r 19955 1996 1997 1998 1999 2000
HA A WA A BMI I WH H
Figuree 3: Catch up growth in Phan Tien, a Vietnamese ethnic minority commune.
MeanMean data of subjects < 23 years in 2000. Error bars indicate the 95% CI.
NutritionalNutritional status after malaria control
'' - i - , 11 5 : .
** * XX X
•• X * * * x x x * X x 8
ii . ' « i . I « * . " 5 »" " x
"" " 1 , M » » " 1 » i-
** x * x
s 5 x 5 -
ÏÏ ! » * " j K X * X X X
"" « I x » I a ! I i I I i I ! I x I « , «
fifi x H g
«« H ft
* _ K
10 0 15 5 200 25
20 0 40 0 60 0
;; J - X , J - -
X " ,, * I 8 X « *
MM . i J » I . i I I S .* " i J i » '
;; »<J; j J » . »"" «»
x « X
xx > > N
' I « i | ;; x > i ! I i <* ;*
... " B Ï *
> n S Ü B * ( > i i * . S > " >
» M " I 1 1 ! ' I! sj t i
x v *
"" « . i 8 I « 1
MM * ii i
"" 1 I
SS X h h
800 0 20
x ** "*
M l " "
xx * *
Figuree 4: Mean annual increase of anthropometric Z-scores from 1995 to 2000, by
thee age in 2000, in the ethnic minority population of Phan Tien, southern Viet Nam.
TheThe mean annual increase of Z-scores was calculated by a linear mixed effects
populationpopulation model of the repeated annual LMS Z-scores.
Too see whether this trend was age dependent, the individual linear regression
coefficientss of the mixed effects model, which reflect the change in growth rate, were
plottedd against the subjects' age in 2000 (Figure 4). As an internal validation it was
takenn that at the age of 24 years in 2000, height growth rate should become zero. The
growthh rate increased most in those girls who were 15 years in 2000 and boys who
weree 16 years in 2000. On average, elderly people did not show an increase of BMI.
Increasee of growth rate showed a negative correlation with the first Z-HA (R2 =
0.119,, p <0.001) , Z-WA (R2 = 0.045, pO.001) and Z-WH (R2 = 0.153, pO.001),
indicatingg that catch-up growth was most outspoken in stunted children. The Z-scores
off the survey at the end of the dry season in 1997 did not deviate from the secular
trendd (data not shown).
Theree was no significant difference between the increase of growth and having
aa positive blood smear for malaria during the surveys. There was also no significant
differencee between the three ethnic groups with respect to LMS Z-scores of all first
Withh respect to our second objective, the BMI is shown for all age groups in
Figuree 5. LMS had difficulties in fitting the BMI-centiles for all aggregated age
groups,, so that Z-BMI for children was taken from the curves shown in Figure 1. The
highestt BMI was observed in young adults. Elderly people had very low values for
BMII was positively correlated to the LMS values of WH Z- in boys (R2 =
0.4072,, pO.001) and girls (R2 = 0.40, P< 0.001). In teenagers this correlation was
weakerr (R2 = 0.32, p<0.001 and R2 = 0.36, pO.001 respectively).
Too study whether in these thin subjects, the BMI, was a measure of
subcutaneouss body fat, the correlation with skin fold thickness was sought. The
calculationn of the percentage of body fat, based on the sum of the four skinfolds,
yieldedd unrealistic values indicating that these equations are not appropriate for this
population.. (22-24) Therefore only the summated skin fold thickness was used for
furtherr analysis. In children, the BMI and sum of skin folds decreased by ageing and
reachedd the lowest value at 8 years in girls and 9 years in boys. During puberty they
increasedd again. In general the correlation between BMI or Z-BMI and the sum of
fourr skin folds was weak. Only in females over 10 years, there was a positive
correlationn between the sum of the four skin folds and the BMI (R2 = 0.5229, p <
0.0011 in teenagers/adolescents and R2 = 0.4892, p < 0.001 in adult females) or Z-
BMII (R2 = 0.4476, p < 0.001 for teenagers/adolescents and R2 = 0.4595, p < 0.001
forr adult females).
NutritionalNutritional status after malaria control
Figuree 5: BMI for age in Phan Tien, a Vietnamese ethnic minority commune. Data
pointspoints represent the first of the measurements between 1995 and 2000
Thiss study uncovered catch up growth, especially in previously stunted
children,, and the development of a pubertal growth spurt, during a period of five
yearss during which the main health intervention was to bring malaria under control.
Thee strength of this study was based on the registration and follow up of
individuall inhabitants, and the survey examiners being not informed of the results of
thee previous years. Other potential sources of bias, e.g. by preferentially attracting
childrenn with complaints or growth retardation, were circumvented by examining all
childrenn of Phan Tien through the primary school.
Thee prevalence of clinical malnutrition was low, with a few cases with slightly
hypopigmentedd hair, a low rate of severe anaemia and no other signs or symptoms of
malnutrition.. Almost all children, though stunted, continued to grow and childhood
mortality,, which usually accompanies severe malnutrition in malarious areas, was
practicallyy nil throughout the study period.(25) In contrast, the international cut-off
valuess of the anthropometric indicators gave an entirely different picture. According
too the US reference tables the prevalence of stunting and wasting as well as severe
malnutritionn would be very high in Phan Tien. This seems incorrect.
Correctt interpretation of anthropometric measures is important because they
aree used to detect subjects or populations at risk for poor health and to design and
followw up health interventions.(26) In industrialized countries, reference tables are
usuallyy available and regularly updated. (27-31) In developing countries, local
referencee tables are often not available and thus reference tables from industrialized
countriess are used instead. Our study shows that the use of US reference tables,
suggestedd unrealistic rates of (severe) malnutrition.
Thee LMS method gave a better description of the anthropometric data in this
population.. It has become the gold standard for constructing growth charts based on
largee reference populations.(32) In this study it served well to normalise the data, to
producee Z-scores and draw growth charts for the entire population of Phan Tien. No
associationn was found between ethnicity and anthropometric indicators. This is valid
forr the two largest groups, Rac Lay and Ta Lop, but a possible difference with the
Kinhh subjects, who definitely have a different physiognomy, cannot be excluded
becausee of the small size of the Kinh group in Phan Tien. Since the majority of
Vietnamesee are Kinh, the Phan Tien chart cannot be extrapolated to the general
Thee difficulties in interpretation of anthropometric data may benefit from
longitudinall follow up. We are not aware of any report using mixed effects population
modelss on LMS Z-scores for this purpose, although Cole already outlined the use of
LMSS for longitudinal studies.(33) However, from a methodological point of view, we
arguee that the analysis of the repeated measures, based on LMS results, is the
appropriatee approach for data sets like ours. In theory, if fitted centiles would not
accuratelyy describe the real distribution in the population, the repeated Z-scores
wouldd show a tendency to correct this over the years, simulating a secular trend. It is
unlikelyy that this also happened in this study for several reasons. Firstly because only
aa systematic fitting error in the initial growth charts, would mimic a secular trend in
alll growing age groups and such errors are unlikely to occur with the LMS
method.(34)) Secondly, the trends in HA and WA and the opposite trends in BMI and
WH,, are meaningful and thus argue against erroneous fitting with LMS. Thirdly, the
precisionn of the LMS method was maximised by using the first measurements of all
subjectss who ever participated and, lastly, the LMS Z-scores showed a strong linear
correlationn with Z-scores based on the US reference populations. Our fear that the
distributionn of the data of Phan Tien would be distorted by squeezing them into the
taill of the distribution of the US reference population did not bear out. But, because of
thee large intercept it seemed more appropriate to use the LMS Z-scores than the US
Thee longitudinal analysis of LMS Z-scores enabled us to recognise catch up
growth.. Catch up growth is an intriguing phenomenon with largely unknown
dynamics.(35)) It appears that child survival benefits from better nutrition and catch up
growth.. (3 6) On the other hand, catch up growth in stunted children is mainly
translatedd into weight gain and increased risk of obesity and thus has a negative
impactt on long term health. (3 7) This is made painfully clear by the steep increase of
obesityy rates in areas of rapid nutrition transition.(38)
Recoveryy from stunting however is a rather flexible process which may extend
welll into preadolescence.(39) Our data show similar results and in addition, an
emergencee of growth spurt at pubertal age in girls, where this was not evident before,
andd its enhancement in boys. Catch up growth was most evident in short and low
weightt children but the changes were small and these findings have to be confirmed.
Thee control of malaria probably contributed to this trend.(8) At the start of the
studyy the local malaria transmission in the village was very high with equal exposure
too all inhabitants. Thereafter the malaria prevalence declined rapidly. At the
NutritionalNutritional status after malaria control
individuall level, no interaction was found between carrying malaria parasites at the
timee of the survey and the Z-scores or acceleration of growth. This is probably related
too the fact that malaria surveys only indicate the point prevalence of malaria but do
nott measure cumulative exposure over time. However, at population level, the
improvementss of the health in Phan Tien, by controlling malaria and helminth
infections,, and the introduction of a primary health care post were the most
significantt changes which took place between 1994 and 2000.
Intakee of nutrients, remained unchanged at a low level throughout the study
period.. Although this is based on our non-quantified observations, it is supported by
thee persistence of high anaemia rates after control of malaria and hookworm
infections.. It is also supported by other data which show that the diet of ethnic
minoritiess and poor people in Viet Nam hardly improved on protein and lipid intake
duringg the same period, despite the socioeconomic changes for other groups in the
population.(40)) We tentatively argue that the improvements of health, and not so
muchh improvements of diet, promoted the catch up growth.
Severee malnutrition was not frequent, and it may have been over estimated
withh the US reference tables, but the population of Phan Tien was indeed very thin
andd clearly at risk for deficiency of macro and micronutrients. The BMI and the
amountt of body fat, assessed by measuring skin folds, were extremely low, and there
wass no indication of gaining body fat, for example by elderly subjects, over the years.
Thee BMI is known to differ between different ethnic groups, especially in the
lowerr ranges and has not been validated as an indicator of malnutrition. (41) (42) In
thiss study, the correlation between skin fold thickness and BMI was weaker in males
thann in females, illustrating that at low values, the muscular body mass contributes
significantlyy to BMI. This study also revealed that the BMI and WH may even mimic
deteriorationn of the nutritional status when recovery of stunting occurs which in itself
iss a clear sign of an improved health situation. Notably, the interpretation of BMI as
ann indicator for obesity associated health risks, is also fraught with difficulties. It was
recentlyy concluded that it is not feasible to define cut offs for optimal body
dimensionss and overweight for the general Asian populations.(43) The summated
skinn folds are by nature a height and weight independent indicator of the amount of
bodyy fat. But, as our results show, also their predictive value for pathologic
conditionss declines when they reach the very low range.
Itt is tempting to speculate on the nutrition transition in Viet Nam. Viet Nam is
aa country with very rapid socioeconomic changes but for the ethnic minorities these
changess are less rapid. In this study cohort we observed catch up growth probably
causedd by controlling malaria without any indication of the obesity pandemic
affectingg this commune. However, prolonged follow up in Phan Tien may teach us
moree about the dynamics of the transition from marginal nutritional status to
overweight.overweight. We tentatively suggest that in areas with marginal food supply, it i
advisablee to start with interventions for infectious diseases, especially malaria control,
becausee that stimulates catch up growth without inducing overweight. This certainly
needss further study.
Inn conclusion, in a chronic state of low food intake, cut-off values for
anthropometricc indicators, based on international reference tables, have limited
diagnosticc and prognostic value. Repeated annual surveys in a Vietnamese ethnic
minorityy population analysed with the LMS method and repeated measures analysis,
uncoveredd catch up growth extending into preadolescent age, without any indication
off the obesity pandemic affecting this community. The almost complete eradication
off malaria in this population was probably the most important cause for these
(1)) Murray CJ, Lopez AD. Alternative projections of mortality and disability by
causee 1990-2020: Global Burden of Disease Study. Lancet 1997;
(2)) Popkin BM. The nutrition transition and obesity in the developing world. J
(3)) Ramakrishnan U. Nutrition and low birth weight: from research to practice.
Amm J Clin Nutr 2004; 79(1):17-21.
(4)) Barker DJP. Maternal nutrition, fetal nutrition, and disease in later life.
Nutritionn 1997; 13(9):807-813.
(5)) Peeters A, Barendregt JJ, Willekens F, Mackenbach JP, Al Mamun A,
Bonneuxx L. Obesity in adulthood and its consequences for life expectancy: a
life-tablee analysis. Ann Intern Med 2003; 138(l):24-32.
(6)) Popkin BM, Horton S, Kim S, Mahal A, Shuigao J. Trends in diet, nutritional
status,, and diet-related noncommunicable diseases in China and India: the
economicc costs of the nutrition transition. Nutr Rev 2001; 59(12):379-390.
(7)) Thang NM, Popkin BM. Patterns of food consumption in Vietnam: effects on
socioeconomicc groups during an era of economic growth. Eur J Clin Nutr
(8)) Hung LQ, Vries PJ, Giao PT, Nam NV, Binh TQ, Chong MT et al. Control of
malaria:: a successful experience from Viet Nam. Bull World Health Organ