Comparison of WHO and Indonesian growth standards in determining prevalence and determinants of stunting and underweight in children under five: a cross-sectional study from Musi sub-district [version 1; peer review: 2 approved with reservations]

Background: Undernutrition among children under five continues to be a critical global public health challenge, especially in developing countries. However, it is believed that Indonesian children are “below” the global standard, thus the WHO standard is not reliable to present the actual prevalence. This study aims to compare the difference between WHO and Indonesian growth standards regarding prevalence of stunting and underweight and its determinants. Methods: This is a cross-sectional study carried out in Musi subdistrict, East Nusa Tenggara province in July 2019. East Nusa Tenggara province had the highest prevalence of stunting and underweight in Indonesia. The study population were children under five, and total sampling method was used for this study. Length/height-for-age and weight-for-age were plotted using WHO and national standards. Univariate and multivariate binomial logistic regression were used for Open Peer Review


Introduction
Undernutrition among children under five continues to be a critical global public health challenge, especially in developing countries 1 .Not only affecting the health of one individual, undernutrition also contributes to many aspects of sustainable development 2 .There are three indicators to measure nutritional imbalance that lead to undernutrition, which are: stunting (low height for age), underweight (low weight for age), and wasting (low weight for height).Stunting is the result of chronic nutritional deprivation, reflecting the cumulative effects of undernutrition and infection.Underweight is a composite indicator and it includes both acute and chronic undernutrition.Wasting is a symptom of acute undernutrition, usually caused by insufficient food intake or high incidence of infectious disease.High prevalence of those indicators reflects poor nutrition and health status among children under five in the population 3 .
According to the data from 2011, the global incidence of stunting, underweight, and wasting were approximately 164.8 million (25.7%), 100.7 million (15.7%), and 51.5 million (8%) among children under five, respectively.Meanwhile, the global deaths attributed to stunting, underweight, and wasting were approximately 1.017 million (14.7%), 999,000 (14.4%), and 875,000 (12.6%) 4 .Until 2018, undernutrition rates remained alarming, although the prevalence was declining.Among continents, Asia has the highest prevalence of stunting (55%) and wasting (68%).Based on country income classification, 65% of all stunted and 73% of all wasted children live in lower-and middle-income countries 5 .However, in the 2018 report, there is no updated data regarding the prevalence of underweight.
The latest basic health survey in Indonesia in 2018 showed that the prevalence of stunting, underweight, and wasting was 30.8%, 17.7%, and 10.2%, respectively.Among other provinces in Indonesia, East Nusa Tenggara province has the highest prevalence of stunting and underweight, at 42.6% and 29.5%, respectively.Meanwhile, the prevalence of wasting was lower, ranked 8 th out of 35 provinces 6 .According to undernutrition severity classification, the severity of stunting is high and underweight is medium in Indonesia.In East Nusa Tenggara province, the severity of stunting is very high, and the severity of underweight is high 7 .
The determinants of child undernutrition are multifaceted and interconnected 8 .Understanding the determinants of childhood undernutrition is important to improve children's nutrition by developing the effective and sustained multi-sectorial nutrition programs and interventions over the long term 9 .Unfortunately, studies evaluating the risk factors of child malnutrition in Indonesia were scarce 10 .A recent review article showed that determinants of stunting in Indonesia were similar to other countries, including maternal height and education, premature birth and birth length, exclusive breastfeeding, and socioeconomic status 11 .
However, determination of undernutrition always uses the WHO growth standard in Indonesia.It is believed that Indonesian children are "below" the global standard in general, thus the WHO standard is not reliable to present the actual prevalence.Therefore, the Indonesia national growth standard was made using data from National Basic Health Survey 12 .To this date, no study has been done to scrutinize the difference between these two standards.This study aims to compare the prevalence and determinants of stunting and underweight using WHO and national standards.We use the data from one of the sub-districts in East Nusa Tenggara province because this province had the highest prevalence of stunting and underweight among children under five in Indonesia.

Ethical statement
This study followed the principles of the Declaration of Helsinki and was approved by the Department of Health Timor Tengah Utara district (approval number: DINKES.440/995/XI/2019).This study also complies with STROBE guidelines 13,14 .All parents gave their written informed consent prior to their children's inclusion in the study.Information for informed consent was given before the informed consent form was signed.Details that might disclose the identity of the study subjects were omitted from the published data file.

Study design and population
This study was an observational cross-sectional study conducted in Musi sub-district, one of the sub-districts in East Nusa Tenggara province.Participant recruitment and data collection were conducted in July 2019.Data analysis was conducted in October -December 2019.There were six villages in Musi sub-district.The study population were children aged less than five years old.Total sampling was used for this study.The children and their parents were approached face-to-face by JF during the monthly growth monitoring program in Posyandu ("Pos Pelayanan Terpadu"), a healthcare program by the Indonesian government.Inclusion criteria were children under five who attended the growth monitoring program during the study period, who were born and live with their parents in Musi sub-district, and had both maternal and child health books (Buku Kesehatan Ibu dan Anak / KIA) and health record card (Kartu Menuju Sehat / KMS) published by the Ministry of Health Republic Indonesia.Children with incomplete KIA and KMS were excluded from the determinants analysis.

Data collection
Both primary and secondary data was used in this study.Primary data for this study consisted of data obtained through interviews with parents, child anthropometry measurements, and maternal height measurements.The interviews took place in the same location as the Posyandu and were conducted by JF using a predetermined questionnaire.The length of the interview was around five minutes.JF is a female general practitioner who worked in primary healthcare in the sub-district where the study took place.She had worked there for seven months when the study was conducted.Interviews with parents was carried out to obtain information regarding village of origin, parents' highest education, number of parities, delivery method, and gender and age of their children.Anthropometry measurements of maternal height and child length/height were done by healthcare workers from Oeolo Primary Healthcare.Secondary data from KIA and KMS was used to obtain data regarding birthweight, gestational age, maternal mid-upper arm circumference, and maternal age during pregnancy.

Categorization of variables
Underweight and stunting were categorized using WHO child growth standards and Indonesian growth standards for the same sex 12,15 .Underweight is defined as weight for age below -2 standard deviations (SD), and severe underweight is defined as weight for age below -3 SD.Stunting is defined as length/height for age below -2 SD, and severe stunting is defined as length/height for age below -3 SD.The cut-off level for maternal mid-upper arm circumference was 23.5 cm, for maternal height was 150 cm, and for children's birthweight was 2500 g.The cut-off level for maternal mid-upper arm circumference was according to the Indonesian national cut-off 16 , while for maternal height and children's birthweight, the cut-off was based on a previous study 17 .Maternal age during pregnancy was categorized to <20 years old, 20-35 years old, and >35 years old.Gestational age and intrauterine growth were categorized based on Lubchenco charts.It categorizes the gestational age to preterm (<37 weeks), term (37-42 weeks), or postterm (>42 weeks) and the intrauterine growth to small for gestational age (SGA) (<10 th percentile), appropriate for gestational age (AGA) (10 th -90 th percentile), or large for gestational age (LGA) (>90 th percentile) 18 .

Statistical analysis
Acquired data was analysed using SPSS Statistic for Windows, version 25.0 (IBM Corp., Armonk, N.Y., USA).Data analysis was conducted in two phases.In the first phase, univariate logistic regression was used to identify independent variables that were associated with stunting or underweight.Variables with p < 0.1 were included in the next phase.In second phase, multivariate logistic regression using backward selection was used.Variables with p <0.05 from multivariate analysis were considered as the determinants.

Results
There was a total of 408 children under five in Musi sub-district.Based on WHO standard, the prevalence of stunting and underweight were 53.9% and 29.17%, respectively 19,20 .Using national standard, the prevalence of stunting and underweight were 10.7% and 17.7%, respectively.There was a significant difference of stunting and underweight between the prevalence from the WHO and national standard (both p <0.001).However, there were only 218 children that fulfilled the criteria to be included for the determinants analysis (Table 1).

Sociodemographic characteristics
The prevalence of stunting and underweight among this study population were 51.4% and 31.7% according to WHO standard and 8.3% and 19.3% according to national standard (Table 1).The number of male and female children was almost equal.More than half of the children were aged between 24 and 59 months old.Majority of the children were born term with a birthweight of more than 2500 g.The education level of both parents was mainly elementary school graduates.Almost half of the mothers had a height of less than 150 cm and more than half of the mothers had a mid-arm circumference of ≤23.5 cm during pregnancy (Table 2).

Determinants of stunting according to WHO and national standards
Based on WHO standard, univariate logistic regression analysis indicated that children with maternal height below 150 cm (OR = 2.844; 95% CI = 1.632 -4.956) were more likely to be stunted (Table 3).In the multivariate logistic regression analysis, other variables with p-value between 0.05 and 0.1 from the univariate analysis (child's birthweight, child's intrauterine growth status, maternal mid-upper arm circumference, and number of parities) were included.Multivariate analysis indicated that children with maternal height below 150 cm   likely to be stunted (Table 3).In multivariate logistic regression analysis, other variables with p-value between 0.05 and 0.1 from the univariate analysis (child's intrauterine growth status and maternal height) were included.No determinant was found in the multivariate analysis (Table 4).

Determinants of underweight according to WHO and national standards
Based on WHO standard, univariate logistic regression analysis indicated that children with birthweight below 2500 g (OR = 3.159; 95% CI = 1.507 -6.622) or intrauterine growth 95% CI = 1.989 -14.186) were also more likely to be underweight (Table 5).In multivariate logistic regression analysis, other variables with p-value between 0.05 and 0.1 from the univariate analysis (paternal education and number of parities) were included.Multivariate analysis indicated that children with intrauterine growth restriction (OR = 3.182; 95% CI = 1.450 -6.980) were more likely to be underweight.Children with maternal age under 20 years old during pregnancy (OR = 6.252; 95% CI = 1.911 -20.457) or with mother that had more than four parities (OR = 4.319; 95% CI = 1.189 -15.689) were also more likely to be underweight (Table 6).
Based on national standard, univariate logistic regression analysis indicated that children with birthweight below 2500 g ) were also more likely to be underweight (Table 6).

Discussion
In our study, the prevalence of both stunting and underweight were significantly lower when measured using Indonesian standard compared to when using WHO standard.It has been suggested that overdiagnoses of stunting or underweight are more likely to occur in developing countries 21 .There are many countries that already proposed their own national growth standard, which are: Korea 22 , Thailand 23 , Argentina 24 , China 25 , India 21 , and 18 European countries 26 .It is argued that the national growth standard of each country is more suitable to reflect the condition in its own population 23 .However, there were only few published studies that compare the difference between national growth standards and WHO growth standard.A comparison study among Thai children in the first two years of life showed that the prevalence of stunting was higher when using WHO standard in both sexes, but at 24 months the only significant difference was in girls.The prevalence of underweight showed a monotonic increment when using WHO standard, but the Thailand national standard showed a fluctuation 23 .In Argentina, the prevalence of underweight using WHO standard was 2 times higher than when using their national standard.Meanwhile for stunting, the prevalence when using WHO standard was 1.5 times higher 24 .In contrary, a comparison study from China showed that the prevalence of stunting and underweight was significantly higher when measured using their national standard 25 .
The marked difference in measurements using Indonesian standard and WHO standard probably stems from the difference in methodology during the development of both growth reference standards.The WHO standard was developed using data from five cities in five different countries: United States, Turkey, Norway, Brazil, and India.The children included in the study were healthy children with suitable sociodemographic conditions for growth.Moreover, all participants agreed to follow the feeding recommendation by WHO 27 .In contrary, the development of Indonesian standard did not have any inclusion and exclusion criteria for study participants.It also did not mention the sociodemographic background of the participants or their feeding habits.The study, however, collected data from all 33 provinces of Indonesia to better reflect the growth of Indonesian children 12 .
Review article by Beal et al. concluded that the determinants of stunting in Indonesia are maternal height and education, child's gender, premature birth and birth length, exclusive breastfeeding for six months, living area, and household socio-economic status 11 .In our study, the determinants of stunting according to WHO standard were maternal height less than 150 cm and maternal upper mid-arm circumference <23.5 cm.In contrast, no determinant was found when Indonesian standard was used.It is because the prevalence of stunting according to Indonesian standard was low.The significant difference in stunting prevalence calculated using Indonesian and WHO standards might be because the WHO standard does not represent local growth appropriately due to population differences in height 26 , and Indonesian people are generally shorter than the rest of the world.
Regarding underweight, the determinants were also different according to the two different standards.However, there was one common determinant: intrauterine growth restriction.The difference of underweight prevalence between the two standards was not as marked as the difference in stunting prevalence; this may explain that there was still one overlapping determinant.The increased odds of undernutrition in SGA infants are more relevant in low-and middle-income countries 28 .SGA children are born with lower intrinsic potential for growth due to the persistent effect of growth restriction in utero 29,30 .SGA is a result of poor maternal nutrition during pregnancy when the child is totally dependent on getting nutrition from the mother through the placenta, hence any nutrition deprivation from the mother will affect the proper growth and development of the fetus 31 .
There were several limitations of this study.We did not discern the feeding habits of the participants of this study.Feeding habit could be an important determinant of malnutrition.For example, introduction of complimentary food earlier than four months increased the likelihood of being underweight and stunted 32 .Data on exclusive breastfeeding and history of immunization cannot be obtained because some of our samples have not yet completed the exclusive breastfeeding and basic immunization period.Data regarding socioeconomic status could not be obtained due to parents' unstable monthly income.Data regarding the frequency of diarrhea could not be obtained because this was not well documented in primary healthcare medical records.These factors should be accounted for in the ensuing studies.Nevertheless, despite all of the limitations, this is the first study that compare the prevalence and determinants of stunting and underweight among Indonesian children under five using Indonesian growth standard and WHO growth standard.

Conclusion
The WHO standard was not suitable to diagnose stunting and underweight in Musi sub-district, since the prevalence was significantly higher when using WHO standard compared to when using Indonesian standard.Future studies should be done to re-evaluate the prevalence and determinants of stunting and underweight nationwide using the Indonesian standard.An Indonesian standard for weight-for-height should also be made to re-evaluate the prevalence and determinants of wasting in Indonesia.The conclusion should consider one more limitation is that the small sample size of the data particularly on child stunning number on national WHO standard in comparing.As a result, the next research should be more explored.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility?Yes Are the conclusions drawn adequately supported by the results?among children under five …".Later it is written: "There are three indicators to measure nutritional imbalance that lead to undernutrition …" suggesting that the manuscript is on nutrition.But where is the data on food?The authors need to provide information on food, or at least on energy deposits in terms of skinfold thickness, arm circumferences etc., or in the case of mal-or undernutrition, show clinical signs of protein or calorie malnutrition, or signs of micronutrient deficiencies.This is not the case.
It is known to the reviewer that there is international confusion about the terms stunting and malnutrition.Calculating the portion of stunted children refers to the question of how many children are below height for age < -2 SD of the WHO Child Growth Standards median.This does not necessarily mean that this portion of children is also undernourished.
The manuscript needs major rethinking and re-evaluating of the measurements of height and weight.It is necessary to describe growth of the children of Musi, and to publish these data.The children of Musi are shorter and lighter than suggested by the WHO standard.But this does NOT mean that these children suffer from food shortage.It rather appears that the WHO standard is not applicable for these children, and thus does NOT indicate malnutrition of this child population.This needs to be stressed.The authors should carefully read some of the recent papers discussing the misinterpretation of stunting as a sign of undernutrition (Hermanussen et al.The impact of social identity and social dominance on the regulation of human growth.A viewpoint.Acta Paediatr.2019 Aug 16 3 ).
The tables show odds ratios.It is much more informative when data are presented as true values.The reader is not so much interested in what proportion of children ranges above or below a certain cut-off, but what are the mean values of height, weight, etc. in this child population.
Minor comments: Correct the term "national standard".What you mean is: "national reference".Correct the term "gender".What you mean is probably "sex".Table 4 needs to be shortened, there is no need to show empty boxes.The same applies for table 6 Discussion: The authors write: "SGA is a result of poor maternal nutrition during pregnancy".This is not quite true.Ample evidence obtained from European countries during periods of war and post-war starvation illustrates that even when pregnant women are severely undernourished, the newborn infants only suffer from minor decreases in body weight (Keys A, Brozek J, Henschel A, Mickelsen O, Longstreet Taylor H.The biology of human starvation.The University of Minnesota Press.Minneapolis.1950.).See various comments in the text.manuscript.
3. It is not clear whether the authors wanted to study size (length and weight), or the nutritional situation of the children of Musi.

Response:
We have revised the manuscript and limited the aim of our study to body size (length and weight) of the Musi children, not about nutritional status because we do not have the data regarding food, energy deposits, or any clinical sign of undernutrition.
4. It is necessary to describe growth of the children of Musi, and to publish these data.
Response: Unfortunately, we do not have the growth record of children in Musi sub-district.The design of our study is cross-sectional, thus we only have one-time height and weight measurement of the Musi children.We have published the measurement data in online repository (https://doi.org/10.6084/m9.figshare.12121938.v5for population data and https://doi.org/10.6084/m9.figshare.12127425.v3for determinants data) 5.The authors should carefully read some of the recent papers discussing the misinterpretation of stunting as a sign of undernutrition Response: We have read the corresponding references, and we have made sure to not misinterpret stunting as the sign of undernutrition because the aim of our study is to compare the prevalence of body size using WHO growth standard and Indonesian growth reference.
6.The tables show odds ratios.It is much more informative when data are presented as true values.The reader is not so much interested in what proportion of children ranges above or below a certain cut-off, but what are the mean values of height, weight, etc. in this child population.

Response:
We would like to give more informative data by presenting the true values of weight and height.However, since the children are not in the same age, we are not sure whether it is appropriate or not.7. Correct the term "national standard".What you mean is: "national reference".Correct the term "gender".What you mean is probably "sex".Table 4 needs to be shortened, there is no need to show empty boxes.The same applies for table 6 Response: We have corrected the term "national standard" to "national reference" and term "gender" to "sex".We also have shortened the table 4 and 6 by omitting the empty boxes.

Table 1 . Prevalence of stunting and underweight of children aged 0 -59 months in Musi sub-district.
#Chi square test was used.*p-value between stunted (and severely stunted) and normal.# p-value between underweight (and severely underweight) and normal.p <0.05 was considered statistically significant.