Height of Iranian Children Age 7-18 in Comparison with WHO (2007), CDC (2000) and NCHS (1977) Global Standards: A Systematic Review and Meta-Analysis

Background: Growth assessment based on standardized height and weight tables and charts is an 2 essential part of healthcare monitoring and services for children and adolescents. The present systematic 3 and meta-analysis study aimed to determine the height of Iranian children aged 7 to 18 years and 4 compare it with the global standards. 5 Methods: A meta-analysis was performed on all relevant studies published until November 2018. The 6 databases Iranmedex, Magiran, SID, IranDoc, Medline, Scopus, PubMed, Science Direct, Web of 7 Knowledge, Cochrane and Google Scholar were searched using mesh and non-mesh keywords. Findings 8 of the reviewed studies were compiled using the random effects model. Data heterogeneity was 9 calculated using the Q statistic and the I 2 index. 10 Results: 37 articles were found eligible for inclusion in the meta-analysis. The total sample size in these 11 studies was 395,210, consisting of 189,985 males and 205,372 females. The mean height of 18-year-old 12 Iranian girls ( 𝑋𝑋� = 157.75, %95CI: 156.79 to 158.71 ) was, respectively, 5.30, 5.38, 5.37 cm lower than 13 the reference figures given by WHO ( 𝑋𝑋� = 163.05 ), CDC ( 𝑋𝑋� = 163.13 ), and NCHS ( 𝑋𝑋� = 163.12 ). 14 Similarly, the mean height of 18-year-old Iranian boys ( 𝑋𝑋� = 171.09, %95CI: 169.72 to 172.47 ) was, 15 respectively, 5.05, 5.09, 5.07 cm lower than the WHO ( 𝑋𝑋� = 176.14 ), CDC ( 𝑋𝑋� = 176.18 ), and NCHS 16 ( 𝑋𝑋� = 176.16 ) standards. Conclusions: Given the slightly different growth pattern and low mean height of Iranian girls and boys 18 compared to the global standards, it is imperative to develop and provide more generalizable growth 19 charts for Iranian children in order to improve the quality of growth assessment undertaken for these 20 age groups.


Background 1
Growth assessments are an essential part of healthcare monitoring and services for children and 2 adolescents (1)(2)(3). Height and weight measurements are highly valued because of the wealth of 3 information they provide for studies on the growth, diet and nutrition, and health of younger people as 4 they develop (4). Growth pattern assessments are excellent measures for multidimensional analysis of 5 growth variations over successive generations and their relationship with the environmental factors.

10
In Iran, as in many other countries, the lack of local standards has led to the use of standardized charts 11 and tables of NCHS as references. Since 1978, WHO has recognized these charts, including the latest 12 version published in 2007, as the global growth standard (11). Considering the variety of genetic, racial, 13 geographic, economic, and social factors involved in the growth process and the fact that reference 14 figures and natural growth patterns vary with the population, these reference charts can lead to erroneous 15 conclusions (13). For example, various studies have shown that the average height of many countries 16 has increased over time (19,20). In Europe, the average height of adults has been on an increase since 17 the middle of the nineteenth century, but while this increase has been about one centimeter per decade 18 in northern Europe (20), it has been three centimeters per decade in southern and eastern Europe (13).

19
In the Netherlands, the average height has increased from 165cm in 1860 to 181cm in 1990, making the 20 Dutch the tallest nation in the world (20). This trend has also occurred in the United States, which has 21 the world's most racially diverse population, but the increase in height-for-age has been about 7 22 centimeters or 4% less than in northern Europe (19). Several anthropometric studies conducted in 23 different parts of Iran show major discrepancies in terms of height percentiles with each other and with 24 the global standards (2,(27)(28)(29). Therefore, a closer examination of the issue is necessary, firstly because 25 of the role of growth factors as major metrics for long-term improvement in the public health, and 26 secondly to illustrate the various dimensions of generational changes in growth indices (34,35).

27
The growth trends in Iranian students aged 7 to 18 years have been the subject of multiple studies, most 28 4 of which have reported some differences from the global standards. However, this is the first 1 comprehensive meta-analysis of all studies carried out so far on this subject. In this systematic review 2 and meta-analysis, the authors screened and compiled the existing reports on the average height of 7 to 3 18-year-old Iranian boys and girls and compared the findings with the global standards so as to 4 contribute to the development of a clear and explicit local measure for the growth pattern assessment of 5 Iranian children and adolescents.

20
The reference list of articles was also reviewed. In the event of any ambiguity, inaccuracy, or missing 21 data, one of the researchers (R.N) contacted the authors via email up to two times. The bibliography of 22 the found articles was also manually searched to identify more possibly relevant articles (A.Y).

24
Inclusion and exclusion criteria 25 Two of the researchers (AY, KS) screened the articles based on the following inclusion criteria: 1) 26 descriptive design, 2) the studied population being 7 to 18 years old, 3) provision of (mean ± standard 27 deviation) or (mean ± standard error of the mean) for the studied population, and 4) being focused on 28 Iran. The exclusion criteria were: 1) non-random sampling, 2) irrelevance to the subject, 3) and article's 1 lack of sufficient data for meta-analysis. The quality of the articles was evaluated using the STROBE checklist (43). The authors adopted a simple 5 procedure for quality scoring. In this procedure, two of the researchers (S.Sh and A.Y), working 6 independently, gave each part of the checklist a score between 0 and 2, and the articles with total scores 7 of 1-15, 16-30, and 31-44 were classified as poor, moderate and high quality respectively. The articles 8 that earned a score of less than 16 (poor quality) were excluded from the meta-analysis process. Out of 9 the 37 works reviewed, 33 articles, which all had a cross-sectional-observational design, were found 10 eligible for data extraction. The required data were extracted by one of the researchers (A. Y) with the 11 help of a prepared checklist. This checklist was designed to facilitate the organized collection of the 12 data pertaining to the year of publication, age, gender, place of study, sample size, and mean and 13 standard deviation (SD) for all subjects. In the cases where the standard error of the mean (SEM) was 14 reported, the researchers converted it to the standard deviation. Stunting was defined as height-for-age 15 shorter than 3rd percentile or 2 SD below the median of NCHS/WHO growth reference (45). Height 16 reports of all studies were compared with the 50th percentile of the global height reference.

26
The total mean height was calculated by two methods. The first method is the weighted averaging using , where ˆi y is the effect size of study i, and i W is the 1 weight of that study, which here is proportional to the size of the studied population, that is, the 2 population of the province where the study has been conducted. The second method is the meta-analysis 3 of fixed and random effects based on data heterogeneity. In this method, instead of the population of the 4 province, the weights are calculated based on the sample size and the mean error of samples. In this 5 method, the 95% confidence interval is calculated using the equation reported in different studies. In this formula, the weight of each study is inversely proportional to the 15 variance of the parameter in that study. In case of using the inverse variance, the weight of each study 16 will be directly proportional to the inverse variance.

17
In the fixed model, it is assumed that ˆi y follows a normal distribution, or: This means that the observations made in the studies have a normal distribution with a fixed mean ( y 20 for the entire population) and a variance of i v . In the random model, it is assumed that: 21 ˆ( , ) and y (y, v )

22
This means that the observations made in each study follow a normal distribution with a mean equal to 23 7 the mean in the same population (y ) i , and the parameters in different populations have a normal 1 distribution with a mean equal to that parameter in all populations (y) and a variance of 2 τ . Hence, the 2 observed differences have two origins 2 ( ,) v τ ). Heterogeneity across studies was evaluated by using the 3 Q statistics at the P < 0.10 level of significance. We also calculated the I 2 statistic, which describes the 4 total variation across studies attributable to heterogeneity rather than chance; an I 2 value greater than 5 50% indicates at least moderate heterogeneity and potential publication bias was assessed by visual 6 inspection of Begg's funnel plots. 7 8

9
In the initial systematic search, a total of 528 articles were identified. Of these, 215 articles were 10 duplicates and 191 were irrelevant, hence leaving 122 articles eligible for full-text review. After this 11 review, 37 of these articles were excluded because of inconsistency in terms of target population, 28 12 were excluded because of missing data, and 17 were excluded because of not meeting the 13 inclusion/exclusion criteria. The remaining 40 articles were subjected to quality evaluation and 37 of 14 them were found eligible for inclusion in the final analysis ( Figure 1).

16
The total sample size of the 37 articles included in the meta-analysis was 395,210, of which 189,985 17 were male and 205,372 were female. These studies were published in Iran between 1991 and 2017. The 18 profile of the studies included in the meta-analysis is presented in Table 1 19 20 As shown in Figure  According to the results presented in Figure 3, the mean height of Iranian girls and boys is lower than 5 the global standards. The results also show that Iranian girls and boys in the age group of 7 to 12 years 6 are only slightly shorter than the global references, but this difference widens from the age of 14 in girls 7 and 13 in boys. For example, Table 2 shows that at the age of 12, Iranian girls are 4.00, 4.25, and 3.96 8 cm shorter than the WHO, NCHS and CDC standards, but at the age of 16, this difference widens to 9 5.11, 5.10, and 5.14 cm, respectively, and remains the same at least until the age of 18 years. For Iranian 10 boys, the difference from the WHO, NCHS and CDC standards is 5.40, 5.76, and 5.44 cm at the age of 11 13, slightly changes to 5.20, 5.74, and 5.63 cm at the age of 16 and remains the same at least until the 12 age of 18. In other words, these differences from the global standards vary with the gender. For girls, 13 the difference starts at the age of 12 and reaches a maximum of about 5 cm at the age of 18. But for 14 boys, the difference from the global standards remains almost constant as they go through childhood 15 and adolescence.

16
According to the results of meta-regression analysis, the final height of 18-year-old Iranian girls and 17 boys has not significantly changed with the year of the study. No statistically significant difference was 18 observed in this regard for boys (P = 0.20) or girls (P = 0.11) ( Table 3).

3
To the best of our knowledge, this is the first systematic review and meta-analysis on the height of 4 Iranians aged 7 to 18 years in comparison with WHO, NCHS, and CDC standards. In this study, the 5 growth patterns of Iranian girls and boys were found to be different from the reference patterns. Meta-6 analysis of studies conducted on 398,758 male and female students showed that Iranian girls of age 10 7 to 13 are averagely taller than Iranian boys of the same age, but this relation reverses at higher ages. The 8 same trend is present in the WHO standard, but in CDC and NCHS, it starts from the age of 11 instead 9 of 10. In the previous studies, this trend has been attributed to the earlier onset of puberty in girls (50).

10
Research has shown that the mean age of menarche in Iran is lower than in developed countries (51).

11
For example, the mean age of menarche is 12.43 in the United States (52), 13.1 in Norway (53), 12.27 12 in Indonesia (54), 12.6 in Colombia (55) and 12.4 in Mexico (56). It is widely known that there are 13 major differences in the age of menarche in different countries and different raced, but the specific 14 reasons for these differences are still not well understood (57). Besides race, multiple factors including 15 biological, social, nutritional, geographical and lifestyle conditions may affect the age of menarche, but 16 overall, early or late onset of menarche cannot be attributed any one factor (58, 59).

17
Since the meta-analysis showed that the average height of Iranian girls and boys exhibit an almost 18 uniform pattern from the age of 16 to 18, the mean height at the age of 18 was chosen for the final 19 comparison with the global standards. In this comparison, it was found that the mean height of 18-year-20 old Iranian girls is, respectively, 4.67, 4.75, and 4.74 cm lower than the WHO, CDC, and NCHS 21 standards. For 18-year-old Iranian boys, the mean height is 4.58, 4.62, and 4.60 cm lower than the WHO, 22 CDC and NCHS standards. In these findings, the mean heights of Iranian girls and boys of all ages were 23 closer to the WHO figures than to other standards. The results also showed a 13.18 cm difference in the 24 mean height of Iranian boys and girls at the age of 18, whereas this difference is respectively 13.09, 25 13.05 and 13.04 cm in WHO, CDC, and NCHS standards. that sampling has been random and not stratified. For example, if a study has taken a sample of 500 1 people from Ilam province, which has a population of 600,000 people, and another study has taken a 2 sample of 500 people from Tehran province, which has a population of 12 million people, then In meta-3 analysis ,weighting to studies is based on inverse variance and sample size we do not give weight to 4 studies base of province population of studies, when we pooled studies we give same weight to all of 5 provinces. But in weighting mean method we use of structure, stratify sampling and population of 6 provinces and sample size of studies accounted in the total mean. As a result, this weighted mean is 7 more accurate than the other mean discussed above. The weighted mean height of 18-year-old Iranian 8 girls was found to be 3.40, 3.48, 3.47 cm lower than the WHO, CDC and NCHS standards. For 18-year-9 old Iranian boys, the weighted mean of the height was 3.76, 3.80, 3.78 cm lower than the WHO, CDC 10 and NCHS standards, respectively. The weighted mean height of girls and boys at the age of 18 was 11 closer to the WHO reference values than to other standards, nevertheless, there is a substantial difference 12 between the growth trends of Iranian girls and boys and those of the global standards.

13
The above findings are consistent with the results of Ulijaszek et al. (2001), which after studying the 14 mean height of 7-year-old students in numerous studies, reported that the mean height of Asian 15 populations is approximately 1.0-1.7 cm lower than that of other demographic groups in Europe, Africa, 16 North America, and South America (60). A study conducted on students aged 6 to 16 in Sagamu,17 Nigeria, showed that the mean height of these students was lower than the WHO and CDC references 18 and generally closer to the former than the latter (61). The findings of the present work are also in 19 agreement with the reports of studies conducted in Italy, Turkey and Saudi Arabia, which showed that, 20 for most ages, stature growths were lower than the reference figures (62-64). A study on the growth of 21 Nigerians aged 1 to 20 also showed that the 50th percentile of children growth chart was lower than that 22 of WHO/FAO standard (65). On the other hand, these findings were inconsistent with the results of a 23 study by Razzaghy et al. (2006), which reported that the height of Iranian boys aged 6 to 15 are 24 comparable to the CDC figures and local standards are only needed for girls of 6 to 17 years old (6).

27
These inconsistencies can be attributed to the populations studied in those works. For example, the study 28 of Razzaghy was conducted on the children of school age in Tehran, which can be hardly generalized 1 to the entire population of Iranian children and adolescents. It should also be noted that the growth rate 2 of girls and boys in different regions is a function of numerous factors. In this regard, some studies have 3 suggested that children who enjoy a better social and economic status are taller than others (45, 68). In 4 another study, race was highlighted as a key factor for the height of children and statistically significant 5 differences were shown between the growth curves of black and white people (69). However, several 6 studies have questioned the role of race in growth indices. For example, Droomers et al. (1995) showed 7 that Indonesian children with high socioeconomic status had higher height and weight than American 8 children (70). Also, the results of a systematic study showed that in low-income and middle-income 9 countries, children living in urban areas are taller than their rural peers (70). The role of health services, 10 education level, and nutritional diet on the height is also well recognized (68) for chairs and tables, which may increase the prevalence of musculoskeletal problems in the local 24 population (73). The results of this study provide evidence in support of the notion that regional growth 25 standards should be developed according to racial, genetic, and geographical specification of each 26 country so that they could be relied upon for use in local studies on health or growth disorders. 12 The main limitation of this study was the skewed geographical distribution of the reviewed studies, 1 which was reflected in the fact that some provinces had multiple representatives while others had no 2 representative in the analyzed data. The main strength of this work is the large size of the compiled 3 sample, which makes the findings more generalizable for the broader population of Iranian children and           Random effects model of the mean height of 18-year-old Iranian girls and boys * Larger circles represent larger sample sizes. * Width of the diagram represents the distance.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.