Differences in Classification Standards For the Prevalence of Overweight and Obesity in Children. A Systematic Review and Meta-Analysis

Introduction

Overweight and obesity are defined as an abnormal or excessive accumulation of fat that can harm health.¹ Obesity has a multifactorial origin that involves both genetic and environmental factors.² Several studies showed that excess body fat increased the risk of multiple comorbidities, such as high blood pressure, coronary heart disease, type 2 diabetes, insulin resistance, and cancer.^3–5

Globally, the prevalence of obesity and overweight tripled in the child and youth populations from the mid-1970s (about 4%) to 2016 (greater than 18%). In 2016, more than 330 million children and adolescents were affected by overweight or obesity globally. It is estimated that by 2030 nearly 30% of all children will be affected by overweight or obesity.⁶ For this reason, childhood obesity is a major public health problem worldwide.⁷

Height and weight are the anthropometric variables that have traditionally been used to evaluate growth, other indicators such as Body Mass Index (BMI) have also been used.^8–10 BMI is a measure of weight relative to height (weight in kilograms divided by the square of height in meters).¹¹ Since BMI is easy to determine and correlates with body fatty tissue, it is the most frequently used parameter to assess excess body fat in children and adolescents.^12–14 In addition, BMI has been routinely used to analyse overweight and obesity in children and adults worldwide, as in addition to being practical and easy to calculate it is universally applicable.^15–18

Currently, there are three classification systems most frequently used to evaluate overweight and childhood obesity. These are: the criteria of the World Health Organization (WHO) published in 2007 (WHO 2007),¹⁹ the tables of the Centers for Disease Control and Prevention (CDC) 2000 (CDC 2000)²⁰ and the International Obesity Task Force (IOTF or Cole-IOTF) standards.¹⁸

The WHO curves published in 2007 show a population of 5 to 19 years and constitute a reconstruction of the previously recommended reference of growth (NCHS/WHO 1977). They include original data from the National Center for Health Statistics (NCHS), complemented by growth pattern data for children up to 5 years of age, based on a multicentre study (MGRS) conducted in six countries (Brazil, Ghana, India, Norway, Oman, and the United States).18–21

WHO classifies overweight as the weight for height greater than 1 standard deviation above the median of the WHO reference growth standard. Obesity is defined as the weight for height greater than 2 standard deviations above the median of the WHO growth reference standard.^18–21

The CDC 2000 references are based on the growth curves developed by the National Center for Health Statistics (NCHS) in 1977, associating data from 5 cross-sectional studies (National Health and Nutrition Examination Surveys - NHES II, III and NHANES I, II, III) conducted between 1963 and 1994, in the American paediatric population. Their cut-off points define overweight for BMI equal to or greater than the 85th percentile and less than the 95th percentile, and obesity for BMI greater than the 95th percentile, according to age and sex.^18,20,21

Authors such as Cole et al²² published BMI curves elaborated by extrapolation of the cut-off points defined for adults to classify overweight as (25–30 kg/m2) and obesity as (>30 kg/m2) according to age group and sex. These curves were constructed from population studies carried out in 6 countries (Brazil, Great Britain, Hong Kong, the Netherlands, Singapore and the United States) between 1963 and 1993 and apply to children and adolescents of both sexes, between 2 and 18 years of age. Due to their international scope, these references were recommended by the IOTF.^18,20–22

Therefore, determining the choice of reference is critical, since various studies indicate that these classifications can provide disparate results and produce a significant bias in the diagnosis of individuals and infant-juvenile populations.^23–28

Since there is currently no consensus on which reference to use, we performed this systematic review and meta-analysis with the main aim of comparing the three most commonly used international classification systems (WHO 2007, CDC 2000 and Cole-IOTF) and the level of agreement between them.

Materials and Methods

When applicable to literature and data, we performed the review in accordance with the PRISMA 2020 guidelines.²⁹ Registration no CRD42022323402 on PROSPERO: International prospective register of systematic reviews (https://www.crd.york.ac.uk/prospero/).

Inclusion Criteria

The works eligible for inclusion in this review were articles from the last 10 years, between the years 2010 to 2021 in which the sample was composed of individuals under 19 years of age. Articles comparing at least two of the international BMI ratings such as WHO 2007, CDC 2000 or Cole-IOTF for underweight, healthy weight, overweight and obesity were considered, regardless of whether they matched their own regional or national references.

Studies were excluded during the title and abstract selection stage if they were research articles whose language was other than English or Spanish and/or if the full text article could not be obtained. During the full text review, we excluded studies with a different age range (over 19 years) than specified above. We also excluded studies that did not study the prevalence of overweight and childhood obesity. In addition, we discarded articles where such prevalence was not compared with any of the three main international classification systems proposed.^18–20

Sources of Information and Search Strategy

The search was carried out for relevant articles in the Medline and Web of Science databases in the period between April and May 2022. The search strategy used the Medical Subject Headings (MeSH) terms: “Body Mass Index”, “Overweight”, “Obesity”, “Children” and “Reference standards”, combined using the Boolean operators “AND”, “OR” and “NOT”. The search strategies used can be found in the Supplementary Material.

Selection Process

All items retrieved from searches were imported into EndNote and duplicates were removed. Each of the authors independently reviewed the remaining articles by title and summary for inclusion and then performed a full text review for methodological reasons of exclusion, based on inclusion criteria.

Data Extraction and Quality Assessment

For each article, one author independently among authors extracted the specific condition, its prevalence or incidence, the classification method or methods used, the age ranges involved and the regions or countries where the study was carried out.

The Joanna Briggs Institute (JBI) checklist³⁰ for analytical cross-sectional studies was used to assess study quality as the percentage achieved on the checklist.

Synthesis Methods

The synthesis process began by analysing all the main findings and conclusions. Once the synthesis was completed, we classified the prevalence of overweight, obesity or excess weight by age, comparing them with the different international classification systems used. Figure 1 shows the flowchart we followed for the systematic literature review process and Table 1 shows the main results from the selected articles.

Table 1 Summary of Articles included

Figure 1 Prisma flow diagram (2020).

Statistical Analysis

We conducted an overall pooled prevalence of childhood overweight and obesity meta-analysis, as well as a subgroup meta-analysis. We assessed heterogeneity between studies using Cochran’s Q test and I² index. In Cochran’s Q test, we interpret significant heterogeneity when p < 0.05. For the I² index, high degree of heterogeneity was considered in values higher than 50%.³¹ As the result of the test for heterogeneity was very high, we decided to use a random-effects meta-analysis to calculate the combined global prevalence of overweight and obesity with 95% CI.³² To examine the individual influence of each study, we applied a leave-1-out sensitivity analysis.³³ Publication bias was assessed qualitatively by visual inspection of funnel plot and quantitatively by Egger’s linear regression test.³⁴ Only the 19 studies that analysed the prevalence of overweight and obesity with the three international standards were included in the meta-analysis.

Results

Figure 1 shows that the literature search identified 729 results. After removing duplicates and performing the screening process for the remaining articles, we selected 84 articles for full-text review, of which 27 studies met the inclusion criteria.^{23–25,35–58} Table 1 summarises the results obtained from the search of the selected studies. Table 2 shows the quality assessment of the studies.

Table 2 Studies Appraised Using the Joanna Briggs institute Critical Appraisal Checklist For Analytical Cross-Sectional Studies

CDC 2000 Vs Cole-IOTF Classification

The study by Mosellekgomo and van Staden³⁵ conducted in South Africa found that the CDC classification indicated a very higher prevalence of overweight and obesity in both males and females compared to the IOTF classification. The same occurred in the work of Goon et al³⁶ in Nigeria, where the prevalence found using the CDC was significantly higher than that found using the IOTF classification.

WHO 2007 Vs CDC 2000 Classifications

The study by Fetuga et al³⁷ in Nigeria, males had a significantly lower mean weight compared to females and a lower BMI compared to females. The prevalence of overweight + obesity was lower in both sexes when determined with CDC standards compared to WHO standards among individuals aged 6 to 16 years.

WHO 2007 Vs Cole-IOTF Classifications

In the study by Phan et al³⁸ in Vietnam, they identified that the height, weight and BMI for age of the male participants were slightly higher than the indices of the female participants. However, in this study population, they found that the percentage of body fat was significantly higher in girls compared to boys. The proportion of children affected by overweight, or obesity was higher in the age range of 11 to 12 years. In the category of obesity in children, the prevalence based on the WHO Z-score was approximately 10–15% higher than the prevalence based on the IOTF and the CDC. Also, the study carried out by Bergel Sanchís et al³⁹ in the countries of Mexico, Venezuela and Spain found that the WHO references overestimated the prevalence of underweight, overweight and obesity in relation to the cut-off points proposed by the IOTF.

WHO 2007 Vs CDC 2000 Vs Cole-IOTF Classifications

The study by Meyer et al⁴⁰ carried out in Argentina, when analysing the prevalence of obesity calculated with the different references, observed that the WHO reference gave the highest rates (29.3% compared to the CDC of 17.6% and the IOTF of 10.1%). For obesity, the differences between references were significant in all comparisons, except between the IOTF and the CDC in both sexes and age groups.

The study by Shields and Tremblay²³ carried out in Canada, where it was found that in children aged 8 to 13 years, in terms of the prevalence of obesity, the cut-off points of the IOTF were uniformly higher than the cut-off points of the WHO. In addition, for children aged 10 and 11, the IOTF’s obesity cut-off points were about 3 BMI units higher than those of WHO.

At all ages, WHO cut-off points assigned higher percentages of boys and girls affected by overweight/obesity than the IOTF and CDC cut-off points.²³ In the study by González-Casanova et al²⁴ carried out in Colombia, it was found that the WHO system generally yielded the highest prevalence estimates, while the IOTF system yields the lowest.

In the study by Hassapidou et al⁴¹ conducted in Greece, it was generally observed that BMI was significantly higher in boys than in girls. Regarding overweight, the rates according to the IOTF criteria were significantly lower compared to those of the WHO and CDC, respectively. However, the CDC references showed a significantly higher prevalence of obesity compared to the IOTF and WHO, respectively. Also, in the study of Misra et al⁴² conducted in India, the IOTF cut-off points showed high agreement with those of the CDC and WHO, respectively.

In the study by Ramírez et al⁴³ in Mexico, in general, the highest rates of prevalence of overweight and obesity in children were assessed with the WHO reference in both ethnic groups in all regions.

In the study by Ma et al²⁵ in China, boys were slightly taller and heavier than girls. According to the IOTF reference, 17.1% of all children were affected by overweight or obesity. According to CDC criteria, it was 22.9% of all children and according to WHO criteria, it was 24.8% of the total. In this study, the WHO and CDC demonstrated a high level of agreement. However, the level of agreement between the WHO and IOTF and CDC was low.

In the study by Silva et al⁴⁴ in Portugal, according to the IOTF, an overall prevalence of overweight was estimated at 18.7% and obesity at 13.5%. The overall prevalence of obesity and overweight was also found to be 14.9% and 20.6% using the WHO curves and 7.2% and 20.2% using the IOTF estimates. In another study carried out in Portugal by Minghelli et al⁴⁵ it was found that the WHO criteria followed by the IOTF estimated the higher prevalence values of overweight and obesity compared to CDC.

In the study by Medehouenou et al⁴⁶ in Canada, the prevalence of obesity was similar among boys and girls according to the CDC and WHO criteria. Also, in the study by Pop et al⁴⁷ in Romania, the prevalence of overweight and obesity was higher in boys for WHO, CDC and IOTF than in girls. Obesity had a higher prevalence with the WHO than with the CDC criteria, with the lowest prevalence being when the IOTF cut-off points were used.

In the study by Dereń et al⁴⁸ in Ukraine it was observed that overweight and obesity occurred more frequently among boys than girls. The IOTF and CDC classifications resulted in a lower incidence of overweight compared to the WHO criteria. In another study carried out in Ukraine by Dereń et al,⁴⁹ significantly more girls were affected by underweight compared to boys. Therefore, boys were more likely to be affected by overweight than girls.

WHO 2007 Vs CDC 2000 Vs Cole-IOTF Vs National Classifications

In the study by Baya et al⁵⁰ carried out in Bolivia both the CDC and WHO results indicated a higher prevalence of overweight than the BAP (national reference). In general, it was suggested that the references of the CDC, the IOTF and the WHO overestimated the prevalence of overweight in adolescents aged 13 to 18 years and the IOTF in children aged 12 to 14 years. Also, in the study by Shan et al⁵¹ conducted in China, the combined prevalence of overweight and obesity according to the WHO reference was the highest among the four criteria (WHO, CDC, IOTF and Local-WGOC).

Another study that also included a national reference was that of Salehi-Abargouei et al⁵² in Iran, where the use of the IOTF criteria revealed overweight among 10.8% of Iranian adolescents, without significant differences between the two sexes (9.7% of boys versus 11.7% of girls). The highest prevalence of obesity was obtained using the WHO criteria.

In the study by Bahk and Khang⁵³ in Korea among the four sets of criteria (WHO, CDC, IOTF and the Korean national reference KCDC), the WHO criteria showed the highest prevalence of overweight and obesity in all age groups. The IOTF criteria produced the lowest prevalence of obesity in all age categories. Also, El-Ghaziri et al⁵⁴ in Kuwait found no significant differences in the prevalence of healthy weight status and the combined prevalence of overweight and obesity among the WHO, CDC, and IOTF. All international approaches showed almost perfect agreement, the highest being between the CDC and WHO. The IOTF criteria showed high agreement with the WHO and CDC. In the study by Wickramasinghe et al⁵⁵ in Sri Lanka, the proportions of children affected by obesity classified by the WHO criteria were slightly higher than the proportion detected by IOTF cut-off values.

WHO 2007 Vs CDC 2000 Vs Cole-IOTF Vs Other Classifications

In the study by Heer et al⁵⁶ conducted in Mexico, the WHO standards showed the highest agreement with local standards. Both WHO and CDC showed better agreement with local scores than with IOTF values.

WHO 2007 Vs CDC 2000 Vs National Classifications

Finally, in the study carried out by Pérez-Bermejo et al⁵⁷ in Spain, significant differences were found in BMI values according to age between the two sexes at some ages, for this reason, the authors argued that, if the WHO reference were used, the rate of overweight and obesity would be double that of national references.

Meta-Analysis Results

Table 3 shows the main results of the meta-analysis. There were 19 studies that reported the prevalence of overweight (Figure 2) and obesity (Figure 3) analysed with the three standards under study. There was significant (Q = 1617.37) and substantial heterogeneity between studies (I² > 99% in all cases). Figure 4 shows the subgroup analysis. The difference between these subgroups was statistically significant. For overweight, Q = 6.27, p = 0.044 and for obesity, Q = 23.27, p = 0.000. In both cases, the values calculated with the WHO 2007 standard are significantly higher with a lower weight in the overall combined effect. Table 4 shows the leave-1-out sensitivity analysis of the influence of single study on the pooled prevalence of overweight and obesity in children. Therefore, no study has been excluded from the meta-analysis. Egger’s test of bias was not significant, indicating that smaller studies not reported larger estimates of the prevalence of overweight and obesity (t = 1.17; p = 0.247). A funnel plot (Figure 5) was used to assess possible publication bias, which showed no substantive asymmetry.

Table 3 Global Prevalence of Childhood Overweight and Obesity Using Random-Effects Meta-Analysis and Subgroup Meta-Analysis

Table 4 Leave-1-Out Sensitivity Analysis of the influence of Single Study On the Pooled Prevalence of Overweight and Obesity in Children

Figure 2 Overall and subgroup prevalence of overweight. Forest plot of the studies documenting prevalence of overweight with the three standards under study. The analysis included 19 studies with a total of 211,739 cases.

Figure 3 Overall and subgroup prevalence of obesity. Forest plot of the studies documenting prevalence of overweight with the three standards under study. The analysis included 19 studies with a total of 211,739 cases.

Figure 4 Meta-analysis of the prevalence of overweight and obesity by subgroups according to the three standards under study.

Figure 5 Funnel plot of standard error by logit event rate. The analysis included 19 studies with the three standards under study (57 points).

Discussion

The results of this literature review suggest that the choice of a BMI reference may significantly influence the decision to provide clinical advice or treatment, as well as estimates of the healthcare resources needed to counteract overweight and obesity. The percentage of children, young people and adolescents affected by overweight varies considerably depending on the BMI cut-off points across classifications and may also vary by age group and ethnicity.^40,59

The prevalence of childhood obesity has increased dramatically all over the world in recent years. It is associated with cardiovascular disease, hypertension, diabetes, osteoarthritis, and cancer. Children affected by obesity also have an increased risk that obesity will persist into adulthood. In addition, childhood obesity has also been associated with mental health problems in young people. While obesity in adults can be easily measured using the BMI calculation, determining overweight and obesity in children is more controversial.⁵⁹

When comparing the three most commonly used international classification systems, WHO 2007, CDC 2000 and Cole-IOTF, and the agreement between them, we observed that each selected study compares and analyses these systems differently.^35,36 The comparison between the WHO 2007 classification and CDC 2000 found that the prevalence of overweight and obesity was lower in both sexes when determined with CDC standards compared to WHO standards.³⁷

The WHO 2007 against Cole-IOTF classification was exclusively investigated in only two studies analysed. For obesity in children, prevalence based on WHO was approximately higher than prevalence based on IOTF.³⁸ WHO references were found to overestimate the prevalence of underweight, overweight, and obesity relative to the IOTF for the diagnosis of malnutrition.³⁹ The comparative studies between the three international classifications of WHO 2007, CDC 2000 and Cole-IOTF found that, when analysing the prevalence of obesity, those obtained with the WHO reference were higher compared to CDC and IOTF.⁴⁰

The differences were generally greater between WHO and IOTF references than between WHO and CDC references. Overall, WHO cut-offs yielded the highest prevalence estimates and IOTF cut-offs the lowest.^{23,40–46,48–50,59} Also BMI classification patterns by age tended to be similar between IOTF cut-offs and CDC in contrast to the WHO cut-offs, for all ages and sexes.³⁸

On the other hand, the authors argue that all international references tend to underestimate underweight, while they overestimate overweight.⁵¹ The highest prevalence of obesity was obtained using the WHO criteria compared to CDC, IOTF and the national criterion in Iran.⁵² Also, among the four sets of WHO, CDC, IOTF and National-KCDC criteria in Korea, the WHO criteria showed the highest prevalence of overweight and obesity in all age groups.⁵³ Studies analysing the WHO 2007 classification versus the national classifications determined that, if the WHO reference was used, the rate of overweight and obesity would double compared to the national reference.⁵⁷

Therefore, statistically significant differences in terms of the comparison of the different international references were found in the studies by González-Casanova et al,²⁴ Mosellekgomo and van Staden,³⁵ Goon et al,³⁶ Fetuga et al,³⁷ Meyer et al,⁴⁰ Hassapidou et al,⁴¹ Pop et al,⁴⁷ Dereń et al,⁴⁸ Salehi-Abargouei et al,⁵² Pérez-Bermejo et al⁵⁷ and López-Sánchez et al.⁵⁹ Furthermore, in the Southern European,⁵⁹ significant differences between obesity and overweight values were demonstrated between the Cole-IOTF classification with 32.3% and WHO 2007 with 37.3%, respectively. The divergences between the three international classification systems are due to the fact that they use different selection criteria, samples and data collection in different time periods, as in the case of the IOTF published in 2000, where the BMI cut-off points are set at ≥25 kg/m2 and <30 kg/m2 for overweight and BMI ≥30 kg/m2 for obesity.²² In contrast, the WHO 2007 reference was composed of a sample collected in the 1970s and analysed between 1997 and 2003 where the diagnosis of overweight was set at BMI >+1 SD and ≤+2 SD and obesity at BMI >+2 SD.¹⁵ Finally, the CDC 2000 classification was elaborated by the National Center for Health Statistics (NCHS) in the US paediatric population between 1963 and 1994 establishing overweight as BMI ≥P85 and <P95 and obesity as BMI >P95.²⁰

BMI is the accepted tool for evaluating overweight and obesity in children. However, it should be used with caution as BMI is not a measure of fat mass or the distribution of adipose tissue in the body.⁵⁹ There are several current trends as to which parameters should be considered more when diagnosing childhood overweight and obesity. Some authors study following the recommendations of the European Childhood Obesity Surveillance Initiative (COSI) of the WHO.⁶⁰ However, other authors prefer to use other anthropometric references or the estimation of compartments and body composition by electrical bioimpedance.

Among the limitations of this study, we consider the different results that can be obtained by using different strategies to determine overweight and obesity, and the uniqueness of each and every child, youth and adolescent when these strategies are applied. We observed differences by age group and ethnicity when evaluating and comparing the application of the various classifications, so this may be a cause of bias in the comparison of the results obtained in each study analysed.

Due to this disparity between the prevalence of childhood overweight and obesity according to the classification used, we must highlight the recent publication by Cole and Lobstein⁶¹ where they developed an algorithm to harmonise the cut-off points according to BMI for age and sex of the WHO, CDC and IOTF and make them comparable. This tool allows the researcher to choose any of these three classifications and compare them with the others, making the results more comparable. This makes it a very interesting tool for minimising the possible biases of comparison between different populations and resolving hypotheses in future lines of research.

Conclusion

The prevalence of childhood overweight and obesity according to BMI was determined to be higher in boys than in girls in most studies, this was considered when analysing the classifications of the WHO 2007, CDC 2000 and Cole-IOTF together. However, there was a higher prevalence of overweight and obesity in girls than in boys when only the CDC 2000 and Cole-IOTF criteria were considered.

The WHO 2007 criteria were those with the highest prevalence of overweight in the child and youth populations compared to CDC 2000, Cole-IOTF and national or local criteria.

Both the results of the review and the great heterogeneity found in the meta-analysis show that it is necessary to unify the criteria for the classification of childhood overweight and obesity. International standards are insufficient for working with the current population. Further studies similar to that described above by Cole and Lobstein⁶¹ comparing or constructing national or local references using BMI as a primary measure are needed or a working group should be created to address this issue and agree on the unification of a gold standard to address the current epidemic of childhood overweight and obesity, taking into account the geographical region, the ethnic groups and the age groups of the child and youth population and above all, the secular growth to stop working with a standard that has been very useful in the absence of others, but that should be updated.