Abstract
Objectives:
To determine which set of BMI cut-offs is the most appropriate to define child and adolescent obesity in urban China.
Design:
A cross-sectional study was carried out between 1 November and 31 December in 2017.
Setting:
Community Healthcare Center in Minhang District, Shanghai, China.
Participants:
A total of 12 426 children and adolescents aged 7–17 years were selected by cluster random sampling. Bioelectrical impedance analysis was the gold standard to measure body composition.
Results:
Comparisons of three sets of BMI cut-offs by sensitivity and κ value revealed that the Working Group on Obesity in China (WGOC) (sensitivity 39·9–84·0 %; κ 0·51–0·79) and WHO (sensitivity 25·5–74·5 %; κ 0·35–0·78) cut-offs were not superior to the International Obesity Task Force (IOTF) (sensitivity 47·9–92·4 %; κ 0·58–0·85) cut-offs across all subgroups. The WGOC and WHO cut-offs yielded higher misclassification rates, in the worst case, categorising 11·2 % of girls with high adiposity as normal and 44·4 % of them as overweight, while the IOTF cut-offs categorised 2·3 % as normal and 30·7 % as overweight. Individuals who were classified by the IOTF cut-offs as overweight had the lowest ratios of high adiposity (4·2–41·6 %) than by the BMI cut-offs for each subgroup. Among pubertal girls, none of the BMI-based cut-offs indicated excellent agreement with body fat percentage, and κ value of the WHO cut-offs (0·35 (95 % CI 0·29, 0·41)) was lower than the other two sets of BMI cut-offs (all P < 0·001).
Conclusions:
The IOTF cut-offs for Asian should be recommended for child obesity screening in urban China. Pubertal individuals need a more accurate indicator of obesity screening.
Keywords: Overweight, Obesity, BMI, Body fat percentage
Child and adolescent obesity may track into adulthood, giving rise to many health issues(1). Great international efforts are underway to control child and adolescent obesity in response to its rapid growth. To accurately define obesity, it is essential to measure excess adiposity(2). Due to the high cost of measuring excess body fat (BF) in clinical settings, the BMI has become a substitute to screen for child and adolescent obesity worldwide. However, it is debatable that BMI is a reliable measure of excessive BF and hence its usefulness of being a tool for obesity screening. Prior studies have used different BMI cut-offs for obesity screening among Chinese children and adolescents, including those recommended by the International Obesity Task Force (IOTF)(3), the WHO(4) and the Working Group on Obesity in China (WGOC)(5). Given the dynamics of BMI in children from birth to adolescence, age- and sex-specific BMI cut-offs are used to identify children’s weight status.
In 2004, WGOC proposed a definition for childhood obesity based on representative data from the 2000 Chinese National Survey on Students Constitution and Health, and cut-offs were constructed based on the 85th and 95th percentiles of the BMI-for-age growth curve suggested by WHO rather than based on the adverse health outcomes(5). Over the past decades, China’s urbanisation and rapid economic growth have led to the increasing prevalence of child and adolescent obesity, especially in cities. WGOC cut-offs might not be applicable for school-aged children nowadays(6,7). Previous studies also used other BMI references such as the WHO cut-offs, while the IOTF cut-offs for Asians have long been infrequently used. To the best of our knowledge, no studies have compared the WGOC cut-offs with either the WHO cut-offs or IOTF cut-offs or the three references concurrently.
This study compared the capacity of the three sets of BMI cut-offs suggested by the WHO, IOTF and WGOC to screen for high adiposity, to determine which is most suitable for screening child and adolescent obesity in urban China.
Methods
Participants
A total of 12 426 school students (6750 males and 5676 females) were enrolled in the study, aged 7–17 years from the Longbai community of Minhang District, Shanghai, China. They were selected from one of the thirteen communities in Minhang District by cluster random sampling. Data were collected from 1 November to 31 December in 2017 and were analysed between 15 January and 7 March in 2018. Children and adolescents were eligible for inclusion if they had no disease associated with pathological obesity, or any cardiac, hepatic, kidney, liver or acute/chronic disease. In general, girls begin puberty around 10–11 years and end puberty around 15–17 years, compared with boys at 11–12 and 16–17 years, respectively(8,9). Therefore, subjects were divided into three age groups (7–10, 11–14 and 15–17 years) which roughly corresponded to pre-pubertal, pubertal and post-pubertal or late-pubertal period.
Body composition assessment
The anthropometric measurements in the study were performed by well-trained nurses or physicians at the Longbai Community Healthcare Center. Before the measures, participants were asked to fast, void and refrain from bathing and vigorous activities for at least 2 h. BF% was estimated by a bioelectrical impedance analysis (BIA) analyser (TANITA BC-420 SMA), which determined total body water by bioelectrical measurement with a high-frequency current (50 Hz, 90 μA); then, each value of BF%, fat mass, fat-free mass, muscle mass and bone mass was calculated by the built-in equations. However, the equations are not provided in official manuals or any literature as they are company confidential (information provided by TANITA Corporation)(10,11). The participant was told to stand on the electrodes with both bare feet parallel without bending knees, so that electric current is supplied from the electrodes on the tips of the toes of both feet, and voltage is measured on the heel of both feet. The TANITA BC-420 SMA was designed for children whose feet were too small for the standard devices used for adults(12). The BIA methodology and Tanita’s accuracy have been supported by research in the previous study and were validated in assessing body composition in children(13–15). Compared with other BF% measurements, BIA is the most suitable method or interchangeable with the dual-energy X-ray absorptiometry method for large-scale investigations(16,17).
Body fat percentage cut-offs
BF% directly reflects fat accumulation, and it was considered the gold standard for obesity screening. BF% cut-offs for obesity were 25 % for boys aged 7–17 years, 30 % for girls aged 7–14 years and 35 % for girls aged 15–17 years. These BF% cut-offs correspond to moderate obesity as defined by the Chinese Guidelines for the Prevention and Control of Overweight and Obesity Among School-Aged Children (the official guidelines designed for Chinese children and adolescents aged 2–18 years), which are associated with increased risks of cardio-metabolic conditions in previous studies(18). They are relatively well matched with adult cut-off values of 25 % for male subjects and 35 % for female subjects. In Japan, another Asian country, the same BF% cut-offs are also recommended for obesity screening in children and adolescents(19). To avoid confusion due to different meanings of the term ‘obesity’, the term ‘high adiposity’ or ‘adiposity’ is sometimes used below to refer to BF%-defined obesity.
BMI cut-offs
Three sets of BMI cut-offs were used to screen for obesity in children and adolescents in the present study, namely the IOTF, WHO and WGOC standards. The IOTF standard was released in 2000, based on data collected between 1963 and 1993 from several large, nationally representative cross-sectional surveys in Brazil, Great Britain, Hong Kong, the Netherlands, Singapore and the USA. Its IOTF BMI cut-offs for overweight and obesity at age 18 years are defined as 25 and 30 kg/m2, and 23 and 27 kg/m2 for Asian children and adolescents, respectively(3). In this paper, we adopted the IOTF cut-offs for Asian populations, and hence all the references to the IOTF cut-offs below refer to Asian cut-offs. The WHO reference (2007 edition) data sets were derived from the Health Examination Survey and the first National Health and Nutrition Examination Survey for the United States between 1963 and 1974(4). At 19 years, the BMI values at +1 sd are identified as the overweight cut-off for boys and girls and the +2 sd value as the obesity cut-off. These values are equivalent to the overweight and obesity cut-off for adults (>25·0 and >30·0 kg/m2). The WGOC cut-offs were based on BMI growth curves developed by the 2000 Chinese National Survey on Student’s Constitution and Health(5). Child and adolescent overweight and obesity were defined as a BMI at or above the sex-specific 85th and 95th percentiles on the WGOC BMI-for-age growth charts and were smooth transition to the adults cut-offs for overweight (>24·0 kg/m2) and obesity (>28·0 kg/m2) at 18 years.
Statistical analysis
To ensure blinding, two different investigators independently determined individual weight status via BMI and BF% measurements. Microsoft Excel 2019 was used to compile the results into a database, and all statistical analyses were performed using SPSS version 20·0 (IBM). Mean and sd were calculated for continuous variables, and ratios and 95 % CI for categorical variables. The χ 2 tests were used to compare obesity rates among different age groups in girls. The McNemar’s test was used to compare the prevalence of obesity between two sets of BMI cut-offs. Diagnostic accuracy of BMI cut-offs for obesity screening was evaluated by the κ statistic, which assesses the level of agreement between BMI and BF% cut-offs by using the term of the chance agreement(20). κ values >0·75 represent excellent agreement beyond chance, values <0·40 represent poor agreement beyond chance and values between 0·40 and 0·75 may be taken to represent fair to a good agreement beyond chance(21,22). Differences between BMI cut-offs on κ values were examined by using z test. Also, the performance of BMI cut-offs was evaluated in terms of sensitivities and specificities, and the McNemar’s test was used to compare sensitivities (Se) and specificities (Sp) between two cut-offs(23). All tests were two-sided, and a P value < 0·01 was considered statistically significant.
Results
Characteristics and obesity prevalence
Characteristics of participants and obesity rates by sex and age are presented in Table 1. There were higher rates of obesity estimated by BF% cut-offs than by BMI-based cut-offs (P < 0·001) except for pre-pubertal girls and post-pubertal boys. The prevalence of high adiposity among pubertal girls (20·1 %) was prominently higher than pre-pubertal girls (9·6 %, χ 2 = 192·583; P < 0·001) and post-pubertal girls (8·7 %, χ 2 = 15·466; P < 0·001) and was approximately 2·0–3·9 times than obesity rates estimated via BMI cut-offs (5·2–10·0 %). However, by using BMI cut-offs, no differences in obesity rates between different age subgroups were significant by pairwise comparisons (all P > 0·01) among girls.
Table 1.
Characteristics of the study population by sex or age
| Characteristics | Values* | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Boys (n 6750) | Girls (n 5676) | |||||||||||||
| Total (N 12 426) | 7–10 years | 11–14 years | 15–17 years | 7–10 years | 11–14 years | 15–17 years | ||||||||
| Mean or ratio | sd or % | Mean or ratio | sd or % | Mean or ratio | sd or % | Mean or ratio | sd or % | Mean or ratio | sd or % | Mean or ratio | sd or % | Mean or ratio | sd or % | |
| Height (cm) | 142·5 | 14·5 | 135·2 | 8·5 | 157·3 | 10·3 | 173·6 | 6·4 | 134·4 | 9·2 | 154·8 | 7·4 | 161·8 | 5·8 |
| Weight (kg) | 38·1 | 14·1 | 33·1 | 9·2 | 49·8 | 13·6 | 69·0 | 16·1 | 30·4 | 8·2 | 46·3 | 11·0 | 55·5 | 10·3 |
| BMI (kg/m2) | 18·2 | 3·7 | 17·9 | 3·3 | 19·9 | 4·0 | 22·8 | 4·8 | 16·6 | 2·8 | 19·1 | 3·6 | 21·1 | 3·5 |
| BF% (%) | 19·1 | 9·5 | 18·4 | 10·6 | 18·4 | 11·1 | 16·8 | 8·5 | 18·0 | 7·1 | 23·2 | 8·3 | 27·7 | 6·2 |
| Fat mass (kg) | 8·1 | 6·8 | 6·9 | 5·9 | 10·3 | 8·8 | 12·8 | 9·7 | 6·0 | 4·3 | 11·6 | 6·9 | 16·0 | 6·7 |
| Fat-free mass (kg) | 30·0 | 9·2 | 26·2 | 4·4 | 39·5 | 7·3 | 56·2 | 7·6 | 24·4 | 4·9 | 34·7 | 5·1 | 39·5 | 4·9 |
| Total body water (kg) | 22·0 | 6·8 | 19·2 | 3·2 | 28·9 | 5·4 | 41·1 | 5·6 | 17·9 | 3·6 | 25·5 | 3·8 | 29·2 | 3·6 |
| Adiposity rate† (%) | 18·5 | 17·9–19·2 | 25·7 | 24·4–27·0 | 26·5 | 24·4–28·5 | 15·7 | 12·2–19·2 | 6·9‡ | 6·1–7·7 | 20·1 | 18·0–22·2 | 11·3‡ | 8·1–14·5 |
| Obesity rate by BMI-based cut-offs (%) | ||||||||||||||
| IOTF | 15·2§ | 14·6–15·8 | 20·5§ | 19·3–21·7 | 18·0§ | 16·2–19·8 | 20·9§ | 17·0–24·8 | 9·6§ | 8·7–10·6 | 10·0§ | 8·5–11·6 | 8·7§ | 5·9–11·5 |
| WHO | 12·8§ | 12·2–13·4 | 20·4§ | 19·2–21·6 | 15·0§ | 13·4–16·7 | 11·6§ | 8·6–14·7 | 6·7 | 5·9–7·5 | 5·2§ | 4·0–6·4 | 3·7§ | 1·8–5·6 |
| WGOC | 13·5§ | 12·9–14·1 | 18·7§ | 17·6–19·9 | 13·4§ | 11·8–15·0 | 15·0 | 11·6–18·4 | 9·6§ | 8·7–10·5 | 8·3§ | 6·9–9·7 | 7·6§ | 5·0–10·3 |
BF%, body fat percentage; IOTF, International Obesity Task Force; WGOC, Working Group of Obesity in China.
Presented as mean and sd or ratio and 95 % CI.
Calculated by BF% cut-offs.
Significant difference from girls aged 11–14 years, P < 0·01 (χ 2 test).
Significantly different from BF% cut-offs in the same sex and age group, P < 0·01 (McNemar’s test).
Sensitivity, specificity and agreement
Table 2 shows the sensitivities, specificities and κ values pertaining to the three sets of BMI cut-offs by sex and age. Very high specificities (92·4–100·0 %) in obesity screening were reported by BMI cut-offs, while sensitivities and κ values varied for BMI cut-offs in different sex and age groups. The comparisons of three sets of BMI cut-offs by sensitivity and κ value revealed that the WGOC (se 39·9–84·0 %; κ 0·51–0·79) and WHO (se 25·5–74·5 %; κ 0·35–0·78) cut-offs were no superior to the IOTF (se 47·9–92·4 %; κ 0·58–0·85) cut-offs across all subgroups. Among boys, the WHO cut-offs did not perform as good as the IOTF cut-offs in κ values (0·64 v. 0·74; P < 0·001) and sensitivity (55·8 v. 66·7 %; P < 0·001) during pubertal growth. The sensitivities and κ values of the WGOC cut-offs were lower than that of the IOTF cut-offs (all P < 0·001) except for post-pubertal boys. Among girls, the WHO cut-offs indicated the worst agreement with BF% cut-offs, especially during puberty (se 25·5 %; κ 0·35) and post-puberty (se 32·6 %; κ 0·46). The WGOC cut-offs had lower sensitivities (v. IOTF) in subgroup aged 11–14 years (39·9 v. 47·9 %; P < 0·01) but showing competitive κ values in all age periods. However, all the BMI cut-offs had found poor sensitivities and lower κ values in the pubertal groups, particularly for girls (se 25·5–47·9 %; κ 0·35–0·58). Among pubertal girls, none of the BMI-based cut-offs indicated an excellent agreement with BF% cut-offs, and κ value of the WHO cut-offs (0·35 (95 % CI 0·29, 0·41)) was lower than the other two sets of BMI cut-offs (all P < 0·001), which exhibited poor agreement with BF%.
Table 2.
Accuracy of BMI-based cut-offs for obesity* screening
| Age | Criteria | Boys | Girls | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Sensitivity | % | Specificity | % | κ† | % se | Sensitivity | % | Specificity | % | κ† | % se | ||
| 7–10 years | IOTF | 74·6 | 72·0–77·1 | 98·2 | 97·7–98·7 | 0·78 | 0·76–0·80 | 85·8 | 80·9–89·7 | 96·0 | 95·3–96·6 | 0·69 | 0·65–0·73 |
| WHO | 74·5 | 71·9–77·0 | 98·3 | 97·8–98·7 | 0·78 | 0·76–0·80 | 72·8‡ | 67·0–77·9 | 98·2 | 97·7–98·6 | 0·72 | 0·68–0·76 | |
| WGOC | 68·4‡ | 65·6–71·0 | 98·4 | 98·0–98·8 | 0·73§ | 0·71–0·76 | 84·0 | 78·9–88·0 | 96·0 | 95·2–96·6 | 0·68 | 0·63–0·72 | |
| 11–14 years | IOTF | 66·7 | 62·3–70·9 | 99·5 | 99·0–99·8 | 0·74 | 0·70–0·78 | 47·9 | 42·0–53·9 | 99·5 | 98·8–99·8 | 0·58 | 0·53–0·64 |
| WHO | 55·8‡ | 51·2–60·3 | 99·6 | 99·1–99·9 | 0·64§ | 0·60–0·68 | 25·5‡ | 20·7–31·1 | 99·9 | 99·4–100·0 | 0·35§ | 0·29–0·41 | |
| WGOC | 49·7‡ | 45·1–54·3 | 99·6 | 99·1–99·9 | 0·59§ | 0·54–0·63 | 39·9‡ | 34·2–45·8 | 99·6 | 99·3–99·9 | 0·51 | 0·45–0·56 | |
| 15–17 years | IOTF | 92·4 | 82·5–97·2 | 92·4 | 89·0–94·8 | 0·75 | 0·66–0·83 | 76·7 | 61·0–87·7 | 100·0 | 98·6–100·0 | 0·85 | 0·77–0·94 |
| WHO | 71·2‡ | 58·6–81·4 | 99·4 | 97·8–99·9 | 0·79 | 0·70–0·88 | 32·6‡ | 19·5–48·7 | 100·0 | 98·6–100·0 | 0·46§ | 0·30–0·62 | |
| WGOC | 77·3‡ | 65·0–86·3 | 96·6 | 94·0–98·2 | 0·75 | 0·66–0·84 | 67·4 | 51·3–80·5 | 100·0 | 98·6–100·0 | 0·79 | 0·68–0·89 | |
IOTF, International Obesity Task Force; WGOC, Working Group of Obesity in China.
Calculated using body fat percentage cut-off points of 25 % in all boys, 30 % in girls aged 7–14 years and 35 % in girls aged 15–17 years.
Significant κ statistic, all κ yielded P < 0·001.
Statistically significant difference in sensitivity from IOTF in the same subgroup, P < 0·01 (McNemar’s test).
Statistically significant difference in κ value from IOTF in the same subgroup, P < 0·001 (z test).
High adiposity classified by BMI cut-offs
The population who had high adiposity were classified by BMI cut-offs, which is shown in Table 3. In general, individuals with high adiposity accurately classified as obese (1248, 73·1 % for boys; 400, 67·0 % for girls) by the IOTF cut-offs were more than by other two sets of BMI cut-offs. The WGOC and WHO cut-offs yielded higher misclassification rates, in the worst-case scenario, by categorising 11·2 % of girls with high adiposity as normal and 44·4 % of them as overweight, while the IOTF cut-offs categorised 2·3 % as normal and 30·7 % as overweight. However, all BMI cut-offs did not perform well among the pubertal group, especially for girls with adiposity as they had very low probabilities of being identified as obesity (25·5–47·9 %).
Table 3.
Different BMI categories among high adiposity
| BMI cut-offs | High adiposity* | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Boys (n 1707) | Girls (n 597) | |||||||||||||||
| All | 7–10 years | 11–14 years | 15–17 years | All | 7–10 years | 11–14 years | 15–17 years | |||||||||
| n | % | n | % | n | % | n | % | n | % | n | % | n | % | n | % | |
| IOTF | ||||||||||||||||
| Normal | 11 | 0·6 | 10 | 0·9 | 1 | 0·2 | 0 | 0·0 | 14 | 2·3 | 2 | 0·7 | 12 | 4·2 | 0 | 0·0 |
| Overweight | 448 | 26·2 | 286 | 24·5 | 157 | 33·1 | 5 | 7·6 | 183 | 30·7 | 36 | 13·4 | 137 | 47·9 | 10 | 23·3 |
| Obese | 1248 | 73·1 | 870 | 76·4 | 317 | 66·7 | 61 | 92·4 | 400 | 67·0 | 230 | 85·8 | 137 | 47·9 | 33 | 76·7 |
| WHO | ||||||||||||||||
| Normal | 15 | 0·9 | 12 | 1·1 | 3 | 0·6 | 0 | 0·0 | 50 | 8·4 | 3 | 1·1 | 47 | 16·4 | 0 | 0·0 |
| Overweight | 511 | 29·9 | 285 | 24·4 | 207 | 43·6 | 19 | 28·8 | 265 | 44·4 | 70 | 26·1 | 166 | 58·0 | 29 | 67·4 |
| Obese | 1181 | 69·2 | 869 | 74·5 | 265 | 55·8 | 47 | 71·2 | 282 | 47·2 | 195 | 72·8 | 73 | 25·5 | 14 | 32·6 |
| WGOC | ||||||||||||||||
| Normal | 51 | 3·0 | 27 | 2·3 | 24 | 5·1 | 0 | 0·0 | 67 | 11·2 | 5 | 1·9 | 62 | 21·7 | 0 | 0·0 |
| Overweight | 572 | 33·5 | 342 | 29·3 | 215 | 45·3 | 15 | 22·7 | 162 | 27·1 | 38 | 14·2 | 110 | 38·5 | 14 | 32·6 |
| Obese | 1084 | 63·5 | 797 | 68·4 | 236 | 49·7 | 51 | 77·3 | 368 | 61·6 | 225 | 84·0 | 114 | 39·9 | 29 | 67·4 |
IOTF, International Obesity Task Force; WGOC, Working Group of Obesity in China.
Defined as body fat percentage cut-off points of 25 % in all boys, 30 % in girls aged 7–14 years and 35 % in girls aged 15–17 years, presented as n and percentages.
High adiposity in overweight children and adolescents
Figure 1 shows the prevalence of high adiposity within different BMI categories. Large numbers of BMI-defined overweight individuals were found with high adiposity; especially during puberty, the prevalence of high adiposity varies from 33·0 to 68·5 % in boys with overweight and 41·6 to 71·0 % in girls with overweight, remaining very high probabilities of being high adiposity in overweight participants. Participants who were overweight classified by the IOTF cut-offs had the lowest ratios of high adiposity (4·2–41·6 %) than by the other BMI cut-offs for each subgroup.
Fig. 1.
Prevalence of high adiposity within different BMI categories by age period: (a) 7–10 years, (b) 11–14 years and (c) 15–17 years. IOTF, International Obesity Task Force; WGOC, Working Group on Obesity in China. 
, normal; 
, overweight; 
, obese
Discussion
This study compared the performances of three sets of BMI cut-offs that are applied in Chinese children and adolescents in urban setting. The IOTF and WGOC cut-offs, recommended for Asian and Chinese populations, respectively, were included. Although many studies used different BMI-based cut-offs such as WHO, IOTF and WGOC to describe the prevalence of childhood obesity, a small number of articles have carried out comparative studies among three or four sets of BMI cut-offs and drawn different conclusions for racial/ethnic groups(24–29). However, few studies made in-depth work on the performances of various BMI cut-offs and examine whether they differ by sex and age groups. Moreover, no articles engaged in comparing IOTF Asian cut-offs with other BMI references to screen for obesity in Chinese child and adolescent populations. Actually, Asian children and adults tend to have a lower BMI for a given body fatness than other ethnic groups(30). Therefore, it is necessary to include the IOTF cut-offs for Asians as well.
Performance of BMI in screening obesity is influenced by age and gender
The large sample size enables the study of performance disparities between sex-specific and age-specific groups. The results show that the three different types of BMI-derived cut-offs exhibited different capacities in various sex- and age-specific groups. This calls into question the conclusions from previous studies that evaluated the diagnostic performance of BMI for childhood obesity, but did not classify subjects by age and gender in specific. Our results help explain the contradictory results in previous studies with some reporting higher sensitivity in boys than in girls and others reporting the opposite(31–34). Consistent with previous studies, our study revealed higher prevalence of obesity in boys than in girls, indicating that urban boys should be key target of obesity prevention(35). Our study also found an underestimation of the prevalence of high adiposity among pubertal girls using BMI cut-offs, another group to be concerned in obesity prevention.
Modification of BMI cut-offs for childhood obesity screening should be considered in urban China
The trade-off between sensitivity and specificity should be considered to evaluate the performance of screening standards when there are no significant differences in κ values between various BMI cut-offs. It is well known that interventions at an early age are essential to prevent childhood obesity. For this reason, low cut-offs yielding higher sensitivity are more conducive to primary prevention of child and adolescent obesity. Conversely, high cut-offs yielding lower sensitivity might reduce public awareness of the prevalence of obesity, missing opportunities to institute appropriate lifestyle interventions for mitigating health risks. The way to prioritise the sensitivity was also used in a similar study since the WHO and IOTF (not Asian cut-offs) severely underestimated the obesity rates among Asian children(25). In addition, parents in China often equate being overweight or obese with being strong and healthy, which would inevitably exacerbate the negligence of children’s obesity. Therefore, provided they have yielded adequate specificity, cut-offs with higher sensitivity are necessary to improve obesity screening. Based on this premise, the IOTF cut-offs rather than the officially recommended WGOC cut-offs are evidently more suitable for obesity screening in urban Chinese children and adolescents. Apart from different data sources, difference also lies between the IOTF and WGOC cut-offs at 18 years of age; the IOTF cut-offs for obesity are based on country-specific centile curves passing through BMI 27 kg/m2 at the age of 18 years, while those of the WGOC are based on curves passing through BMI 28 kg/m2. This may account for the better obesity screening performance of the IOTF cut-offs observed in our study.
Attach importance to girls whose fatness dynamically increase
BMI is now irreplaceable in most low- to middle-income countries, due to its convenience. Previous studies have attributed high specificity but moderate or low sensitivity to the BMI standards in children and adolescents(19). Not only are there disparities in sensitivity between different BMI cut-offs but also for the same BMI cut-offs, the sensitivity varies between different ages and sexes groups, which has not been previously reported. Our study suggests that none of the BMI-derived cut-offs investigated is adequate for obesity screening in pubertal groups, especially for girls. A possible explanation is that the increases in adiposity may be higher than that indicated by BMI during puberty(36–38). Evidently, fat accumulation resumes in both sexes but that it is twice as rapid in pubertal girls(39). We therefore utilised conservative BF% cut-offs in girls (cut-offs for moderate obesity for girls aged 7–17 years according to Chinese guidelines), yet surprisingly the prevalence of BF%-defined obesity is still 100–287 % higher than that determined by the various BMI-derived cut-offs. This suggests that the high prevalence of obesity in pubertal girls should not be solely attributed to physiological fat accumulation. Over half (52·1–74·5 %) of the individuals with high adiposity were misclassified into non-obese groups whatever the BMI cut-offs were used, arising poor agreement between BF% and BMI cut-offs in girls aged 11–14 years. Even using IOTF to classify the girls aged 11–14 years, the highest prevalence (41·6 %) of high adiposity was found in overweight when compared with other age subgroups (4·7–33·0 %). Therefore, more accurate indicators or techniques should be incorporated into obesity screening in girls during puberty.
Notably, the emergence of large numbers of high adiposity among children and adolescents arises with rapid economic growth in China, but only 39·9–84·0 % of them are identified as obese by the WGOC cut-offs. Those findings should not be overlooked, and stronger measures are necessary to control obesity in future, particularly in economically developed cities in China. In addition, previously reported sex disparities in which boys exhibited greater rates of high adiposity than girls are no longer evident during puberty(40). Therefore, attention should be paid to changes in body weight and fat accumulation during growth in both boys and girls. The public health agencies need to initiate innovative weight management and behavioural interventions, especially for girls who are overweight.
Limitations in our study
The current study has some limitations. First, we used BIA to determine BF%, rather than dual-energy X-ray absorptiometry, which is generally regarded as the gold standard modality for measuring BF%(41). However, previous studies have shown strong correlations between BIA and dual-energy X-ray absorptiometry, and BIA is more suitable than dual-energy X-ray absorptiometry for use in large population-based studies(42,43). Second, our study recommends to lower the cut-offs in order to screen for child and adolescent obesity, but it also gives rise to a possibility that those with normal weight status are labelled as overweight or obesity by this stricter standard, which might have negative psychosocial impact(44).
Conclusion
The IOTF cut-offs for Asian perform better than the cut-offs by WHO and WGOC and should be recommended for child and adolescent obesity screening in urban China. Individuals during puberty need an even more accurate indicator to screen for obesity. The public health agencies should implement appropriate weight management, behavioural and environmental interventions to reduce the obesity prevalence, especially for boys and pubertal girls.
Acknowledgements
Acknowledgements: The authors express gratitude to all the participants of the study and to the staff of Qibao Community Health Service Center of Minhang District. This work was supported by grants-in-aid from The National Natural Science Foundation of China and The Innovation Program of Shanghai Municipal Education Commission. Financial support: This work was supported by the National Natural Science Foundation of China (grant no. 71303156) and the Innovation Program of Shanghai Municipal Education Commission (grant no.14YS019). Conflict of interest: There are no conflicts of interest. Authorship: K.Q. and L.T. conceptualised and designed the study, drafted the initial manuscript, and reviewed and revised the manuscript. Z.L., F.Y., S.G. and X.R. collected data and carried out the initial analyses. Professor H.L. and S.L. reviewed and revised the manuscript. Dr J.Z. and Professor Z.Z. conceptualised and designed the study, coordinated and supervised data collection and critically reviewed the manuscript for important intellectual content. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work. Ethics of human subject participation: This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving research study participants were approved by the institutional ethics boards of Shanghai Jiao Tong University, School of Public Health. Written informed consent was obtained from all subjects/patients.
References
- 1. Grossman DC, Bibbins-Domingo K, Curry SJ et al. (2017) Screening for obesity in children and adolescents. JAMA 317, 2417. [DOI] [PubMed] [Google Scholar]
- 2. De Lorenzo A, Romano L, Di Renzo L et al. (2019) Triponderal mass index rather than body mass index: an indicator of high adiposity in Italian children and adolescents. Nutrition 60, 41–47. [DOI] [PubMed] [Google Scholar]
- 3. Cole TJ & Lobstein T (2012) Extended international (IOTF) body mass index cut-offs for thinness, overweight and obesity. Pediatr Obes 7, 284–294. [DOI] [PubMed] [Google Scholar]
- 4. de Onis M (2007) Development of a WHO growth reference for school-aged children and adolescents. World Health Organ 85, 660–667. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Group of China Obesity Force (2004) Body mass index reference norm for screening overweight and obesity in Chinese children and adolescents. Zhonghua Liu Xing Bing Xue Za Zhi 25, 97–102. [PubMed] [Google Scholar]
- 6. Hu L, Huang X, You C et al. (2017) Prevalence of overweight, obesity, abdominal obesity and obesity-related risk factors in southern China. PLoS One. Published online: 14 September 2017. doi: 10.1371/journal.pone.0183934. [DOI] [PMC free article] [PubMed]
- 7. Liu D, Hao YX, Zhao TZ et al. (2019) Childhood BMI and adult obesity in a Chinese sample: a 13-year follow-up study. Biomed Environ Sci 32, 162–168. [DOI] [PubMed] [Google Scholar]
- 8. Kail RV & Cavanaugh JC (2010) Human Development: A Life-Span View, 5th ed., 768 p. Boston: Cengage Learning. [Google Scholar]
- 9. Phillips DC (editor) (2014) Encyclopedia of Educational Theory and Philosophy, Adolescent Development, pp. 18–19. Thouand Oaks: Sage Publishing. [Google Scholar]
- 10. Kalra S, Mercuri M & Anand SS (2013) Measures of body fat in South Asian adults. Nutr Diabetes. Published online: 27 May 2013. doi: 10.1038/nutd.2013.10. [DOI] [PMC free article] [PubMed]
- 11. Luque V, Escribano J, Zaragoza-Jordana M et al. (2014) Bioimpedance in 7-year-old children: validation by dual X-ray absorptiometry – part 2: assessment of segmental composition. Ann Nutr Metab 64, 144–155. [DOI] [PubMed] [Google Scholar]
- 12. Bammann K, Huybrechts I, Vicente-Rodriguez G et al. (2013) Validation of anthropometry and foot-to-foot bioelectrical resistance against a three-component model to assess total body fat in children: the IDEFICS study. Int J Obes (Lond) 37, 520–526. [DOI] [PubMed] [Google Scholar]
- 13. Kabiri LS, Hernandez DC & Mitchell K (2015) Reliability, validity, and diagnostic value of a pediatric bioelectrical impedance analysis scale. Child Obes 11, 650–655. [DOI] [PubMed] [Google Scholar]
- 14. Meredith-Jones KA, Williams SM & Taylor RW (2015) Bioelectrical impedance as a measure of change in body composition in young children. Pediatr Obes 10, 252–259. [DOI] [PubMed] [Google Scholar]
- 15. Houtkooper LB, Lohman TG, Going SB et al. (1989) Validity of bioelectric impedance for body composition assessment in children. J Appl Physiol (1985) 66, 814–821. [DOI] [PubMed] [Google Scholar]
- 16. Chen W, Jiang H, Yang JX et al. (2017) Body composition analysis by using bioelectrical impedance in a young healthy Chinese population: methodological considerations. Food Nutr Bull 38, 172–181. [DOI] [PubMed] [Google Scholar]
- 17. Achamrah N, Colange G, Delay J et al. (2018) Comparison of body composition assessment by DXA and BIA according to the body mass index: a retrospective study on 3655 measures. PLoS One. Published online: 12 July 2018. doi: 10.1371/journal.pone.0200465. [DOI] [PMC free article] [PubMed]
- 18. Williams DP, Going SB, Lohman TG et al. (1992) Body fatness and risk for elevated blood pressure, total cholesterol, and serum lipoprotein ratios in children and adolescents. Am J Public Health 82, 358–363. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Javed A, Jumean M, Murad MH et al. (2015) Diagnostic performance of body mass index to identify obesity as defined by body adiposity in children and adolescents: a systematic review and meta-analysis. Pediatr Obes 10, 234–244. [DOI] [PubMed] [Google Scholar]
- 20. Cohen J (1960) A coefficient of agreement for nominal scales. Educ Psychol Meas xx, 37–46. [Google Scholar]
- 21. Landis JR & Koch GG (1977) The measurement of observer agreement for categorical data. Biometrics 33, 159–174. [PubMed] [Google Scholar]
- 22. Fleiss JL (1971) Measuring nominal scale agreement among many raters. Psychol Bull 76, 378–382. [Google Scholar]
- 23. Newcombe RG (2001) Simultaneous comparison of sensitivity and specificity of two tests in the paired design: a straightforward graphical approach. Stat Med 20, 907–915. [DOI] [PubMed] [Google Scholar]
- 24. Shan X, Xi B, Cheng H et al. (2010) Prevalence and behavioral risk factors of overweight and obesity among children aged 2–18 in Beijing, China. Int J Pediatr Obes 5, 383–389. [DOI] [PubMed] [Google Scholar]
- 25. Liu A, Byrne NM, Kagawa M et al. (2011) Ethnic differences in the relationship between body mass index and percentage body fat among Asian children from different backgrounds. Br J Nutr 106, 1390–1397. [DOI] [PubMed] [Google Scholar]
- 26. Jiang XX, Hardy LL, Baur LA et al. (2014) High prevalence of overweight and obesity among inner city Chinese children in Shanghai, 2011. Ann Hum Biol 41, 469–472. [DOI] [PubMed] [Google Scholar]
- 27. Mendoza PP, Valdes J & Ortiz-Hernandez L (2015) Accuracy of body mass index for age to diagnose obesity in Mexican schoolchildren. Nutr Hosp 31, 2668–2675. [DOI] [PubMed] [Google Scholar]
- 28. Gordon-Larsen P, Wang H & Popkin BM (2014) Overweight dynamics in Chinese children and adults. Obes Rev 15, Suppl. 1, 37–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Adom T, Kengne AP, De Villiers A et al. (2019) Diagnostic accuracy of body mass index in defining childhood obesity: analysis of cross-sectional data from Ghanaian children. Int J Environ Res Public Health. Published online: 19 December 2019. doi: 10.3390/ijerph17010036. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Deurenberg-Yap M, Schmidt G, van Staveren WA et al. (2000) The paradox of low body mass index and high body fat percentage among Chinese, Malays and Indians in Singapore. Int J Obes Relat Metab Disord 24, 1011–1017. [DOI] [PubMed] [Google Scholar]
- 31. Neovius MG, Linne YM, Barkeling BS et al. (2004) Sensitivity and specificity of classification systems for fatness in adolescents. Am J Clin Nutr 80, 597–603. [DOI] [PubMed] [Google Scholar]
- 32. Li S, Zhang M, Yang S et al. (2005) Age- and sex-specific body composition of Chinese children. Acta Paediatr 94, 1139–1142. [DOI] [PubMed] [Google Scholar]
- 33. Gaskin PS & Walker SP (2003) Obesity in a cohort of black Jamaican children as estimated by BMI and other indices of adiposity. Eur J Clin Nutr 57, 420–426. [DOI] [PubMed] [Google Scholar]
- 34. Wickramasinghe VP, Cleghorn GJ, Edmiston KA et al. (2005) Validity of BMI as a measure of obesity in Australian white Caucasian and Australian Sri Lankan children. Ann Hum Biol 32, 60–71. [DOI] [PubMed] [Google Scholar]
- 35. Song Y, Wang HJ, Dong B et al. (2016) 25-year trends in gender disparity for obesity and overweight by using WHO and IOTF definitions among Chinese school-aged children: a multiple cross-sectional study. BMJ Open. Published online: 22 September 2016. doi: 10.1136/bmjopen-2016-011904. [DOI] [PMC free article] [PubMed]
- 36. McCarthy HD, Ellis SM & Cole TJ (2003) Central overweight and obesity in British youth aged 11–16 years: cross sectional surveys of waist circumference. BMJ 326, 624. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Wells JC, Coward WA, Cole TJ et al. (2002) The contribution of fat and fat-free tissue to body mass index in contemporary children and the reference child. Int J Obes Relat Metab Disord 26, 1323–1328. [DOI] [PubMed] [Google Scholar]
- 38. Ruxton CH, Reilly JJ & Kirk TR (1999) Body composition of healthy 7-and 8-year-old children and a comparison with the ‘reference child’. Int J Obes Relat Metab Disord 23, 1276–1281. [DOI] [PubMed] [Google Scholar]
- 39. Taylor RW, Grant AM, Williams SM et al. (2010) Sex differences in regional body fat distribution from pre- to postpuberty. Obesity (Silver Spring) 18, 1410–1416. [DOI] [PubMed] [Google Scholar]
- 40. Zhang J, Li X, Hawley N et al. (2018) Trends in the prevalence of overweight and obesity among Chinese school-age children and adolescents from 2010 to 2015. Child Obes 14, 182–188. [DOI] [PubMed] [Google Scholar]
- 41. Hung S, Chen C, Guo F et al. (2017) Combine body mass index and body fat percentage measures to improve the accuracy of obesity screening in young adults. Obes Res Clin Pract 11, 11–18. [DOI] [PubMed] [Google Scholar]
- 42. Pietrobelli A, Andreoli A, Cervelli V et al. (2003) Predicting fat-free mass in children using bioimpedance analysis. Acta Diabetol 40, Suppl. 1, S212–S215. [DOI] [PubMed] [Google Scholar]
- 43. Tyrrell VJ, Richards G, Hofman P et al. (2001) Foot-to-foot bioelectrical impedance analysis: a valuable tool for the measurement of body composition in children. Int J Obes Relat Metab Disord 25, 273–278. [DOI] [PubMed] [Google Scholar]
- 44. Sagar R & Gupta T (2018) Psychological aspects of obesity in children and adolescents. Indian J Pediatr 85, 554–559. [DOI] [PubMed] [Google Scholar]