Percentile Curves for Anthropometric Measures for Canadian Children and Youth

Body mass index (BMI) is commonly used to assess a child's weight status but it does not provide information about the distribution of body fat. Since the disease risks associated with obesity are related to the amount and distribution of body fat, measures that assess visceral or subcutaneous fat, such as waist circumference (WC), waist-to-height ratio (WHtR), or skinfolds thickness may be more suitable. The objective of this study was to develop percentile curves for BMI, WC, WHtR, and sum of 5 skinfolds (SF5) in a representative sample of Canadian children and youth. The analysis used data from 4115 children and adolescents between 6 and 19 years of age that participated in the Canadian Health Measures Survey Cycles 1 (2007/2009) and 2 (2009/2011). BMI, WC, WHtR, and SF5 were measured using standardized procedures. Age- and sex-specific centiles were calculated using the LMS method and the percentiles that intersect the adult cutpoints for BMI, WC, and WHtR at age 18 years were determined. Percentile curves for all measures showed an upward shift compared to curves from the pre-obesity epidemic era. The adult cutoffs for overweight and obesity corresponded to the 72nd and 91st percentile, respectively, for both sexes. The current study has presented for the first time percentile curves for BMI, WC, WHtR, and SF5 in a representative sample of Canadian children and youth. The percentile curves presented are meant to be descriptive rather than prescriptive as associations with cardiovascular disease markers or outcomes were not assessed.


Introduction
Childhood obesity is associated with adverse health, psychosocial, and economic outcomes in childhood and adulthood [1]. Body mass index (BMI) is the most commonly used method to assess a child's weight status. The drawbacks of BMI are that it cannot differentiate between lean and fat mass and does not provide information about the distribution of body fat [2][3][4]. Since the cardiovascular disease (CVD) risks associated with obesity are related to the amount and distribution of body fat [5][6][7][8], measures that assess visceral or subcutaneous fat may provide a better risk assessment than the BMI. Waist circumference (WC) and waist-to-height ratio (WHtR) have both been shown to be associated with CVD risk in children and adults [6,[9][10][11]; skinfolds thickness (SF) can potentially better identify children and adults with excess total body fatness or adverse CVD risk factor levels than BMI [12][13][14]. As body composition in childhood is dependent on age, sex, and ethnicity, no single universal cut-off point exists for either of these measures in children and youth. Percentile curves have been developed for WC, WHtR, and SF in different populations [15][16][17][18] but, with the exception of waist circumference in youth aged 11 to 18 years [19], there are no Canadian reference data available. Therefore, the objective of the present paper was to develop percentile curves for anthropometric measures in a representative sample of Canadian children and youth using the LMS method [20].

Materials and Methods
The current study used data from the Canadian Health Measures Survey (CHMS) Cycles 1 and 2. The CMHS is a representative, cross-sectional survey assessing indicators of health and wellness in Canadians between 3 and 79 years [21,22]. The survey consisted of a household interview to obtain sociodemographic and health information, and a visit to mobile examination centre to perform a number of physical measurements and tests. The sampling frame of the Canadian Labour Force Survey was used to identify the collection sites for the mobile examination centres. Within each collection site, households were selected using the 2006 Census as the sampling frame. Interviews and examinations for the CHMS Cycle 1 and 2 were performed between 2007 and 2009, and 2009 and 2011, respectively. The overall response rate in the two cycles was 51.7% and 55.7%, respectively. Data from the two cycles was combined as per Statistics Canada guidelines [23]. A total of 11,999 persons participated in physical examination part of the survey. The present analysis uses data from 4115 children and adolescents (2089 males and 2026 females) between 6 and 19 years of age.

Anthropometric measures
All anthropometric measurements were performed by trained health professionals at the mobile examination centres. Body mass index was calculated from measured weight and height using the formula weight/height 2 [kg/m 2 ]. Weight was measured using a calibrated digital scale (Mettler Toledo, Mississauga, ON, Canada) to the nearest 0.1 kg. Standing height was measured using a fixed stadiometer with a vertical backboard and a moveable headboard to the nearest 0.01 cm. Waist circumference measurement was based on the Canadian Physical Activity, Fitness, and Lifestyle Approach (CPAFLA) protocol [24] using a 150 cm or a 200 cm Gulick tape measure. The WC was measured at the mid-point between the bottom of the rib cage and the top of the iliac crest at the end of a normal expiration to the nearest 0.1 cm. Waist to Height Ratio was calculated as waist circumference over height. Sum of 5 skinfolds (SF5) was determined using the five site method of the CPAFLA protocol [24] with a Harpenden skinfold caliper and a 150 cm Gulick tape measure to the nearest 0.2 mm. Each SF was measured twice. Triceps SF was measured on the midline of the back of the arm at the mid-point level between the acromium process and the tip of the olecranon process. Biceps SF was measured over the biceps at the same level as the midpoint for the triceps. Subscapular SF was measured below the inferior angle of the scapula at an angle of 45 degrees to the spine. Iliac crest SF was measured in the mid-axillary line above the crest of the ilium. Medial calf SF was measured at the medial side of the calf at the point of the largest circumference. Body mass index and WC were not measured in pregnant women, and SF measurements were not done on individuals with a BMI ! 30 kg/m 2 . Height was based on self-report for participants who were unable to stand unassisted.

Statistical analysis
The data were stratified by sex and summarized using the LMS method by Cole and Green [20]. This method assumes that the data are normalized after the Box Cox transformation The age-specific distribution expresses the mean, coefficient of variation, and skewness as parameters that change smoothly as a function of age by modeling them as cubic splines. These functions can be plotted as smooth curves over age and are referred to as the M (mean μ), S (variance σ), and L (skewness λ) curves. Centiles are computed by using the values of the three parameters curves for a corresponding age with the formula where z α is the upper α quantile for the truncated standard normal distribution.
The LMS method allows us to assess the likelihood of an individual observation with the formula The 3 rd , 10 th , 25 th , 50 th , 75 th , 90 th , and 97 th centile curves were computed for BMI, WC, SF5, and WHtR. To avoid unusual behaviours of the spline functions near the end of the age range, data from respondents up to age 30 years was used to fit the models. This modification produced smoother curves that more accurately reflect the population characteristics. In addition, the z-score and percentile that intersects the adult cutpoints for BMI (18.5, 25, and 30 kg/m 2 ) [25,26], WC (102 cm for males and 88 cm for females) [27], and WHtR (0.5 for both sexes) [28] at age 18 years were determined. Residual quantile plots ("worm plots") [29] were used to assess the goodness of fit of each component of the LMS models.
All calculations were performed using the sampling weights provided by Statistics Canada [23] to account for design effect and non-response bias. The CHMS uses a multistage sampling design with two sampling frames to select its sample. The probability of an individual to be selected for the survey is determined as the product of the probability of selection at each stage. To correct for non-response the weight of non-respondent households and individuals is redistributed to respondents within homogeneous response groups based on characteristics that are available for both respondents and non-respondents as determined from the Census of Canada (such as dwelling type or household income). A detailed description of the weighting procedure can be found elsewhere [22].
The statistical software package R [30] with the gamlss package [31] was used to perform the statistical analyses. the child provided his or her assent to participate [21,22]. The current project was approved by the IWK Health Centre Research Ethics Board, Halifax, NS, Canada (File # 1014413).

Results
Descriptive statistics for BMI, WC, SF5, and WHtR by age and sex are shown in Table 1. The prevalence of overweight and obesity in the sample based on the IOTF (International Obesity Task Force) growth reference [32] was 17.0 and 9.6%, respectively. Characteristics of the sample are shown in S1 Table. Body mass index and WC increased throughout childhood and percentile cutpoints were consistently higher in males compared to females, albeit the differences were small (Tables 2 and 3, Figs 1 and 2). The adult cutoffs for overweight and obesity approximately corresponded to the 72 nd and 91 st percentile, respectively, for both sexes. The 75 th percentile and lower for WHtR in girls showed a slight decline until 12 years, after which it increased, while the 90 th  and 97 th percentile increased throughout childhood and adolescence; a similar pattern was observed for boys (Table 4 and Fig 3). The adult WHtR cutoff of 0.5 corresponded to the 81 st and 78 th percentile in males and females, respectively. Sum of 5 skinfolds increased until puberty in girls and plateaued afterwards. Among boys, the 50 th percentile for SF5 increased slightly until puberty and remained fairly constant afterwards, while percentile levels above the median exhibited a sharp increase with a peak around 11 years of age and subsequent drop ( Table 5 and Fig 4). All three components of the models for females showed extremely good fit for every measure and a good fit for males with the exception of SF5, where the worm plots showed a moderate fit due to some kurtosis that the model could not account for. Details on the percentiles of BMI, WC, and WHtR by age and sex that intersect adult cutpoints at age 18 years can be found in Table 6 and S2 Table.

Discussion
The current study has presented for the first time percentile curves for the most commonly used anthropometric measures for body composition assessment based on a representative sample of Canadian children and youth aged 6 to 19 years. The percentile curves presented are meant to be descriptive rather than prescriptive as associations with cardiovascular disease markers or outcomes were not assessed. The data may be used by researchers as reference data for future studies.
The general shape of the BMI centile curves in males and females was comparable to existing growth curves from the IOTF [32], WHO (World Health Organization) [33], and CDC (Centers for Disease Control) [34] with the BMI increasing steadily until puberty, after which the slope begins to level off. Since the data for the CHMS were collected during the obesity epidemic, the percentile cutpoints are higher than those from the IOTF, WHO, and CDC curves, which were based on data that were for the most part collected before the 1980s. While the 72 nd and 91 st percentiles for both sexes in our sample intersected the adult cutpoints for overweight and obesity at age 18 years, the corresponding percentiles were 88 th (females) and 90 th (males) for overweight, and 99 th for obesity (both sexes) using the IOTF reference [32], and 82 nd (females) and 81 th (males) for overweight, and 95 th (females) and 96 th (males) for obesity using the CDC reference [34]. These differences underline the importance of relating percentile curves and cutoffs to actual health outcomes as the associated health risks of distribution-based cutoffs may change considerably with the reference population.
Waist circumference has been found to be a better predictor of CVD risk factors in children than BMI [6,10,35] but the method still has not seen widespread use, neither clinically or for research purposes. Percentiles for WC in children have been published for more than 20  countries but there is no universally accepted WC cutoff that is based on health outcomes or CVD risk markers in children. Establishment of WC cutoffs is further hampered by the use of different method to measure WC between studies. While the CHMS and other recent studies consistently measured the WC at the midpoint between the lower end of the rib cage and the iliac crest at the end of an expiration [16,18], studies in samples from the pre-obesity epidemic era often used the umbilicus or the point of maximum waist narrowing [17,19,36]. For this reason, comparison of WC percentiles determined in this study with existing Canadian reference data [19] from the 1981 Canadian Fitness Survey (which measured WC at "the point of noticeable waist narrowing") was not possible. The shape of the WC curves in boys and girls in the present study is similar to previously published centile curves [16], including the Canadian reference curves [19]. The intersection of the WC curves at age 18 years with recommended cutoff points for adults suggested the 94 th percentile (males) and 86 th percentile (females), respectively, as WC cutoffs in childhood. The only other study that related their centile curves to adult cutoffs was a German study using data collected between 2003 and 2006; the investigators found that the 98 th and 97 th percentile corresponded to the adult cutoffs in males and females, respectively [16]. A WHtR greater than 0.5 has been proposed as an indicator of CVD risk in adults [28] and two recent meta-analyses concluded that WHtR provides a better discrimination for CVD outcomes in adults than WC or BMI [9,11]. The same cutoff has been suggested for use throughout childhood [37,38], which would offer the great advantage of obviating the need for agerelated reference values and providing the simple universal public health message "keep your waist circumference to less than half your height" [38]. Centiles for WHtR in the present study, with the exception of the higher percentiles, decreased slightly until around puberty and increased again thereafter as reported previously in samples from other countries [18,39]. The median WHtR varied by less than 3% throughout childhood for both sexes. The percentiles that corresponded to the 0.5 cutoff at age 18 years were comparable between males and females (81 st and 78 th percentile, respectively). These cutoffs were lower than those found in representative samples of Norwegian [18] and German children and youth [39], where the cutoffs fell above the 90 th percentile.
Compared to the other methods of body fat measurement discussed above, skinfold thickness has the highest measurement error [40], which may further increase with the degree of adiposity [41]. This shortcoming makes interpretation of skinfolds thickness challenging and, along with the need for specialized measurement equipment (calipers), may account for its limited use. In the largest study to date, Addo and Himes determined reference curves for triceps and subscapular skinfold thickness based on five large representative samples of US children between 1963 and 1994 [15]. They found a distinct difference in skinfold trajectories between sexes, with boys exhibiting a dramatic increase in the percentile levels above the median before puberty while girls showed a more steady increase throughout childhood and adolescence. Sex differences were also evident for SF5 in the current study: Percentile levels above the 75 th percentile showed a steep increase before puberty for boys followed by a sharp decline as result of an increase and decline in the μ parameter that is amplified by a large variance and large negative value of the skewness curve in this age range. The percentiles for girls increased slightly until age 12 years and then remained fairly constant. Almost identical sex-specific patterns for  triceps and subscapular skinfold thickness were found in a large representative sample of German children (n = 17,158) [42].
The strengths of the current study include the nationally representative sample of children and youth, and the use of sample weighting to account for non-response and design effect. Due to the physical burden of the assessments used in the survey, and the need to travel to the mobile examination clinics, there may have been a self-selection toward more mobile, healthier and fitter individuals. Our study is limited by the relatively small sample size, and the cross-sectional nature of the data; longitudinal data may more accurately reflect how body fatness changes with age. Lastly, while the LMS method provides a very flexible and powerful tool to model data whose dispersion and skewness change over age, it lacks a mechanism for modeling kurtosis, an issue addressed by extensions of the LMS method [43,44]. Its flexibility also means that the curves may differ considerably based on the parameter choices made by the researcher; e.g., the choice of a higher smoothing parameter allows the elimination of features that are not representative of the underlying population. While choice of smoothing parameters and model selection adds some arbitrariness to the process it allows the researcher to balance mathematical rigidity with clinical usefulness.

Conclusions
This study has presented percentile curves for measures of body fatness in a representative sample of Canadian children and youth. Our findings indicate a substantial upward shift of the percentile curves for all measures compared to data from the pre-obesity epidemic era. Since we did not examine any relationships with health outcomes or disease markers, the data should be considered as a reference for future studies not as a growth standard.
Supporting Information S1