Cynthia L. Ogden

Characterizing extreme values of body mass index–for-age by using the 2000 Centers for Disease Control and Prevention growth charts

BACKGROUNDnThe 2000 Centers for Disease Control and Prevention (CDC) growth charts included lambda-mu-sigma (LMS) parameters intended to calculate smoothed percentiles from only the 3rd to the 97th percentile.nnnOBJECTIVEnThe objective was to evaluate different approaches to describing more extreme values of body mass index (BMI)-for-age by using simple functions of the CDC growth charts.nnnDESIGNnEmpirical data for the 99th and the 1st percentiles of BMI-for-age were calculated from the data set used to construct the growth charts and were compared with estimates extrapolated from the CDC-supplied LMS parameters and to various functions of other smoothed percentiles. A set of reestimated LMS parameters that incorporated a smoothed 99th percentile were also evaluated.nnnRESULTSnExtreme percentiles extrapolated from the CDC-supplied LMS parameters did not match well to the empirical data for the 99th percentile. A better fit to the empirical data was obtained by using 120% of the smoothed 95th percentile. The empirical first percentile was reasonably well approximated by extrapolations from the LMS values. The reestimated LMS parameters had several drawbacks and no clear advantages.nnnCONCLUSIONSnSeveral approximations can be used to describe extreme high values of BMI-for-age with the use of the CDC growth charts. Extrapolation from the CDC-supplied LMS parameters does not provide a good fit to the empirical 99th percentile values. Simple approximations to high values as percentages of the existing smoothed percentiles have some practical advantages over imputation of very high percentiles. The expression of high BMI values as a percentage of the 95th percentile can provide a flexible approach to describing and tracking heavier children.

Associations of Relative Handgrip Strength and Cardiovascular Disease Biomarkers in U.S. Adults, 2011-2012

INTRODUCTIONnAlthough decline in muscle mass and quality and resulting declines in muscle strength are associated with aging, more research is needed in general populations to assess the utility of handgrip strength as an indicator of muscle strength and cardiovascular disease risk.nnnMETHODSnData from 4,221 participants aged ≥20 years in the 2011-2012 cycle of National Health and Nutrition Examination Survey were analyzed during 2014-2015. Standing isometric relative handgrip strength (calculated as maximal absolute handgrip strength from both hands divided by BMI) was used to predict cardiovascular biomarkers, including blood pressure (measured systolic and diastolic blood pressure); serum lipids (total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, and triglycerides); and plasma insulin and glucose.nnnRESULTSnResults from regression analyses showed that higher relative grip strength was significantly associated with lower systolic blood pressure, triglycerides, and plasma insulin and glucose, and higher high-density lipoprotein cholesterol in male and female participants (p<0.05 for all). Secondary descriptive analyses found that absolute handgrip strength increased significantly with increasing weight status, but relative handgrip strength decreased significantly with increasing weight status.nnnCONCLUSIONSnResults suggest that increased relative handgrip strength may be associated with a better profile of cardiovascular health biomarkers among U.S. adults. Relative grip strength, which both adjusts for the confounding of mass and assesses concomitant health risks of increased body size and low muscle strength, may be a useful public health measure of muscle strength.

Validity of the WHO cutoffs for biologically implausible values of weight, height, and BMI in children and adolescents in NHANES from 1999 through 2012

BACKGROUNDnThe WHO cutoffs to classify biologically implausible values (BIVs) for weight, height, and weight-for-height in children and adolescents are widely used in data cleaning.nnnOBJECTIVESnWe assess 1) the prevalence of these BIVs, 2) whether they were consistent with information on waist circumference, arm circumference, and leg lengths, and 3) the effect of their exclusion on the estimated prevalence of obesity in 2- to 19-y-olds in the NHANES, which is a study in which extreme values were verified when recorded.nnnDESIGNnWe conducted cross-sectional analyses in 26,480 children and adolescents in the NHANES from 1999-2000 through 2011-2012.nnnRESULTSnThe overall prevalence for a BIV for any body-size measure was 0.9% (n = 277), and almost all BIVs were due to extremely high, rather than low, values. Of 186 subjects who had a high BIV for weight or body mass index (BMI), all but one subject had both arm and waist circumferences that were greater than the sex- and age-specific 95th percentiles; 75% of subjects had circumferences greater than the 99th percentile. Of 63 subjects with a high height BIV, 75% of them had a leg length that was greater than the 95th percentile. The exclusion of children and adolescents with a BIV reduced the overall prevalence of obesity by ∼0.5 percentage points and by 1.7% in non-Hispanic blacks.nnnCONCLUSIONSnMost of the extremely high values of weight, height, and BMI flagged as BIVs in the NHANES are very likely correct. The increase of z score cutoffs or the use of an alternative method to detect possible errors could improve the balance between removing incorrect values and retaining extremely high, but accurate, values in other data sets.

The Epidemiology of Childhood Obesity in Canada, Mexico and the United States

Published reports based on different definitions indicate that in Canada, Mexico and the United States childhood overweight and obesity have increased dramatically since 1980, with the US leading the way. The prevalence of overweight, using the International Obesity Task Force (IOTF) definitions (Cole et al. 2000) in 7–13 year old girls doubled in Canada between 1981 and 1996 and tripled in boys (Tremblay et al. 2002). In 2004 in Canada, 26% of children and adolescents aged 2–17 were overweight or obese and 8% were obese (Shields 2006). Among children under 5 years of age in Mexico, overweight prevalence (z-score of weight-for-height above +2 of World Health Organization/National Center for Health Statistics/Centers for Disease Control and Prevention (WHO/NCHS/CDC) references (Dibley et al. 1987)) increased from 4.2 to 5.3% between 1988 and 1999 (Rivera et al. 2002). In the US, between 1980 and 2006 the prevalence of high body mass index (BMI ≥95th percentile of the sex specific 2000 CDC growth charts) increased from 6 to 16% among children and teens 2 through 19 years of age (Ogden et al. 2002, 2003, 2007, 2008).

Cardiovascular Disease Risk Factors Among Male Veterans, U.S., 2009–2012

INTRODUCTIONnCardiovascular disease remains an important cause of death in the U.S. where veterans of the U.S. Armed Forces represent a significant segment of the population. Limited national estimates of cardiovascular disease risk factors using physical measurements and reported veteran status in the U.S. civilian population have been reported. The purpose of this study was to compare the prevalence of cardiovascular disease risk factors among veteran and non-veteran men in the U.S. civilian population.nnnMETHODSnUsing data from the 2009-2012 National Health and Nutrition Examination Surveys, 1,107 veteran and 3,972 non-veteran men were identified for this study (analyzed in 2014-2015). Differences in hypertension, dsylipidemia, diabetes, obesity, and smoking between veterans and non-veterans were compared using chi-square and t-tests. Predicted prevalence from multivariable logistic regression models adjusted for age, race/Hispanic origin, and poverty level were used to assess whether previous military service was associated with having a cardiovascular disease risk factor.nnnRESULTSnVeteran men were older than non-veteran men (59.9 years vs 43.4 years) and were more likely to be non-Hispanic white (79.9% vs 65.7%). Adjusted predicted prevalence estimates show that veterans were more likely than non-veterans to be obese (42.6% vs 33.7%, p<0.01). After adjustment for obesity, there was no difference in hypertension, dyslipidemia, diagnosed diabetes, or smoking between veteran and non-veteran men.nnnCONCLUSIONSnThis study identified a segment of the U.S. civilian population-veteran men-who have a higher prevalence for obesity, a risk factor associated with increased risk for other cardiovascular disease risk factors.

Seafood consumption and blood mercury concentrations in adults aged ≥20 y, 2007–2010

BACKGROUNDnSeafood is part of a healthy diet, but seafood can also contain methyl mercury-a neurotoxin.nnnOBJECTIVEnThe objective was to describe seafood consumption in US adults and to explore the relation between seafood consumption and blood mercury.nnnDESIGNnSeafood consumption, obtained from a food-frequency questionnaire, and blood mercury data were available for 10,673 adults who participated in the 2007-2010 NHANES-a cross-sectional nationally representative sample of the US population. Seafood consumption was categorized by type (fish or shellfish) and by frequency of consumption (0, 1-2, 3-4, or ≥5 times/mo). Linear trends in geometric mean blood mercury concentrations by frequency of seafood consumption were tested. Logistic regression analyses examined the odds of blood mercury concentrations ≥5.8 μg/L (as identified by the National Research Council) based on frequency of the specific type of seafood consumed (included in the model as continuous variables) adjusted for sex, age, and race/Hispanic origin.nnnRESULTSnIn 2007-2010, 83.0% ± 0.7% (±SE) of adults consumed seafood in the preceding month. In adults consuming seafood, the blood mercury concentration increased as the frequency of seafood consumption increased (P < 0.001). In 2007-2010, 4.6% ± 0.39% of adults had blood mercury concentrations ≥5.8 μg/L. Results of the logistic regression on blood mercury concentrations ≥5.8 μg/L showed no association with shrimp (P = 0.21) or crab (P = 0.48) consumption and a highly significant positive association with consumption of high-mercury fish (adjusted OR per unit monthly consumption: 4.58; 95% CI: 2.44, 8.62; P < 0.001), tuna (adjusted OR: 1.14; 95% CI: 1.10, 1.17; P < 0.001), salmon (adjusted OR: 1.14; 95% CI: 1.09, 1.20; P < 0.001), and other seafood (adjusted OR: 1.12; 95% CI: 1.08, 1.15; P < 0.001).nnnCONCLUSIONnMost US adults consume seafood, and the blood mercury concentration is associated with the consumption of tuna, salmon, high-mercury fish, and other seafood.

More Than Half of US Youth Consume Seafood and Most Have Blood Mercury Concentrations below the EPA Reference Level, 2009–2012

BACKGROUNDnConsuming seafood has health benefits, but seafood can also contain methylmercury, a neurotoxicant. Exposure to methylmercury affects children at different stages of brain development, including during adolescence.nnnOBJECTIVEnThe objective was to examine seafood consumption and blood mercury concentrations in US youth.nnnMETHODSnIn the 2009-2012 NHANES, a cross-sectional nationally representative sample of the US population, seafood consumption in the past 30 d and blood mercury concentrations on the day of examination were collected from 5656 youth aged 1-19 y. Log-linear regression was used to examine the association between frequency of specific seafood consumption and blood mercury concentration, adjusting for race/Hispanic origin, sex, and age.nnnRESULTSnIn 2009-2012, 62.4% ± 1.4% (percent ± SE) of youth consumed any seafood in the preceding month; 38.4% ± 1.4% and 48.5% ± 1.5% reported consuming shellfish and fish, respectively. In 2009-2012, the geometric mean blood mercury concentration was 0.50 ± 0.02 μg/L among seafood consumers and 0.27 ± 0.01 μg/L among those who did not consume seafood. Less than 0.5% of youth had blood mercury concentrations ≥5.8 μg/L. In adjusted log-linear regression analysis, no significant associations were observed between frequency of breaded fish or catfish consumption and blood mercury concentrations, but frequency of consuming certain seafood types had significant positive association with blood mercury concentrations: high-mercury fish (swordfish and shark) [exponentiated β coefficient (expβ): 2.40; 95% CI: 1.23, 4.68]; salmon (expβ: 1.41; 95% CI: 1.26, 1.55); tuna (expβ: 1.38; 95% CI: 1.29, 1.45); crabs (expβ: 1.35; 95% CI: 1.17, 1.55); shrimp (expβ: 1.12; 95% CI: 1.05, 1.20), and all other seafood (expβ: 1.23; 95% CI: 1.17, 1.32). Age-stratified log-linear regression analyses produced similar results.nnnCONCLUSIONnFew US youth have blood mercury concentrations ≥5.8 μg/L, although more than half of US youth consumed seafood in the past month.

Effects of trimming weight-for-height data on growth-chart percentiles

BACKGROUNDnBefore estimating smoothed percentiles of weight-for-height and BMI-for-age to construct the WHO growth charts, WHO excluded observations that were considered to represent unhealthy weights for height.nnnOBJECTIVEnThe objective was to estimate the effects of similar data trimming on empirical percentiles from the CDC growth-chart data set relative to the smoothed WHO percentiles for ages 24-59 mo.nnnDESIGNnWe used the nationally representative US weight and height data from 1971 to 1994, which was the source data for the 2000 CDC growth charts. Trimming cutoffs were calculated on the basis of weight-for-height for 9722 children aged 24-71 mo. Empirical percentiles for 7315 children aged 24-59 mo were compared with the corresponding smoothed WHO percentiles.nnnRESULTSnBefore trimming, the mean empirical percentiles for weight-for-height in the CDC data set were higher than the corresponding smoothed WHO percentiles. After trimming, the mean empirical 95th and 97th percentiles of weight-for-height were lower than the WHO percentiles, and the proportion of children in the CDC data set above the WHO 95th percentile decreased from 7% to 5%. The findings were similar for BMI-for-age. However, for weight-for-age, which had not been trimmed by the WHO, the empirical percentiles before trimming agreed closely with the upper percentiles from the WHO charts.nnnCONCLUSIONnWHO data-trimming procedures may account for some of the differences between the WHO growth charts and the 2000 CDC growth charts.

Secular trends for skinfolds differ from those for BMI and waist circumference among adults examined in NHANES from 1988–1994 through 2009–2010

BACKGROUNDnAlthough the prevalence of a body mass index [BMI (in kg/m2)] ≥30 has tripled among US adults since the 1960s, BMI is only moderately correlated with body fatness. Because skinfolds can more accurately estimate body fatness than can BMI, it is possible that skinfolds could be useful in monitoring secular trends in body fatness.nnnOBJECTIVEnWe examined whether there were similar secular trends for skinfolds (triceps and subscapular), BMI, and waist circumference between US adults.nnnDESIGNnThis study was an analysis of 45,754 adults who participated in the NHANES from 1988-1994 through 2009-2010. Approximately 19% of the subjects were missing ≥1 skinfold-thickness measurement. These missing values were imputed from other characteristics.nnnRESULTSnTrends in mean levels and in the prevalence of high levels of the 4 body size measures were fairly similar between men, with mean levels increasing by ≥5% from 1988-1994 through 2009-2010. Slightly larger increases were seen in women for BMI and waist circumference (7-8%), but trends in skinfolds were markedly different. The mean triceps skinfold, for example, increased by 2 mm through 2003-2004, but subsequently decreased so that the mean in 2009-2010 did not differ from that in 1988-1994. Compared with obese women in 1988-1994, the mean BMI of obese women in 2009-2010 was 1 higher, but mean levels of both skinfolds were 5-10% lower.nnnCONCLUSIONSnAlthough there were fairly similar trends in levels of BMI, waist circumference, and skinfold thicknesses in men in the United States from 1988-1994 through 2009-2010, there were substantial differences in women. Our results indicate that it is unlikely that skinfold thicknesses could be used to monitor trends in obesity.

Secular trends in pediatric BMI

Studies that use cross-sectional data have shown an increase in the prevalence of childhood obesity, as assessed by BMI (weight in kilograms divided by height in meters squared), over the past 30 y in the United States (1). Between the 1960s and 2000 the distribution of BMI among children became more skewed, whereas between 2000 and 2010 there was relatively little change among either boys or girls (1). Further elucidation of the trend in obesity can be understood on the basis of both longitudinal and cross-sectional analyses of longer time periods such as Johnson et al (2) have published in the current issue of the Journal with the use of data from the Fels Longitudinal Growth Study. The Fels study began enrolling pregnant women who lived in the Yellow Springs, Ohio, region in 1929 (3). Longitudinal data on weights and heights, among other measures, have been collected on participants. The authors explored differences in BMI growth velocity between 3 birth cohorts (1929–1953, 1954– 1972, and 1973–1999) in the Fels study, comparing various parameters of growth curves. The differences observed between the birth cohorts show increased BMI growth velocity among children born during 1973–1999 compared with the earlier cohorts. This change is generally consistent with the pattern of increases in obesity prevalence seen in national cross-sectional data from the 1980s and 1990s. As the authors state, their analyses of the European-American children in the Fels study may not be applicable to other populations. The Fels cohorts of white children from the Ohio area do not represent the race-ethnic and geographic diversity in the US pediatric population. Because there are substantial race-ethnic differences in obesity prevalence and in body fat composition (4), changes in some aspects of the BMI growth patterns in the white Fels population may not reflect the trends in the US population. Much of the discussion in the article by Johnson et al (2) focuses on the age of adiposity rebound, or the BMI nadir, which has been shown to be related to BMI later in life (5). On the basis of multilevel models that accounted for the repeated measurements in the Fels study, the BMI curves show, and the Abstract states, that the BMIs of children under 5 y (2–5 y) born during 1973–1999 (the most recent period) were lower than those of children in the earlier cohorts. Differences across cohorts in the estimated age of the adiposity rebound, however, were significant, however, only among girls (Tables 3 and 4 in their article). National cross-sectional data do not show similar results: 2to 5-y-olds examined in recent national studies do not have lower BMIs than do those who were examined decades ago. For example, among 3-y-olds in the United States, mean BMIs (in kg/m) were 15.7 in 1971–1974 (n 1⁄4 292) and 16.0 in 1999–2002 (n 1⁄4 173) among girls and 16.0 in 1971–1974 (n 1⁄4 308) and 16.2 in 1999–2002 (n 1⁄4 209) among boys (6). In addition, the mean BMIs of 50,000 3-y-olds in England were found to increase by ;0.6 between 1988 and 2003 (7). These time periods represent the period of increase in obesity prevalence seen in both the United States and England. The difference between the United States, England, and the Fels results may be due to the fact that the data for the United States and England (6, 7) are crosssectional, whereas the BMIs of 2to 5-y-olds in the Fels data were estimated from longitudinal data. However, in the analysis of the Fels data, 29 of 855 subjects were excluded, not because of invalid data but because the fitted, multilevel models yielded implausible results, such as negative ages at the BMI nadir. This raises the possibility that the models may be misspecified. Another explanation for the differences may be that the Fels data are not nationally representative, which highlights the need for additional analyses of data from various ethnic groups and geographic regions in the United States. The authors of the current study do note that there are questions concerning the significance of the adiposity rebound. For example, a previous article from the Fels study (8) found that, among women, age at BMI rebound was associated with adult BMI but not with total or percentage body fat; among men, the rebound age was not associated with adult BMI or body fatness. It has been suggested that an early rebound simply identifies those children who already have a high BMI or have a BMI that is increasing (9). An evaluation of the changes in BMI growth velocity between the Fels birth cohorts might also have considered infant feeding. Infants born between 1929 and 1974 were primarily formula fed (10). Infant feeding practices affect weight gain, and breastfeeding appears to have a small protective effect against obesity (11). The distribution of breastfeeding and its influence on BMI