Obesity and blood pressurefresults from the examination of 2365 schoolchildren in Germany

Transcription

1 (2003) 27, & 2003 Nature Publishing Group All rights reserved /03 $ PAPER Obesity and blood pressurefresults from the examination of 2365 schoolchildren in Germany A Reich 1,GMüller 1, G Gelbrich 2, K Deutscher 1,RGödicke 3 and W Kiess 1 * 1 University Hospital for Children and Adolescents Leipzig, Leipzig, Germany; 2 Coordination Center for Clinical Trials Leipzig, Leipzig, Germany; and 3 Public Health Department of the City of Leipzig, Leipzig, Germany OBJECTIVES: To investigate the relationship between different indices of body fat and blood pressure in children and adolescents. DESIGN: Cross-sectional cohort study along with regular public health service examinations in school classes two, five and nine. PARTICIPANTS: A total of 2365 healthy schoolchildren aged 8 16 y. MEASUREMENTS: Body mass index (BMI), skinfolds, waist hip ratio, body fat determined by bioelectric impedance analysis and blood pressure. RESULTS: In comparison with recently published normative data, a significant increase of obesity was found. There is a higher prevalence of obesity in children with lower education. A positive association between body fat and hypertension was observed in children aged above y. The BMI had the strongest association with blood pressure among the indices of body fat considered here. CONCLUSION: Obesity is an increasing problem even among schoolchildren. This observation should be treated seriously as the relationship of body fat to cardiovascular risk is detectable already at a young age. Attention should be paid to the dependence of obesity on the level of education. The analysis suggests that BMI should be the preferred index to assess body fat. (2003) 27, doi:.38/sj.ijo Keywords: body fat; body mass index; hypertension; children; adolescents Introduction The prevalence of obesity in childhood and adolescence is increasing rapidly worldwide. 1 6 Obesity at young age predicts the likelihood of obesity in adulthood. 7 Obesity in childhood is an important risk factor for several disorders 8 such as the metabolic syndrome, 9 hypertension,,11 and cardiovascular disease, as well as for morbidity and mortality in the later life.,16 One of the aims of our study was to assess the prevalence of obesity in schoolchildren compared to recently published regional data, 6 and its relationship to the level of education. A recent work 17 has examined the (in)consistency of different indices of body fat in children, concluding with the question about their ability to predict morbidity. The present study is giving a partial answer, providing estimates for the influence of body fat, assessed by body mass index (BMI), skinfolds, waist hip ratio (WHR), and bioelectric *Correspondence: Dr W Kiess, University Hospital for Children and Adolescents Leipzig, Oststrae 21-25, Leipzig, Germany. kiw@medizin.uni-leipzig.de Received 25 October 2002; accepted 27 March 2003 impedance analysis (BIA), on the risk of hypertension. Furthermore, these indices of body fat are compared with respect to their ability to distinguish between hypertensive and normotensive subjects. Methods Participants and measurements Participating schools were chosen by random selection according to the geographical location in all city areas of Leipzig. A total of 2365 schoolchildren in the age groups 8/9, 11/12, and /16 y were recruited along with regular examinations taking place in the second, fifth and ninth classes. Written consent was obtained from both children and parents. The following measurements were obtained: body height (mobile digital stadiometer System Dr Keller III, Günter GmbH, Tauscha, Germany; precision 72 mm), weight (digital SECA s -scale, Vogel & Halke GmbH, Hamburg, Germany; precision 70 g), right-sided biceps, triceps, subscapular and suprailiac skinfolds (Harpenden skinfold calliper, 18 Holtain Ltd, Crosswell, Crymych, UK; resolution

2 mm), blood pressure (Riva-Rocci-Korotkoff method, 12 cm cuff, supine position after rest), circumferences of waist (smallest circumference between hip and chest) and hip (measured by a nonstretchable metric band), and bioelectric impedance (portable Holtain Body Composition Analyser, Holtain Ltd, Crosswell, Crymych, UK; 800 ma/ 50 khz; electrodes on right wrist and right ankle). Cutoff values and formulas Overweight and obesity were defined as BMI above the 90th and 97th percentiles, 3 respectively, using the reference percentiles of Hesse et al, 6 which had been based upon a cohort of children in the same geographical region of Germany in the early 1990s. Hypertension was defined as diastolic blood pressure above the 95th reference percentile, 19,20 using the reference values of Sörgel et al. 21 Body fat was calculated from BIA according to Wabitsch et al. 22 Statistical methods Comparisons of frequencies were made by w 2 methods. BMI percentiles are graphically represented by telescope plots (see Figure 1 for description) with underlying reference centile curves. In order to discuss whether the effects found for BMI are attributable to differences in height, the increase of BMI with body height was determined by linear regression, accounting for age and gender. For graphical representation of the association between BMI and blood pressure, participants were divided into five groups with respect to BMI percentiles computed from our sample, accounting for age and gender. Percentages of hypertension depending on BMI are depicted with Scale of measurement maximum 97th percentile 90th percentile 75th percentile median 25th percentile th percentile 3rd percentile minimum Figure 1 We propose the telescope plot as an extension of the box-andwhiskers plot. Besides the interquartile box, additional boxes are drawn to depict further intercentile intervals. We suggest the widths of the boxes to be proportional to the amount of data lying outside the respective interval (ie 50:20:6 in our case). Sequences of telescope plots are suitable to illustrate changes of distributions along with a covariate that is ordinal but not metric, or that has discontinuous range not allowing for continuous centile curves. confidence intervals. The ordinal trend was tested by Kendall s t b. Estimates for the risk of hypertension depending on BMI, stem fat (logarithm of the sum of the subscapular and suprailiac skinfolds), WHR and body fat obtained from BIA were calculated by logistic regression. Increased subcutaneous fat in the upper arm in obese might lead to a measurement bias towards higher blood pressure. In order to eliminate this effect from the risk estimates, arm fat (logarithm of the sum of the biceps and triceps skinfolds) was included as a covariate. Regression analysis was carried out separately in each of the six groups determined by age (8/9, 11/12, and /16 y) and gender. The regression coefficients were then tested for homogeneity, and aggregated to a common estimate when appropriate, using methods of meta-analysis. 23 To assess which of the fat indices showed the largest contrast between hypertensive and normotensive subjects, the four indices of body fat measured on different scales had to be made comparable. Standard deviation scores (SDS) were computed for the hypertensive subjects by SDS ¼ (index mean reference )/s.d. reference, where index was either of individual BMI, stem fat, WHR or BIA body fat, and mean reference and s.d. reference were computed from the normotensive subjects representing the reference population. The four SDS variables obtained thereby were then compared by repeated measurements analysis of variance, including age and gender as between-subject factors. All statistical computations were carried out by SPSS 9.0 (SPSS Inc., Chicago, USA). Results Prevalence of obesity We found an increased amount of overweight and obese children in comparison to recently published studies also carried out on East German schoolchildren 6 (Table 1). Overweight and obesity were observed more frequently in children with lower education. This contrast was stronger in the older subjects and was present in both genders. Despite this inhomogeneity, the trend towards higher BMI was significant in all groups distinguished by age, gender and type of school. The prevalence of obesity in our sample was two to five times higher than expected. Figure 2 shows that almost all percentiles are above the reference values, so the upward shift includes the entire distribution. Is the BMI rise due to secular trend? Our sample has been drawn about 6 y later than Hesse s data set. Owing to a secular trend phenomenon, the body height in our sample, and so body weight and BMI, might have possibly been increased. In fact, the body heights almost coincide with those reported from the reference data. 24 On average, our boys are 0.7 cm taller (P ¼ 0.053), and our girls are 0.6 cm shorter (P ¼ 0.092) than the references.

3 1461 Table 1 Frequencies of normal weight, overweight and obesity BMI centile group a P-value for test Age (y) School type b Normal Overweight Obese BMI c School d Boys 8/9 Primary 264 (78%) 39 (12%) 35 (11%) o0.001 NA 11/12 Higher 45 (75%) 8 (13%) 7 (12%) o Lower 2 (67%) 60 (19%) 48 (%) o0.001 /16 Higher 143 (87%) (6%) 11 (7%) Lower 200 (79%) 22 (9%) 32 (13%) o0.001 Girls 8/9 Primary 267 (71%) 39 (13%) 59 (16%) o0.001 NA 11/12 Higher 69 (79%) 13 (%) 5 (6%) Lower 236 (71%) 59 (18%) 36 (11%) o0.001 /16 Higher 180 (85%) 17 (8%) 14 (7%) o0.001 Lower 147 (67%) 41 (19%) 33 (%) o0.001 NA ¼ not applicable. a Normal BMI: p90 th ; overweight: 490th 97 th ; obesity: 497th reference percentile 6. Percentages are given within each group determined by age and gender (their sum is equal to 0% up to differences due to rounding). b In Germany, the secondary school is splitted into higher ( Gymnasium ) and lower educational tracks ( Hauptschule, Realschule ). c Tests the null hypothesis that the percentages of the BMI centile groups are 90, 7 and 3%. d Tests the null hypothesis that the trend to overweight/obesity does not depend on the school type. a Boys b 40 Girls BMI [kg/m ² ] BMI [kg/m ² ] Age [y] Age [y] Figure 2 Increased percentage of overweight and obese boys (a) and girls (b) in age groups 8/9, 11/12 and /16 y, BMI compared to data of Hesse et al 6. The telescope plots represent our data, curves represent the reference centiles (dark greyf3rd, 50th and 97th percentiles, light greyfth, 25th, 75th and 90th percentiles). Almost all percentiles shown by the telescopes are higher than the reference values, and the medians of our data are on average about 1 kg/m 2 above the reference medians

4 1462 Moreover, within-groups regression analysis showed that, on average, a 1 cm difference in height is associated with a 0.1 kg/m 2 (95% CI: [0.08,0.012]) rise in BMI. In contrast to this, groupwise BMI medians were on average about 1 kg/m 2 above the reference values (th percentiles: 0.5 kg/m 2 ; 90th percentiles: 2.5 kg/m 2 ). Thus, the secular trend in height increments did not explain the observed differences in BMI. BMI and blood pressure There was a significant (Po0.001) trend towards higher prevalence of hypertension in subjects with high BMI (Figure 3). Interestingly, the positive correlation of hypertension to BMI was not only caused by overweight subjects. When the high-risk group (above the 90th percentile) was excluded, the trend remained significant (Po0.001 for boys, P ¼ for girls). This implies to treat the BMI as a continuous covariate rather than to fix an arbitrary cutoff when giving a quantitative description of the risk. Frequency of hypertension [%], with 95% C.I <P25 P25...<P50 P50...<P75 P75...<P90 P90+ BMI percentiles (computed age and gender specific) Gender male female Figure 3 Association of hypertension with BMI. While subjects above the 90th BMI percentile appear to form a high-risk group, there is a continuous rise of blood pressure with BMI even in the lower centile groups. Risk estimates Logistic regression predicting hypertension from BMI, adjusted for arm skinfolds, yielded inhomogeneous estimates in the groups distinguished by age and gender. No association could be detected in girls and boys aged 8/9 y, while there were considerable effects of similar size in the other four groups. The common estimates are presented in Table 2. As a consequence of the odds ratio of 1.23 per kg/m 2, a rise of 3.3 kg/m 2 in BMI (95% CI: [2.4,5.3]) is associated with doubling the risk of hypertension. This would be realized, for example, by a 7.5 kg increase in body weight in an average schoolchild of the fifth class. Similar effects were observed for stem skinfolds, WHR and percentage of body fat calculated from BIA. The respective odds ratios are also shown in Table 2. Comparison of the indices of body fat The comparison of the contrasts between hypertensive and normotensive subjects with respect to the four indices of body fat involved only the age groups of 11/12 and / 16 y, as no significant contrast was found in the children aged 8/9 y. Regarding normotensive subjects that form the reference population, the average SDS of BMI of hypertensive children was 1.21, which was significantly higher than their SDS of stem fat (mean 0.99; P ¼ 0.008), WHR (mean 0.79; P ¼ 0.002) and BIA body fat (mean 0.84; Po0.001). This coincides with the finding that in logistic regression for hypertension, given two of these indices of body fat as competing regressors, BMI was preferred to any other of these variables to be the stronger predictor. Given BMI, no significant information on hypertension was added by any of the other measures of body fat. The converse was not true. Discussion Our observations are consistent with the results of studies from other countries: obesity among children is increas- Table 2 Risk of hypertension a (HT) depending on increased body fat Effect Age group Increase in risk of HT P-value for difference b Odds ratio c 95% CI Rise of BMI by 1 kg/m 2 8/ [0.87,1.16] /12+/ [1.14,1.34] Double size of stem skinfolds 8/ [0.57,2.17] /12+/ [1.61,4.03] Rise of WHR by / [0.66,1.32] /12+/ [1.09,1.66] Rise of BIA body fat by 5% 8/ [0.65,1.] /12+/ [1.21,1.88] NA ¼ not applicable. a Hypertension is considered to be a diastolic blood pressure above the 95th reference percentile with respect to age, gender and body height 21. b Tests the null hypothesis that the two odds ratios for the age groups 8/9 and 11/12+/16 are equal. The P-values indicate that the age groups do not behave homogeneously. c The odds ratios describe the increase of the risk of HT associated with the effect given in the first column (adjusted for upper arm skinfolds).

5 ing, 4,5 is related to the socioeconomic status 25 (reflected by the level of education in our study) and is associated with hypertension even in young age.,11 In addition, it has been pointed out that the rise in BMI is something more than the normal secular trend seen in the body height. Furthermore, it was shown that not only obesity is related to hypertension. The risk is increasing already within a span of BMI values commonly considered to be normal range. Hence, having a BMI below some widely accepted cutoff value does not mean the absence of increased cardiovascular risk resulting from body fat. We were able to demonstrate that the significant relationship of obesity to blood pressure in childhood can be described using any of four different established indices of body fat, providing quantitative estimates of the risk. However, our data suggest that BMI is superior to the other measures in distinguishing the hypertensive subpopulation. Taking into account the need for special devices to measure skinfolds or bioelectric impedance, observer-specific techniques in measuring skinfolds, the number of formulas converting impedance measurements into estimates of body fat, and the simplicity of the BMI on the other hand, it seems that BMI should be the preferred index of body fat for research in this area. As a limitation of this field study, single measurements of blood pressure are less reliable than repeated controls, 19,20 which were not available in our study. Less erroneous diagnostics would allow for more precise risk estimates. A second limitation arises from the way of adjustment of the risk estimates for arm fat. Arm skinfolds reflect the body fat content, and therefore predict hypertension as other markers of body fat do. They also increase the mechanical pressure on the measurement device, leading to a measurement bias. When adjusting the odds ratios for the arm skinfolds, not only the mechanical but also the cardiovascular component of information contained in the arm fat was eliminated, so our estimates might be too small. In particular, our results do not guarantee the absence of increased risk even in the age of 8/9 y. Nonetheless, the situation is dramatic enough even with our conservative estimates. Regarding the mean rise of BMI by about 1 kg/m 2 compared to the reference data, and the odds ratio for hypertension of 1.23 per kg/m 2, one may assume that additionally one out of 0 schoolchildren aged above y is hypertensive due to the increase of BMI during the past few years. As the main consequence, surveillance of obesity among children is important in view of the clinical consequences of overweight, which are measurable and quantifiable already in school age. BMI should be preferred as the most simple and powerful variable to assess the degree of obesity. As the increase of health risk is continuous, intervention and prevention studies should include a wider population rather than only subjects above the established cutoff values. Acknowledgements We express our sincere gratitude to the nurses and physicians of the community health authorities, in particular Bodo Gronemann of the Public Health Department of the city Leipzig, for the logistic support. We are in debt of the children and their parents for their agreement and participation in the study. The help of Hella de Paly, Alexandra Keller and Alexander Lammert (University Hospital for Children and Adolescents Leipzig) is also gratefully acknowledged. Part of the work was supported by unrestricted grants from the pharmaceutical companies Novo Nordisk, Lilly and Pharmacia. The fourth author was supported by a grant from the Federal Ministry of Education and Research, IZKF Leipzig, Germany. References 1 World Health Organisation. Obesity: preventing and managing the global epidemic, Report of a WHO consultation, Geneva, 3 5 June WHO: Geneva; 1998 [WHO/Nut/98/1]. 2 Pi-Sunyer FX, Laferrère B, Aronne LJ, Bray GA. Therapeutic controversy: obesityfa modern day epidemic. JCEM 1999; 84: Cole TJ, Bellizzi MC, Flegal KM, Dietz WH. Establishing a standard definition for child overweight and obesity worldwide: international survey. BMJ 2000; 320: Rolland-Cachera MF, Sempé M, Gouilloud-Bataille M, Patois E, Pequignot-Guggenbuhl F, Fautrad V. Adiposity indices in children. Am J Clin Nutr 1982; 36: Cole TJ, Freeman JV, Preece MA. Body mass index reference curves for the UK, Arch Dis Child 1995; 73: Hesse V, Bartezky R, Jaeger U, Kromeyer-Hauschild K, Zellner K, Vogel H, Bernhardt I, Hofmann A. Körper Massen-Index: Perzentilen deutscher Kinder im Alter von 0 18 Jahren. Kinderärztl Prax 1999; 8: Guo SS, Roche AF, Chumela WC, Gardner JD, Siervogel RM. The predictive value of childhood body-mass-index values for overweight at age 35 y. Am J Clin Nutr 1994; 59: Kiess W, Reich A, Müller G, Meyer K, Galler A, Bennek J, Kratzsch J. Clinical aspects of obesity in childhood and adolescencefdiagnosis, treatment and prevention. Int J Obes Relat Metab Disord 2001; 25 (Suppl 1): S75 S79. 9 Young-Hyman D, Schlundt DG, Herman L, D Luca F, Counts D. Evaluation of the insulin resistance syndrome in 5- to -year-old overweight/obese African-American children. Diabetes Care 2001; 24: Burke V, Beilin LJ, Dunbar D. Tracking of blood pressure in Australian children. J Hypertens 2001; 19: Chu NF, Wang DG, Shieh SM. Obesity, leptin and blood pressure among children in Taiwan: the Taipei Children s Heart Study. Am J Hypertens 2001; 14: Srinivasan SR, Myers L, Berenson GS. Rate of change in adiposity and ist relationship to concomitant changes in cardiovascular risk variables among biracial (black white) children and young adults: The Bogalusa Heart Study. Metabolism 2001; 50: Geiss HC, Parhofer KG, Schwandt P. Parameters of childhood obesity and their relationship to cardiovascular risk factors in healthy prepubescent children. Int J Obes Relat Metab Disord 2001; 25: Frontini MG, Bao W, Elkasabany A, Srinivasan SR, Berenson G. Comparison of weigth-for-height indices as a measure of adiposity and cardivascular risk from childhood to young adulthood. The Bogalusa Heart study. J Clin Epidemiol 2001; 54:

6 1464 Maffeis C, Tato L. Long-term effects of childhood obesity on morbidity and mortality. Horm Res 2001; 55 (Suppl 1): Power C, Lake JK, Cole TJ. Measurement of long-term health risks of child and adolescent fatness. Int J Obes Relat Metab Disord 1997; 21: Mast M, Sönnichsen A, Langnäse K, Labitzke K, Bruse U, Preu U, Müller MJ. Inconsistencies in bioelectrical impedance and anthropometric measurements of fat mass in a field study of prepubertal children. Br J Nutr 2002; 87: Durnin JV, Rahaman MM. The assessment of the amount of fat in the human body from measurements of skinfold thickness. Br J Nutr 1967; 21: Flynn JT. Evaluation and management of hypertension in childhood. Prog Pediatr Cardiol 2001; 12: Goonasekera CDA, Dillon MJ. Measurement and interpretation of blood pressure. Arch Dis Child 2000; 82: Sörgel M, Kirschstein M, Busch C, Danne T, Gellermann J, Holl R, Krull F, Reichert H, Reusz GS, Rascher W. Oscillometric twentyfour-hour ambulatory blood pressure values in healthy children and adolescents: a multicenter trial including 1141 subjects. J Pediatr 1997; 130/2: Wabitsch M, Braun U, Heinze E, Muche R, Mayer H, Teller W, Fusch C. Body composition in 5 18-y-old obese children and adolescents before and after weight reduction as assessed by deuterium dilution and bioelectrical impedance analysis. Am J Clin Nutr 1996; 64: Whitehead A, Whitehead J. A general parametric approach to the meta-analysis of randomized clinical trials. Stat Med 1991; : Hesse V, Jaeger U, Vogel H, Kromeyer K, Zellner K, Bernhardt I, Hofmann A, Deichl A. Wachstumsdaten deutscher Kinder von der Geburt bis zu 18 Jahren. Sozialpädiatrie 1997; 20: Robinson TN, Chang JY, Haydel KF, Killen JD. Overweight concerns and body dissatisfaction among third-grade children: the impacts of ethnicity and socioeconomic status. J Pediatr 2001; 138: