Introduction

In recent years, an increasing prevalence of type 2 diabetes mellitus and its intermediate traits has been observed in children.^{1, 2, 3} Previous studies have demonstrated inverse relationships between measures of insulin resistance and subjectively measured physical activity in the paediatric population.^{4, 5, 6, 7} However, questionnaire-based assessment of physical activity, which is the most common subjective method, is imprecise, particularly in children. Consequently, it is normally recommended that in children of 11 y of age or younger, self-report methods are not used.^{8, 9, 10, 11} A positive correlation between fasting insulin and doubly labelled water measured physical activity has been reported in a small sample of premenarcheal girls, although this relationship disappeared when outliers were excluded.¹² No epidemiological study has so far related insulin resistance in children to an objective measure of physical activity, for example, accelerometry. Therefore, the dose–response relationship between physical activity and insulin resistance in children remains unclear, making it difficult to derive informed public health recommendations. Thus, the aim of this study was to investigate the association between indices of insulin resistance and an accelerometry-derived measure of physical activity. These relationships were examined at multiple levels of adjustment for potential confounders, including age, gender, puberty, adiposity, weight, height, ethnicity, birth weight, socioeconomic grouping, parental smoking, and accelerometer unit. Furthermore, because of increasing interest in the association between birth weight and insulin resistance,^{13, 14, 15} we also investigated whether birth weight modifies the relationship between physical activity and insulin resistance.

Materials and methods

Design

The study was a school-based, cross-sectional study of prepubertal children, randomly selected by a two-stage sampling strategy. Data were collected in the academic year 1997–1998 in the county of Odense, Denmark, as part of the European Youth Heart Study (EYHS).

Participants

Target population: Altogether 711 8- to 10-y-old children, who attended schools in the county of Odense, Denmark, were invited to participate in the study, along with their parents.

Sampling: Schools were stratified according to location (urban, suburban, rural) and socioeconomic profile of uptake area (high, middle, low). From each stratum, a proportional, two-stage cluster sample of children was selected. The primary units were the schools. The sampling frame for schools was a complete list of public schools in Odense, from which schools were sampled using probability proportional to school size. The secondary units were the children within the schools, and equal numbers of children were sampled from each school. The children were allocated code numbers and randomly selected using random number tables.

Study sample: In all, 28 out of 35 schools were sampled, and 25 agreed to participate. The only two urban low-income area schools in Odense were sampled, 20 out of 24 schools from middle-income areas, two out of four schools in high-income areas, and four out of five rural area schools were sampled. Of the three nonparticipating schools, one was rural, one was urban from middleclass area, and one was urban from a low-income area. All these three schools gave interference with the educational process as reason for not participating. A total of 711 individuals (58% of the total eligible population) were sampled and invited to participate in the study, of which 690 children (97%) responded to the invitation. A total of 589 (83% of the sampled and 48% of the total eligible population) participated in the study together with their parents. Among the participating children, there were 310 females and 279 males, which reflected the gender distribution of the Odense region.

Ethics

The study was approved by the local scientific ethics committee (case no. 96/272) and performed in accordance with the Helsinki declaration. All parents gave written informed consent for their child to participate and all children gave verbal consent. The measurement procedures were carefully explained during meetings held at each school as well as in the written invitation.

Data collection

Physical examination: Height and weight were measured with a stadiometer and a calibrated scale, respectively. The level of adiposity was assessed by the skinfold technique, which has been shown to correlate highly with DEXA measured body fat percentage in similarly aged children.¹⁶ Harpenden callipers were used to measure skinfold thicknesses at the positions of the m.triceps brachii, m.biceps brachii, subscapularly, and superior to the spina iliaca anterior superior. Measurements were performed on the left side of the body with the child standing. Two measurements were taken on each position. If there was a difference of more than 2 mm, a third measurement was taken, and the mean of the two closest measurements was then used. Measurements were made in rotation (one measurement on each site, then repeated). All measurements were performed by trained personnel and quality control was frequently performed. The same male researcher measured all boys, whereas the girls were measured by two female researchers. Intra- and inter-researcher reliability was r>0.925 and r>0.835, respectively. Puberty status was assessed by examination according to Marshall and Tanner.^{17, 18}

Blood samples: All children and their parents received written information and the children were asked to be fasted when attending the school for blood sampling. The majority of children when asked reported that they had not eaten breakfast. Two children stated that they had eaten breakfast. Therefore, arrangements were made to draw blood from these children the following morning, and the requirement that they attended the school fasted was reiterated. The fasting samples were taken between 0800 and 2030 in the morning. They were immediately stored at -80°C and analysed at a central certified laboratory for insulin and glucose. Glucose was analysed using the Hexokinase method, measured on an Olympus AU600 autoanalyser (Olympus Diagnostica GmbH, Hamburg, Germany). Insulin was analysed using enzyme immunoassay (microtitre plate format, Dako Diagnostics Ltd, Ely, England). The coefficients of variation between batches were as follows: glucose 1.2% (at 3.26 and 14.67 mmol l^-1) and insulin 6.9% (at 110.4 pmol l^-1) and 5.9% (at 356.2 pmol l^-1).

Physical activity: Habitual physical activity was assessed with the Computer Science and Applications (CSA) accelerometer, now also known as the MTI actigraph (Manufacturing Technology Inc., Fort Walton Beach, FL, USA). The CSA is a uniaxial piezoelectric accelerometer with a dynamic range of 2.13g and a frequency-dependent bandwidth filter, which can be regarded as a mathematical weighting function.^{19, 20} The CSA samples acceleration at 10 Hz and integrates this over the user-defined epoch. In this study, the CSA was mailed to the children, preprogrammed to start at 0800 the following morning. The children were instructed to wear it for 4 days at all times, except during water-based activities or when sleeping. The epoch was set at 1 min, comprising an integral of 600 measurements for each data point. The CSA was returned by the children and data downloaded on the day of the physical examination. In order to distinguish true zeros from the zeros recorded when the monitor had been taken off, the field data were cleaned in the following manner: All CSA files were screened for periods of zero activity. Zero activity periods of 10 min or longer were interpreted as 'CSA not worn' and these periods were thus removed from the summation of activity. Given these criteria, the data were included if the child had accumulated more than 10 h of activity data per day for at least 3 days. Data are expressed as total counts per registered time (counts min^-1) to yield a measure of average physical activity intensity. Although the CSA exhibits good intrainstrument reliability in mechanical settings, interinstrument differences have been reported.^{20, 21} Analyses are therefore also adjusted for CSA unit as a categorical variable. There were 64 CSA units in use in this study.

Other information: The socioeconomic group (SEG) of each parent was assessed via questionnaire. Both the education and the income level of each of the parents were coded according to the National Statistic Registry and then recoded into five-level scores, ranging from 1 to 5. A higher number indicates a higher level of education and income. These two scores were then averaged into the SEG score, which therefore has nine levels (1, 1.5, 2, ..., 5). The parent questionnaire also contained questions on birth weight (kg), ethnicity (1=Caucasian, 0=Other), and parental smoking (coded as 1=one or both parents smoke, and 0=none of the parents smoke).

Statistics

The ethical committee did not approve any collection of information on nonparticipants. Hence, dropout analyses are performed by comparing differences between subjects with complete data on the primary exposure and outcome and subjects with one or more missing values. All variables were checked for normality. Insulin and average CSA counts were normalised by square root, body weight, glucose, and skinfold thickness were normalised by the reciprocal square root, age was logarithmically transformed, and birth weight was raised to the power of two.

Following a bivariate correlation analysis, multiple linear regression analyses were used for assessing the relationship between physical activity and indices of insulin resistance. The relationships were adjusted for age, gender, puberty, ethnicity, socioeconomic status, parental smoking, and birth weight. Additional adjustments were made for body mass index (BMI) and the sum of four skinfolds, separately. The regression analyses were clustered on school (primary sampling unit) to produce robust standard errors. To investigate the possibility of effect modification by birth weight, the analyses were also stratified by below and above the median birth weight. The software package STATA, Version 8.0 (Stata Corp, TX, USA) was used for all statistical analyses. The level of significance was set at P=0.05.

Results

Representativeness

In comparison with the National Statistic Registry, the distributions of both educational level and income of the parents were representative of the country as a whole. Likewise, the BMI of both genders did not differ from the age-specific BMI, estimated from school census data.²²

Data completeness

Owing to a limited number of accelerometers available in this study, it was only possible to assess physical activity in 427 children. Of this group, 384 (179 boys, 205 girls) had valid physical activity data, as determined by the criteria described in the Materials and methods section. Of the outstanding 43 observations, data were unavailable due to download error (n=9) or were excluded due to instrument breakage (n=23), too little registered time (n=10), or distinct unphysiological pattern (ie >9 standard deviations higher than the median physical activity, n=1). Table 1 displays descriptives of all measured variables for the sample as a whole and for the subsample with complete data in all variables. Biochemistry data were available in 525 children (247 boys, 278 girls). The mean fasting glucose was 5.1 mmol l^-1, ranging from 4.1 to 6.1 mmol l^-1. Complete data for age, gender, puberty, physical activity, biochemistry, skinfold thickness, ethnicity, SEG, parental smoking, and birth weight were available in 313 children. There were proportionally fewer non-Caucasians in the subsample with complete data (P=0.002).

Table 1 - Baseline characteristics of Danish 9- to 10-y-old children in the EYHS.

Full table

Unadjusted associations

In the unadjusted correlation analyses, fasting insulin was inversely associated with physical activity (r=-0.23, P<0.001), SEG of the father (r=-0.18, P<0.001) and the mother (r=-0.17, P<0.001), and age (r=-0.10, P=0.017). Insulin was positively correlated with body weight (r=0.38, P<0.001), glucose (r=0.35, P<0.001), skinfold thickness (r=0.24, P<0.001), puberty (r=0.23, P<0.001), and height (r=0.12, P=0.005). Fasting glucose was inversely correlated with physical activity (r=-0.12, P=0.028), SEG of the father (r=-0.13, P=0.003) and the mother (r=-0.10, P=0.022), and age (r=-0.11, P=0.010). On average, boys had 0.90 pmol l^-1 lower insulin levels (P=0.010), 0.11 mmol l^-1 higher glucose levels (P<0.001), and had accumulated 127 CSA counts min^-1 more than girls (P<0.001). Fasting insulin was 19.3 pmol l^-1 higher in the non-Caucasians (P=0.022).

Association between physical activity and indices of insulin resistance

The PA regression coefficients with 95% confidence intervals for fasting insulin and glucose are displayed in Table 2. There was an inverse relationship between physical activity and insulin after adjustment for potential confounders. This remained after adjustment for both BMI (standardised PA =-0.179, P=0.039) and when skinfold thickness was used as indicator of adiposity (P=0.049). Puberty status was the only other significant predictor of fasting insulin (P0.048). There were no significant associations between fasting glucose and physical activity. The only significant predictor of fasting glucose was ethnicity (lower in Caucasian children, P0.009). The associations were similar when the analyses were restricted to the Caucasian children and in the strata below and above the median birth weight (3.40 kg). However, when we stratified for gender, the relationships in the girls (n=163) between activity and insulin resistance were significant for both insulin and glucose, and also following adjustment for adiposity. In boys (n=150), the relationships were no longer significant.

Table 2 - Prediction of insulin resistance from physical activity.

Full table

Discussion

Physical activity sampled over 3–4 days represents only a fraction of the time that may be relevant for influencing insulin resistance. Nonetheless, the variance in fasting insulin was significantly associated with the variance in accelerometry measured physical activity after adjustment for potential confounders. This association was independent of BMI and skinfold thickness. Gender-specific analyses revealed that this was strongest in the girls, whereas none of the associations we explored were significant in the boys. However, this may be due to a combination of effect modification and insufficient statistical power. In contrast to fasting insulin, the association between physical activity and glucose was insignificant, when all confounders were controlled for. However, in girls this relationship was significant and persisted even after adjustment for adiposity.

The precision of objective assessment of physical activity energy expenditure in children is superior to subjective methods,^{8, 9, 10, 11, 23} but there are nonetheless some limitations, which should be highlighted. For example, cycling and swimming is not captured. CSA output is related to acceleration but not in a linear manner, as it also depends on movement frequency.²⁰ Step frequency therefore influences validity of the CSA as a measure of intensity in running and to a lesser extent in walking.^{24, 25, 26} Even so, CSA output is highly correlated with intensity within the walking–jogging range, although accelerometry generally underestimates higher running intensities due to biomechanical limitations.^{25, 26} Furthermore, since larger individuals have lower step frequencies,^{27, 28} this may cause differential bias because body size also relates to puberty status and adiposity, which are both linked to insulin resistance.^{29, 30, 31} Interestingly, when adjusting for BMI, the relationship between activity and insulin remained virtually the same, whereas when skinfold was used as the measure of adiposity, this association was attenuated. This may well be explained by a BMI correction for measurement error in the activity variable, as opposed to an adjustment for adiposity per se. Adiposity is likely to be on the causal pathway between activity and insulin resistance and adjustment will therefore remove some of the true association. Although chance, bias, and confounding could not explain this association, residual confounding from unmeasured factors may be important. Additionally, genetic factors may well modify the relationship between physical activity and insulin resistance in children.

The coefficient for insulin was -0.085 on the transformed scale (square root). This equates to -1.0 pmol l^-1, -1.2 pmol l^-1, and -1.4 pmol l^-1, on the 25th, 50th, and 75th percentiles of fasting insulin, respectively. One unit increment on the transformed physical activity scale equates into increments of 46 counts min^-1, 51 counts min^-1, and 58 counts min^-1 for the 25th, 50th, and 75th percentiles, respectively. Using doubly labelled water data,²³ an increase of one unit on the transformed square root of counts min^-1 scale results in an increase of 6.5 kJ kg^-1 of daily physical activity energy expenditure (age and gender adjusted) or 211 kJ day^-1 for a child weighing 32.4 kg (median body weight of the sample). The dose–response estimate from this calculation agrees reasonably well with the estimate obtained in a recent cross-sectional study in 10- to 16-y-old individuals, in which a 26.3 pmol l^-1 decrease in fasting insulin for a questionnaire-estimated increase in physical activity energy expenditure of 6231 kJ was reported.⁷ A 211 kJ increase in daily energy expenditure from activity in this population would thus correspond to a 0.9 pmol l^-1 decrease in fasting insulin.⁷ The small difference between the two studies may be explained by underestimation of physical activity by the CSA, overestimation by the questionnaire, and/or differences between other characteristics of the two cohorts, such as pubertal status and body composition.

Other studies in children have also reported inverse relationships between physical activity and insulin resistance indices.^{5, 32} Only a single study has reported a positive relationship between doubly labelled water measured energy expenditure and insulin.¹² However, the relationship in this study disappeared when two outliers were excluded from the analysis of this relatively small sample. Moreover, the estimates of energy expenditure were not standardised for body size and the analyses were not adjusted for puberty or adiposity. Thus, the positive association may be attributable to confounding.

Several trials have demonstrated improvements in insulin sensitivity following exercise intervention programmes^{33, 34, 35, 36, 37, 38, 39, 40, 41} and, moreover, that cessation of activity reverses the training-induced changes.³⁷ However, in one trial involving 10 weeks of exercise training in 24 obese black girls, no significant change in fasting insulin was observed but this may be explained by compensatory behaviour, since only self-reported vigorous physical activity and not overall volume of activity was increased.⁴²

The cross-sectional nature of our study limits inference about direction of causality. Nonetheless, the observation that fasting insulin level is more consistently associated with physical activity, whereas fasting glucose level is less consistently associated with physical activity, may be explained by the way in which normal levels of blood glucose are maintained when insulin resistance develops. Although end-stage insulin resistance is characterised by -cell exhaustion, in the very early stages, such as in childhood insulin resistance, the pancreas compensate for tissue resistance by excreting greater quantities of insulin.⁴³ The association between activity and insulin resistance is most likely explained by enhanced insulin action and noninsulin-stimulated glucose uptake through physical activity, whereby the insulin pathway is off-loaded. Indeed, improved insulin action and glycaemic control through physical activity are usually the product of synergism between numerous mechanisms involving the muscle, adipose tissue, liver, and endothelium. These mechanisms include upregulation of key proteins in the insulin cascade, for example, insulin receptor substrates, phospho-inositol-3-kinase, and glucose transporting proteins.⁴⁴ Other mechanisms are upregulation of the rate-limiting enzymes hexokinase, citrate synthase, and glycogen synthase, which would facilitate maintenance of the concentration gradient during the time when glucose transporting proteins are incorporated in the cell membrane. Another pathway may be the lowering of lipid levels in blood and muscle, which could otherwise impair the insulin cascade.^{45, 46} Existence of a causal relationship between physical activity and insulin resistance is further supported by the observation that immobilisation results in worsening of the glucose tolerance.⁴⁷

In conclusion, physical activity is inversely associated with insulin resistance in glucose-tolerant Danish children and most evidently in girls. Our observations support the view that in children physical activity is an important behaviour for metabolic function. Therefore, the development and implementation of strategies to prevent sedentary behaviour in children is likely to be of paramount importance if metabolic disease is to be averted. Such lifestyle intervention strategies have successfully produced beneficial secular trends in insulin parameters in adolescents in a high school setting.⁴⁸

References

1. American Diabetes Association. Type 2 diabetes in children and adolescents. American Diabetes Association. Diabetes Care 2000; 23: 381–389. | PubMed | ISI |
2. Ehtisham S, Barrett TG, Shaw NJ. Type 2 diabetes mellitus in UK children—an emerging problem. Diabetic Med 2000; 17: 867–871. | Article | PubMed | ISI | ChemPort |
3. Goran MI, Ball GD, Cruz ML. Obesity and risk of type 2 diabetes and cardiovascular disease in children and adolescents. J Clin Endocrinol Metab 2003; 88: 1417–1427. | Article | PubMed | ChemPort |
4. Anavian J, Brenner DJ, Fort P, Speiser PW. Profiles of obese children presenting for metabolic evaluation. J Pediatr Endocrinol Metab 2001; 14: 1145–1150. | PubMed |
5. Ku CY, Gower BA, Hunter GR, Goran MI. Racial differences in insulin secretion and sensitivity in prepubertal children: role of physical fitness and physical activity. Obes Res 2000; 8: 506–515. | PubMed | ISI | ChemPort |
6. Raitakari OT, Taimela S, Porkka KV, Telama R, Valimaki I, Akerblom HK, Viikari JS. Associations between physical activity and risk factors for coronary heart disease: the Cardiovascular Risk in Young Finns Study. Med Sci Sports Exerc 1997; 29: 1055–1061. | PubMed |
7. Schmitz KH, Jacobs Jr DR, Hong CP, Steinberger J, Moran A, Sinaiko AR. Association of physical activity with insulin sensitivity in children. Int J Obes Relat Metab Disord 2002; 26: 1310–1316. | Article | PubMed | ChemPort |
8. Kohl III HW, Fulton JE, Caspersen CJ. Assessment of physical activity among children and adolescents: a review and synthesis. Prev Med 2000; 31: S54–S76. | Article | ISI |
9. Baranowski T, Dworkin R, Cieslik C, Hooks P, Clearman D, Ray L. Reliability and validity of self report of aerobic activity: Family Health Project. Res Q 1984; 55: 309–317.
10. Saris WH. The assessment and evaluation of daily physical activity in children. A review. Acta Paediatr Scand Suppl 1985; 318: 37–48. | PubMed |
11. Sallis JF, Buono MJ, Freedson PS. Bias in estimating caloric expenditure from physical activity in children. Implications for epidemiological studies. Sports Med 1991; 11: 203–209. | PubMed |
12. Craig SB, Bandini LG, Lichtenstein AH, Schaefer EJ, Dietz WH. The impact of physical activity on lipids, lipoproteins, and blood pressure in preadolescent girls. Pediatrics 1996; 98 (Part 1): 389–395. | PubMed | ChemPort |
13. Ozanne SE, Hales CN. Early programming of glucose–insulin metabolism. Trends Endocrinol Metab 2002; 13: 368–373. | Article | PubMed | ChemPort |
14. Li C, Johnson MS, Goran MI. Effects of low birth weight on insulin resistance syndrome in caucasian and African–American children. Diabetes Care 2001; 24: 2035–2042. | Article | PubMed | ISI | ChemPort |
15. Harder T, Kohlhoff R, Dorner G, Rohde W, Plagemann A. Perinatal 'programming' of insulin resistance in childhood: critical impact of neonatal insulin and low birth weight in a risk population. Diabetic Med 2001; 18: 634–639. | Article | PubMed |
16. Gutin B, Litaker M, Islam S, Manos T, Smith C, Treiber F. Body-composition measurement in 9–11-y-old children by dual-energy X-ray absorptiometry, skinfold-thickness measurements, and bioimpedance analysis. Am J Clin Nutr 1996; 63: 287–292. | PubMed | ChemPort |
17. Marshall W, Tanner JM. Variations in pattern of pubertal changes in girls. Arch Dis Child 1969; 44: 291–301. | PubMed | ISI | ChemPort |
18. Marshall W, Tanner JM. Variations in the pattern of pubertal changes in boys. Arch Dis Child 1970; 45: 13–23. | PubMed | ISI | ChemPort |
19. Tryon WW, Williams R. Fully proportional actigraphy: A new instrument. Behav Res Methods Instr Comput 1996; 28: 392–403.
20. Brage S, Brage N, Wedderkopp N, Froberg K. Reliability and validity of the Computer Science and Applications accelerometer in a mechanical setting. Meas Phys Educ Exerc Sci 2003; 7: 101–119. | Article |
21. Metcalf BS, Curnow JS, Evans C, Voss LD, Wilkin TJ. Technical reliability of the CSA activity monitor: the EarlyBird Study. Med Sci Sports Exerc 2002; 34: 1533–1537. | Article | PubMed |
22. Petersen TA, Rasmussen S, Madsen M. BMI of Danish school children measured during the periods 1986/1987–1996/1997 compared to Danish measurement in 1971/1972. Ugeskr Laeger 2002; 164: 5006–5010. | PubMed |
23. Ekelund U, Sjostrom M, Yngve A, Poortvliet E, Nilsson A, Froberg K, Wedderkopp N, Westerterp K. Physical activity assessed by activity monitor and doubly labeled water in children. Med Sci Sports Exerc 2001; 33: 275–281. | Article | PubMed | ISI | ChemPort |
24. Trost SG, Ward DS, Moorehead SM, Watson PD, Riner W, Burke JR. Validity of the computer science and applications (CSA) activity monitor in children. Med Sci Sports Exerc 1998; 30: 629–633. | PubMed | ISI | ChemPort |
25. Brage S, Wedderkopp N, Andersen LB, Froberg K. Influence of step frequency on movement intensity predictions with the CSA accelerometer: a field validation study in children. Ped Exerc Sci 2003; 15: 277–287. | ISI |
26. Brage S, Wedderkopp N, Franks PW, Andersen LB, Froberg K. Reexamination of validity and reliability of the CSA monitor in walking and running. Med Sci Sports Exerc 2003; 35: 1447–1454. | Article | PubMed | ISI |
27. Cavagna GA, Franzetti P. The determinants of the step frequency in walking in humans. J Physiol (Lond) 1986; 373: 235–242. | PubMed |
28. Cavagna GA, Franzetti P, Heglund NC, Willems P. The determinants of the step frequency in running, trotting and hopping in man and other vertebrates. J Physiol (Lond) 1988; 399: 81–92. | PubMed | ISI | ChemPort |
29. Caprio S, Plewe G, Diamond MP, Simonson DC, Boulware SD, Sherwin RS, Tamborlane WV. Increased insulin secretion in puberty: a compensatory response to reductions in insulin sensitivity. J Pediatr 1989; 114: 963–967. | PubMed | ISI | ChemPort |
30. Lestradet H, Deschamps I, Giron B. Insulin and free fatty acid levels during oral glucose tolerance tests and their relation to age in 70 healthy children. Diabetes 1976; 25: 505–508. | PubMed |
31. Johnson MS, Figueroa-Colon R, Huang TTK, Dwyer JH, Goran MI. Longitudinal changes in body fat in African American and Caucasian children: influence of fasting insulin and insulin sensitivity. J Clin Endocrinol Metab 2001; 86: 3182–3187. | Article | PubMed | ChemPort |
32. Matsui I, Nambu S, Baba S. Evaluation of fasting serum insulin levels among Japanese school-age children. J Nutr Sci Vitaminol (Tokyo) 1998; 44: 819–828. | PubMed |
33. Campaigne BN, Gilliam TB, Spencer ML, Lampman RM, Schork MA. Effects of a physical activity program on metabolic control and cardiovascular fitness in children with insulin-dependent diabetes mellitus. Diabetes Care 1984; 7: 57–62. | PubMed |
34. Kahle EB, O'Dorisio TM, Walker RB, Eisenman PA, Reiser S, Cataland S, Zipf WB. Exercise adaptation responses for gastric inhibitory polypeptide (GIP) and insulin in obese children. Possible extra-ancreatic effects. Diabetes 1986; 35: 579–582. | Article | PubMed | ISI | ChemPort |
35. Kahle EB, Zipf WB, Lamb DR, Horswill CA, Ward KM. Association between mild, routine exercise and improved insulin dynamics and glucose control in obese adolescents. Int J Sports Med 1996; 17: 1–6. | PubMed |
36. Hardin DS, Hebert JD, Bayden T, Dehart M, Mazur L. Treatment of childhood syndrome X. Pediatrics 1997; 100: E5. | Article | PubMed |
37. Ferguson MA, Gutin B, Le NA, Karp W, Litaker M, Humphries M, Okuyama T, Riggs S, Owens S. Effects of exercise training and its cessation on components of the insulin resistance syndrome in obese children. Int J Obes Relat Metab Disord 1999; 23: 889–895. | Article | PubMed |
38. Escobar O, Mizuma H, Sothern MS, Blecker U, Udall Jr JN, Suskind RM, Hilton C, Vargas A. Hepatic insulin clearance increases after weight loss in obese children and adolescents. Am J Med Sci 1999; 317: 282–286. | Article | PubMed | ISI | ChemPort |
39. McMurray RG, Bauman MJ, Harrell JS, Brown S, Bangdiwala SI. Effects of improvement in aerobic power on resting insulin and glucose concentrations in children. Eur J Appl Physiol 2000; 81: 132–139. | PubMed |
40. Sudi KM, Gallistl S, Trobinger M, Payerl D, Weinhandl G, Muntean W, Aigner R, Borkenstein MH. The influence of weight loss on fibrinolytic and metabolic parameters in obese children and adolescents. J Pediatr Endocrinol Metab 2001; 14: 85–94. | PubMed |
41. Kang HS, Gutin B, Barbeau P, Owens S, Lemmon CR, Allison J, Litaker MS, Le NA. Physical training improves insulin resistance syndrome markers in obese adolescents. Med Sci Sports Exerc 2002; 34: 1920–1927. | Article | PubMed | ISI | ChemPort |
42. Gutin B, Cucuzzo N, Islam S, Smith C, Stachura ME. Physical training, lifestyle education, and coronary risk factors in obese girls. Med Sci Sports Exerc 1996; 28: 19–23. | Article | PubMed | ISI | ChemPort |
43. Le Stunff C, Fallin D, Schork NJ, Bougneres P. The insulin gene VNTR is associated with fasting insulin levels and development of juvenile obesity. Nat Genet 2000; 26: 444–446. | Article | PubMed | ISI | ChemPort |
44. Zierath JR. Invited review: EXERCISE training-induced changes in insulin signaling in skeletal muscle. J Appl Physiol 2002; 93: 773–781. | PubMed |
45. Greco AV, Mingrone G, Giancaterini A, Manco M, Morroni M, Cinti S, Granzotto M, Vettor R, Camastra S, Ferrannini E. Insulin resistance in morbid obesity: reversal with intramyocellular fat depletion. Diabetes 2002; 51: 144–151. | Article | PubMed | ISI | ChemPort |
46. Ashley MA, Buckley AJ, Criss AL, Ward JA, Kemp A, Garnett S, Cowell CT, Baur LA, Thompson CH. Familial, anthropometric, and metabolic associations of intramyocellular lipid levels in prepubertal males. Pediatr Res 2002; 51: 81–86. | PubMed |
47. Derave W, Eijnde BO, Verbessem P, Ramaekers M, Van Leemputte M, Richter EA, Hespel P. Combined creatine and protein supplementation in conjunction with resistance training promotes muscle GLUT-4 content and glucose tolerance in humans. J Appl Physiol 2003; 94: 1910–1916. | PubMed |
48. Ritenbaugh C, Teufel-Shone NI, Aickin MG, Joe JR, Poirier S, Dillingham DC, Johnson D, Henning S, Cole SM, Cockerham D. A lifestyle intervention improves plasma insulin levels among Native American high school youth. Prev Med 2003; 36: 309–319. | Article | PubMed |

Acknowledgements

We acknowledge the statistical assistance of Dr Jian'an Luan, MRC Epidemiology Unit, Cambridge, UK. The study received financial support from the Danish Medical Research Council and the Danish Heart Foundation.