[Skip to Content]
[Skip to Content Landing]
Figure.
Path Diagram for the Total and Indirect Effects of Physical Activity on Insulin Sensitivity
Path Diagram for the Total and Indirect Effects of Physical Activity on Insulin Sensitivity

For every additional 10 minutes of moderate-to-vigorous physical activity daily, insulin sensitivity increased by approximately 4% (for both the homeostatic model assessment of insulin resistance and the Matsuda insulin sensitivity index) 2 years later. Every additional hour of screen time daily was associated with a 5% decrease in insulin sensitivity (figure not shown). Every additional 10 minutes of moderate-to-vigorous physical activity was associated with a 4.3% decrease in adiposity, and every hour of screen time was associated with a 3.1% increase in adiposity. In the multivariable model adjusted for adiposity, each 1% increase in adiposity predicted a 3% decrease in insulin sensitivity (for both the homeostatic model assessment of insulin resistance and the Matsuda insulin sensitivity index); moderate-to-vigorous physical activity predicted only a 1.8% increase in insulin sensitivity, and screen time predicted only a 2% decrease in insulin sensitivity. Results from the Sobel tests for mediation were significant, with all P < .02.

Table 1.  
Baseline and Follow-up Characteristics of Participants Aged 8 to 10 Years at Baseline in the QUALITY Cohort
Baseline and Follow-up Characteristics of Participants Aged 8 to 10 Years at Baseline in the QUALITY Cohort
Table 2.  
Multivariate Linear Regression Models Examining the Associations Among Physical Activity, Fitness, Sedentary Behavior, and Insulin Sensitivity in Children in the QUALITY Cohorta
Multivariate Linear Regression Models Examining the Associations Among Physical Activity, Fitness, Sedentary Behavior, and Insulin Sensitivity in Children in the QUALITY Cohorta
Table 3.  
Multivariable Regression Models Using Fractional Polynomials to Adjust for Insulin Sensitivity Examining the Associations Among Physical Activity, Fitness, Sedentary Behavior, and First-Phase Insulin Secretion in Children in the QUALITY Cohorta
Multivariable Regression Models Using Fractional Polynomials to Adjust for Insulin Sensitivity Examining the Associations Among Physical Activity, Fitness, Sedentary Behavior, and First-Phase Insulin Secretion in Children in the QUALITY Cohorta
Table 4.  
Multivariable Regression Models Using Fractional Polynomials to Adjust for Insulin Sensitivity Examining the Associations Among Physical Activity, Fitness, Sedentary Behavior, and Second-Phase Insulin Secretion in Children in the QUALITY Cohorta
Multivariable Regression Models Using Fractional Polynomials to Adjust for Insulin Sensitivity Examining the Associations Among Physical Activity, Fitness, Sedentary Behavior, and Second-Phase Insulin Secretion in Children in the QUALITY Cohorta
1.
Knowler  WC, Barrett-Connor  E, Fowler  SE,  et al; Diabetes Prevention Program Research Group.  Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin.  N Engl J Med. 2002;346(6):393-403.PubMedGoogle ScholarCrossref
2.
Pan  XR, Li  GW, Hu  YH,  et al.  Effects of diet and exercise in preventing NIDDM in people with impaired glucose tolerance: the Da Qing IGT and Diabetes Study.  Diabetes Care. 1997;20(4):537-544.PubMedGoogle ScholarCrossref
3.
Honig  CR, Odoroff  CL, Frierson  JL.  Capillary recruitment in exercise: rate, extent, uniformity, and relation to blood flow.  Am J Physiol. 1980;238(1):H31-H42.PubMedGoogle Scholar
4.
Goodyear  LJ, Kahn  BB.  Exercise, glucose transport, and insulin sensitivity.  Annu Rev Med. 1998;49:235-261.PubMedGoogle ScholarCrossref
5.
Ivy  JL.  Role of exercise training in the prevention and treatment of insulin resistance and non-insulin-dependent diabetes mellitus.  Sports Med. 1997;24(5):321-336.PubMedGoogle ScholarCrossref
6.
Earnest  CP.  Exercise interval training: an improved stimulus for improving the physiology of pre-diabetes.  Med Hypotheses. 2008;71(5):752-761.PubMedGoogle ScholarCrossref
7.
Blüher  M.  Adipose tissue dysfunction in obesity.  Exp Clin Endocrinol Diabetes. 2009;117(6):241-250.PubMedGoogle ScholarCrossref
8.
Cusi  K.  The role of adipose tissue and lipotoxicity in the pathogenesis of type 2 diabetes.  Curr Diab Rep. 2010;10(4):306-315.PubMedGoogle ScholarCrossref
9.
Soumaya  K.  Molecular mechanisms of insulin resistance in diabetes.  Adv Exp Med Biol. 2012;771:240-251.PubMedGoogle Scholar
10.
Zhao  YF, Feng  DD, Chen  C.  Contribution of adipocyte-derived factors to beta-cell dysfunction in diabetes.  Int J Biochem Cell Biol. 2006;38(5-6):804-819.PubMedGoogle ScholarCrossref
11.
Li  C, Ford  ES, Zhao  G, Mokdad  AH.  Prevalence of pre-diabetes and its association with clustering of cardiometabolic risk factors and hyperinsulinemia among U.S. adolescents: National Health and Nutrition Examination Survey 2005-2006.  Diabetes Care. 2009;32(2):342-347.PubMedGoogle ScholarCrossref
12.
Lambert  M, Delvin  EE, Levy  E,  et al.  Prevalence of cardiometabolic risk factors by weight status in a population-based sample of Quebec children and adolescents.  Can J Cardiol. 2008;24(7):575-583.PubMedGoogle ScholarCrossref
13.
Active Healthy Kids Canada.  Are We Driving Our Kids to Unhealthy Habits? The 2013 Active Healthy Kids Canada Report Card on Physical Activity for Children and Youth. Toronto, ON: Active Healthy Kids Canada; 2014.
14.
Thomas  AS, Greene  LF, Ard  JD, Oster  RA, Darnell  BE, Gower  BA.  Physical activity may facilitate diabetes prevention in adolescents.  Diabetes Care. 2009;32(1):9-13.PubMedGoogle ScholarCrossref
15.
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(7):506-515.PubMedGoogle ScholarCrossref
16.
Rizzo  NS, Ruiz  JR, Hurtig-Wennlöf  A, Ortega  FB, Sjöström  M.  Relationship of physical activity, fitness, and fatness with clustered metabolic risk in children and adolescents: the European Youth Heart Study.  J Pediatr. 2007;150(4):388-394.PubMedGoogle ScholarCrossref
17.
Kasa-Vubu  JZ, Lee  CC, Rosenthal  A, Singer  K, Halter  JB.  Cardiovascular fitness and exercise as determinants of insulin resistance in postpubertal adolescent females.  J Clin Endocrinol Metab. 2005;90(2):849-854.PubMedGoogle ScholarCrossref
18.
Ruiz  JR, Rizzo  NS, Ortega  FB, Loit  HM, Veidebaum  T, Sjöström  M.  Markers of insulin resistance are associated with fatness and fitness in school-aged children: the European Youth Heart Study.  Diabetologia. 2007;50(7):1401-1408.PubMedGoogle ScholarCrossref
19.
Ball  GD, Shaibi  GQ, Cruz  ML, Watkins  MP, Weigensberg  MJ, Goran  MI.  Insulin sensitivity, cardiorespiratory fitness, and physical activity in overweight Hispanic youth.  Obes Res. 2004;12(1):77-85.PubMedGoogle ScholarCrossref
20.
Lee  S, Bacha  F, Gungor  N, Arslanian  SA.  Cardiorespiratory fitness in youth: relationship to insulin sensitivity and beta-cell function.  Obesity (Silver Spring). 2006;14(9):1579-1585.PubMedGoogle ScholarCrossref
21.
Hardy  LL, Denney-Wilson  E, Thrift  AP, Okely  AD, Baur  LA.  Screen time and metabolic risk factors among adolescents.  Arch Pediatr Adolesc Med. 2010;164(7):643-649.PubMedGoogle ScholarCrossref
22.
Danielsen  YS, Júlíusson  PB, Nordhus  IH,  et al.  The relationship between life-style and cardio-metabolic risk indicators in children: the importance of screen time.  Acta Paediatr. 2011;100(2):253-259.PubMedGoogle ScholarCrossref
23.
Jago  R, Wedderkopp  N, Kristensen  PL,  et al.  Six-year change in youth physical activity and effect on fasting insulin and HOMA-IR.  Am J Prev Med. 2008;35(6):554-560.PubMedGoogle ScholarCrossref
24.
Telford  RD, Cunningham  RB, Shaw  JE,  et al.  Contrasting longitudinal and cross-sectional relationships between insulin resistance and percentage of body fat, fitness, and physical activity in children: the LOOK study.  Pediatr Diabetes. 2009;10(8):500-507.PubMedGoogle ScholarCrossref
25.
Henderson  M, Gray-Donald  K, Rabasa-Lhoret  R,  et al.  Insulin secretion and its association with physical activity, fitness and screen time in children.  Obesity (Silver Spring). 2014;22(2):504-511.PubMedGoogle ScholarCrossref
26.
Henderson  M, Gray-Donald  K, Mathieu  ME,  et al.  How are physical activity, fitness, and sedentary behavior associated with insulin sensitivity in children?  Diabetes Care. 2012;35(6):1272-1278.PubMedGoogle ScholarCrossref
27.
Lambert  M, Van Hulst  A, O’Loughlin  J,  et al.  Cohort profile: the Quebec adipose and lifestyle investigation in youth cohort.  Int J Epidemiol. 2012;41(6):1533-1544.PubMedGoogle ScholarCrossref
28.
Colley  R, Connor Gorber  S, Tremblay  MS.  Quality control and data reduction procedures for accelerometry-derived measures of physical activity.  Health Rep. 2010;21(1):63-69.PubMedGoogle Scholar
29.
Troiano  RP, Berrigan  D, Dodd  KW, Mâsse  LC, Tilert  T, McDowell  M.  Physical activity in the United States measured by accelerometer.  Med Sci Sports Exerc. 2008;40(1):181-188.PubMedGoogle ScholarCrossref
30.
Evenson  KR, Catellier  DJ, Gill  K, Ondrak  KS, McMurray  RG.  Calibration of two objective measures of physical activity for children.  J Sports Sci. 2008;26(14):1557-1565.PubMedGoogle ScholarCrossref
31.
Trost  SG, Loprinzi  PD, Moore  R, Pfeiffer  KA.  Comparison of accelerometer cut points for predicting activity intensity in youth.  Med Sci Sports Exerc. 2011;43(7):1360-1368.PubMedGoogle ScholarCrossref
32.
Skinner  JS, ed.  Exercise Testing and Exercise Prescription for Special Cases: Theoretical Basis and Clinical Application. 3rd ed. Baltimore, MD: Lippincott Williams & Wilkins; 2005.
33.
Docherty  D, ed.  Measurement in Pediatric Exercise Science. Champaign, IL: Human Kinetics; 1996.
34.
Allard  P, Delvin  EE, Paradis  G,  et al.  Distribution of fasting plasma insulin, free fatty acids, and glucose concentrations and of homeostasis model assessment of insulin resistance in a representative sample of Quebec children and adolescents.  Clin Chem. 2003;49(4):644-649.PubMedGoogle ScholarCrossref
35.
Matthews  DR, Hosker  JP, Rudenski  AS, Naylor  BA, Treacher  DF, Turner  RC.  Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man.  Diabetologia. 1985;28(7):412-419.PubMedGoogle ScholarCrossref
36.
Matsuda  M, DeFronzo  RA.  Insulin sensitivity indices obtained from oral glucose tolerance testing: comparison with the euglycemic insulin clamp.  Diabetes Care. 1999;22(9):1462-1470.PubMedGoogle ScholarCrossref
37.
Henderson  M, Rabasa-Lhoret  R, Bastard  JP,  et al.  Measuring insulin sensitivity in youth: how do the different indices compare with the gold-standard method?  Diabetes Metab. 2011;37(1):72-78.PubMedGoogle ScholarCrossref
38.
Henderson  M, Baillargeon  JP, Rabasa-Lhoret  R, Chiasson  JL, Hanley  J, Lambert  M.  Estimating insulin secretion in youth using simple indices derived from the oral glucose tolerance test.  Diabetes Metab. 2012;38(4):309-315.PubMedGoogle ScholarCrossref
39.
Goran  MI.  Measurement issues related to studies of childhood obesity: assessment of body composition, body fat distribution, physical activity, and food intake.  Pediatrics. 1998;101(3, pt 2):505-518.PubMedGoogle Scholar
40.
Centers for Disease Control and Prevention. 2000 Centers for Disease Control and Prevention growth charts. http://www.cdc.gov/growthcharts/. Accessed July 12, 2012.
41.
Marshall  WA, Tanner  JM.  Variations in pattern of pubertal changes in girls.  Arch Dis Child. 1969;44(235):291-303.PubMedGoogle ScholarCrossref
42.
Marshall  WA, Tanner  JM.  Variations in the pattern of pubertal changes in boys.  Arch Dis Child. 1970;45(239):13-23.PubMedGoogle ScholarCrossref
43.
Raghunathan  T, Solenberger  P, Van Hoewyk  J. IVEware: Imputation and Variance Estimation Software user guide. http://www.isr.umich.edu/src/smp/ive/. Accessed May 19, 2008.
44.
White  IR, Royston  P, Wood  AM.  Multiple imputation using chained equations: issues and guidance for practice.  Stat Med. 2011;30(4):377-399.PubMedGoogle ScholarCrossref
45.
Royston  P, Sauerbrei  W.  Multivariable Model-Building: A Pragmatic Approach to Regression Analysis Based on Fractional Polynomials for Continuous Variables. Hoboken, NJ: Wiley; 2008.
46.
Bélanger  M, Gray-Donald  K, O’Loughlin  J, Paradis  G, Hanley  J.  Influence of weather conditions and season on physical activity in adolescents.  Ann Epidemiol. 2009;19(3):180-186.PubMedGoogle ScholarCrossref
47.
Cole  TJ.  Sympercents: symmetric percentage differences on the 100 log(e) scale simplify the presentation of log transformed data.  Stat Med. 2000;19(22):3109-3125.PubMedGoogle ScholarCrossref
48.
Baron  RM, Kenny  DA.  The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations.  J Pers Soc Psychol. 1986;51(6):1173-1182.PubMedGoogle ScholarCrossref
49.
Guy  S, Ratzki-Leewing  A, Gwadry-Sridhar  F.  Moving beyond the stigma: systematic review of video games and their potential to combat obesity.  Int J Hypertens. 2011;2011:179124.PubMedGoogle ScholarCrossref
50.
Hannon  TS, Janosky  J, Arslanian  SA.  Longitudinal study of physiologic insulin resistance and metabolic changes of puberty.  Pediatr Res. 2006;60(6):759-763.PubMedGoogle ScholarCrossref
Original Investigation
March 2016

Influence of Adiposity, Physical Activity, Fitness, and Screen Time on Insulin Dynamics Over 2 Years in Children

Author Affiliations
  • 1Centre Hospitalier Universitaire Sainte-Justine Research Center, University of Montreal, Montreal, Quebec, Canada
  • 2Department of Pediatrics, University of Montreal, Montreal, Quebec, Canada
  • 3Department of Epidemiology, Biostatistics, and Occupational Health, McGill University, Montreal, Quebec, Canada
  • 4Department of Medicine, McGill University, Montreal, Quebec, Canada
  • 5Respiratory Epidemiology and Clinical Research Unit, McGill University Health Centre, Montreal, Quebec, Canada
  • 6Epidemiology and Biostatistic Unit, INRS–Institut Armand-Frappier, Laval, Quebec, Canada
  • 7Department of Kinesiology, University of Montreal, Montreal, Quebec, Canada
  • 8School of Dietetics and Human Nutrition, McGill University, Montreal, Quebec, Canada
JAMA Pediatr. 2016;170(3):227-235. doi:10.1001/jamapediatrics.2015.3909
Abstract

Importance  Despite extensive evidence showing that lifestyle habits play a critical role in preventing or delaying the onset of type 2 diabetes in adults, little is known regarding the impact of lifestyle habits on type 2 diabetes risk in childhood.

Objective  To assess whether adiposity, fitness, moderate-to-vigorous physical activity, and screen time predict insulin sensitivity or insulin secretion during a 2-year period in children with a family history of obesity.

Design, Setting, and Participants  This is a prospective longitudinal cohort study of 630 children, having at least 1 obese parent, recruited from schools in Quebec, Canada, between July 2005 and December 2008 in the Quebec Adipose and Lifestyle Investigation in Youth (QUALITY) cohort. Children were assessed at baseline (ages 8-10 years) and 2 years later. Fitness was measured by peak oxygen consumption, percentage of body fat (adiposity) by dual-energy x-ray absorptiometry, moderate-to-vigorous physical activity using accelerometry, and screen time by average daily hours of self-reported television, video game, or computer use. Regression models were adjusted for age, sex, season, and pubertal stage. The current analysis was completed in October 2015.

Main Outcomes and Measures  Insulin sensitivity was measured by the homeostatic model assessment of insulin resistance and an oral glucose tolerance test–based index (Matsuda insulin sensitivity index). Insulin secretion was measured using the area under the curve of insulin to glucose during the first 30 minutes of the oral glucose tolerance test and using the area under the curve of insulin to glucose over 2 hours.

Results  Of 630 children evaluated at baseline (mean [SD] age, 9.6 [0.9] years; 54.4% male; 56.2% normal weight, 19.2% overweight, and 22.7% obese), 564 were evaluated at 2-year follow-up. Adiposity and changes in adiposity were the central predictors of insulin dynamics over time. Every additional 1% of body fat at ages 8 to 10 years decreased insulin sensitivity by 2.9% (95% CI, −3.3% to −2.5%; P < .001) and led to a 0.5% (95% CI, 0.09% to 0.8%; P = .02) increased requirement in the area under the curve of insulin to glucose during the first 30 minutes of the oral glucose tolerance test 2 years later. Higher levels of moderate-to-vigorous physical activity and lower screen time appear to be beneficial to insulin sensitivity in part through their effect on adiposity levels.

Conclusions and Relevance  Adiposity plays a determining role in cardiometabolic health at a young age. Public health strategies that promote healthy body weight, notably physical activity, need to target school-aged and possibly younger children.

Introduction

In adults, extensive evidence shows that physical activity (PA) prevents or delays the transition from prediabetes states, such as impaired fasting glucose and impaired glucose tolerance, to type 2 diabetes with up to 58% decreased progression.1,2 Physical activity influences insulin dynamics by increasing glucose transporter type 4 proteins, muscle mass, and capillary recruitment and proliferation as well as enhancing cellular insulin signaling.3,4 Improved cardiorespiratory fitness is thought to have beneficial effects on insulin homeostasis through increased muscle respiratory capacity, improved mitochondrial function, and enhanced Krebs cycle and beta oxidation pathways.5,6 Moreover, adipokines, secreted by adipose tissue, are thought to have a central role in insulin homeostasis.7-10 Conversely, little is known regarding mechanisms underlying the emerging associations between sedentary behavior and insulin dynamics. However, the impact of PA, fitness, and sedentary behavior on glucose homeostasis remains poorly understood in children and adolescents. Recent data from American children estimate that the prevalence of prediabetes has increased by 87% during a 5-year-period,11 and Canadian figures put obese children and adolescents at highest risk for prediabetes.12 Moreover, only 4% to 7% of Canadian children and adolescents are currently meeting PA guidelines.13 In addition, 31% of Canadian children and 69% of adolescents exceed the recommended maximum of 2 hours of screen time (ST) daily.13 Several cross-sectional studies have reported conflicting results for the association of PA,14-16 fitness,17-20 and sedentary behavior21,22 with insulin sensitivity. Only 2 longitudinal studies have examined these associations in children. Jago et al23 reported that decreased PA was associated with decreased insulin sensitivity during a 6-year period in both boys and girls. Similarly, Telford et al24 found that increased PA and fitness were associated with improved insulin sensitivity during a 2-year period but in boys alone. In these studies, however, only fasting-based indices of insulin sensitivity, reflecting predominantly hepatic insulin sensitivity, were examined and insulin secretion was not considered. Neither study examined the influence of sedentary behavior.

Understanding the link between adiposity, PA, sedentary behavior, fitness, and abnormal glucose metabolism is essential to the development of effective prevention strategies. This project builds on our previous work that found lifestyle habits to be associated with insulin dynamics cross sectionally25,26 in the Quebec Adipose and Lifestyle Investigation in Youth (QUALITY) cohort, a cohort of children with a family history of obesity. The objectives of the current analysis are to determine the extent to which these potentially modifiable risk factors predict insulin sensitivity and insulin secretion during a 2-year period in the QUALITY cohort and to determine whether these relationships differ by sex.

Box Section Ref ID

Key Points

  • Question: How do physical activity, fitness, and screen time influence insulin sensitivity and secretion over 2 years in children?

  • Findings: Adiposity and changes in adiposity were the central predictors of insulin dynamics over time. Every additional 1% of body fat at ages 8 to 10 years decreased insulin sensitivity by approximately 3% and led to a 0.5% increased requirement in first-phase insulin secretion 2 years later.

  • Meaning: Higher moderate-to-vigorous physical activity and lower screen time appear to be beneficial to insulin sensitivity over time, in part through their effect on adiposity levels.

Methods
Cohort

Data from the baseline evaluation (July 2005 to December 2008) and first follow-up assessment (July 2007 to March 2011) of the QUALITY cohort were used.27 The QUALITY study is an ongoing longitudinal study of white children aged 8 to 10 years at recruitment, designed to study the natural history of obesity and the development of type 2 diabetes and cardiovascular disease in children.27 The current analysis was completed in October 2015.

At baseline, 630 families were evaluated; of these, 564 families were followed up 2 years later. All participants underwent the following assessments at both times: blood samples, height and weight, body composition, pubertal stage, questionnaires on lifestyle habits, oral glucose tolerance test (OGTT), fitness, and PA using accelerometers. Written informed assent and consent were obtained from all participants and their parents. This study was approved by the ethics boards of the Centre Hospitalier Universitaire Sainte-Justine and Université Laval.

Assessments

The participants’ PA was assessed during a 7-day period using an Actigraph LS 7164 activity monitor. Accelerometry data were downloaded as 1-minute epochs and underwent standardized quality control and data reduction procedures28; participants with a minimum of 4 days with a minimum of 10 hours of wear time were retained for analyses.26 Nonwear time was defined as any period of 60 minutes or more of 0 counts, accepting 1 to 2 consecutive minutes where count values were higher than 0 and lower than or equal to 100.29 Moderate-to-vigorous PA (MVPA) was computed by adding the total minutes spent daily in moderate PA and in vigorous PA and averaging over the total number of valid days of wear.30,31 Screen time was assessed by interviewer-administered questionnaire, documenting daily hours of television viewing and leisure computer or video game use; average daily ST was tabulated.

Fitness was estimated using peak oxygen consumption during an adapted standard incremental exercise test32 on an electromagnetic bicycle to volitional exhaustion with indirect calorimetry measurements throughout the test. Peak oxygen consumption was considered as a true maximum value if at least 1 of the following criteria was attained: (1) a respiratory exchange ratio (carbon dioxide production to oxygen consumption) higher than 1.0; or (2) a heart rate of 185 beats/min or higher.33 Peak oxygen consumption was expressed as a function of lean body mass.

All participants underwent a 2-hour OGTT after a 12-hour overnight fast. Blood samples were collected at 30-, 60-, 90-, and 120-minute intervals after an oral glucose dose of 1.75 g/kg of body weight (maximum 75 g). Plasma insulin was measured using the ultrasensitive Accessimmunoassay system (Beckman Coulter, Inc), which has no cross-reactivity with proinsulin or C-peptide.34 Plasma glucose concentrations were computed on the Beckman Coulter Synchron LX20 automat using the glucose oxidase method. Analyses were performed in batches at the Centre Hospitalier Universitaire Sainte-Justine Clinical Biochemistry laboratory twice monthly.

Fasting-based indices (homeostatic model assessment of insulin resistance [HOMA-IR]) and OGTT-derived indices (Matsuda insulin sensitivity index [Matsuda ISI]) were used to measure insulin sensitivity. The HOMA-IR index was used as a measure of fasting insulin resistance.35 Matsuda ISI, derived from the OGTT, is computed as 10 000/√[(fasting glucose × fasting insulin) × (mean OGTT glucose × mean OGTT insulin)].36 We have previously validated both measures of insulin sensitivity.37 The OGTT-derived measures of insulin secretion included the ratio of the area under the curve (AUC) of insulin to the AUC of glucose during the first 30 minutes (AUC I/Gt30min) of the OGTT (first-phase insulin secretion) and the ratio of the AUC of insulin to the AUC of glucose during the full 2 hours (AUC I/Gt120min) of the OGTT (second-phase insulin secretion). The AUC I/Gt30min correlates well with the acute insulin response to glucose in healthy children.38

Body composition was determined with dual-energy x-ray absorptiometry (Prodigy Bone Densitometer System, DF+14664; GE Lunar Corp).39 We expressed adiposity using the fat mass index (total fat mass in kilograms divided by height in meters squared) and percentage of body fat.

Age- and sex-adjusted body mass index (calculated as weight in kilograms divided by height in meters squared) z scores were calculated based on the Centers for Disease Control and Prevention growth charts.40 A trained nurse evaluated children’s pubertal development according to Tanner stages.41,42 Children were classified as prepubertal (Tanner stage 1) or pubertal (Tanner stages 2-5).

Statistical Analysis

Means (with standard deviations), medians (with interquartile ranges), or proportions were used to describe characteristics of the participants at baseline and follow-up. Multiple imputation via chained equations was used to account for missing data (IVEware software; Institute for Social Research, University of Michigan).43,44 We used unadjusted and multivariable linear regressions to examine the associations between PA, fitness, sedentary behavior, and insulin sensitivity and secretion. The association between adiposity and insulin sensitivity was statistically significantly nonlinear in some models.45 However, given that other estimates remained unchanged and that the overall shape of the association was generally linear (as well as for ease of interpretation), we report the results from the linear regressions for insulin sensitivity. On the other hand, the association between insulin secretion and insulin sensitivity was significantly nonlinear and was accounted for by using second-degree fractional polynomials.45

Due to reduced PA in winter,46 we adjusted for season (November through March vs April through October) in models that included PA. Potential confounders included sex, age, Tanner stage, and percentage of body fat. Outcome variables had nonsymmetric distributions and were log transformed (100 × lnVAR); thus, β coefficients represent the percentage of change in the outcome associated with a 1-unit increase in the exposure.47

We first modeled how baseline lifestyle behaviors predicted adiposity 2 years later, and then examined how they predicted insulin sensitivity and secretion 2 years later. Specifically, we tested whether lifestyle variables or attributes were directly associated with insulin sensitivity or secretion, or whether these associations were mediated by adiposity. Mediating effects were tested using the method proposed by Baron and Kenny.48 Finally, we examined how changes in lifestyle behaviors predicted change in insulin sensitivity or secretion across the 2-year period. We tested for effect modification by sex for PA, fitness, ST, and adiposity.

The α level (2-sided) was set at .05. All analyses were performed using SAS version 9.2 statistical software (SAS Institute, Inc).

Results
Participant Characteristics

Participant characteristics are outlined in Table 1. A total of 630 children were evaluated at baseline (mean [SD] age, 9.6 [0.9] years; 54.4% male). Sixty-six children were lost to follow-up during the 2-year period. Those lost to follow-up had higher body fat, lower insulin sensitivity, and higher insulin secretion at baseline than those retained. More than half of the participants were of normal weight for their age and sex at both times, while approximately 20% were overweight and another 20% were obese. Adiposity tracked across the 2-year period, with a correlation coefficient of 0.90 (Table 1). Lifestyle behaviors and fitness showed more variability across time, with correlation coefficients ranging from 0.42 for ST to 0.57 for PA. Overall, PA declined during the 2-year period in these children, with ST and fitness increasing slightly (Table 1). In keeping with the established decline in insulin sensitivity during puberty, indices of insulin sensitivity decreased during the 2-year period, while measures of insulin secretion increased (Table 1). We observed a hyperbolic association between Matsuda ISI and AUC I/Gt30min and AUC I/Gt120min, as expected.

PA at Baseline and Adiposity 2 Years Later

We examined how lifestyle habits at ages 8 to 10 years predicted adiposity 2 years later and found that every additional 10 min/d of MVPA was associated with 3.5% lower body fat at ages 10 to 12 years (95% CI, −5.1 to −1.9; P < .001), even after adjusting for fitness and ST (data not shown). We also found that every 1-hour increase in ST at baseline predicted a 2.9% increase in body fat (95% CI, 1.1 to 4.8; P = .002), even after adjusting for PA and fitness. Given that these analyses and all subsequent analyses were similar when using the fat mass index as a measure of adiposity, we present results using percentage of body fat herein.

Adiposity at Baseline and Insulin Sensitivity 2 Years Later

Adiposity appeared to be the most important predictor of insulin sensitivity across time. For every additional 1% of body fat at baseline, HOMA-IR increased by 3.2% (95% CI, 2.8% to 3.6%; P < .001) (eTable in the Supplement) and Matsuda ISI decreased by 2.9% (95% CI, −3.3% to −2.5%; P < .001) (Table 2) after adjusting for PA, fitness, ST, and covariates.

Adiposity Partially Mediates the Effect of PA and ST

The Figure illustrates the total effect of PA on insulin sensitivity. For every additional 10 minutes of MVPA daily at baseline, insulin sensitivity increased by approximately 4.8% (95% CI, 2.5% to 7.1%; P < .001) (Table 2). Similarly, every additional hour of ST daily was associated with a decrease in insulin sensitivity of approximately 4.5% (95% CI, −7.0% to −2.1%; P < .001) (Table 2). The relationship between lifestyle behavior and insulin sensitivity was partially mediated by adiposity, with the direct effects of PA and ST attenuated when adjusting for adiposity (Figure).

Adiposity at Baseline and First-Phase Insulin Secretion 2 Years Later

Adiposity appeared to be the most important predictor of first-phase insulin secretion across time. For every additional 1% of body fat at baseline, AUC I/Gt30min increased by 0.5% (95% CI, 0.09% to 0.8%; P = .02) (Table 3) after adjusting for PA, fitness, ST, and covariates.

While fitness predicted a 0.7% decrease (95% CI, −1.2% to −0.1%; P = .02) in AUC I/Gt120min during the 2-year period after adjusting for PA, ST, and adiposity in the case-complete analyses (Table 4), this did not reach statistical significance in the imputed data sets.

Changes in Adiposity Over Time and Changes in Insulin Sensitivity and Secretion

We also attempted to examine how changes in lifestyle habits predicted change in insulin dynamics and found that an increase in adiposity predicted deterioration in insulin sensitivity using both measures (HOMA-IR and Matsuda ISI) and increased first-phase insulin secretion (AUC I/Gt30min) over 2 years (data not shown). We found that for every 1-hour increase in ST during the 2 years, AUC I/Gt30min increased by 1.2 units (P = .005). Moreover, every 1-unit increase in peak oxygen consumption over the 2 years predicted a 0.4-unit decrease in AUC I/Gt120min (P = .04). However, these models did not exhibit adequate fit. The estimates were very small and adjusted R2 values were extremely low.

No Significant Differences Between Boys and Girls Overall

Finally, all the reported associations herein did not differ between boys and girls.

Discussion

To our knowledge, this is the first study to examine how lifestyle habits and attributes jointly predict insulin sensitivity and insulin secretion in youth. This work provides evidence that in 8- to 10-year-old children with an obese parent, baseline adiposity and changes in adiposity remain the central predictors of insulin sensitivity and insulin secretion (first phase) over time. Nonetheless, lifestyle habits also play an important role, particularly in insulin sensitivity: in the 8- to 10-year-old child, higher MVPA and lower ST predict better insulin sensitivity 2 years later. These associations appear to be partially mediated by adiposity, consistent with our previous work.26 In our cross-sectional studies, ST appeared deleterious to insulin sensitivity and insulin secretion in girls but not in boys; however, in longitudinal analyses, ST appears deleterious only to insulin sensitivity during the 2-year period, albeit in both sexes.

The exact mechanisms through which ST influences adiposity and insulin sensitivity remain unclear. Screen time is a crude indicator of sedentary behavior, and more detailed and objective measures of sedentary behavior in general and of ST in particular are warranted.49

Fitness appeared to be an independent predictor of second-phase insulin secretion in case-complete analyses (but not in imputed data sets), potentially reflecting higher cellular efficiency. Having a higher peak oxygen consumption reflects an increased ability of the cardiovascular system to deliver oxygen and of the muscular system to consume oxygen, and it appears intimately related to the body’s ability to actively produce insulin after an oral challenge (second-phase insulin secretion). Similarly, when examining change in fitness over 2 years, we found that an increase in fitness was associated with decreasing second-phase insulin secretion requirements (AUC I/Gt120min), although these models exhibited poor fit (discussed later). These findings are consistent with our previous cross-sectional analyses,25 and future studies are required to confirm the influence of fitness on second-phase insulin secretion in childhood.

Overall, our findings suggest that establishing and maintaining a highly physically active lifestyle early in life, while minimizing sedentary behavior (specifically ST), may result in a more favorable cardiometabolic profile as children and adolescents enter puberty.

We also examined how changes in lifestyle habits predicted change in insulin sensitivity and secretion. An increase in adiposity predicted deterioration in insulin sensitivity using both measures (HOMA-IR and Matsuda ISI) and increased requirements of first-phase insulin secretion (AUC I/Gt30min) over 2 years; however, these models did not perform well. Changes in lifestyle habits (with the exception of ST and first-phase insulin secretion) may not have been associated with change in insulin sensitivity and secretion because children did not change their lifestyle habits sufficiently during the 2-year period of observation to substantially affect insulin dynamics. Moreover, any such effects may be minor compared with other important influences in this age group such as puberty. Indeed, only 14% of children changed their activity by more than 30 minutes daily, and only 4% of children changed adiposity by 10% during the 2-year period. Nevertheless, the fact that baseline lifestyle habits predicted insulin sensitivity 2 years later reinforces the notion that setting healthy lifestyle habits early in life might be an effective strategy to maintaining optimal cardiometabolic health later in life.

Our findings are generally in keeping with the sparse data available in the literature. Jago et al23 examined how PA in third grade was associated with HOMA-IR 6 years later in a group of 384 Danish students and found that both overall PA levels and moderate- to vigorous-intensity activity levels were associated with better insulin sensitivity over time. The authors did not include measures of fitness or sedentary behavior in their analyses, however, and they adjusted for waist circumference as a marker of adiposity. Moreover, pubertal stage was not included in their models in this critical period of development wherein even small changes in pubertal stage can have an important influence on insulin sensitivity.50

Telford et al24 examined how changes in PA (measured by step count) and fitness levels were associated with changes in HOMA-IR during a 2-year period in 8-year-old children from the LOOK study. They found that increases in PA and fitness, individually, were associated with better hepatic insulin sensitivity as measured by HOMA-IR. Surprisingly, they did not find an association between changes in adiposity and changes in insulin sensitivity. These discrepancies with the present study might be due to the fact that Telford and colleagues did not consider the combined effect of changes in adiposity, PA, and fitness on insulin sensitivity, nor did they include measures of sedentary behavior. Nevertheless, the pathophysiological mechanisms underlying the association of adiposity with insulin sensitivity,7-9 and to a lesser degree with insulin secretion,10 are well studied in both animal and human models. Our findings confirm the central role adiposity plays in cardiometabolic health, even in childhood.

The QUALITY cohort was designed to study cardiovascular disease risk factors, including abnormal glucose homeostasis, in children at the key time when children transition from childhood to early puberty. The unique nature of our cohort, with its comprehensive and state-of-the art assessment of exposures and outcomes of interest, allowed us to examine how lifestyle habits taken together influence insulin sensitivity and secretion in children. We did not measure insulin sensitivity and secretion using the gold-standard methods, given that these are not feasible in large epidemiologic studies; however, we used measures previously validated in children and adolescents. Our findings apply to white children and adolescents with an obese parent and may not be generalizable to other populations. Furthermore, children lost to follow-up during the 2-year period of this study tended to be more insulin resistant and have higher adiposity than those retained for analyses. Given the significant adverse effect of adiposity on insulin dynamics, it is likely that the associations noted would be stronger had this group been included, but further studies examining these associations specifically in obese children and adolescents might better answer this question.

Conclusions

In conclusion, adiposity plays a critical role in cardiometabolic health at a young age, and public health strategies that promote a healthy body weight through lifestyle habits and likely lower energy intake need to target children early. Establishing and maintaining a physically active lifestyle early in life may be an effective strategy to prevent the development of type 2 diabetes in at-risk children and adolescents.

Back to top
Article Information

Corresponding Author: Mélanie Henderson, MD, FRCPC, PhD, Division of Endocrinology, Centre Hospitalier Universitaire Sainte-Justine, 3175 Côte Ste-Catherine, Montreal, QC H3T 1C5, Canada (melanie.henderson.hsj@gmail.com).

Accepted for Publication: October 25, 2015.

Published Online: February 8, 2016. doi:10.1001/jamapediatrics.2015.3909.

Author Contributions: Dr Henderson had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Henderson, Gray-Donald.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Henderson.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Henderson, Benedetti, Barnett.

Obtained funding: Henderson.

Administrative, technical, or material support: Henderson, Barnett, Mathieu.

Study supervision: Gray-Donald.

Conflict of Interest Disclosures: None reported.

Funding/Support: The QUALITY cohort is supported by grants OHF-69442, NMD-94067, MOP-97853, and MOP-119512 from the Canadian Institutes of Health Research, by grant PG-040291 from the Heart and Stroke Foundation of Canada, and by the Fonds de la recherche en santé du Québec. Drs Henderson and Benedetti are supported by Junior 1 salary awards and Dr Barnett is supported by a Junior 2 salary award from the Fonds de la recherche en Santé du Québec.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: Marie Lambert, MD (July 1952 to February 2012), pediatric geneticist and researcher, initiated the QUALITY cohort. Her leadership and devotion to QUALITY will always be remembered and appreciated. The cohort integrates members of TEAM PRODIGY, an interuniversity research team including Université de Montréal, Concordia University, INRS–Institute-Armand Frappier, Université Laval, and McGill University. The research team is grateful to all the children and their families who took part in this study, as well as the technicians, research assistants, and coordinators involved in the QUALITY cohort project.

References
1.
Knowler  WC, Barrett-Connor  E, Fowler  SE,  et al; Diabetes Prevention Program Research Group.  Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin.  N Engl J Med. 2002;346(6):393-403.PubMedGoogle ScholarCrossref
2.
Pan  XR, Li  GW, Hu  YH,  et al.  Effects of diet and exercise in preventing NIDDM in people with impaired glucose tolerance: the Da Qing IGT and Diabetes Study.  Diabetes Care. 1997;20(4):537-544.PubMedGoogle ScholarCrossref
3.
Honig  CR, Odoroff  CL, Frierson  JL.  Capillary recruitment in exercise: rate, extent, uniformity, and relation to blood flow.  Am J Physiol. 1980;238(1):H31-H42.PubMedGoogle Scholar
4.
Goodyear  LJ, Kahn  BB.  Exercise, glucose transport, and insulin sensitivity.  Annu Rev Med. 1998;49:235-261.PubMedGoogle ScholarCrossref
5.
Ivy  JL.  Role of exercise training in the prevention and treatment of insulin resistance and non-insulin-dependent diabetes mellitus.  Sports Med. 1997;24(5):321-336.PubMedGoogle ScholarCrossref
6.
Earnest  CP.  Exercise interval training: an improved stimulus for improving the physiology of pre-diabetes.  Med Hypotheses. 2008;71(5):752-761.PubMedGoogle ScholarCrossref
7.
Blüher  M.  Adipose tissue dysfunction in obesity.  Exp Clin Endocrinol Diabetes. 2009;117(6):241-250.PubMedGoogle ScholarCrossref
8.
Cusi  K.  The role of adipose tissue and lipotoxicity in the pathogenesis of type 2 diabetes.  Curr Diab Rep. 2010;10(4):306-315.PubMedGoogle ScholarCrossref
9.
Soumaya  K.  Molecular mechanisms of insulin resistance in diabetes.  Adv Exp Med Biol. 2012;771:240-251.PubMedGoogle Scholar
10.
Zhao  YF, Feng  DD, Chen  C.  Contribution of adipocyte-derived factors to beta-cell dysfunction in diabetes.  Int J Biochem Cell Biol. 2006;38(5-6):804-819.PubMedGoogle ScholarCrossref
11.
Li  C, Ford  ES, Zhao  G, Mokdad  AH.  Prevalence of pre-diabetes and its association with clustering of cardiometabolic risk factors and hyperinsulinemia among U.S. adolescents: National Health and Nutrition Examination Survey 2005-2006.  Diabetes Care. 2009;32(2):342-347.PubMedGoogle ScholarCrossref
12.
Lambert  M, Delvin  EE, Levy  E,  et al.  Prevalence of cardiometabolic risk factors by weight status in a population-based sample of Quebec children and adolescents.  Can J Cardiol. 2008;24(7):575-583.PubMedGoogle ScholarCrossref
13.
Active Healthy Kids Canada.  Are We Driving Our Kids to Unhealthy Habits? The 2013 Active Healthy Kids Canada Report Card on Physical Activity for Children and Youth. Toronto, ON: Active Healthy Kids Canada; 2014.
14.
Thomas  AS, Greene  LF, Ard  JD, Oster  RA, Darnell  BE, Gower  BA.  Physical activity may facilitate diabetes prevention in adolescents.  Diabetes Care. 2009;32(1):9-13.PubMedGoogle ScholarCrossref
15.
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(7):506-515.PubMedGoogle ScholarCrossref
16.
Rizzo  NS, Ruiz  JR, Hurtig-Wennlöf  A, Ortega  FB, Sjöström  M.  Relationship of physical activity, fitness, and fatness with clustered metabolic risk in children and adolescents: the European Youth Heart Study.  J Pediatr. 2007;150(4):388-394.PubMedGoogle ScholarCrossref
17.
Kasa-Vubu  JZ, Lee  CC, Rosenthal  A, Singer  K, Halter  JB.  Cardiovascular fitness and exercise as determinants of insulin resistance in postpubertal adolescent females.  J Clin Endocrinol Metab. 2005;90(2):849-854.PubMedGoogle ScholarCrossref
18.
Ruiz  JR, Rizzo  NS, Ortega  FB, Loit  HM, Veidebaum  T, Sjöström  M.  Markers of insulin resistance are associated with fatness and fitness in school-aged children: the European Youth Heart Study.  Diabetologia. 2007;50(7):1401-1408.PubMedGoogle ScholarCrossref
19.
Ball  GD, Shaibi  GQ, Cruz  ML, Watkins  MP, Weigensberg  MJ, Goran  MI.  Insulin sensitivity, cardiorespiratory fitness, and physical activity in overweight Hispanic youth.  Obes Res. 2004;12(1):77-85.PubMedGoogle ScholarCrossref
20.
Lee  S, Bacha  F, Gungor  N, Arslanian  SA.  Cardiorespiratory fitness in youth: relationship to insulin sensitivity and beta-cell function.  Obesity (Silver Spring). 2006;14(9):1579-1585.PubMedGoogle ScholarCrossref
21.
Hardy  LL, Denney-Wilson  E, Thrift  AP, Okely  AD, Baur  LA.  Screen time and metabolic risk factors among adolescents.  Arch Pediatr Adolesc Med. 2010;164(7):643-649.PubMedGoogle ScholarCrossref
22.
Danielsen  YS, Júlíusson  PB, Nordhus  IH,  et al.  The relationship between life-style and cardio-metabolic risk indicators in children: the importance of screen time.  Acta Paediatr. 2011;100(2):253-259.PubMedGoogle ScholarCrossref
23.
Jago  R, Wedderkopp  N, Kristensen  PL,  et al.  Six-year change in youth physical activity and effect on fasting insulin and HOMA-IR.  Am J Prev Med. 2008;35(6):554-560.PubMedGoogle ScholarCrossref
24.
Telford  RD, Cunningham  RB, Shaw  JE,  et al.  Contrasting longitudinal and cross-sectional relationships between insulin resistance and percentage of body fat, fitness, and physical activity in children: the LOOK study.  Pediatr Diabetes. 2009;10(8):500-507.PubMedGoogle ScholarCrossref
25.
Henderson  M, Gray-Donald  K, Rabasa-Lhoret  R,  et al.  Insulin secretion and its association with physical activity, fitness and screen time in children.  Obesity (Silver Spring). 2014;22(2):504-511.PubMedGoogle ScholarCrossref
26.
Henderson  M, Gray-Donald  K, Mathieu  ME,  et al.  How are physical activity, fitness, and sedentary behavior associated with insulin sensitivity in children?  Diabetes Care. 2012;35(6):1272-1278.PubMedGoogle ScholarCrossref
27.
Lambert  M, Van Hulst  A, O’Loughlin  J,  et al.  Cohort profile: the Quebec adipose and lifestyle investigation in youth cohort.  Int J Epidemiol. 2012;41(6):1533-1544.PubMedGoogle ScholarCrossref
28.
Colley  R, Connor Gorber  S, Tremblay  MS.  Quality control and data reduction procedures for accelerometry-derived measures of physical activity.  Health Rep. 2010;21(1):63-69.PubMedGoogle Scholar
29.
Troiano  RP, Berrigan  D, Dodd  KW, Mâsse  LC, Tilert  T, McDowell  M.  Physical activity in the United States measured by accelerometer.  Med Sci Sports Exerc. 2008;40(1):181-188.PubMedGoogle ScholarCrossref
30.
Evenson  KR, Catellier  DJ, Gill  K, Ondrak  KS, McMurray  RG.  Calibration of two objective measures of physical activity for children.  J Sports Sci. 2008;26(14):1557-1565.PubMedGoogle ScholarCrossref
31.
Trost  SG, Loprinzi  PD, Moore  R, Pfeiffer  KA.  Comparison of accelerometer cut points for predicting activity intensity in youth.  Med Sci Sports Exerc. 2011;43(7):1360-1368.PubMedGoogle ScholarCrossref
32.
Skinner  JS, ed.  Exercise Testing and Exercise Prescription for Special Cases: Theoretical Basis and Clinical Application. 3rd ed. Baltimore, MD: Lippincott Williams & Wilkins; 2005.
33.
Docherty  D, ed.  Measurement in Pediatric Exercise Science. Champaign, IL: Human Kinetics; 1996.
34.
Allard  P, Delvin  EE, Paradis  G,  et al.  Distribution of fasting plasma insulin, free fatty acids, and glucose concentrations and of homeostasis model assessment of insulin resistance in a representative sample of Quebec children and adolescents.  Clin Chem. 2003;49(4):644-649.PubMedGoogle ScholarCrossref
35.
Matthews  DR, Hosker  JP, Rudenski  AS, Naylor  BA, Treacher  DF, Turner  RC.  Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man.  Diabetologia. 1985;28(7):412-419.PubMedGoogle ScholarCrossref
36.
Matsuda  M, DeFronzo  RA.  Insulin sensitivity indices obtained from oral glucose tolerance testing: comparison with the euglycemic insulin clamp.  Diabetes Care. 1999;22(9):1462-1470.PubMedGoogle ScholarCrossref
37.
Henderson  M, Rabasa-Lhoret  R, Bastard  JP,  et al.  Measuring insulin sensitivity in youth: how do the different indices compare with the gold-standard method?  Diabetes Metab. 2011;37(1):72-78.PubMedGoogle ScholarCrossref
38.
Henderson  M, Baillargeon  JP, Rabasa-Lhoret  R, Chiasson  JL, Hanley  J, Lambert  M.  Estimating insulin secretion in youth using simple indices derived from the oral glucose tolerance test.  Diabetes Metab. 2012;38(4):309-315.PubMedGoogle ScholarCrossref
39.
Goran  MI.  Measurement issues related to studies of childhood obesity: assessment of body composition, body fat distribution, physical activity, and food intake.  Pediatrics. 1998;101(3, pt 2):505-518.PubMedGoogle Scholar
40.
Centers for Disease Control and Prevention. 2000 Centers for Disease Control and Prevention growth charts. http://www.cdc.gov/growthcharts/. Accessed July 12, 2012.
41.
Marshall  WA, Tanner  JM.  Variations in pattern of pubertal changes in girls.  Arch Dis Child. 1969;44(235):291-303.PubMedGoogle ScholarCrossref
42.
Marshall  WA, Tanner  JM.  Variations in the pattern of pubertal changes in boys.  Arch Dis Child. 1970;45(239):13-23.PubMedGoogle ScholarCrossref
43.
Raghunathan  T, Solenberger  P, Van Hoewyk  J. IVEware: Imputation and Variance Estimation Software user guide. http://www.isr.umich.edu/src/smp/ive/. Accessed May 19, 2008.
44.
White  IR, Royston  P, Wood  AM.  Multiple imputation using chained equations: issues and guidance for practice.  Stat Med. 2011;30(4):377-399.PubMedGoogle ScholarCrossref
45.
Royston  P, Sauerbrei  W.  Multivariable Model-Building: A Pragmatic Approach to Regression Analysis Based on Fractional Polynomials for Continuous Variables. Hoboken, NJ: Wiley; 2008.
46.
Bélanger  M, Gray-Donald  K, O’Loughlin  J, Paradis  G, Hanley  J.  Influence of weather conditions and season on physical activity in adolescents.  Ann Epidemiol. 2009;19(3):180-186.PubMedGoogle ScholarCrossref
47.
Cole  TJ.  Sympercents: symmetric percentage differences on the 100 log(e) scale simplify the presentation of log transformed data.  Stat Med. 2000;19(22):3109-3125.PubMedGoogle ScholarCrossref
48.
Baron  RM, Kenny  DA.  The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations.  J Pers Soc Psychol. 1986;51(6):1173-1182.PubMedGoogle ScholarCrossref
49.
Guy  S, Ratzki-Leewing  A, Gwadry-Sridhar  F.  Moving beyond the stigma: systematic review of video games and their potential to combat obesity.  Int J Hypertens. 2011;2011:179124.PubMedGoogle ScholarCrossref
50.
Hannon  TS, Janosky  J, Arslanian  SA.  Longitudinal study of physiologic insulin resistance and metabolic changes of puberty.  Pediatr Res. 2006;60(6):759-763.PubMedGoogle ScholarCrossref
×