Morbidity Patterns Among the Underweight, Overweight and Obese Between 2 and 18 Years: Population-Based Cross-Sectional Analyses

Home , Underweight

PEDIATRIC ORIGINAL ARTICLE Morbidity patterns among the underweight, overweight and obese between 2 and 18 years: population-based cross-sectional analyses

M Wake1,2,3, SA Clifford2, GC Patton1,2,3, E Waters3,4, J Williams2, L Canterford2 and JB Carlin1,2,3

CONTEXT: No study has documented how symptomatic morbidity varies across the body mass index (BMI) spectrum (underweight, normal weight, overweight and obese) or across the entire child and adolescent age range. OBJECTIVE: To (1) quantify physical and psychosocial morbidities experienced by 2–18-year-olds according to BMI status and (2) explore morbidity patterns by age. DESIGN, SETTING AND PARTICIPANTS: Cross-sectional data from two Australian population studies (the Longitudinal Study of Australian Children and the Health of Young Victorians Study) were collected during 2000–2006. Participants were grouped into ﬁve age bands: 2–3 (n ¼ 4606), 4–5 (n ¼ 4983), 6–7 (n ¼ 4464), 8–12 (n ¼ 1541) and 13–18 (n ¼ 928) years. MAIN MEASURES: Outcomes—Parent- and self-reported global health; physical, psychosocial and mental health; special health- care needs; wheeze; asthma and sleep problems. Exposure—measured BMI (kg m À 2) categorised using standard international cutpoints. ANALYSES: The variation in comorbidities across BMI categories within and between age bands was examined using linear and logistic regression models. RESULTS: Comorbidities varied with BMI category for all except sleep problems, generally showing the highest levels for the obese category. However, patterns differed markedly between age groups. In particular, poorer global health and special health-care needs were associated with underweight in young children, but obesity in older children. Prevalence of poorer physical health varied little by BMI in 2–5-year-olds, but from 6 to 7 years was increasingly associated with obesity. Normal-weight children tended to experience the best psychosocial and mental health, with little evidence that the U-shaped associations of these variables with BMI status varied by age. Wheeze and asthma increased slightly with BMI at all ages. CONCLUSIONS: Deviation from normal weight is associated with health differences in children and adolescents that vary by morbidity and age. As well as lowering risks for later disease, promoting normal body weight appears central to improving the health and well-being of the young.

International Journal of Obesity (2013) 37, 86–93; doi:10.1038/ijo.2012.86; published online 12 June 2012 Keywords: overweight; underweight; comorbidity; child; adolescent; cross-sectional studies

INTRODUCTION functioning domains vary relatively little with increasing body 10,11 Child and adolescent obesity is a priority health concern mass index (BMI) throughout childhood and adolescence. internationally,1 as it strongly predicts adult obesity,2 which is However, older children with high BMI tend to have poorer 10,11 clearly associated with a heavy health burden.3 However, the physical and social functioning with these associations range of physical and psychosocial health problems that may be marginal in overweight children, modest in community samples associated with overweight and obesity during childhood and of obese children and very marked in help-seeking tertiary clinical 10–12 adolescence has not been confirmed or quantified. Such samples. information must be considered alongside projections of future No study has comprehensively documented the symptomatic adult outcomes4 to understand the total burden of child and physical or mental health problems that might prompt individuals, adolescent obesity in developed countries. families or health professionals to treat or manage BMI in children To date, the literature has largely focused on asymptomatic and adolescents. Health problems putatively related to high BMI 13,14 15 complications (such as insulin resistance and hypertension),5,6 include asthma, sleep problems and special health-care 16–18 16,19–21 associated health-risk behaviours (such as dieting and eating- needs, all of which are very prevalent and incur high 16,21 disordered patterns)7 and self-perceptions (such as body image levels of health burden and cost. However, the extent to and self esteem)8 in school-aged children and adolescents. which they may cluster among obese children and at what ages An emerging but piecemeal literature suggests that health- remains unclear, and all these findings need to be extended related quality of life (HRQoL) is not affected by overweight/ and compared in population studies that span the entire child and obesity in 4–5-year-olds,9 and that HRQoL’s emotional and school adolescent age range.

1Royal Children’s Hospital, Melbourne, Victoria, Australia; 2Murdoch Childrens Research Institute, Melbourne, Victoria, Australia; 3Department of Paediatrics, The University of Melbourne, Melbourne, Victoria, Australia and 4School of Population Health, The University of Melbourne, Melbourne, Victoria, Australia. Correspondence: Professor M Wake, Centre for Community Child Health, Royal Children’s Hospital, Flemington Road, Melbourne, Victoria 3052, Australia. E-mail: [email protected] Received 8 November 2011; revised 12 April 2012; accepted 18 April 2012; published online 12 June 2012 Comorbidities of child underweight and overweight M Wake et al 87 A further issue is that the epidemiology of childhood BMI may participant was excluded to constitute a 13–18-year age band (n ¼ 928). be changing across its entire range. Underweight may also be on The baseline characteristics of those retained and lost to follow-up at Wave the rise in preschoolers (Wake et al, under preparation), primary 3 were similar with regard to gender and socio-economic status (Socio- school children22 and adolescents,23 the net effect being reduced Economic Indexes for Areas Disadvantage Index, see below), but those lost had less educated mothers, were older, had higher BMI Z-score and a proportions of normal-weight children. Other than the small 33 24 higher proportion were overweight or obese. numbers with anorexia nervosa, little is known about the health Both studies were approved by the relevant Ethics Committees, with status of thin children and adolescents. In Dutch 5–14-year-olds, HOYVS also approved by relevant education authorities of participating no association was found between mental health and BMI status schools. Written informed consent was provided by parents in both 25 including underweight, whereas obesity, but not underweight, studies, and also by adolescents in wave 3 of HOYVS. was associated with poorer global and physical health, more Anthropometric measures (primary exposure) were taken by trained health visits and more school absenteeism.26 field workers using similar protocols in both studies. Weight was measured The advent of standardised cut-off points for childhood in light clothing using digital scales to the nearest 50 g in LSAC (Salter underweight27 using the same metric as the International Australia (Springvale, Victoria, Australia) code 79985 and HoMedics Obesity TaskForce (IOTF) overweight/obesity classification28 now (Melbourne, Victoria, Australia) digital BMI bathroom scales) and to the nearest 100 g in HOYVS (Tanita (Tokyo, Japan), Model 1597 in 2000, Model makes it possible to study how common morbidities may vary THD-646 in 2005). Height was measured to the nearest 0.1 cm using across the 2–18-year age range and the full BMI spectrum. Here, portable rigid stadiometers (Invicta (Leicester, UK), Model IPO955). Only we address this question drawing on two contemporaneous large- one height measurement was taken in wave 2 of HOYVS. In LSAC and wave scale, population-based Australian longitudinal studies with 3 of HOYVS, height was measured twice and the two measurements comparable physical and psychosocial health measures averaged. If they differed by 40.5 cm, a third measurement was taken; spanning the entire 2–18-year-old age range and the full BMI LSAC used the average of the two closest, whereas HOYVS wave 3 used the 2 spectrum. We aimed to: median of all three measurements. BMI was then calculated (kg m À ) and participants classified as underweight, normal, overweight or obese according to the international age- and gender-specific cutpoints defined 1. Quantify the physical and psychosocial health of underweight, by Cole et al.27,28 overweight and obese children and adolescents aged 2–18 Potential comorbidities (outcomes) are summarised in Table 1 and cover years, compared with their normal-weight peers and global health, physical and psychosocial health status, mental health, 2. Explore whether and how patterns of associations between BMI special health-care needs, asthma and wheeze, and sleep problems. These status and morbidity patterns vary by age. outcomes were selected because of their availability and consistency across all the ages and both studies, and because all are putatively associated with BMI status as outlined in the Introduction. Demographic variables were the child/adolescent’s sex (male/female), age and the Socio-Economic Indexes for Areas Disadvantage Index,34 a MATERIALS AND METHODS measure of neighbourhood disadvantage, for the most recent postcode of Design and sample residence categorised into quintiles. Socio-Economic Indexes for Areas To gain complete age coverage of childhood and adolescence, we used Indexes are standardised scores compiled from census data for geographic data from (i) waves 1 and 2 (2004 and 2006) of the Longitudinal Study of areas to numerically summarise the distribution of Australian social and economic conditions (national mean 1000, s.d. 100; higher values Australian Children (LSAC) and (ii) waves 2 and 3 (2000 and 2005) of the represent greater advantage). Health of Young Victorians Study (HOYVS). LSAC is a national study that recruited 5107 Australian infants aged 0–1 year and 4983 preschool children aged 4–5 years in 2004. This paper Statistical analysis draws on wave 2 infant cohort data (that is, at age 2–3 years, the first wave For each age band, the distribution of comorbidities across the four BMI that included BMI data) and waves 1 (4–5 years) and 2 (6–7 years) data categories was summarised into percentages for categorical variables and from the preschool cohort. LSAC used a two-stage cluster sampling design means for continuous variables and presented in tabular and graphical form. with Australian postcodes as the primary sampling units, stratified by state Regarding aim 1, we examined variation in morbidity level within each of residence and urban versus rural status, and children enrolled on the 29 age band across the four BMI categories, using unadjusted linear Medicare Australia database as the secondary sampling units. Of those regression (continuous outcomes) and logistic regression (categorical who were resident in the sampled postcodes and contactable, response outcomes) methods to assess the evidence for (unstructured) differences rates were 64% for the infant cohort and 59% for the preschool cohort. between categories. In all, 4606 of the infants (90.2%) and 4464 (89.6%) of the preschool Regarding aim 2 (exploring whether and how morbidity patterns vary by children participated in wave 2. The LSAC sample is considered broadly age), we used interaction terms in linear and logistic regression models to representative of the Australian population, although children with highly test for differences in the pattern of variation with BMI category between educated parents are slightly overrepresented in wave 1, whereas children age categories, adjusting for multiple measures on the same individual by in single-parent families, non-English speaking families and families living using robust ‘information sandwich’ standard errors.35 Comparisons were in rental accommodation are underrepresented in wave 1 and have lower 30,31 adjusted for sex and socio-economic indexes for area, but we report the retention in wave 2. simpler unadjusted results as adjustment made no substantial difference, HOYVS was a longitudinal population-based cohort study established 32 because there was no major imbalance on these factors between the BMI in 1997. Sampling and methods have been reported previously. Briefly, categories. Analyses were conducted using Stata release 11.1 (StataCorp 24 elementary schools were selected from across the state of Victoria, (College Station, TX, USA), 2007). Australia (population 4.6 million in 1997), using a stratified two-stage random sampling design based on school education sector (government, Catholic or independent) and school class level. The baseline response rate for students in the first (‘Prep’) through the fourth (grade 3) school year in RESULTS 1997 was 83.2% (1943 out of 2336 identified children, age range: 5.0–10.7 Of those retained within the specified age, BMI was available for years). The achieved sample mirrored Victorian census data for age 4522 (98.2%) children aged 2–3 years, 4934 (99.0%) aged 4–5 distribution, sex, ethnicity (parental county of birth) and proportion of years, 4423 (99.1%) aged 6–7 years, 1540 (99.9%) aged 8–12 years indigenous persons. and 920 (99.1%) aged 13–18 years. Table 2 shows the Children were resurveyed 3 years later (wave 2, 2000) when in grades 3 characteristics of the sample. The prevalence of underweight through 6 (n 1575; response rate 81.0%; age range: 8.4–13.8 years). We ¼ was highest in the toddlers (5.3%) and lowest in the teenagers excluded children with missing age data (n ¼ 8) or aged X13 (n ¼ 26), resulting in 1541 children in the 8–12-year age band. (4.6%), whereas, conversely, obesity was most prevalent in Wave 3 was conducted a further five years later (September 2005– teenagers (6.1%) and least prevalent in toddlers (4.4%). December 2006) when adolescents were in grades 8 through 12 (n ¼ 929; Morbidity levels across the BMI categories within each age band response rate 47.8%; age range: 13.6–19.4 years). One 19-year-old are presented in Table 3 (continuous data outcomes) and Table 4

& 2013 Macmillan Publishers Limited International Journal of Obesity (2013) 86 – 93 Comorbidities of child underweight and overweight M Wake et al 88 Table 1. Measures of potential comorbidities (‘outcomes’) of child and adolescent overweight/obesity

Construct Measure Additional information

Continuous outcomes Health status Pediatric Quality Of Life Inventory Version 4.0 23 items. Physical and psychosocial health summary scores have (PedsQL 4.0);48 parent report possible ranges of 0–100, with 100 representing best-possible health48 Mental healtha Strengths and Difficulties Questionnaire (SDQ);49 25 items. Total difficulties score has possible range of 0–40 with parent report (ages 4–5 and 6–7) or self-report higher scores indicative of more difficulties49 (ages 13–18)

Categorical outcomes Global health From the Child Health Questionnaire;50 parent report Single question (‘in general, how would you say your child’s current health is?’ or ‘in general, would you say your teenager’s health is:’); dichotomised as ‘excellent/very good’ vs ‘good/ fair/poor’ Special Children with Special Health-Care Needs Screener Probes whether or not a child/adolescent has any condition health-care 2-item version;51 parent report expected to last for at least 12 months that need or use more care needsb than parents would consider usual; dichotomized as yes/no Asthma/ Questions derived from ISAAC52 and the National LSAC (ages 2–3, 4–5 and 6–7 years): Doctor diagnosed asthma with wheezeb Population Health Survey 1998/1999;53 parent report medication use in the last 12 months). HOYVS wave 3 (age 13–18): (ages 2–3, 4–5 and 6–7) or self-report (ages 13–18) Ever had asthma with medication use in the last 12 months Sleepb Parent report (ages 2–3, 4–5 and 6–7) or Pittsburgh LSAC (ages 2–3, 4–5 and 6–7 years): Single question (‘How much is Sleep Quality Index;54 self-report (ages 13–18) child’s sleeping pattern or habits a problem for you?’); dichotomised into ‘no problem/mild’ vs ‘moderate/severe’ problem. Parent report is an established marker of problematic child sleep patterns and there are strong indications that parents can reliably report a child sleep problem. HOYVS wave 3 (age 13–18 years): Single question (‘During the past month, how would you rate your sleep quality overall (how well do you sleep)?’); dichotomised as ‘fairly good/very good’ vs ‘fairly bad/very bad’55 Abbreviations: HOYVS, Health of Young Victorians Study; LSAC, Longitudinal Study of Australian Children. aData unavailable for ages 2–3 and 8–12 years. bData unavailable for age 8–12 years.

Table 2. Demographic characteristics of children and adolescents with body mass index data

Characteristic Toddlers Preschoolers Early primary Late primary Adolescents (2–3 years)a (4–5 years)b (6–7 years)c (8–12 years)d (13–18 years)e n ¼ 4522 n ¼ 4934 n ¼ 4423 n ¼ 1540 n ¼ 920

Male (%) 51.0 50.8 51.0 50.1 50.7 Age (years), mean (s.d.) 2.8 (0.2) 4.7 (0.2) 6.8 (0.2) 10.8 (1.2) 15.9 (1.2)

BMI status (%) Underweight 5.3 5.0 5.0 4.9 4.4 Normal weight 72.0 74.5 76.3 70.5 69.6 Overweight 18.2 15.3 13.4 19.6 20.0 Obese 4.6 5.2 5.4 5.0 6.1

SEIFA, mean (s.d.) 1009.5 (60.7) 1011.1 (58.7) 1009.9 (62.7) 1025.5 (57.2) —

Maternal education, completed 20.2 27.9 26.5 25.0 — p10 years of school (%) Abbreviation: BMI, body mass index; SEIFA, Socio-Economic Indexes for Areas Disadvantage Index. aLSAC B cohort wave 2. bLSAC K cohort wave 1. cLSAC K cohort wave 2. dHOYVS wave 2. eHOYVS wave 3.

(categorical data outcomes). These tables report (i) P-values from poor physical health (Pediatric Quality of Life Inventory) across tests of association between BMI and morbidity within age bands BMI categories varied with age: although there was little and (ii) P-values for interaction between BMI and age in their difference in physical health between BMI categories amongst association with morbidity. younger children, poorer physical health was associated with There was strong evidence (Po0.001) that the pattern of risk obesity in older children. of poorer global health across BMI categories varied between age Children in the normal-weight group generally had the best groups, with these differing patterns very evident visually in psychosocial (Pediatric Quality of Life Inventory) and mental Figure 1. In the youngest age groups, risk was primarily (Strengths and Difﬁculties Questionnaire) health, with poorer elevated in the underweight category, while as age increased health for the obese category and, to a lesser extent, the an elevated risk of poorer global health progressively emerged overweight and underweight categories. There was no evidence in the obese and to some extent the overweight categories. of differential trends across age bands for either psychosocial There was similarly strong evidence (Po0.001) that the risk of (P ¼ 0.55) or mental (P ¼ 0.51) health.

International Journal of Obesity (2013) 86 – 93 & 2013 Macmillan Publishers Limited Comorbidities of child underweight and overweight M Wake et al 89 Table 3. Morbidity levels (continuous outcomes) by BMI status and age

Measure n BMI status (mean (s.e.)) P-valuea P-valueb

Underweight Normal weight Overweight Obese

Physical health (PedsQL physical summary)c o0.001 2–3 years 3474 81.2 (1.0) 83.1 (0.2) 83.8 (0.4) 82.1 (1.1) 0.03 4–5 years 4171 83.1 (0.9) 82.9 (0.2) 82.6 (0.5) 81.3 (1.0) 0.27 6–7 years 3454 80.0 (1.2) 83.1 (0.3) 81.8 (0.7) 78.3 (1.4) o0.001 8–12 years 1300 88.1 (2.0) 87.7 (0.5) 83.1 (1.0) 76.3 (2.2) o0.001 13–18 years 878 84.6 (2.6) 86.0 (0.6) 82.6 (1.3) 76.5 (2.7) o0.001

Psychosocial health (PedsQL psychosocial summary)c 0.55 2–3 years 3456 80.2 (1.0) 81.2 (0.2) 81.4 (0.5) 80.2 (1.0) 0.48 4–5 years 4171 79.7 (0.9) 80.1 (0.2) 79.2 (0.5) 78.0 (0.9) 0.04 6–7 years 3454 76.9 (1.0) 78.3 (0.3) 77.3 (0.6) 74.7 (1.2) 0.003 8–12 years 1299 75.5 (2.1) 77.6 (0.5) 76.3 (0.9) 74.0 (1.9) 0.16 13–18 years 868 77.0 (2.3) 79.9 (0.6) 78.1 (1.2) 72.5 (2.5) 0.003

Mental health (SDQ total) 0.51 2–3 years — — — — — — 4–5 years 4922 9.7 (0.3) 9.2 (0.9) 9.4 (0.2) 10.2 (0.3) 0.02 6–7 years 4308 7.8 (0.3) 7.8 (0.1) 8.0 (0.2) 8.8 (0.3) 0.04 8–12 years — — — — — — 13–18 years 920 9.5 (0.8) 9.0 (0.2) 10.0 (0.4) 10.5 (0.7) 0.03 Abbreviations: BMI, body mass index; PedsQL; Pediatric Quality Of Life Inventory; SDQ, Strengths And Difﬁculties Questionnaire. SDQ total data not available for 2–3- or 8–12-year-olds. aP-value from omnibus (3 degrees of freedom) test for differences between BMI categories, obtained via linear regression. bP-value for interaction test based on four categories of BMI crossed with age categories, using linear regression. cLower LSAC numbers reﬂect missing data, with PedsQL collected by leave-behind questionnaire rather than in the interview itself.

There was a suggestive evidence (P ¼ 0.03) of different patterns overweight and obesity in older children. Poorer physical health of age-related risk of special health-care needs across weight varied little by BMI in children aged 2–5 years, but from 6 to 7 categories but the variation in prevalence of this outcome was years was increasingly associated with overweight and obesity. difficult to interpret. In the two preschool age groups, under- Wheeze and asthma increased slightly with increasing BMI weight children had the highest rates of special health-care needs, category at all ages. Associations with psychological health were but in the school-aged children, underweight children had the weaker but, at all ages, the best psychosocial and mental health lowest rates. In all the groups, special needs were more prevalent was experienced by normal-weight children, and the worst by in overweight and obese than in normal-weight children. obese children. The pattern of association between BMI categories and both wheeze and asthma were similar at every age, that is, increasing Strengths of the study modestly across the BMI categories, with underweight children These analyses utilised large, contemporary Australian population- experiencing the lowest incidence. There was little evidence of an based samples spanning ages 2–18 years, all studied within a age interaction for either wheeze (P ¼ 0.88) or asthma (P ¼ 0.20). narrow time window (2000–2006) and all recruited using clustered Rates of sleep problems were not clearly associated with BMI geographic sampling. Height and weight were measured using status in any age band. virtually identical protocols in the three cohorts, with BMI We ran several post-hoc analyses (data not shown) to examine classified using a single metric for the four categories and the the possibility that the differing patterns of morbidity by age entire age spectrum. This high degree of consistency provides might reflect sampling and/or secular trend differences in the confidence that the changing patterns of association are not due cohorts. First, the older (HOYVS) children were from the single to measurement error in the primary exposure of BMI. Similarly, state of Victoria, whereas the younger (LSAC) children were a there was a very strong consistency across the entire age range in national Australian sample; conclusions were unchanged when we the standardised outcome measures of potential comorbidity. reran the analyses with the LSAC Victorian subsample only. Second, the HOYVS 8–12-year-old data were collected in 2000, whereas the LSAC data were collected in 2004–2006. To check Study limitations whether secular trend might be an issue, we examined patterns of Five age bands comprised three separate groups of individuals morbidity by BMI for 8–9-year-olds in both HOYVS and the measured once or twice; however, the analytical approach subsequent 2008 LSAC wave. Again, patterns were very similar. ensured that there was a benefit from these repeated measures on the same individuals, in the form of enhanced precision for the comparisons between age groups. Loss to follow-up resulted in some underrepresentation of families with a single parent, DISCUSSION mothers with lower education levels, non-English speaking Principal findings background or living in rental accommodation in the later waves; Patterns of comorbidity varied across BMI categories for all although morbidity prevalence rates may not fully generalise to outcomes except sleep problems, with obese children and adole- these groups, substantial numbers in these groups were still scents generally showing the highest levels. However, patterns present in the sample and internal associations between weight varied greatly both by morbidity type and age. In particular, and morbidities are unlikely to be biased. All associations were poorer global health and special health-care needs were cross-sectional, limiting the scope for conclusions about causality. associated with underweight in young children, but with It remains possible that some of the age differences reflected

& 2013 Macmillan Publishers Limited International Journal of Obesity (2013) 86 – 93 Comorbidities of child underweight and overweight M Wake et al 90 Table 4. Morbidities (categorical data) by BMI status and age

Measure n BMI status (% (s.e.)) P-valuea P-valueb

Underweight Normal weight Overweight Obese

Global health—good/fair/poor o0.001 2–3 years 4522 20.6 (2.6) 13.9 (0.6) 13.0 (1.1) 14.0 (2.4) 0.03 4–5 years 4934 22.1 (2.7) 11.8 (0.5) 11.5 (1.2) 14.3 (2.2) o0.001 6–7 years 4423 22.1 (2.8) 10.9 (0.5) 11.5 (1.3) 16.9 (2.4) o0.001 8–12 years 1303 8.5 (3.7) 9.8 (1.0) 15.8 (2.3) 31.2 (6.0) o0.001 13–18 years 854 10.8 (5.6) 11.1 (1.2) 17.8 (2.9) 28.9 (6.3) 0.002

Special health-care needs—yes 0.03 2–3 years 4304 14.9 (2.4) 10.6 (0.6) 12.7 (1.2) 13.2 (2.4) 0.09 4–5 years 4889 17.2 (2.4) 12.1 (0.5) 14.0 (1.3) 20.5 (2.5) o0.001 6–7 years 4274 10.7 (2.1) 14.5 (0.6) 16.8 (0.2) 16.1 (0.2) 0.15 8–12 years — — — — — — 13–18 years 840 7.9 (4.4) 14.9 (1.5) 19.2 (3.1) 19.2 (5.5) 0.27

Wheeze—yes 0.88 2–3 years 4515 24.4 (2.8) 23.6 (0.7) 27.5 (1.6) 32.5 (3.3) 0.01 4–5 years 4915 12.7 (2.1) 14.3 (0.6) 14.7 (1.3) 21.5 (2.6) 0.01 6–7 years 4423 16.7 (0.3) 15.3 (0.6) 15.2 (1.5) 20.7 (2.6) 0.17 8–12 years — — — — — — 13–18 years 920 20.0 (6.4) 18.0 (1.5) 22.3 (3.1) 25.0 (5.8) 0.40

Asthma—yes 0.20 2–3 years 4494 7.6 (1.7) 10.1 (0.5) 15.2 (1.3) 13.5 (2.4) o0.001 4–5 years 4917 11.9 (2.1) 14.5 (0.6) 14.9 (1.3) 19.1 (2.5) 0.13 6–7 years 4413 10.5 (2.1) 15.4 (0.6) 16.1 (1.5) 19.4 (2.6) 0.07 8–12 years — — — — — — 13–18 years 913 2.5 (2.5) 16.2 (1.5) 19.8 (3.0) 23.6 (5.8) 0.07

Sleep problem—yes 0.77 2–3 years 4521 10.9 (2.0) 12.2 (0.6) 11.2 (1.1) 17.9 (2.7) 0.07 4–5 years 4933 13.5 (2.2) 13.2 (0.6) 14.1 (1.3) 16.3 (2.3) 0.52 6–7 years 4420 5.0 (1.5) 5.6 (0.3) 5.4 (0.9) 9.8 (1.9) 0.06 8–12 years — — — — — — 13–18 years 918 12.5 (5.3) 13.9 (1.4) 15.9 (2.7) 12.5 (4.4) 0.88 Abbreviation: BMI, body mass index. aP-value from omnibus (3 degrees of freedom) test for differences between BMI categories, obtained via logistic regression. bP-value for interaction test based on four categories of BMI crossed with age categories, using logistic regression. Special health-care needs, wheeze, asthma and sleep data not available for 8–12-year-olds.

secular trend and/or the use of more than one cohort; however, very young children, emerges convincingly only in the school our post-hoc analyses suggest that our findings would probably years and then steadily strengthens with age. Our findings are have been similar had we been able to study our 11 000 consistent with the more fragmented literature reporting that participants in a single national sample, combining all morbidities overweight/obese older children and adolescents report poorer and measured BMI across the age range of 2–18 years. global health,26 greater primary health-care needs26 and higher Although our primary exposure (BMI) was measured, the prevalence of wheeze/asthma36 than children of normal weight. outcome measures of potential comorbidities were all either The lack of association between sleep problems and BMI status in parent- or self-reported. This is appropriate for subjective this study may reflect either limitations with our sleep measure constructs, such as HRQoL and special health-care needs, but it (problems rather than duration) or a true lack of association, with is possible that, had we measured asthma and sleep objectively, recent research producing conflicting findings as to whether short our conclusions might have differed. However, it would be difficult sleep duration does15,37,38 or does not39,40 predict childhood if not impossible to objectively measure these consistently given obesity. the limitations of their measurement in field studies, the very large Underweight was associated with substantially greater morbid- number of participants and the wide age range (including very ity than obesity in preschoolers. This may well reflect broader young children) that we see as a particular strength of the study. societal stereotypes of health and might partly explain why All our measures are standardised, validated and widely used in parents and practitioners seem less concerned about obesity in epidemiological research. Further, parent- and self-reported their preschool children;41 fear of underweight might actively morbidities are likely to be highly relevant to help-seeking and hinder attempts to address excess adiposity in this age group. thus the health-care burden and costs of BMI-related morbidity in Conversely, underweight school-aged children and adolescents children and adolescents. were among the most healthy in their age groups. Thus, underweight 13–18-year-olds had the least special health-care needs and asthma and the best global health. Their physical and Interpretation in light of other studies mental health was not dissimilar to normal-weight individuals, but Our study confirms previous findings11,12 that obese children they showed slightly lower psychosocial HRQoL from the late experience lower HRQoL than their normal-weight peers, but goes primary school years onwards. The long-term outcomes of further in demonstrating that this association is weak or absent in thinness in healthy individuals are as yet unknown. On the one

International Journal of Obesity (2013) 86 – 93 & 2013 Macmillan Publishers Limited Comorbidities of child underweight and overweight M Wake et al 91

Figure 1. Morbidities by BMI status for the ﬁve age groups. (a) Physical health (PedsQL Physical Summary). (b) Psychosocial health (PedsQL Psychosocial Summary). (c) Mental health (SDQ Total). (d) Global health good/fair/poor. (e) Special health-care needs. (f) Wheeze. (g) Asthma. (h) Sleep problem. P-value for interaction test based on four categories of BMI crossed with age categories, using linear regression. Special health-care needs, wheeze, asthma and sleep data not available for 8–12-year-olds. SDQ total data not available for 2–3- or 8–12-year-olds. SDQ, Strengths and Difﬁculties Questionnaire; PedsQL, Pediatric Quality of Life Inventory.

hand, deliberate caloric restriction in healthy young adults may CONCLUSIONS 42 enhance health and longevity, with trials currently under way. These ﬁndings have obvious implications for policy. In young Conversely, those with underweight in the context of eating children, lack of obesity impacts coupled with heightened concern disorders are known to have worse mental health that continues about underweight is likely to impede efforts to systematically 43 well into young adulthood, and several large longitudinal address early-onset obesity. Conversely, reductions in global and studies have indicated that healthy underweight (baseline BMI physical health are already strongly evident among obese À 2 o18.5 kg m ), as well as obese, adults have higher subsequent adolescents, reinforcing the need for effective preventive strate- 44–46 cardiovascular mortality. gies throughout childhood47 and indeed adolescence.

& 2013 Macmillan Publishers Limited International Journal of Obesity (2013) 86 – 93 Comorbidities of child underweight and overweight M Wake et al 92 The great variation of relationships with BMI status by type of 2 Singh AS, Mulder C, Twisk JW, van Mechelen W, Chinapaw MJ. Tracking of morbidity and by age throughout childhood and adolescence childhood overweight into adulthood: a systematic review of the literature. suggests that a range of different mechanisms are involved. Obes Rev 2008; 9: 474–488. Future longitudinal analyses within these and other cohorts 3 Jia H, Lubetkin EI. The impact of obesity on health-related quality-of-life in the should shed light on the temporal nature of these associations, general adult US population. J Public Health (Oxf) 2005; 27: 156–164. their causal pathways and their specific mechanisms. This informa- 4 Ludwig DS. Childhood obesity—the shape of things to come. N Engl J Med 2007; tion may provide population-based insights into the intricate 357: 2325–2327. emerging balance between human metabolism, growth, health 5 Freedman DS, Katzmarzyk PT, Dietz WH, Srinivasan SR, Berenson GS. Relation of and ageing from the earliest years. body mass index and skinfold thicknesses to cardiovascular disease risk factors in children: the Bogalusa Heart Study. Am J Clin Nutr 2009; 90: 210–216. Taken as a whole, this study shows how profoundly deviation 6 Wilkin TJ, Metcalf BS, Murphy MJ, Kirkby J, Jeffery AN, Voss LD. The relative con- from normal weight is linked to changes in health in children and tributions of birth weight, weight change, and current weight to insulin resistance in adolescents. Promoting a normal body weight has the potential contemporary 5-year-olds: the EarlyBird Study. Diabetes 2002; 51: 3468–3472. to affect not just risks for later life disease but appears central to 7 Puder JJ, Munsch S. Psychological correlates of childhood obesity. Int J Obes improving the health and well-being of the young. (Lond) 2010; 34(Suppl 2): S37–S43. 8 Gibson LY, Byrne SM, Blair E, Davis EA, Jacoby P, Zubrick SR. Clustering of psychosocial symptoms in overweight children. Aust NZ J Psychiatry 2008; 42: 118–125. CONFLICT OF INTEREST 9 Wake M, Hardy P, Sawyer MG, Carlin JB. Comorbidities of overweight/obesity in Australian preschoolers: a cross-sectional population study. Arch Dis Child 2008; The authors declare no conflict of interest. 93: 502–507. 10 Tsiros MD, Olds T, Buckley JD, Grimshaw P, Brennan L, Walkley J et al. Health-related quality of life in obese children and adolescents. Int J Obes (Lond) ACKNOWLEDGEMENTS 2009; 33: 387–400. We thank all the parents and children who took part in LSAC and HOYVS, and the 11 Williams J, Wake M, Hesketh K, Maher E, Waters E. Health-related quality of life of major contributions of all field workers in both studies. This paper uses unit record overweight and obese children. JAMA 2005; 293: 70–76. data from Growing Up in Australia, the Longitudinal Study of Australian Children. The 12 Schwimmer JB, Burwinkle TM, Varni JW. Health-related quality of life of severely study is conducted in partnership between the Department of Families, Housing, obese children and adolescents. JAMA 2003; 289: 1813–1819. Community Services and Indigenous Affairs (FaHCSIA), the Australian Institute 13 Reilly JJ, Methven E, McDowell ZC, Hacking B, Alexander D, Stewart L et al. of Family Studies (AIFS) and the Australian Bureau of Statistics (ABS). The third wave Health consequences of obesity. Arch Dis Child 2003; 88: 748–752. of the Health of Young Victorians Study (HOYVS) was funded by the Australian 14 Flaherman V, Rutherford GW. A meta-analysis of the effect of high weight on National Health and Medical Research Council (NHMRC) Project Grant 334303, and asthma. Arch Dis Child 2006; 91: 334–339. the second wave by grants-in-aid from the National Heart Foundation, Murdoch 15 Chen X, Beydoun MA, Wang Y. Is sleep duration associated with childhood Childrens Research Institute and the Financial Markets Foundation for Children. MW is obesity? A systematic review and meta-analysis. Obesity (Silver Spring) 2008; 16: supported by the NHMRC Career Development Award 546405, GP by the NHMRC 265–274. Senior Principal Research Fellowship 454360 and EW by the Jack Brockoff Foundation 16 Perrin JM, Bloom SR, Gortmaker SL. The increase of childhood chronic conditions and NHMRC Child and Adolescent Obesity Prevention Capacity Building Grant. MCRI in the United States. JAMA 2007; 297: 2755–2759. research is supported by the Victorian Government’s Operational Infrastructure 17 Rimmer JH, Yamaki K, Lowry BM, Wang E, Vogel LC. Obesity and obesity-related Support Program. secondary conditions in adolescents with intellectual/developmental disabilities. J Intellect Disabil Res 2010; 54: 787–794. 18 Yamaki K, Rimmer JH, Lowry BD, Vogel LC. Prevalence of obesity-related chronic AUTHOR CONTRIBUTIONS health conditions in overweight adolescents with disabilities. Res Dev Disabil All the authors had access to the data and take responsibility for the integrity of 2011; 32: 280–288. the data and the accuracy of the data analysis. MW is the study guarantor. 19 Asher MI, Montefort S, Bjorksten B, Lai CK, Strachan DP, Weiland SK et al. MW, EW, GP, JW, Kylie Hesketh and Timothy Olds conceived the third wave of Worldwide time trends in the prevalence of symptoms of asthma, allergic rhinoconjunctivitis, and eczema in childhood: ISAAC Phases One and Three repeat the Health of Young Victorians Study, obtained funding and directed the study. multicountry cross-sectional surveys. Lancet 2006; 368: 733–743. LC carried out the preliminary analyses and SC carried out the final analyses 20 Wake M, Sanson A, Berthelsen D, Hardy P, Mission S, Smith K et al. How Well are under the supervision of JC. MW and SC wrote the paper, with critical input Australian Infants and Children Aged 4 to 5 Years Doing? Findings of the Long- from the other authors. itudinal Study of Australian Children Wave 1. Australian Government Department of Families, Housing, Community Services and Indigenous Affairs: Canberra, Australia, 2008. DISCLAIMER 21 Owens J. Classification and epidemiology of childhood sleep disorders. Prim Care The findings and views reported are those of the authors and should not be 2008; 35: 533–546vii. attributed to FaHCSIA, AIFS or the ABS. The funding bodies had no role in the 22 Martinez-Vizcaino V, Sanchez Lopez M, Moya Martinez P, Solera Martinez M, Notario Pacheco B, Salcedo Aguilar F et al. Trends in excess weight and thinness conduct of the studies or development of this manuscript. The researchers were among Spanish schoolchildren in the period 1992–2004: the Cuenca study. independent of the funders. Public Health Nutr 2009; 12: 1015–1018. 23 Bjornelv S, Lydersen S, Mykletun A, Holmen TL. Changes in BMI-distribution from 1966–1969 to 1995–1997 in adolescents. The Young-HUNT study, Norway. BMC ETHICAL APPROVAL Public Health 2007; 7: 279–284. LSAC: Written informed consent was obtained at wave 1, and the study was 24 Braun DL, Sunday SR, Halmi KA. Psychiatric comorbidity in patients with eating approved by the Australian Institute of Family Studies Ethics Committee. disorders. Psychol Med 1994; 24: 859–867. HOYVS: Each wave was approved by the Royal Children’s Hospital Ethics in 25 Drukker M, Wojciechowski F, Feron FJ, Mengelers R, Van Os J. A community study Human Research Committee and the authorities responsible for each education of psychosocial functioning and weight in young children and adolescents. Int J Pediatr Obes 2009; 4: 91–97. sector, and parents provided written consent (plus written adolescent consent 26 Wijga AH, Scholtens S, Bemelmans WJ, de Jongste JC, Kerkhof M, Schipper M et al. for wave 3). Comorbidities of obesity in school children: a cross-sectional study in the PIAMA birth cohort. BMC Public Health 2010; 10: 184–193. 27 Cole TJ, Flegal KM, Nicholls D, Jackson AA. Body mass index cut offs to REFERENCES define thinness in children and adolescents: international survey. BMJ 2007; 335: 1 Olds T, Maher C. Global trends in childhood overweight and obesity in developed 194–201. countries. In: O’Dea J, Eriksen M (eds). Childhood Obesity Prevention—International 28 Cole TJ, Bellizzi MC, Flegal KM, Dietz WH. Establishing a standard definition for Research, Controversies and Interventions. 1st edn. Oxford University Press: Oxford, child overweight and obesity worldwide: international survey. BMJ 2000; 320: UK, 2010, pp 69–83. 1240–1243.

International Journal of Obesity (2013) 86 – 93 & 2013 Macmillan Publishers Limited Comorbidities of child underweight and overweight M Wake et al 93 29 Soloff C, Lawrence D, Johnstone R. LSAC Technical Paper Number 1: Sample Design. 42 Redman LM, Martin CK, Williamson DA, Ravussin E. Effect of caloric restriction in Australian Institute of Family Studies: Melbourne, Australia, 2005. non-obese humans on physiological, psychological and behavioral outcomes. 30 Mission S, Sipthorp M. LSAC Technical Paper Number 5: Wave 2 Weighting Physiol Behav 2008; 94: 643–648. and Non-response. Australian Institute of Family Studies: Melbourne, Australia, 43 Patton GC, Coffey C, Carlin JB, Sanci L, Sawyer S. Prognosis of adolescent partial 2007. syndromes of eating disorder. Br J Psychiatry 2008; 192: 294–299. 31 Soloff C, Lawrence D, Mission S, Johnstone R. LSAC Technical Paper Number 3: 44 Funada S, Shimazu T, Kakizaki M, Kuriyama S, Sato Y, Matsuda-Ohmori K et al. Wave 1 Weighting and Non-response. Australian Institute of Family Studies: Mel- Body mass index and cardiovascular disease mortality in Japan: the Ohsaki Study. bourne, Australia, 2006. Prev Med 2008; 47: 66–70. 32 Wake M, Salmon L, Waters E, Wright M, Hesketh K. Parent-reported health status 45 Ringback Weitoft G, Eliasson M, Rosen M. Underweight, overweight and obesity as risk of overweight and obese Australian primary school children: a cross-sectional factors for mortality and hospitalization. Scand J Public Health 2008; 36: 169–176. population survey. Int J Obes Relat Metab Disord 2002; 26: 717–724. 46 Flegal KM, Graubard BI, Williamson DF, Gail MH. Cause-specific excess 33 Wake M, Canterford L, Patton GC, Hesketh K, Hardy P, Williams J et al. deaths associated with underweight, overweight, and obesity. JAMA 2007; 298: Comorbidities of overweight/obesity experienced in adolescence: longitudinal 2028–2037. study. Arch Dis Child 2010; 95: 162–168. 47 Waters E, de Silva-Sanigorski A, Hall BJ, Brown T, Campbell KJ, Gao Y et al. 34 Australian Bureau of Statistics. Socio-Economic Indexes for Areas (SEIFA)—Technical Interventions for preventing obesity in children. Cochrane Database Syst Rev 2011, Paper, 2006. ABS: Canberra, Australia, 2008. (12): CD001871. 35 Carlin JB, Wolfe R, Coffey C, Patton GC. Analysis of binary outcomes in 48 Varni JW, Seid M, Kurtin PS. PedsQL 4.0: reliability and validity of the Pediatric longitudinal studies using weighted estimating equations and discrete-time Quality of Life Inventory version 4.0 generic core scales in healthy and patient survival methods: prevalence and incidence of smoking in an adolescent cohort. populations. Med Care 2001; 39: 800–812. Stat Med 1999; 18: 2655–2679. 49 Goodman R. The Strengths and Difficulties Questionnaire: a research note. J Child 36 Noal RB, Menezes AM, Macedo SE, Dumith SC. Childhood body mass index and Psychol Psychiatry 1997; 38: 581–586. 50 Landgraf JM, Abetz L, Ware JE. Child Health Questionnaire (CHQ): A User’s Manual. risk of asthma in adolescence: a systematic review. Obes Rev 2011; 12: 93–104. 37 Lumeng JC, Somashekar D, Appugliese D, Kaciroti N, Corwyn RF, Bradley RH. The Health Institute, New England Medical Centre: Boston, MA, USA, 1996. 51 Bethell CD, Read D, Stein RE, Blumberg SJ, Wells N, Newacheck PW. Identifying Shorter sleep duration is associated with increased risk for being overweight at children with special health care needs: development and evaluation of a short ages 9 to 12 years. Pediatrics 2007; 120: 1020–1029. screening instrument. Ambul Pediatr 2002; 2: 38–48. 38 Marshall NS, Glozier N, Grunstein RR. Is sleep duration related to obesity? A critical 52 Asher MI, Keil U, Anderson HR, Beasley R, Crane J, Martinez F et al. International review of the epidemiological evidence. Sleep Med Rev 2008; 12: 289–298. Study of Asthma and Allergies in Childhood (ISAAC): rationale and methods. 39 Hassan F, Davis MM, Chervin RD. No independent association between Eur Respir J 1995; 8: 483–491. insufficient sleep and childhood obesity in the National Survey of Children’s 53 1998 National Population Health Survey Content for June, 1998. Statistics Canada, Health. J Clin Sleep Med 2011; 7: 153–157. 2000. 40 Sung V, Beebe DW, Vandyke R, Fenchel MC, Crimmins NA, Kirk S et al. Does sleep 54 Buysse DJ, Reynolds 3rd CF, Monk TH, Berman SR, Kupfer DJ. The Pittsburgh Sleep duration predict metabolic risk in obese adolescents attending tertiary services? Quality Index: a new instrument for psychiatric practice and research. Psychiatry A cross-sectional study. Sleep 2011; 34: 891–898. Res 1989; 28: 193–213. 41 Campbell MW, Williams J, Hampton A, Wake M. Maternal concern and 55 Hiscock H, Bayer J, Gold L, Hampton A, Ukoumunne OC, Wake M. Improving infant perceptions of overweight in Australian preschool-aged children. Med J Aust 2006; sleep and maternal mental health: a cluster randomised trial. Arch Dis Child 2007; 184: 274–277. 92: 952–958.

& 2013 Macmillan Publishers Limited International Journal of Obesity (2013) 86 – 93