http://www.diva-portal.

org

This is the published version of a paper published in International Journal of Obesity.

Citation for the original published paper (version of record):

Hayward, J., Millar, L., Petersen, S., Swinburn, B., Lewis, A. (2014)
When ignorance is bliss: weight perception, body mass index and quality of life in adolescents.
International Journal of Obesity, 38(10): 1328-1334
http://dx.doi.org/10.1038/ijo.2014.78

Access to the published version may require subscription.

N.B. When citing this work, cite the original published paper.

Permanent link to this version:

http://urn.kb.se/resolve?urn=urn:nbn:se:umu:diva-90923
 International Journal of Obesity (2014) 38, 1328–1334 OPEN
© 2014 Macmillan Publishers Limited All rights reserved 0307-0565/14
www.nature.com/ijo

PEDIATRIC ORIGINAL ARTICLE

When ignorance is bliss: weight perception, body mass
index and quality of life in adolescents
J Hayward1, L Millar1, S Petersen1,2, B Swinburn1,3 and AJ Lewis4

BACKGROUND/OBJECTIVES: Body weight is negatively associated with adolescent Health-Related Quality of Life (HRQoL).
Despite this well-established relationship, some adolescents with obesity do not display the expected HRQoL decreases.
This study hypothesised weight perception as a moderator of the association between weight status and adolescent HRQoL.
SUBJECTS/METHODS: Subjects were secondary school students from an obesity prevention project in the Barwon South-West
region of Victoria, Australia, entitled It’s Your Move (N = 3040). Measures included standardised body mass index (BMI-z; World
Health Organization growth standards), weight perception and HRQoL, measured by the Paediatric Quality of Life Inventory. Linear
regression and average marginal effect analyses were conducted on cross-sectional baseline data to determine the significance
of any interaction between weight perception and measured weight status in shaping adolescent HRQoL.
RESULTS: The BMI-z/perceived weight status interaction was significantly associated with adolescent HRQoL outcomes.
Adolescents with BMI z-scores in the overweight/obesity range who perceived themselves as overweight had lower HRQoL than
those who perceived themselves as ‘about right.’ Conversely, adolescents with BMI scores in the lower end of the normal range
or in the thinness range who perceived themselves as underweight had lower HRQoL than those with ‘about right’ perceptions.
CONCLUSIONS: This was the first study to report third-variable impacts of a body-perception variable on the relationship between
adolescent weight status and HRQoL. Adolescents’ weight perceptions significantly moderated the relationship between
overweight/obesity and reduced HRQoL. Adolescents who were outside the normal weight range and misperceived their
objectively measured weight status enjoyed a higher HRQoL than adolescents whose weight perception was concordant with their
actual weight status. These findings suggest that practitioners may need to exercise caution when educating adolescents about
their weight status, as such ‘reality checks’ may negatively impact on adolescent HRQoL. It is suggested that more research be
conducted to examine this potential effect.
International Journal of Obesity (2014) 38, 1328–1334; doi:10.1038/ijo.2014.78

INTRODUCTION adolescent obesity.4 HRQoL is a subjective experience influenced

The negative impact of obesity on the physical and psychiatric by the individual’s unique perspective on their own lifestyle,
health and well-being of individuals has been well documented.1,2 hopes, dreams and past experiences.7 Generally, HRQoL covers at
Adolescent obesity is a chronic condition and a known risk factor least the following three core domains physical, emotional and
for the later development of other diseases.3 Recently, studies social functioning, and well-being.8
have broadened the focus to include physical as well as There is a well-established negative association between body
psychological outcomes associated with obesity in adolescence. weight and HRQoL in adolescents.4,9,10 In rare instances, studies
These studies consistently identify negative associations between have failed to reproduce this association; however, these studies
weight status and subjective measures of functioning and well- have either been underpowered,11,12 have used unconventional
being.4 measures13 or have analysed data from samples with highly
The 2011 Youth Risk Behaviour Surveillance survey from the restricted body mass index (BMI) ranges.11
United States reported high rates of overweight and obesity Despite the apparent strength and consistency of the associa-
among high-school students. The survey found 15.2% of students tion between adolescent obesity and HRQoL, some studies have
to be obese and 13% to be overweight.5 Similar prevalence figures found that some adolescents with obesity do not experience an
have been recorded among Australian adolescents; combined associated decrease in HRQoL, suggesting that there may be
overweight/obesity prevalence was 28% in 2008.6 Obesity during significant moderators of this relationship.14 Identifying individual
this developmental period is associated with increases in blood differences that may protect adolescents from obesity-related
pressure, high cholesterol levels, metabolic risk, cardiovascular reductions in HRQoL could aide in the development of targeted
disease and asthma.1 intervention and in the optimal public health approach to obesity
Recently, studies have gone beyond the medical sequelae to in adolescents.15
investigate outcomes of perceived functional health and well- In this study, we focus on weight perception as a potential
being, or Health-Related Quality of Life (HRQoL), associated with moderator of the relationship between overweight/obesity and

1
WHO Collaborating Centre for Obesity Prevention, Faculty of Health, Deakin University, Geelong, Victoria, Australia; 2Child and Adolescent Psychiatry, Clinical Sciences, Umeå
University, Umeå, Sweden; 3School of Population Health, University of Auckland, Auckland, New Zealand and 4School of Psychology, Faculty of Health, Deakin University,
Geelong, Victoria, Australia. Correspondence: J Hayward, Faculty of Health, WHO Collaborating Centre for Obesity Prevention, Deakin University, Waterfront Campus,
1 Gheringhap Street, Geelong, Victoria 3220, Australia.
E-mail: [email protected]
Received 4 October 2013; revised 14 April 2014; accepted 30 April 2014; accepted article preview online 14 May 2014; advance online publication, 17 June 2014
 Adolescent weight, wellbeing and weight perception
J Hayward et al
1329
HRQoL. Weight perception is defined as the subjective appraisal of Seca, Chino, CA, USA). Weight was measured to the nearest 0.1 kg, using a
actual weight status, and is impacted by social and ethnic TANITA body composition analyser (Model BC 418). BMI was calculated as
factors.16 Weight perceptions may be conceptualised as a person’s weight (kg)/height (m2), and BMI-z was calculated using the World Health
perception of whether they are heavier than (overweight Organization (WHO) Growth Reference 2007.25 The WHO Reference 2007
perception), lighter than (underweight perception) or about equal age-specific BMI cutoffs were also used to classify children’s weight status
as thinness o − 2 s.d., normal weight between − 2 and 1 s.d., overweight
to (‘right weight’ perception) the right body weight.17
between 1 and 2 s.d., and obesity 42 s.d.25
Children with objectively high BMI commonly describe their
own weight inaccurately and avoid using labels that reflect Weight perception. Weight perception was measured using one item from
extreme obesity.18 These individuals are found to often mis- the Adolescent Behaviour, Attitudes and Knowledge Questionnaire, which
perceive their weight as ‘about right’, which may serve to protect was designed for the purpose of this study. Participants responded to the
the adolescent from negative stigma. Cross-cultural studies have question ‘How would you describe your weight?’ by choosing a response
explored the global expansion of weight-related stigma, finding out of the following levels: 1 = very underweight, 2 = slightly underweight,
that ‘fat-stigmatising’ beliefs not only persist among western 3 = about the right weight, 4 = slightly overweight or 5 = very overweight.
societies but also have spread to other cultures that were The five levels were collapsed into three: 0 = about the right weight,
previously considered to hold ‘fat-positive’ beliefs.19 People often 1 = underweight and 2 = overweight. The weight perception item was
designed specifically for this study, and the Adolescent Behaviour,
describe individuals with obesity as personal and social failures,
Attitudes and Knowledge questionnaire was piloted in a sample of 95
and the victims of disease. If an adolescent presents with an students from Australia.24 The pilot sample commented on the compre-
extreme weight status (obesity or extreme thinness), they are hensibility and readability of the questionnaire while the range of
likely to be subject to these stigma in everyday life.20 Accurate responses was checked. The survey was modified in response to this
perception of these weight categories, combined with identifica- feedback.
tion with these stigmatised, non-ideal body types, may lead to
internalisation of the negative thoughts surrounding abnormal Health-Related Quality of Life. HRQoL was assessed using the adolescent
body weight and an associated low HRQoL. HRQoL may therefore form of the Paediatric Quality of Life Inventory (PedsQL 4.0).26,27 The
be highest where adolescents misperceive their objective extreme PedsQL 4.0 comprises 23 items, with responses assessed on a five-point
weight statuses, and lowest where they accurately perceive it. Likert scale ranging from 0 (never a problem) to 4 (almost always a
problem). Items are combined into a global measure of HRQoL,
Previous studies have shown similar effects in Canadian adults, encompassing a physical subdomain (8 items) and a psychosocial
with objectively overweight participants who perceived them- subdomain (15 items), the latter capturing information about emotional,
selves as overweight had poorer self-rated health and life social and school functioning and well-being. The composite and
satisfaction,21 and in Mexican University students, where partici- subdomain scores range from 0 to 100, with higher scores indicating
pants who reported perceptions of overweight had comparatively greater HRQoL. In the interest of parsimony, only global, psychosocial and
low HRQoL.22 physical domains were analysed in this study.
Accordingly, this study examines whether weight perception
moderates the association between BMI-z, and HRQoL. It was Covariates. Covariates were participant’s gender and age. All regression
hypothesised that: models accounted for the clustering of data by school.
1. Adolescents with high BMI-z who accurately perceive their
weight status as overweight/very overweight would report lower Procedure
HRQoL than adolescents with high BMI-z who inaccurately Students completed the questionnaires in class, providing demographic
perceive their weight status as normal. information via paper questionnaire and completing HRQoL with the
2. Adolescents with low BMI-z who accurately perceive their use of Personal Digital Assistant devices (hand-held computers,
weight status as underweight/very underweight would report Hewlett Packard iPAC Pocket PC, Hewlett Packard, Palo Alto, CA, USA).24
lower HRQoL than adolescents with low BMI-z who inaccurately Questionnaires generally took students 30–40 min to complete. Comple-
perceive their weight status as normal. tion of questionnaires was overseen by trained research staff who also
undertook the direct measurement of anthropometric data.

MATERIALS AND METHODS Statistical analysis

Participants All statistical analyses were carried out using the Stata 12 (StataCorp.,
The data analysed in this study were collected during the baseline phase College Station, TX, USA) statistics package. All continuous variables were
of It’s Your Move (IYM), a large-sample, school-based, obesity prevention checked for normality and any responses 43 s.d. from the mean were
intervention programme conducted in 12 secondary schools from the excluded casewise from the relevant analysis. Demographic variables are
Barwon South-West region of Victoria, Australia. IYM aimed to prevent presented using descriptive statistics, and gender differences were tested
childhood obesity by building the capacity of the community to promote using t-tests for continuous variables and χ2 analyses for categorical
healthy eating and physical activity. Data were collected from N = 3040 variables. Some or all data were missing from 92 participants who were
students (original sample N = 5961, response rate 51%) with a mean age of excluded casewise from analysis (3.03% missing data).
14.6 years, between 2005 and 2006. A series of linear regression analyses were devised to examine
The sample mostly contained state schools (n = 7) and also included associations between the main effects and multiplicative interactions
Christian schools (n = 2), Catholic schools (n = 2) and one independent between the independent variables, BMI-z and weight perceptions, and
school.23 Schools involved in the study were similar in size and the dependent variables and PedsQL outcome (global score, physical score
socioeconomic status compared with the remaining 37 schools in the and psychosocial score). Multiplicative models are consistent with the
region.23 Parents or guardians provided written consent before the study, theoretical perspective of the study.28
and participants were provided assent on the day of data collection. Ethics An additional analysis estimated HRQoL differences, or average marginal
approval was granted by the Deakin University Human Research Ethics effects (AMEs), between individuals with different weight perceptions for a
Committee (EC 37-2004). The IYM project was a registered trial (Australian given discrete BMI z-score between − 2 and 3. This analysis generated
New Zealand Clinical Trials Registry Number 12607000257460). AMEs for both overweight perception and underweight perception,
compared against right weight perception for all HRQoL outcomes.
Significant findings in this analysis can be interpreted as a meaningful
Measures HRQoL difference between individuals with a specified BMI z-score who
Body mass index. Anthropometric data were measured by trained perceive their weight as ‘about right’ and individuals of the same weight
research staff by using a standardised procedure.24 Participants’ heights status who perceived themselves to be either overweight or underweight.
were measured to the closest 0.1 cm by using portable stadiometers The differences in the slope of BMI-z over the three levels of the
(Surgical and Medical PE87 or Leicester Height Measure, Seca Model 214, categorical variable (weight perceptions) were tested for significance.

© 2014 Macmillan Publishers Limited International Journal of Obesity (2014) 1328 – 1334
 Adolescent weight, wellbeing and weight perception
J Hayward et al
1330
The predicted margins for the interactions between the three-level AME analysis (Supplementary Information available at the
categorical variable (weight perception) and the continuous variable International Journal of Obesity’s website) estimated HRQoL AMEs
(BMI-z) were calculated and graphed. For all tests of statistical significance, between right weight perceiving individuals and underweight or
an alpha level of 0.05 was adopted. overweight perceiving individuals at discrete BMI z-scores
between − 2 and 3. For the purposes of further interpretation,
graphs of the predicted margins for Global PedsQL score (whole
RESULTS
sample) and Physical PedsQL score (male and female participants
The sample included 3040 secondary school students, surveyed separately) are provided in Figures 1–3.
during the baseline phase of IYM. Demographic characteristics
along with tests of significance for gender differences are reported
in Table 1. All outcome variables were found to differ by gender, Regression analyses
except age, BMI-z and weight status. The regression analyses for Global HRQoL (Table 3) showed that
there was a significant interaction between BMI z-score and
weight perception when participants with underweight percep-
Demographic characteristics
tions were compared with those who perceived themselves as
The levels of concordance between measured weight status and ‘right weight’ (P = 0.04). A significant interaction was also identified
self-reported weight perception are presented in Table 2, together in the physical subscale when participants with overweight
with χ2 analysis to test the association between the two variables. perceptions were compared with those who perceived themselves
Although there was a significant association between measured as ‘right weight’ (P = 0.006).
weight status and weight perception, there remained a clear When observed by gender, significant interactions were observed
proportion of the sample that did not accurately perceive their between females perceiving underweight vs right weight for Global
weight status; overall, 21.8% of participants with a BMI in the HRQoL (P = 0.02), females perceiving underweight vs right weight
normal/thinness range thought they were underweight ( o2% of for Psychosocial HRQoL (P = 0.04), and in males perceiving over-
the sample had a BMI in the thinness range) and 9.4% of normal weight vs right weight for Physical HRQoL (P = 0.006).
weight participants thought they were overweight. The degree of Notably, for each significant interaction between the right
distortion increased substantially for the participants with a BMI in weight perceivers and the relevant comparison group, the
the overweight range, where 48.7% perceived themselves as opposite comparison groups did not reach significance in the
being the right weight. In contrast, almost all the participants regression (that is, for Global HRQoL, BMI z-score × underweight vs
(87.5%) who had a BMI in the obesity category correctly perceived right weight was significant, whereas BMI z-score × overweight vs
themselves to be overweight. The patterns were largely similar right weight was not significant). Inspection of the regression
among males and females, with the notable exception of females coefficients and robust standard error shows that in these cases
whose BMI was categorised as normal/thinness and who more the effect coefficients were lower, and the error is often higher
often perceived themselves as overweight (13.9%) compared with than the significant interaction terms. This suggests that the non-
males (6.0%). significant interactions may have resulted from a lack of statistical
There was a significant association between HRQoL and BMI power for these comparisons.
z-score in this sample, although the strength of the association
was weak by Cohen’s criteria29 (Cohen’s r = 0.07, P o 0.01).
AME analyses—global HRQoL, underweight vs right weight
A significant correlation was also found between HRQoL and
perception
weight perception (r = 0.10, P o 0.01).
To test the hypothesis regarding interactions, we used Examination of AMEs showed that at BMI z-scores in the thinness
regression analyses and AMEs analyses, and results are displayed or the lower end of the normal weight range (BMI-z ⩽ 0),
in Table 3 and in Supplementary Tables. Supplementary participants who perceived themselves as underweight had lower
Information is available at the International Journal of Obesity’s global HRQoL than those who perceived themselves as ‘about
website. right’ (Figure 1). At BMI z-scores above 0, the two weight
perception groups reported similar global HRQoL.
Table 1. Demographic characteristics of study participants
AME analyses—global HRQoL, overweight vs right weight
Variable Total Male Female P perception
(N = 3040) (n = 1706) (n = 1334)
There was a different pattern for those who perceived themselves
M (s.d.) M (s.d.) M (s.d.) to be overweight, when BMI z-scores were in the higher end of the
Age
Height
14.62 (1.38) 14.59 (1.34) 14.65 (1.42) 0.23
164.84 (9.68) 167.50 (10.55) 161.44 (7.10) o0.001
normal range or in the overweight/obesity range (BMI-z ⩾ 0),
Weight 59.61 (13.40) 61.15 (14.33) 57.63 (11.84) o0.001 participants had lower global HRQoL than those who perceived
BMI 21.78 (3.81) 21.59 (3.72) 22.03 (3.91) o0.01 themselves to be ‘about right’ (Figure 1). There was no significant
BMI-z 0.55 (1.04) 0.58 (1.06) 0.52 (1.03) 0.18
global HRQoL difference at BMI z-scores − 2 or − 1. Thus, when
PedsQL perceptions of overweight were concordant with measured
Global 78.77 (10.40) 79.57 (10.28) 77.74 (10.47) o0.001 weight, global HRQoL was lower than in participants of similar
74.08 (12.38) o0.01
Psychosocial
Physical
74.76 (12.33)
86.38 (9.57)
75.30 (12.26)
87.82 (8.90) 84.56 (10.01) o0.001
weight who perceived their weight as being ‘about right’.
Examination of AMEs by gender replicated the above results in
Weight status (3-category) N (%) N (%) N (%) males and females separately.
Thinness/normal 2026 (68.58) 1139 (68.61) 887 (68.55)
Overweight 649 (21.97) 352 (21.20) 297 (22.95)
Obesity 279 (9.44) 169 (10.18) 110 (8.50) 0.50 AME analyses—psychosocial functioning, underweight vs right
Weight perception
weight perception
Underweight 469 (15.88) 312 (18.80) 157 (12.13) For psychosocial functioning, the AMEs showed that participants
About the right weight 1734 (58.70) 979 (58.98) 755 (58.35)
Overweight 751 (25.42) 369 (22.23) 382 (29.52) o0.001
who reported perceptions of underweight had significantly lower
scores than those who reported being about the right weight
Abbreviations: BMI, body mass index; BMI-z, standardised body mass for those in the BMI-z range from − 2 to 0 (lower end of normal
index; M, mean; PedsQL, Paediatric Quality of Life Inventory. Bolded values weight range), but there was no significant difference from BMI
are significant at Po0.01. z-scores of 0–3.

International Journal of Obesity (2014) 1328 – 1334 © 2014 Macmillan Publishers Limited
 Adolescent weight, wellbeing and weight perception
J Hayward et al
1331
AME analyses—psychosocial functioning, overweight vs right participants at BMI z-scores − 2 and − 1 (low end of normal range)
weight perception were not significantly different from those who considered their
Psychosocial scores for adolescents who reported overweight weight to be ‘about right.’ Therefore, when weight perception was
perceptions were significantly lower than those who reported concordant with measured weight, psychosocial scores were
their weight as ‘about right’ at BMI z-scores 0–3 (high end of lower than when participants described their weight as being
normal to overweight/obesity range). Estimated AMEs for these ‘about right.’ Participants who were in the high end of the normal
weight range and reported overweight perceptions also had lower
psychosocial scores. When considered by gender, this pattern was
Table 2. Concordance between measured weight status and self- consistent with both male and female participants.
reported weight perception

Measured weight status Weight perception

AME analyses—physical functioning
For physical scores, examination of AMEs showed that participants
Underweight, Right weight, Overweight, P who described themselves as underweight recorded lower
n (%) n (%) n (%) physical functioning at BMI z-scores − 2 to 1 (lower end of normal
weight range) than those who described their weight as ‘about
All participants (n = 2954)
Normal/thinness (n = 2026) 441 (21.8) 1394 (68.8) 191 (9.4) right.’ For those participants who described themselves as
Overweight (n = 649) 17 (2.6) 316 (48.7) 316 (48.7) overweight compared with those who described themselves
Obesity (n = 279) 11 (3.9) 24 (8.6) 244 (87.5) o0.001
about the right weight, lower physical functioning scores were
Males (n = 1660) recorded at BMI z-scores 1–3 (overweight/obesity range). Again
Normal/thinness (n = 1139) 292 (25.6) 779 (68.4) 68 (6.0) concordance between perceived weight and actual weight was
Overweight (n = 352) 13 (3.7) 181 (51.4) 158 (44.9)
Obesity (n = 169) 7 (4.1) 19 (11.2) 143 (84.6) 0.001 associated with lower physical functioning scores.
There were differences between male and female patterns of
Females (n = 1294) physical functioning scores. Males who perceived themselves as
Normal/thinness (n = 887) 149 (16.8) 615 (69.3) 123 (13.9)
Overweight (n = 297) 4 (1.3) 135 (45.5) 158 (53.2) underweight demonstrated significantly lower physical function-
Obesity (n = 110) 4 (3.6) 5 (4.6) 101 (92.8) o0.001 ing scores at BMI z-scores − 2 to 0 (lower end of normal weight
Bolded values are significant at Po0.01.
range) compared with those who perceived themselves to be
the right weight. However, females perceiving themselves as

Table 3. Regression analysis for BMI z-score × weight perception interaction

PedsQL All Males Females

Coef Robust s.e.m. P Coef Robust s.e.m. P Coef Robust s.e.m. P

Global
BMI z-score − 0.23 0.38 0.56 0.48 0.44 0.29 − 1.12 0.64 0.11

Weight perception (right weight–base)

Underweight − 2.25 0.47 0.001 − 2.24 0.8 0.018 − 1.94 0.68 0.02
Overweight − 2.68 1.1 0.03 − 1.76 1.43 0.24 − 3.29 1.2 0.02

BMI z-score × weight description (right weight–base)

Underweight 0.95 0.4 0.04 0.37 0.71 0.6 1.73 0.66 0.02
Overweight − 0.72 0.81 0.39 − 1.7 0.98 0.11 0.39 1.15 0.74

Psychosocial
BMI z-score − 0.33 0.44 0.47 0.42 0.54 0.45 − 1.23 0.7 0.1

Weight description (right weight–base)

Underweight − 2.58 0.52 o 0.001 − 2.41 0.87 0.02 − 2.48 0.65 0.003
Overweight − 4.1 1.43 0.01 − 2.96 1.76 0.12 − 4.8 1.51 0.009

BMI z-score × weight description (right weight–base)

Underweight 1.2 0.57 0.06 0.66 0.89 0.47 1.84 0.82 0.04
Overweight − 0.04 0.93 0.97 − 0.97 1.15 0.41 0.88 1.27 0.5

Physical
BMI z-score 0.12 0.22 0.57 0.41 0.33 0.24 − 0.28 0.44 0.53

Weight description (right weight–base)

Underweight − 1.7 0.41 0.002 − 2.28 0.62 0.004 − 0.55 0.82 0.52
Overweight − 0.69 0.71 0.35 − 0.6 0.97 0.55 − 0.66 0.78 0.41

BMI z-score × weight description (right weight–base)

Underweight 0.37 0.48 0.45 0.4 0.74 0.59 0.49 0.59 0.42
Overweight − 2.04 0.59 0.006 − 2.46 0.72 0.006 −1.43 0.71 0.07
Abbreviations: BMI, body mass index; BMI-z, standardised body mass index; Coef, coefficient. Bolded values are significant at Po 0.05.

© 2014 Macmillan Publishers Limited International Journal of Obesity (2014) 1328 – 1334
 Adolescent weight, wellbeing and weight perception
J Hayward et al
1332

Figure 1. Plotted marginal effect between categorical weight Figure 3. Plotted marginal effect between categorical weight
perception and continuous BMI z-score and Global HRQoL. The perception and continuous BMI z-score and Physical HRQoL (female
figure shows the different effects of weight status on adolescents participants). The figure shows the different effects of weight status
who have different perceptions of their weight. The outcome in this on female adolescents who have different perceptions of their
figure is Global HRQoL, measured in both male and female weight. The outcome in this figure is Physical HRQoL. The unbroken
participants. The unbroken line represents adolescents who line represents adolescents who perceive themselves to be the ‘right
perceive themselves to be the ‘right weight,’ whereas the long- weight,’ whereas the long-dashed line represents adolescents
dashed line represents adolescents who perceive themselves to be who perceive themselves to be ‘overweight’ and the short-dashed
‘overweight’ and the short-dashed line represents adolescents who line represents adolescents who perceive themselves to be
perceive themselves to be ‘underweight’. ‘underweight’.

the current study investigated weight perception as a potential

moderator of the association between BMI z-score and HRQoL in
adolescence, with consideration of gender differences. It was
hypothesised that adolescents whose weight perceptions were
concordant with an extreme objective weight status would report
lower HRQoL than their peers who reported discordant weight
perceptions.
To date, no studies have identified perception-related
moderators of the relationship between adolescent obesity
and HRQoL. The findings of this study partially supported the
hypothesised moderation effect, suggesting that the association
between obesity and HRQoL in adolescence significantly differs
according to weight perception. Regression analyses suggested
that the hypothesised interaction effect was significant only in
Figure 2. Plotted marginal effect between categorical weight females for global HRQoL and psychosocial functioning, and in
perception and continuous BMI z-score and Physical HRQoL (male males for physical functioning, whereas examination of AMEs
participants). The figure shows the different effects of weight status supported the existence of the interaction effect in both males
on male adolescents who have different perceptions of their weight. and females for global HRQoL and psychosocial functioning,
The outcome in this figure is Physical HRQoL. The unbroken line
represents adolescents who perceive themselves to be the ‘right differing by gender only for physical functioning.
weight,’ whereas the long-dashed line represents adolescents Plotted estimates of the AMEs of the interaction terms revealed
who perceive themselves to be ‘overweight’ and the short-dashed that, among adolescents, those who accurately perceive their
line represents adolescents who perceive themselves to be objective overweight or obesity have lower global, psychosocial
‘underweight’. and physical HRQoL than those who perceive themselves as being
‘about the right weight.’ Adolescents who perceived themselves
as overweight and were at the higher end of the normal BMI
underweight showed no significant difference in physical z-score range (that is, z-scores 0–1) also appeared to have lower
functioning compared with those who perceived themselves as HRQoL than those who perceived themselves as about the right
the right weight for any discrete BMI z-score. weight. Results suggested that the gap in HRQoL for these
These results show that adolescents who are objectively at the adolescents may increase as objective BMI z-score increases. In the
lower end of the BMI z-score spectrum and describe themselves as lower BMI z-score range, there was no detectable difference
underweight differ according to gender. Perceptions of under- between the HRQoL of adolescents who perceived themselves as
weight are associated with lower physical functioning in male but ‘about right’ vs overweight.
not in female participants. The plotted AMEs of the interaction are Among adolescents who had BMI z-scores in the lower end of
presented in Figure 2 (males) and Figure 3 (females). the normal weight range (that is, z-scores − 2 to 0), those who
Effect size analysis showed that the interaction models used perceived themselves as underweight had lower HRQoL than
explained between 1.94 and 4.9% of the variance in HRQoL scores. those who perceived themselves as ‘about right’ weight, although
examination of gender differences on physical functioning
revealed that this effect was evident for males only. Again, results
DISCUSSION suggest that the HRQoL divide between these groups may widen
This study aimed to build upon current understanding of the as BMI z-score decreases.
relationship between adolescent obesity and HRQoL. In order to The findings of this study support the argument that adolescent
understand individual differences that might attenuate the males must not be ignored in body image and perceptions
negative association between overweight/obesity and HRQoL, research. In this sample, the proportions of female participants

International Journal of Obesity (2014) 1328 – 1334 © 2014 Macmillan Publishers Limited
 Adolescent weight, wellbeing and weight perception
J Hayward et al
1333
who were objectively in the normal/thinness weight category, but Data analysed in this study were cross-sectional that prevent
who described themselves as either underweight or overweight, conclusions being drawn about causality. Although there is an
were similar (16.8 and 13.9%, respectively). By comparison, males association between awareness of obesity and lower HRQoL, these
in this category described themselves more often as underweight data do not allow a conclusion to be made regarding whether
(25.6%) than overweight (6.0%). This finding may be explained by awareness of obesity leads to decreased HRQoL, or low HRQoL
previous literature that found that a high proportion of adolescent predisposes an adolescent to be more cognisant of their true
males wish to increase the size of their muscles.30 The under- weight status. Further studies involving longitudinal cohorts
weight perceptions of these students may reflect a desire to should investigate these alternatives.
increase body weight by accumulating lean muscle mass. There is In the context of an increasingly overweight society, it is
a significant body of evidence identifying points of difference important to understand individual differences that may influence
between adolescent male and female body ideals;31,32 however, in the degree to which adolescents with overweight and obesity
some areas of research there remains an exclusive focus experience diminished HRQoL. This study highlights the impor-
on females. This female-centric focus has fostered a gap in tance of weight perception as one of these differences, and
research that examines both male and female participants highlights the possibility that obesity awareness may lead
simultaneously.33 This study suggests that females and males individuals to internalise negative stigma. Adequate social support
are both affected by body-perception issues relating to weight. during this critical developmental period may be important for
Future efforts to address adolescents’ weight perception and body improving the long-term outcomes for these adolescents.
image problems should not focus solely on females.
In this sample, adolescents whose weight perceptions matched
their objective overweight or underweight reported the lowest CONFLICT OF INTEREST
HRQoL. This may suggest that the effects of actual weight status on The authors declare no conflict of interest.
adolescent HRQoL involve social mechanisms. As adolescents
become aware of their extreme weight status, HRQoL may be
impacted by the internalisation of the various stigma that adolescents ACKNOWLEDGEMENTS
attribute to different or non-ideal bodies.19,20 This highlights the We would like to thank the many people involved in the Pacific OPIC Project
including co-investigators, other staff and postgraduate students and partner
importance of adequate support networks for adolescents during this
organisations, and especially the schools, students, parents and communities. We
critical developmental period. Where adolescents become aware of
also thank Lucy Busija for her assistance in interpreting the regression results. The
their objective weight status, there must be support available to funding for the project was from the Victorian Department of Health, the National
offset the associated reductions in HRQoL. Health and Medical Research Council (in conjunction with the Health Research
Two further implications for practise arose from this study. First, Council (New Zealand) and the Wellcome Trust (UK) as part of their innovative
programs aiming to educate adolescents about healthy weight must International Collaborative Research Grant Scheme), and AusAID.
consider the potential negative effects of altering adolescent weight
perceptions. Second, obesity prevention efforts should consider
whether it is appropriate to specifically target obese or at-risk REFERENCES
adolescents, when population or community interventions may 1 Pulgarón ER. Childhood obesity: a review of increased risk for physical and
serve as a more indirect path to health promotion, without the risk of psychological comorbidities. Clin Ther 2013; 35: A18–A32.
damaging adolescent HRQoL. Many of the determinants surrounding 2 Schienkiewitz A, Mensink GBM, Scheidt-Nave C. Comorbidity of overweight and
obesity in a nationally representative sample of German adults aged 18-79 years.
adolescent weight may be environmental, and therefore largely
BMC Public Health 2012; 12: 658–658.
beyond the control of individual adolescents. Population-based 3 Neef M, Weise S, Adler M, Sergeyev E, Dittrich K, Körner A et al. Health impact in
programmes allow for promotion of healthy lifestyle and environ- children and adolescents. Best Pract Res Clin Endocrinol Metab 2013; 27: 229–238.
mental change across the community, making behaviour change 4 Tsiros MD, Olds T, Buckley JD, Grimshaw P, Brennan L, Walkley J et al.
easier, and more socially desirable for at-risk individuals.34 Health-related quality of life in obese children and adolescents. Int J Obes 2009;
Data collected in the IYM project were strong in its use of 33: 387–400.
directly measured anthropometry, self-reported HRQoL measured 5 Eaton DK, Kann L, Kinchen S, Shanklin S, Flint KH, Hawkins J et al. Youth Risk
with a valid and reliable instrument and self-reported information Behavior Surveillance — United States, 2011. Surveill Summ 2012; 61(SS04):
1–162.
on weight perception. There is a well-documented tendency for
6 Millar L, Kremer P, de Silva-Sanigorski A, McCabe M, Mavoa H, Moodie M et al.
individuals to under-report weight and over-report height when Reduction in overweight and obesity from a 3 year community-based interven-
self-report BMI is adopted.35 Missing data was also minimal, with tion in Australia: The “It’s Your Move!” project. Obes Rev 2011; 12(s2): 20–28.
no more than 3.03% of cases missing from any of the regression 7 Eiser C, Morse R. The measurement of quality of life in children: past and future
models. Schools involved in the IYM project were representative of perspectives. J Dev Behav Pediatr 2001; 22: 248–256.
other schools in their region. 8 De Civita M, Regier D, Alamgir AH, Anis AH, FitzGerald MJ, Marra CA. Evaluating
This study was limited in interpretation of the interaction effect health-related quality-of-life studies in paediatric populations: some conceptual,
identified. The discussions of Baron and Kenny36 of moderation methodological and developmental considerations and recent applications.
PharmacoEconomics 2005; 23: 659–685.
methodology states that to maximise the interpretability of an
9 Griffiths LJ, Parsons TJ, Hill AJ. Self-esteem and quality of life in obese children and
interaction, independent variables should not be correlated; adolescents: A systematic review. Internat J Pediatr Obes 2010; 5: 282–304.
however, weight perception was weakly correlated to PedsQL 10 Ul-Haq Z, Mackay DF, Fenwick E, Pell JP. Meta-analysis of the association between
outcomes in this sample. Although this does not violate the body mass index and health-related quality of life among children and
conditions for a moderation effect, it suggests that the effect may adolescents, assessed using the pediatric quality of life inventory index. J Pediatr
contain additional complexity beyond that considered in this 2013; 1622: e1.
analysis. Effect sizes in the analysis were also small, ranging 11 Ingerski LM, Janicke DM, Silverstein JH. Brief report: quality of life in overweight
between 1.94 and 4.9%. This effect size range may suggest that youth--the role of multiple informants and perceived social support. J Pediatr
the significance of the findings is because of the large sample Psychol 2007; 32: 869–869.
12 Dalton WTdee Schetzina KE, Pfortmiller DT, Slawson DL, Frye WS. Health
used. The study was further limited by the low number of behaviors and health-related quality of life among middle school children in
participants who had objective thinness, and some non-significant Southern Appalachia: data from the Winning with Wellness Project. J Pediatr
results may be a reflection of this lack of thin participants. Results Psychol 2011; 36: 677–686.
of analyses at this end of the weight distribution should therefore 13 Sato H, Nakamura N, Sasaki N. Effects of bodyweight on health-related quality
be interpreted with appropriate caution. of life in school-aged children and adolescents. Pediatr Int 2008; 50: 552–556.

© 2014 Macmillan Publishers Limited International Journal of Obesity (2014) 1328 – 1334
 Adolescent weight, wellbeing and weight perception
J Hayward et al
1334
14 Modi AC, Loux TJ, Bell SK, Harmon CM, Inge TH, Zeller MH. Weight-specific health- 27 Varni JW, Seid M, Rode CA. The PEDSQL: measurement model for the pediatric
related quality of life in adolescents with extreme obesity. Obesity 2008; 16: 2266–2271. quality of life inventory. Med Care 1999; 37: 126–139.
15 Fairchild AJ, MacKinnon DP. A general model for testing mediation and 28 Kendler KS, Gardner CO. Interpretation of interactions: guide for the perplexed.
moderation effects. Prev Sci 2009; 10: 87–99. Br J Psychiatry 2010; 197: 170–171.
16 Gillen MM, Lefkowitz ES. Body size perceptions in racially/ethnically diverse men 29 Cohen J. A power primer. Psychol Bull 1992; 112: 155–159.
and women: implications for body image and self-esteem. Am J Psychol 2011; 13: 30 McCabe MP, Ricciardelli LA. Body image and body change techniques among
447–467. young adolescent boys. Eur Eat Disord Rev 2001; 9: 335–347.
17 Chang VW, Christakis NA. Self-perception of weight appropriateness in the 31 Maruf FA, Akinpelu AO, Nwankwo MJ. Perceived body image and weight:
United States. Am J Prev Med 2003; 24: 332–339. discrepancies and gender differences among university undergraduates.
18 Saxton J, Hill C, Chadwick P, Wardle J. Weight status and perceived body size in Afr Health Sci 2012; 12: 464–472.
children. Arch Dis Child 2009; 94: 944–949. 32 Stanford JN, McCabe MP. Body image ideal among males and females: sociocultural
19 Brewis AA, Wutich A, Falletta-Cowden A, Rodriguez-Soto I. Body norms and fat influences and focus on different body parts. Health Psychol 2002; 7: 675–684.
stigma in global perspective. Curr Anthropol 2011; 52: 269–276. 33 McCabe MP, Butler K, Watt C. Media influences on attitudes and perceptions
20 Puhl RM, Luedicke J, Heuer C. Weight-based victimization toward overweight toward the body among adult men and women. J Appl Biobehav Res 2007; 12:
adolescents: observations and reactions of peers. J Sch Health 2011; 81: 101–118.
696–703. 34 Kumanyika SK, Obarzanek E, Stettler N, Bell R, Field AE, Fortmann SP, Franklin BA,
21 Herman KM, Hopman WM, Rosenberg MW. Self-rated health and life satisfaction Gillman MW, Lewis CE, Poston WC II et al. Population-based prevention of obesity:
among Canadian adults: associations of perceived weight status versus BMI. Qual the need for comprehensive promotion of healthful eating, physical activity, and
Life Res 2013; 22: 2693–2705. energy balance: a scientific statement from American Heart Association council
22 Hidalgo-Rasmussen CA, Hidalgo-San Martín A, Rasmussen-Cruz B, on epidemiology and prevention, interdisciplinary committee for prevention
Montaño-Espinoza R. Quality of life according to self-perceived weight, weight (formerly the Expert Panel on Population and Prevention Science). Circulation
control behaviors, and gender among adolescent university students in Mexico. 2008; 118: 428–464.
Cad Saude Publica 2011; 27: 67–77. 35 Gorber SC, Tremblay M, Moher D, Gorber B. A comparison of direct vs. self-report
23 Mathews LB, Moodie MM, Simmons AM, Swinburn BA. The process evaluation of measures for assessing height, weight and body mass index: a systematic review.
It's Your Move!, an Australian adolescent community-based obesity prevention Obes Rev 2007; 8: 307–326.
project. BMC Public Health 2010; 10: 448–448. 36 Baron RM, Kenny DA. The moderator–mediator variable distinction in social
24 Mathews LB, Kremer P, Sanigorski A, Simmons AM, Nichols M, Moodie M et al. psychological research: Conceptual, strategic, and statistical considerations. J Pers
Nutrition and Physical Activity in Children and Adolescents: Report 1: Methods Soc Psychol 1986; 51: 1173–1182.
and Tools. Sentinal Site Series. Department of Human Services: Melbourne,
Victoria, Australia, 2009. This work is licensed under a Creative Commons Attribution 3.0
25 de Onis M, Onyango AW, Borghi E, Siyam A, Nishida C, Siekmann J. Development Unported License. The images or other third party material in this
of a WHO growth reference for school-aged children and adolescents. Bull World article are included in the article’s Creative Commons license, unless indicated
Health Organ 2007; 85: 660–667. otherwise in the credit line; if the material is not included under the Creative Commons
26 Varni JW, Seid M, Kurtin PS. PedsQL 4.0: reliability and validity of the Pediatric license, users will need to obtain permission from the license holder to reproduce
Quality of Life Inventory version 4.0 generic core scales in healthy and patient the material. To view a copy of this license, visit http://creativecommons.org/
populations. Med Care 2001; 39: 800–812. licenses/by/3.0/

Supplementary Information accompanies this paper on International Journal of Obesity website (http://www.nature.com/ijo)

International Journal of Obesity (2014) 1328 – 1334 © 2014 Macmillan Publishers Limited