Trends in Adolescent Overweight Perception and its Association with Psychosomatic 1 Health 2002-2014: Evidence from 33 Countries

Perceiving oneself as overweight is common and strongly associated with adolescents’


Conclusions
Evidence is presented which suggests that for adolescent girls in 12 Northern and Western European countries, and for boys in four perceiving oneself as overweight may be increasingly deleterious for psychosomatic health Keywords: body image; body size perception; overweight; adolescents; mental well-being; psychosomatic symptoms; perceived body fatness Implications and Contribution: We use a unique dataset to examine trends in adolescents' body image and mental well-being in 33 countries.We present evidence suggesting that the influence of body image on adolescent well-being is increasing over time.This may play a role in the observed worsening of mental well-being in adolescent girls.

Trends in Adolescent Overweight Perception and its Association with Psychosomatic Health 2002-2014: Evidence from 33 Countries
Whilst several factors contribute to body image satisfaction, self-perception of body weight plays a particularly important role, especially amongst adolescent girls (1).As normal physical development during adolescence involves rapid and conspicuous somatic changes, including weight gain, it is common for adolescents to monitor changes to their weight during this period (2) and make comparisons against peers (3).This age group also tend to compare their weight against body shapes propagated by media outlets (3), which for decades have portrayed a thin body shape as optimal, especially for females (4), while for males a slim but muscular build dominates the media (5).
Adolescents frequently evaluate themselves as overweight relative to either their perception of normal weight or subjective ideal body size (6).These perceptions are common even among those with a healthy body mass index (BMI), with over one quarter of adolescent girls incorrectly judging themselves as overweight (7).The perception that oneself is overweight, whether accurate or not, is associated with deleterious behaviours and outcomes in adolescence including maladaptive weight-loss strategies (7,8) and weight gain (9).
Overweight perceptions are also strongly and consistently associated with reduced subjective well-being among adolescents, especially internalising disorders such as depressive symptoms, anxiety and social withdrawal (7,10).
The subjective well-being of adolescents in many developed nations has worsened over the past two decades, particularly for girls (11).Amongst these findings is evidence that the proportion of adolescents reporting psychosomatic health complaints has increased across Europe and North America (6,12).These changes have been particularly large for 15-year old girls; in many countries the prevalence experiencing more than one weekly health complaint rose by over 15 percentage points between 2002 and 2014, with a sharp increase between 2010 and 2014.It is imperative to investigate potential determinants of these trends as subjective well-being represents a principal influence on illness and disability during adolescence (13), with impacts on long-term health and prosperity (14,15).
It is necessary to consider the role of body weight perception, due to the aforementioned links with subjective well-being in adolescence.The prevalence of overweight perceptions and the concomitant impact on subjective well-being may have increased in recent years for several reasons.Firstly, the global obesity epidemic has increased the salience of weight-related issues including the role of personal responsibility, body weight scrutiny, stigmatization and pressures to maintain a thin body shape (16,17).Secondly, a dramatic increase in adolescents' consumption of digital visual media (18) has facilitated the proliferation of idealised, yet extreme body shapes amongst this age group.Thirdly, over the past decade many countries in Europe and North America have experienced changes in socio-cultural factors known to influence adolescents' body image, particularly family structure and peer and family support (6,19,20).
International variation in the relationship between body image and subjective well-being is likely given significant cross-national differences in both of these measures (6).There is also international variability in the societal and cultural factors that could affect body weight perception and its impact on subjective well-being (6,(16)(17)(18).Using a unique dataset collected from 33 countries by the Health Behaviour in School-aged Children (HBSC) study, we examine the hypothesis that the prevalence of overweight perceptions, and their concomitant impact on psychosomatic complaints has increased amongst adolescents in Europe and North America between 2002 and 2014.Whilst the HBSC study collects data on 11-, 13-, and 15year olds, 15-year olds are focused on in the present study as this age group has seen the greatest deterioration in psychosomatic health in recent years (6).This group are also at particularly high risk of overweight perceptions (6).

Methods
Data from four rounds of the international HBSC study were used, covering the period 2002-2014.HBSC is a cross-sectional study of adolescent health carried out every four years in line with a standardised research protocol which specifies sampling methods and questionnaire content across 44 participating countries (6).For each survey round, countries collect a nationally representative sample of 15-year olds, with the timing of fieldwork arranged to achieve a mean age 15.5.
Participants were recruited via stratified random cluster sampling, with whole school classes as the sampling unit.Adolescents completed questionnaires in classroom settings, and were able to leave any question blank.Questionnaires were translated from English into respective national languages with back-translation checks.Institutional ethical consent was gained in each participating country, with schools and adolescents each giving informed consent.
Participating countries were eligible for the present analysis if they had collected data on body size perception, psychosomatic complaints, height and weight from 15-year olds in the 2002, 2006, 2010 and 2014 HBSC surveys.A total of 220,805 individual participants were recruited by eligible countries, of which 15.0% (N=33,139) were excluded due to missing responses on one or more of the below items.

Overweight Perception
Participants were asked "Do you think your body is: Much too thin, A bit too thin, About the right size, A bit too fat or Much too fat".The latter two response options were recoded as 'perceived overweight'.As perceived underweight is also associated with reduced subjective well-being, especially in boys (21), those responding "about the right size" are utilised as the reference category in regression analyses.

Psychosomatic Complaints
Psychosomatic health complaints are used here as an indicator of subjective well-being.
Participants indicated the frequency with which they had experienced the following eight health complaints over the last six months; "feeling low", "irritability or bad temper", "feeling nervous", "difficulties in getting to sleep", "feeling dizzy", "headache", "stomach ache" and "backache" (0= "Rarely or never", 1= "About every month", 2= "About every week", 3= "More than once a week", 4= "About every day").Responses across all eight complaints were summed to generate a single score between 0 and 32, with higher values reflecting a greater psychosomatic complaint burden.This scale has undergone extensive qualitative (22) and quantitative (23) validation and shows good test-retest reliability (22), unidimensionality (24) and external validity (25,26).

Body Mass Index
Participants self-reported their height and weight, which were used to calculate BMI (kg/m 2 ).
Those with BMI values less than 12 or greater than 45 were considered outliers and excluded from analyses (0.1%, N=155).BMI was used to categorise participants as either overweight (including obese) or not overweight according to age-and gender-appropriate International Obesity Taskforce cut-offs (27).

Statistical Analysis
Analyses were stratified by country to allow international comparison.Analyses were also stratified by gender within country.Dataset weights were applied as appropriate to achieve national representativeness of each country at each time point.
Regression analyses were conducted using the SPSS v.22 complex samples toolkit, allowing shared variance within sampling units to be accounted for.Logistic regression was used to quantify changes in the prevalence of overweight perceptions over time (Tables A1 and A2).
The linear effect of survey year on psychosomatic complaint burden was evaluated using general linear modelling (Tables A3 and A4).The association between perceived overweight and psychosomatic complaints was estimated at each time point using general linear modelling (Tables 1 and 2).This analysis was adjusted for overweight status, to investigate the association between perceived overweight and psychosomatic complaints independent of actual body size.Lastly, general linear modelling was used to test whether the relationship between perceived overweight and psychosomatic complaints (again adjusting for overweight status) has changed over time by including an interaction term between survey year (zeroed on 2002) and perceived overweight (Figures 1 and 2).

Results
Data were available from 187,511 participants (51.6% girls) from 33 countries over the period 2002-2014, after exclusion of countries without four consecutive waves of data, and individuals with missing responses.The number of respondents per country ranged from 832 to 11,815 (median 5,410; see Table A5).Within this sample, 20.9% (20.6, 21.2) of boys and 41.4% (41.0, 41.8) of girls perceived themselves as overweight (± 95% CI).According to self-reported height and weight, 16.7% (16.4,16.9) of boys and 9.7% (9.4,9.9) of girls were classified as overweight or obese.Misperception of overweight status was common, with 10.5% (10.3, 10.7) of boys and 32.9% (32.5, 33.3) of girls classified as normal weight or lower (according to self-reported height and weight) perceiving themselves as overweight.

Trends in Psychosomatic Complaints 2002-2014
In 10 of the 33 countries, a linear increase in boys' psychosomatic complaint burden was seen between 2002 and 2014, with France (b=0.14

Association between Psychosomatic Complaints and Overweight Perception
The coefficients in Tables 1 and 2 represent for boys and girls, respectively, increase in the 32-point psychosomatic symptom score for those that perceive their body is too fat, relative to those that feel their body is 'about right' (adjusting for overweight status).For boys, a positive association was seen in 29 of 33 countries in 2014, such that those perceiving themselves as overweight reported a higher burden of psychosomatic complaints.This association was less widespread across countries prior to 2014; 15  and 2014, relative to those perceiving that their body is 'about right'.For example, Scottish girls feeling their body is overweight have, relative to those feeling their body is 'about right' increased by 0.324 points per annum along the 32-point psychosomatic symptom scale.This reflects an increase equivalent to 12.15% of the entire scale over the period 2002-2014.
As self-reported BMI was used to indicate adolescents' overweight status it is possible that a self-selection bias was introduced.However, the results presented here were largely similar when removing the control for overweight status and reinstating those participants (12.1%,N=25,828) that had failed to report height and/or weight.For boys and girls, the observed changes in the relationship between psychosomatic symptoms and overweight perception were substantively identical when controlling for BMI as a continuous, rather than binary variable.

Discussion
This study presents twelve-year trends (2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014) in perceived overweight and its association with psychosomatic complaint burden among adolescents in 33 countries in Europe and North America.Among boys there was an increase in the prevalence of perceived overweight in one third of countries, particularly in Russia, Estonia and Latvia, where historically the prevalence of boys' overweight perception has been very low (12).However, for the vast majority of countries, there was little change in the already high prevalence of overweight perceptions among girls.Despite this stability, girls remain more likely than boys to believe they are overweight in all observed countries.
In line with recent research (11) widespread increases were seen in adolescents' psychosomatic complaints between 2002 and 2014, particularly for girls, with increases in psychosomatic complaint burden in two-thirds of the observed countries.Given that in 2002 44% of 15-year old girls in Europe and North America already exhibited more than one weekly psychosomatic complaint (12), the magnitude of change is a cause for concern in a number of countries.This is particularly true in Scotland, Ireland and the Netherlands which each saw girls' complaint burden rise in 2014 to over 130% of their respective levels in 2002.
Boys' complaint burden was lower than girls' over this period in the majority of countries, and whilst increases in boys' complaints were seen in some countries, these changes were less widespread and of a smaller magnitude than those seen for girls.As such, the findings of the present study indicate that the gender gap in psychosomatic complaints has widened since 2002 in many countries.
As overweight perceptions are common and psychosomatic complaints are increasingly burdensome for adolescents, changes in the known association between complaint burden and overweight body perception were examined between 2002 and 2014.For girls in 12 out of 33 countries, and for boys in four, the association between overweight perception and adolescents' health complaints strengthened between 2002 and 2014.In these countries, young people that feel that their body is too fat have experienced a relative deterioration in psychosomatic health compared to those that feel that their body is about the right size.For girls there is an apparent geographical divide in the degree of change in this association.
Broadly, countries in Northern and Western Europe (and Canada) have seen a strengthening association between overweight perception and psychosomatic complaints, whereas countries in Southern and Eastern Europe have seen no significant change.This international variation mirrors differences in the trajectory of population-level BMI.

Whereas adult population BMI has increased since the 1980s in Northern and Western
European countries, it has until the past decade remained relatively stable in Southern and Eastern Europe (28,29).Whilst Italy and Belgium saw little change in the association between girls' overweight perception and psychosomatic complaint burden, these countries have also seen relatively little change over time in population BMI, particularly for females (29).This apparent association with population-level BMI trajectory may reflect increases in the salience of obesity and weight-based scrutiny which have accompanied national public health efforts designed to combat long-term population weight gain (30).The absence of an equivalent geographic pattern among boys may indicate that females' psychosomatic health is disproportionately affected by an increase in the public conspicuity of body weight.
International differences in the changing association between adolescents' weight perception and psychosomatic health may also be due to cross-national variation in internet usage.
Countries that witnessed a strengthening relationship tended to be those that embraced the internet at an earlier point in history (31) and those that currently have higher levels of internet (32) and social media usage (31,33).Recent evidence indicates that internet exposure and social media use play a particularly strong role in the development of body image concerns (i.e.internalisation of the thin body ideal, body surveillance and the drive for thinness) among girls, with users being more likely than non-users to exhibit body weight concerns (34).
Observed trends in the basic prevalence of self-perceived overweight may provide insight into the comparative judgements that adolescents make when assessing their own body size, typically in reference to media figures or peers (3).In countries where self-perceived overweight was stable between 2002 and 2014, adolescents' perceptions of what constitutes a desirable weight is unlikely to have changed substantially, given the importance of perceived norms in this context (3).The widespread stability of adolescent girls' perceived overweight status may indicate that same-age peers are a particularly important comparison group for girls in most countries, as between 2002 and 2014 the actual weight of adolescent girls' changed relatively little (6,12,35).In contrast, the stability of girls' self-perceived weight is despite adolescents in Europe and North America being increasingly exposed to unrealistic body shapes propagated by online outlets over the past decade (18).This may indicate that figures propagated by media outlets are a less important comparator group in many of the observed countries.The role of the media is, however, likely to be stronger among populations that have seen increases in the prevalence of perceived overweight (particularly Russia, Estonia and Latvia in the case of boys, and Russia and Ukraine for girls).This is potentially due to low exposure to Westernised body ideals in these countries (36) prior to the recent worldwide proliferation in adolescents' use of digital visual media (18).
Whilst the findings reported here are consistent with a worsening impact of overweight perceptions on girls' psychosomatic health in many countries, it is not possible to make causal inferences given the cross-sectional nature of the HBSC study.A further limitation is that BMI was calculated on the basis of self-reported height and weight, which may result in underestimation.It is possible that the reported association between perceived weight status and psychosomatic complaints would be attenuated when controlling for objective BMIbased overweight status.Additionally, excluding participants that did not self-report BMI potentially introduced a selection bias.However, the results presented here were largely similar when removing the control for BMI-based overweight status and including those that failed to report height and/or weight.Finally, there may exist international differences in the extent to which our measure of perceived body weight, specifically the term "fat" elicited stigma.This may influence the basic prevalence of perceived overweight and its association with psychosomatic symptoms.
Despite these limitations, this study presents a unique cross-national examination of recent trends in adolescents' perception of overweight status, and its association with psychosomatic health.Whilst the prevalence of overweight perceptions remained largely static between 2002 and 2014, the present findings may suggest that such perceptions are increasingly damaging for adolescents' psychosomatic healthfor females in Northern and Western European countries.
The results of this study should be heeded as a cautionary tale as some countries may yet observe a change in the association between overweight perceptions and psychosomatic health.It is possible that recent increases in adult population BMI (28,29), a surge in internet use (18), and increases in the prevalence of overweight perceptions will have deleterious consequences for mental health in these regions, particularly those in Southern and Eastern Europe.
It is important for further research to consider potential mediators of the relationship between overweight perceptions and psychosomatic complaints, and changes in their role over time.
One such mediator may involve maladaptive weight-loss strategies including binge-eating and purging which are likely to be associated with physical pains including headache and stomach ache (37).The present study also highlights that it is critical for future work to consider how to restore objectivity into adolescent body weight perception, and encourage adolescents to recognise positive attributes of their bodies, including strength, fitness and the ability to express oneself through movement.It is also necessary to develop and utilise intervention approaches to incentivise weight loss amongst those that are overweight without damaging self-perception and mental health.For instance a recent physical activity intervention amongst obese adolescents has shown that resistance training can achieve improvements in both body image and mental health (38).
This study indicates that the association between overweight perceptions and psychosomatic complaints increased in many countries between 2002 and 2014, especially for girls in Northern and Western Europe.As such, the current scrutiny of body size and weight may represent an increasing burden on mental health.This burden may extend to physical health, given links between poor subjective well-being and low engagement in health-promoting behaviours (9,39,40).

Figure 1 .
Figure 1.Changes over the period 2002-2014 in the association between perceived

Figure 2 .
Figure 2. Changes over the period 2002-2014 in the association between perceived
of the 33 countries observed no association at one or more time points.Combining data from years between 2002 and 2014, the strongest associations among boys were seen in Russia (b=3.22,