Trends in the perceived body size of adolescent males and females in Scotland, 1990–2014: changing associations with mental well-being

For survey rounds between 1990 and 2010, ethical approval was granted by the Moray House School of Education Ethics Committee, University of Edinburgh. The 2014 survey was approved by the University of St Andrews Teaching and Research Ethics Committee. Prior informed consent was obtained at local authority, school, parent, and pupil levels.

For all seven survey cycles, pupils were asked to report whether they perceived their body to be “Much too thin”, “A bit too thin”, “About the right size”, “A bit too fat”, or “Much too fat”. The former two options were considered here as ‘perceived underweight’ with the latter two as ‘perceived overweight’. Between 1990 and 1998, participants could also select “I don’t think about it”. For the purposes of analysis, “I don’t think about it” and “About the right size” are assumed to indicate the absence of perceived over- or underweight (henceforth ‘about right’).

BMI (kg/m²) was calculated from self-reported weight and height, available from a total of 8160 pupils from 2006 onwards.

All analyses are stratified by gender as the nature of body weight perception differs between boys and girls (Inchley et al. 2016; Voelker et al. 2015). The SPSS (v.22) complex samples toolkit was used to adjust for survey design (clustering of pupils within local authority and school each within survey year). For all regression analyses, data set weights were applied as appropriate per survey to achieve national representativeness with respect to local authority area, school denomination, gender, and school urban–rural classification. Where age categories are combined, additional weighting is applied to ensure equal representation. Survey year is treated as a linear continuous variable throughout.

To examine age differences in adolescents’ weight perceptions, a series of binary logistic regressions were conducted combining data across all survey years (see Supplementary Tables S1 and S2, models 1a and 1b). Separate analyses were conducted with either over- or underweight perception as the binary-dependent variable. These binary variables coded either over- or underweight perception against all other responses. The sole independent variable, age category, was dummy coded as 11, 13, and 15 years, with one category omitted to act as reference, depending on the pairs of age categories being compared. The overall (all age categories combined) effect of survey year on over- and underweight perceptions was tested using separate binary logistic regressions, with survey year (treated as continuous) as the only independent variable (see Supplementary Tables S1 and S2, model 2). To test for change over time in any association between age and over- or underweight perception, further logistic regressions were conducted, allowing for interactions between age and survey year (see Supplementary Tables S1 and S2, models 3a and 3b). Where a significant interaction was observed, model 2 was repeated per age category to further elucidate any effect of survey year.

Associations between body size perception and categorical mental well-being outcomes (confidence and happiness) were tested in a series of ordinal regressions, treating these outcomes as ordinal-dependent variables. For these analyses, the main effects of body size perception (dummy coded as perceived underweight, perceived overweight or ‘about right’, with the latter as reference category) and survey year (centred on 1994) were entered as independent variables. To examine whether any association between well-being outcomes and perceived body size has changed over time, interaction terms between over- and underweight perception and survey year were also entered. The associations between body size perception and psychological symptoms (a continuous variable) were tested using similar structured general linear models. Parameter estimates for all models are given in Table 3.

Overall 10.0% of girls and 15.8% of boys perceived that they are underweight (Table 1) (see Table 2 for the prevalence of perceived underweight disaggregated by gender, age, and survey year and also see Supplementary Table S1 for further details on the models used to examine age differences and time trends in perceived underweight). The likelihood of perceived underweight was equivalent between 11- and 13-year-old girls [χ ² (1) = 0.84, p = .361]. Fifteen-year-old girls exhibited lower levels of perceived underweight relative to both groups of younger girls (both χ ² > 19.51, p < .001). Among girls, the prevalence of perceived underweight reduced slightly over time [F (1, 1617) = 9.86, p = .002], with no relative change between age groups (all interactions F < 0.08, p > .774)

Among boys, the likelihood of perceived underweight increased with age (all comparisons χ ² > 10.08, p < .002). No effect of survey year was seen on perceived underweight among boys [F (1, 1617) = 1.052, p = .305] with no relative change in perceived underweight between age groups (all interactions F < 2.97, p > .085).

Overall, 42.5% of girls and 24.1% of boys perceived themselves as overweight (Table 1) (see Table 2 for the prevalence of perceived overweight disaggregated by gender, age, and survey year and also see Supplementary Table S2 for further details on the models used to examine age differences and time trends in perceived overweight). The likelihood of perceived overweight increased with age among girls (all comparisons χ ² > 75.2, p < .001). Since 1990, 11-year-old girls showed a linear decrease in perceived overweight [F (1, 1617) = 34.54, p < .001]. Overweight perceptions among 13- and 15-year-old girls did not change over this period (both F < 0.44, p > .51).

Among boys, the likelihood of perceived overweight was higher among 13- and 15-year-olds relative to 11-year-olds (both χ ² > 35.54, p < .001). Fifteen-year-old boys exhibited lower levels of perceived overweight relative to 13-year-old boys χ ² (1 = 4.50, p = .034), but whereas 11- and 13-year-old boys did not change over time (both F < 2.57, p > .109), 15-year-old boys showed a linear increase in perceived overweight between 1990 and 2014 [F (1, 1617) = 20.17, p < .001].

The following analyses are further stratified by age group as the analysis above indicates heterogeneity in perceived body size over time.

A possible explanation for relative deterioration of mental well-being outcomes among those with over- or underweight perception is that the actual weight of these individuals has changed over time compared to those perceiving themselves as ‘about right’. To investigate this, time trends in body mass index (BMI) were examined in post hoc analyses (see Supplementary Table S3). The mean BMI of 15-year-old boys perceiving themselves as overweight reduced between 2006 and 2014 relative to those perceiving themselves as ‘about right’ [F (1771) = 4.54, p = .033]. For all other age/gender permutations, no relative changes in mean BMI were seen between 2006 and 2014 for those with over- or underweight perceptions (all F < 1.64, p > .201).

Among boys, there was little change in the prevalence of under- or overweight perceptions, the only substantive change being for 15-year-old boys, who, since 1990, have become more likely to report that they are overweight. Conversely, since 1990, 11-year-old girls have become less likely to report that they are either underweight or overweight. Whilst 13- and 15-year-old girls have also become less likely to perceive themselves as underweight, they have consistently remained at high risk of perceived overweight with little change in the proportion reporting that they are overweight over this period. These findings may suggest that national body image initiatives such as the UK government’s Body Confidence Campaign (established November 2010) have been most effective in widening the perception of normative body size among pre-, but not peri- or post-pubertal girls. Our results may also indicate either that post-pubertal boys are increasingly sensitive to cultural influences on body image perception or that the nature of cultural norms is changing for this age group of males.

Corroborating prior research (Cash and Fleming 2002; Atlantis and Ball 2008; Williams and Currie 2000), we demonstrate that perceived overweight is strongly associated with reduced confidence and happiness, and increased psychological symptoms for boys and girls at all observed ages. Consequently, the high and largely unabating prevalence of perceived overweight among young people in Scotland is concerning, with currently over 40% of 11–15-year-old girls and a quarter of boys feeling that they are overweight. Moreover, our findings suggest that the recently observed deterioration in Scottish adolescents’ mental well-being (Currie et al. 2015) may be precipitated in part by a worsening association between mental well-being and perceived overweight, especially among girls. Whilst for some groups, we did not observe evidence of this worsening association across all mental well-being outcomes (predominantly boys; e.g., 11- and 13-year-old boys’ psychological symptoms), this tended to be among groups experiencing lesser deterioration in these outcomes over time (Currie et al. 2015).

Whilst the prevalence of perceived underweight has remained comparatively low since 1990, it too was associated with reduced mental well-being, particularly among boys. These findings are concordant with indications elsewhere (Leone et al. 2005) that muscle dysmorphia is an issue that affects adolescent boys’ mental well-being. We present some evidence that perceived underweight may impose an increasingly negative effect on adolescents’ mental well-being; however, this evidence is much less consistent than that for perceived overweight.

Perceived underweight was associated with increased psychological health symptoms for both boys (except 13-year-olds) and girls. Late pubertal development is generally held to be a factor in adolescent boys’ underweight perceptions (Siegel et al. 1999). It may also be the case that perceived underweight amongst girls has a similar aetiology, arising due to physical discrepancy between self and more physically developed peers (Petersen and Taylor 1980). Further work with cross-sectional and longitudinal data sets will elucidate links between pubertal timing, perceived body size and mental well-being, and whether any change over time in the role of pubertal timing is associated with changes in mental well-being outcomes.

For some groups, well-being outcomes have worsened over time even among those reporting normal body size (e.g., the confidence and psychological symptoms of 15-year-old boys and girls). Whilst there are likely to be determinants of declining mental well-being beyond those concerning body image, it is important to recognise that broader aspects of self-perception may also be relevant in this context. Low levels of overall appearance satisfaction exist among Scottish adolescents, even for those perceiving normal weight. In 2014, 42% of Scottish girls aged 13 and 15 reported that their body is about the right size, but only 12% reported that they have good looks (Currie et al. 2015). To resolve their relative importance, it is necessary for further work to simultaneously examine the influence of weight-dependent and weight-independent appearance concerns on mental well-being outcomes, particularly in Scotland where girls’ self-perceived attractiveness has reduced substantially since 2002 (Currie et al. 2015).

Given the cross-sectional design of the HBSC study, a limitation of this study is that we cannot be certain that the changing association between body weight perception and mental well-being is causative, although our results are consistent with this explanation. There may exist unexamined factors which have changed over the studied time period, (e.g., quality of peer relationships) which confound the apparent effect of time in this analysis. There also may exist factors associated both with perceived body size and mental well-being that are the true drivers of worsening well-being. One such factor is BMI (Gray and Leyland 2008; Quick et al. 2014). Our post hoc analyses suggested that the worsening association between weight perception and mental well-being is not due to those perceiving themselves as overweight experiencing a relative increase in weight. However, data on adolescents’ BMI were only available from 2006 onwards. In addition, we were restricted to self-reported height and weight, with a large amount of missing, and potentially biased data (Elgar and Stewart 2008), which could obscure any relative increase in the weight of those perceiving they are overweight; however, we cannot definitively say that this is the case given the available data. Socioeconomic status is also a potential confound which has previously been related to both body weight perception and mental well-being (Inchley et al. 2016; McLaughlin et al. 2012). Future work in this context is warranted amongst adolescents from nations with positive or negative associations between affluence and perceived overweight, such as Turkey and Iceland, respectively (Inchley et al. 2016).

Whilst there has been little change since 1990 in the prevalence of overweight perceptions, our findings suggest that perceiving oneself as overweight has a larger impact on Scottish adolescents’ mental well-being today than it did 20 years ago. In addition to an increasing impact on psychological well-being, we speculate that perceived overweight may also be increasingly responsible for poor somatic health. Adolescents experiencing overweight perceptions and mental health symptoms are less likely to engage in health-promoting activities, including those important for maintaining a healthy weight (Fulkerson et al. 2004; Loth et al. 2015; Neumark-Sztainer et al. 2006; Robinson et al. 2015). It is possible, then, that the focus on weight-related issues accompanying the global obesity epidemic (Brownell et al. 2010; Taylor 2011) is increasingly responsible for exacerbating, rather than ameliorating weight-related morbidity.

When targeting obesity via public health campaigns, it is crucial to consider how to incentivize weight loss without damaging adolescents’ self-perception (O’Dea 2005). Those individuals who are overweight may be particularly likely to perceive a wide gap between current and ideal body image. Therefore, these individuals may benefit from more achievable weight-loss goals in conjunction with techniques to enhance self-efficacy (Linde et al. 2006) and psychological resilience. Future observation of national and international trends in body perceptions will help evaluate whether contemporary body image campaigns (e.g., the ‘Be Real’ campaign) are effective in both widening the definition of normative body weight and in encouraging positive body image perception among young people, particularly non-overweight individuals considering themselves as overweight.

Adolescence presents particular challenges as a time of marked physical development and identity formation. Efforts are required to help support young people through this transition period to protect and enhance self-perceptions and well-being. It will be valuable to explore potential mediators and moderators of the changing relationship between body image and mental well-being outcomes, such as digital media use, physical activity, bullying, and socio-cultural assets (Fenton et al. 2010) to shape future intervention efforts.