Obesity has been steadily rising among adolescents in the United States (US) in the previous three decades with almost a fifth of all US youth aged 12–19 years now being obese [1]. Deviations from a healthy weight additionally include being underweight, which is also detrimental to health [2]. In the US prevalence of adolescent obesity, overweight, and underweight is an estimated 17.2%, 16.2%, and 3.8% respectively. Obesity is associated with metabolic syndrome and several cancers [3], while underweight is associated with higher rates of respiratory, intestinal, and reproductive problems [4]. Obesity has been found to be more prevalent among males and ethnic minorities, especially Black/African Americans and Hispanics/Latinos [1]. Meta-analyses suggest that current weight change programmes based on diet and exercise education vary in effectiveness and show high heterogeneity in results [5], with comprehensive behavioral family lifestyle interventions showing promising results [6]. Weight status perceptions are strong motivators of intention to lose weight among adults [7, 8] and weight status misperception is a stronger predictor of weight change intentions (WCIs) than actual body weight [9] among adolescents. It has been documented that adolescents who perceive themselves as overweight are more likely to report trying to lose weight and those who perceive themselves as underweight more likely to report trying to gain weight, regardless of their actual weight status [10].

The Centre for Disease Control and Prevention (CDC) estimate that close to one in two high school students in the US are trying to lose weight [11]. Intentions to lose weight are closely associated with unhealthy dieting behaviours (UDBs), including fasting or purging after meals via vomiting and taking laxatives [12]. UDBs have been suggested to be risk factors of several eating disorders (EDs), namely anorexia nervosa and bulimia nervosa, which result in undesirable weight loss [13, 14]. At the same time, individuals with obesity with the intention to lose weight are also more likely to attempt it through UDBs [15]. These UDBs are linked to potentially dangerous outcomes for individuals who are underweight, and are ineffective for long-term weight loss for persons with overweight [16, 17].

Hence monitoring trends and associations of UDBs over time is important to inform public health policies. The most recent data from the Youth Risk Behaviour Surveillance System (YRBSS) have not been used for this purpose [18, 19]. Additionally, the majority of studies have focused on people with overweight and obesity and so there is a paucity of literature analysing effects on underweight people [20, 21]. This study attempts to fill these gaps by using the latest YRBSS data to examine the associations of self-perceptions of weight, weight change intentions (WCIs) and UDBs with sex, age and race, as well as trends of UDBs in American adolescents across the weight spectrum.

This nationally representative study has found that prevalence of Unhealthy Dieting Behaviours is persistently high among US adolescents and was highest among females across all racial groups. Types of UDBs differed by racial groups, although the prevalence of overall UDBs were comparable across the races, and there was a lack of significant differential time trends among races. These findings are largely consistent with other large scale adolescent studies and support current literature in highlighting the sex and racial differences in UDBs [16]. These differences are of concern as UDBs are associated with undesired weight gain and weight loss in individuals with obesity/overweight and underweight respectively, thus exacerbating weight mismanagement. This study also showed that weight status underestimation and desire to gain weight were not protective against UDBs when compared to normal weight accurate weight status perceivers who did not want to do anything about their weight.

UDBs paradoxically lead to further weight gain as short-term UDBs may divert adolescents away from more effective long-term weight management practices [16, 17]. It has been hypothesised that the weight gain is due to the association of UDBs with ineffective weight-loss behaviours [16]. These include skipping breakfast, decreasing fruit and vegetable consumption, and reducing physical activity, all of which have been shown to be linked to weight gain [26–28]. This might be particularly detrimental to adolescents who are overweight since further weight gain may trigger even more UDB practices if desire to lose weight intensifies. As UDBs are predictors of both EDs and weight loss/gain, they are potentially detrimental to adolescents who are both underweight or obese/overweight.

This analysis also found significant racial disparities in UDBs. The generally higher socioeconomic status (SES) of Whites [29] may provide them with greater access to diet PPL than other races due to the cost of those products, compared to fasting and vomiting which involve no monetary cost. These UDB disparities are troubling since fasting and purging behaviours have been reported to be important precursors of anorexia nervosa and bulimia nervosa respectively [30]. Cultural beliefs and body images in different cultures may also be a factor, although we found no racial disparities in the overall prevalence of UDBs. We found that there were no racially differential time trends in UDBs from 1999 to 2013, with the exception of Hispanic/Latino females who had worsening trends in vomiting/taking laxatives compared to Whites. The lack of significant differential time trends among races indicates a persistence in racial disparities, which predominantly disadvantage minority groups.

The association between wanting to lose weight and UDBs was much stronger in females than in males, which may indicate a difference in the magnitude of the impact that weight loss intentions have between sexes. Interestingly, we found that with the exception of fasting in males, all those who wanted to gain weight also displayed higher odds of UDBs compared to the WCI referent. This could be due to adolescent males wanting to gain muscle mass but lose fat mass, resulting in them participating in UDBs to lose fat despite wanting to gain weight overall. However, this result requires further investigation. Overall though, it is consistent with the message that those not intending to do anything about weight are the least likely to exhibit UDBs.

Prior research has shown that in adolescents with overweight/obesity weight loss attempts are less likely among those who underestimate their weight status, compared to those with accurate perceptions [31, 32]. Underestimating the weight status has also been found to be protective against a number of disordered eating behaviours among adolescents with overweight/obesity [33]. Our study included individuals across the weight spectrum and we have further adjusted for WCIs; hence, similar direct comparisons within the group of adolescents with overweight/obesity are not straightforward. However, we found that the desire to lose weight was associated with all UDBs in both males and females, which may highlight a pathway that could explain both our findings and those from the studies mentioned above.

This study found weight status misperceptions in adolescents across all weight statuses and WCIs to be associated with higher UDBs. Apart from a slight decline in diet PPL use, these behaviours have been persistent between 1999 and 2013. Racial disparities in the prevalence of different UDB types were also observed, with no improvements over time. As UDBs have been previously established to increase undesired weight gain and weight loss in individuals who are obese/overweight and underweight respectively, its reduction is likely to address weight mismanagement at both ends. Hence, current trends are concerning, and concerted action will be needed on a number of levels. Public health policies might aim for adolescents to improve weight status perception accuracy to potentially reduce inappropriate weight preoccupation and subsequently UDBs.