This study identified a number of personal, behavioral and socio-environmental factors during adolescence, and changes in these factors from adolescence to young adulthood, that are predictive of the incidence of overweight at 10-year follow-up. Among females and males higher levels of body dissatisfaction, weight concerns, unhealthy weight control behaviors (e.g., fasting), dieting, binge eating, weight-related teasing, and parental weight-related concerns and behaviors during adolescence, and/or increases in these factors over the 10-year study period predicted the incidence of overweight. Only a few healthy dietary behaviors (i.e., increased whole grain and vegetable intake) or eating patterns (e.g., eating breakfast on a more regular basis) were found to decrease the odds of becoming overweight 10 years later. For instance, among female participants, increased whole grain intake and regular consumption of breakfast and dinner during adolescence protected against the incidence of overweight, while increases in fast food consumption over the 10-year study period increased their odds of becoming overweight. These findings suggest the importance of intervening early and preventing the endorsement or progression of unhealthy behaviors during the adolescent years, by helping young people feel better about their bodies and avoiding unhealthy weight control behaviors, and by working with parents to provide a home environment in which conversations about weight, in particular weight-related teasing, are minimized.
There were some interesting gender-specific associations between weight-related variables and the incidence of overweight in the 10-year change and baseline models that have important implications for the timing of interventions. For instance, among females, endorsing unhealthy weight control behaviors during adolescence (baseline) and 10-year changes in these behaviors (either starting or continued use of unhealthy weight control behaviors) over time were both predictive of overweight incidence. Among males, 10-year changes in unhealthy weight control behaviors significantly predicted overweight incidence but endorsement of unhealthy weight control behaviors during adolescence was not predictive of overweight incidence in young adulthood. These findings among males may indicate that we are unable to predict if male adolescents will be become overweight based on their adolescent unhealthy weight control behaviors; instead what matters is where these males are ending up with endorsement of these behaviors as a young adult. These findings indicate that regardless of gender, initiating the endorsement of unhealthy weight control behaviors will place individuals at increased risk for weight gain and odds for becoming overweight as a young adult compared to those who never endorse these behaviors, and discontinuing these behaviors during the transition from adolescence to young adulthood may help to prevent the onset of overweight.
Our findings build on previous shorter-term studies showing that weight concerns [35
], body dissatisfaction [36
], and unhealthy weight control behaviors [8
] during adolescence increase risk for weight gain over time. Adolescents with high levels of weight concerns and body dissatisfaction may use weight control behaviors, such as dieting, as a means to reach their ideal body weight or shape [38
]. However, dieting and the use of other restrictive, unhealthy weight control behaviors (e.g., fasting, taking laxatives) may lead to overeating or binge eating and result in weight gain [39
]. For instance, a 4-year longitudinal study of children and early adolescents (ages 6–12) found binge eating and dieting to be strongly predictive of increases in body fat; children who reported binge eating gained, on average, 15% more fat mass compared to children who did not report binge eating [8
]. Thus, findings suggest that unhealthy weight control behaviors are ineffective and potentially harmful strategies. Support is needed to help adolescents adopt healthier eating behaviors that emphasize a non-dieting approach to weight management and to be more accepting of their bodies.
Previous research examining dietary variables have found significant associations between measures of dietary intake such as sugar-sweetened beverage intakes and body weight [40
] although findings are not always consistent nor strong in magnitude [40
]. In the current study, few dietary intake behaviors during adolescence were consistently found across analyses and gender to be predictive of the incidence of overweight 10 years later. When interpreting these results, it is important to consider the methodological challenges in the assessment of dietary intake, particularly the assessment of energy intake. Although analyses of dietary variables in this study were adjusted for caloric intake to help account for reporting bias, residual confounding may still be present, and there may be differences by weight status [42
]. Additionally, there were challenges in examining 10-year changes in dietary intake; however, age-appropriate food frequency questionnaires developed along similar lines were utilized [28
]. Nevertheless, findings from the present study indicate that some healthy eating behaviors (i.e., increased whole grain and vegetable intake) during adolescence are protective against overweight incidence in young adulthood and need to be considered as important aspects in obesity prevention and treatment interventions.
Meal frequency findings were inconsistent between female and male adolescents, but suggest that female adolescents who frequently consume breakfast and dinner are protected against excessive weight gain in young adulthood, which supports other prospective [43
] and cross-sectional studies [45
]. Unexpectedly, among male adolescents, increases in the frequency of lunch and dinner during the transition from adolescence to young adulthood increased their risk of becoming overweight 10 years later. Over this same time period, male adolescents increased their fast food consumption; thus, these additional meals consumed by male adolescents were potentially unhealthy food choices that led to weight gain and obesity risk. These findings suggest the importance of encouraging male adolescents to make healthy meal choices.
Higher levels of weight-related teasing and parental weight-related concerns and behaviors during adolescence and, in some instances, increases in these factors from adolescence to young adulthood were found to be predictive of incidence of overweight 10 years later. These findings suggest that parental weight-related comments and pressures to lose weight during adolescence may be harmful and should be discouraged. Research has shown weight-related teasing during adolescence to be associated with disordered eating [46
] and poor psychological well-being [47
] and, as found in the present study, predictive of overweight risk. Thus, obesity prevention and treatment interventions should also involve educating adolescents and their surrounding network of families and friends about the importance of avoiding negative weight talk and strategies to promote positive, supportive conversations. Findings from this study were comparable to the 5-year longitudinal study findings [10
], further supporting the notion that socio-environmental factors experienced during adolescence influence weight status into young adulthood.
Several strengths of the current study enhance the utility of the findings. First, because this was a prospective study, there was reduced chance of recall bias. Second, this study was conducted in a large, ethnically/racially and socio-economically diverse population that was similar in terms of racial/ethnic composition to the U.S. population of adolescents and youth adults [48
], improving our ability to generalize the findings. Third, variable selection was guided by Social Cognitive Theory and a myriad of personal, behavioral and socio-environmental factors were assessed, allowing for a comprehensive examination of potential predictors of overweight incidence.
It is also important to take study limitations into consideration when interpreting the findings. The attrition of participants and use of self-reported BMI and dietary intake data may have introduced bias to the study findings. Compared with the original sample, adolescents who completed both the baseline and follow-up survey were more likely to be female, white and in the upper SES categories. However, sampling weights correcting for non-response bias were used in all analyses. Additionally, high correlations between self-reported and measured BMI values were found at Time 1 [19
] and Time 3 [37
]. Due to the limited number of physical activity measures assessed at Time 1, we were unable to adequately address factors such as perceived barriers to physical activity, social support for physical activity, and physical activity self-efficacy in terms of their ability to predict overweight incidence in this study. More comprehensive measures of these potential predictors of physical activity were added later in study waves. Future research will be needed to more thoroughly examine factors relevant to energy expenditure versus energy intake. In addition, we were unable to fully explore the role of life events that commonly occur during the transition to young adulthood such as finishing school, moving out of parent’s house, marriage and starting a full-time job; in future studies, markers of transition to young adulthood, and their timing, should be measured and examined in relation to weight changes overtime. Also, when interpreting our 10-year change findings, it is important to consider that we are unable to determine whether the change of certain behaviors or attitudes occurred before an individual became overweight over the 10-year time period. Thus, we are unable to rule out the possibility of reverse causation. Testing for interactions between personal, behavioral and socio-environmental factors that could further help to determine how these factors interact with one another to influence weight gain among adolescents was beyond the scope of this study, but could be considered in future analyses.