The prevalence of childhood obesity and overweight has been increasing in the United States since the 1980s. This trend appears to be stabilizing but remains high (1,2). Adolescent obesity has been particularly difficult to address: although the obesity rate for children aged 2 to 5 years has recently declined, a similar decline among children aged 6 years or older has not been seen (2). In 2012 about 21% of US youths aged 12 to 19 years were obese (2).

Childhood obesity appears to be a result of low energy expenditure and high energy intake (3). Childhood obesity interventions have focused on promoting child behavior changes for a more healthful diet, a higher level of physical activity, or both. A family-based approach is increasingly emphasized to control childhood obesity, because parents play a critical role in providing a home environment that fosters healthful eating and physical activity for children and adolescents, ultimately affecting their weight status (4,5). Recently, attention has also been given to the role of physicians in adolescents’ weight management (6). Because behavior risks are interrelated, children and adolescents who are at risk for unhealthy eating are also more likely to have less physical activity and more screen time (7). A multiple health behavior change approach for obesity prevention is needed to address multiple risk behavior domains at multiple levels (8,9).

Interest is increasing in the interconnections among multiple overweight and obesity risk factors and in tailoring interventions for different population subgroups. However, studies to date have focused on identifying the main effects of single risk factors and, to a lesser extent, of 2- or 3-way interactions. For dietary behaviors, studies showed an inconsistent relationship between fruit and vegetable consumption and child body mass index (kg/m²) (BMI) (10,11) and a negative association between regular breakfast consumption and childhood obesity outcomes (12,13). Studies of child physical activity and weight status also showed mixed results, although most studies found an inverse association of physical activity or sports team participation with overweight and obesity outcomes or a direct association of inactivity and sedentary behavior with overweight and obesity outcomes (14–17). Research using various measures of parenting practices regarding diet and physical activity yields inconsistent results for adolescents. One measure of parenting practice, letting a youth decide what to eat, was found to be associated with unhealthy eating behaviors, such as skipping breakfast (18). Although 1 study found that rules restricting sedentary time were associated with less TV watching (19), another study found that such rules were associated with undesirable behaviors and increases in children’s BMI (20). An authoritative parenting style, where parents are both responsive to children’s needs and controlling of children’s behaviors, was associated with healthier eating, more physical activity, and lower BMI among children (21). Physician advice has a positive effect on adolescents’ weight loss attempts (22), but no evidence has been reported on its association with adolescents’ weight change. Some studies included both physical activity and dietary behavior indicators in their analysis but did not show a consistent pattern of how these behaviors simultaneously affect adolescent obesity outcomes (23–25). To date, studies rarely considered the combination of individual behavior risk factors with parenting practices and access to physician advice, and the patterns in which these factors congregate, possibly because of methodological restrictions and unavailable data. The objective of this study was to examine the joint influence of multiple risk factors on adolescent overweight status.

We used latent class analysis (LCA) to categorize youths aged 12 to 18 years in a Midwestern city into several mutually exclusive groups based on patterns of modifiable obesity risk factors in 4 domains: dietary behavior, physical activity, parenting practices, and physician advice. To examine the joint influence of multiple risk factors on adolescent overweight, we also conducted logistic regression to see whether each of the 4 risk-factor groups was associated with overweight status, after controlling for each youth’s sociodemographic factors. The study contributes to the field in 3 ways. First, the study enhances the understanding of the unique patterns of adolescent dietary and physical activity behavior as well as parenting practices and physician involvement. Second, the study provides valuable information for stakeholders to use in prioritizing interventions and developing programs for youths with different behavior patterns. Tailored interventions for youths exposed to particular sets of risk factors use scarce health care resources more effectively. Third, this study addresses a lack of data on adolescent risk factors for obesity, especially data on parenting practices and physician advice, by providing relevant evidence to assess the importance of these factors in youths’ overweight status and to inform interventions in this high-risk group.

Forty-six percent of participants were aged 12 to 14 years, 53% were boys, 85% were non-Hispanic white, 12% had no parent who attended college or university, and 87% lived with both parents (Table 1).

Over 61% reported eating breakfast every day, and 63% reported eating 5 or more servings of fruits and vegetables per day for 4 days or more a week. Eighty percent engaged in at least 60 minutes of physical activity for 4 days or more a week, 77% participated in at least 1 sports team, and 67% watched television or played video games for less than 2 hours a day during weekdays. Eighty percent of participants reported that their parents let them choose what to eat, and 66% reported that their parents let them decide how much time to spend watching television or playing video games. During their last routine physical examination within the previous12 months, 57% received physician advice on nutrition and 61% received physician advice on physical activity. About 24% were overweight, which is lower than the national average (2).

Our results indicated that adolescents in our study were differently exposed to the 4 obesity risk factors we considered: dietary behavior, physical activity, parenting practices, and physician advice. Three latent classes emerged: youths with healthy dietary behavior and physical activity, less permissive parenting practices, and physician advice (Group I); youths with moderately healthy dietary behavior and physical activity, less permissive parenting practices, and physician advice (Group II); and youths with unhealthy dietary behavior and physical activity, permissive parenting practices, and physician advice (Group III). The group patterns identified in our analysis suggest that unhealthy behaviors are not only interrelated, as suggested in the literature (7); they also co-occur with permissive parenting practices and physician advice to youths.

The logistic regression analysis results suggested that categorizing these youths into 3 groups was meaningful in explaining youth weight status. Youths from Group III were more likely to be overweight than those from Group I and II. This finding highlights the need to make addressing obesity risk factors a priority for adolescents with unhealthy eating behaviors, inadequate exercise, more permissive parenting practices, and physician advice. Compared with youths in Group I, youths in Group III were likely to be older, female, and from a minority race; to have non-college educated parents; and less likely to live with both parents. Interventions targeting this group should be the priority for allocating limited resources to address adolescent obesity and overweight. This finding also supports the call for interventions to target multiple risk behaviors that address multiple domains of obesity risk factors (8). Multidimensional interventions in all 4 areas should be jointly implemented to improve the weight status of youths in Group III.

This study adds to the current understanding of adolescent obesity by illustrating how risk factors jointly influence the weight status of youths. Healthy individual behaviors and less permissive parenting practices together had a strong association with a better weight outcome for adolescents. Because parents in Group I and II were less permissive regarding sedentary behavior than parents in Group III, our results reinforce previous findings that some parental control over behaviors may be effective in maintaining healthy weight among youths (18,19). A family-based approach that considers family background and stresses the role of parenting in shaping children’s obesity-related behaviors may be most effective (4,5).

Interestingly, we did not observe much difference in weight status between Group I and Group II. Although youths in both groups were physically active and received similar parenting, youths in Group II showed less healthy dietary behavior and were less likely to receive physician advice on nutrition and physical activity than those in Group I. One possible explanation for the lack of difference between these 2 groups might be the lack of specificity in some behavior indicators, such as total calorie or total sugar and fat intake (11). The results could be due to Group II having an elevated risk for obesity but having not yet gained excessive weight. If this explanation is accurate, then this group should also be targeted for future intervention to prevent obesity. Another plausible explanation could be that a physician did not discuss nutrition or physical activity during a consultation with healthy-weight youths. Because only about half of pediatricians calculate adolescents’ BMI during a routine visit (30), there is still room for physicians to play a more active role in obesity prevention.

This study has several limitations. First, given that the study sample was from a single Midwestern city and the response rate was relatively low, the results may not be readily generalizable to a larger population. Second, the cross-sectional data used do not allow us to infer causality, and the data did not include information on some obesity risk factors such as inadequate sleep and poor mental health status. Third, the parenting practices and physician advice measures used were relatively crude, and validated measures should be used in future studies. Lastly, we used self-reported BMI to measure overweight status, which may have underestimated the likelihood of overweight. However, we do not expect that these limitations are sufficiently severe to change study results on the youth groups and their relationship to overweight status.

These limitations notwithstanding, our study showed a need to simultaneously examine the complex set of factors and their patterned congregations in relation to adolescent obesity. Concrete efforts in individual, family, and primary care settings are important to shape adolescent behaviors and to help adolescents maintain a healthy weight. The physician’s role may need to be expanded from a predominant focus on management of obesity-related conditions to a more active approach to obesity prevention. The 3 youth groups identified in our study and their different risks of overweight also suggest that obesity interventions will be more effective if they are customized for particular behavioral and contextual phenotypes of youth.