Theoretical frameworks

The health belief model and socio-ecological model guided this study. The health belief model posits that an individual will make choices about their health behavior based on how severe they perceive their risk of illness/injury [27]. As a characteristic of adolescence, youth may have a heightened sense of invincibility, coupled with a potential lack of knowledge about the risks associated with binge drinking—potentially increasing their risk of the behavior [28]. Several studies have documented that adolescents who perceive risk or susceptibility to alcohol use or binge drinking are less likely to drink [6–8], and alcohol preventive interventions have targeted perceived susceptibility to alcohol use as one potential mechanism for influencing alcohol use [29].

The socio-ecological model proposes that the environment where an adolescent resides contributes to problematic behaviors such as alcohol use. The environment includes the following domains: individual (e.g., presence of a mental disorder, perception of risk of harm from binge drinking), peer (e.g., peer beliefs about substance use), family (e.g., parental support), and community-level (e.g., access to substances) [30,31]. The interaction of these domains rather than a single factor alone contributes to the risk for engaging in problematic behavior [32]. As such, understanding how specific factors (e.g., extracurricular activity) within their ecological domains predispose adolescents to problematic alcohol use should consider the influence of other potentially salient factors. The variables included in this study reflect the theoretical underpinnings of each of these conceptual frameworks.

Ethics

This study was deemed exempt human subjects research by the Yale Institutional Review Board (IRB) and Research Ethics Board (REB) #2000036003. The data used in this study is publicly available data from the National Survey on Drug Use and Health (NSDUH) (see description below). Staff of Research Triangle Institute International, an independent, nonprofit research organization commissioned to conduct the NSDUH study funded by the Substance Abuse and Mental Health Administration (SAMHSA), obtained informed consent from participants [33]. Verbal informed consent was obtained from the parents/guardians of adolescents aged 12-17 years or from adolescents aged 17 years who were living independently [33]. Verbal informed assent was obtained from adolescents aged 12-17 living with their parents [33]. A more detailed explanation about the process of obtaining informed assent from adolescents can be found in the NSDUH Field Interviewer Manual (pages 232-236) [33]. Briefly, staff used an approved script to inform the parent/guardian that their child has been selected for the study, provided details about the study and informed them that their child’s participation is voluntary. Once parental permission was granted, the staff confirmed that an adult will be present for the entire duration of the interview with the child. Although an adult must be present in the house during the interview, the interview with the child is still held in a private space to ensure confidentiality. Informed assent is obtained from the child using an approved script. The child is informed that the study is voluntary, that their responses are confidential, and that they will be compensated with $30 after completing the interview. A description of the study procedures is then described to the child. Interview proceeds only after informed assent is obtained from the child.

Sample population

This study used data from the publicly available 2019 NSDUH dataset [34]. The NSDUH is a national survey administered each year to approximately 70,000 US civilians aged 12 years and older. It employs a complex, multistage probability sampling design to allow for a representative selection of the U.S. population that collects national- and state-level data concerning tobacco, alcohol, and other drug use as well as mental health. Individuals who are active members of the military, hold no household address, or reside in institutional group quarters such as prisons, nursing homes, treatment centers, and hospitals are excluded from survey participation. The present study conducted a cross-sectional secondary analysis of questionnaire data collected from the 2019 NSDUH. This year was chosen due to concern that data from subsequent years would likely have been influenced by the COVID-19 pandemic. Inclusion criteria were participants between 12-17 years of age (N=13,397) who responded to questions about perception of risk of harm from weekly binge drinking (N=13,133). From this sample, we excluded participants who reported a history of binge drinking (N=1,263) given that this was a cross-sectional analysis and including participants with binge drinking, which lies distal to perception of risk of harm, is likely to cloud any existing relationship between extracurricular activities and perception of risk of harm. Thus, the final analytical sample consisted of 11,870 adolescents.

Perceived risk of harm from weekly binge drinking (dependent variable)

The primary outcome variable was perceived risk of harm from weekly binge drinking, which was assessed through the question, “How much do people risk harming themselves physically and in other ways when they have five or more drinks of an alcoholic beverage once or twice a week?” [1,35–40]. Survey respondents were presented with four choices ranging from no risk to great risk, which we dichotomized into two categories for the purpose of this analysis: no risk of harm weekly binge drinking (“no risk”) and any risk of harm from weekly binge drinking (“great risk”; “moderate risk” “slight risk”) coded as 0/1. This dichotomization is consistent with previous studies that utilized the NSDUH to explore risk perception of marijuana use [41,42] and the team’s consideration that any perception of risk from binge drinking is favorable to perceiving no risk from this problematic behavior. S1 Appendix includes all variables from the 2019 NSDUH dataset used in this study.

Participation in extracurricular activities (independent variables)

The NSDUH 2019 dataset contains four questions (corresponding to four different activity-types) asking respondents to report the number of activities they had participated in during the past 12 months (e.g., During the past 12 months, in how many different kinds of school-based activities, such as team sports, cheerleading, choir, band, student government, or clubs, have you participated?). Options included: “None,” “One,” “Two,” or “Three or More.” The dataset differentiates the four types of extracurriculars as 1) school-based [“such as team sports, cheerleading, choir, band, student government, or clubs”]; 2) community-based [“such as volunteer activities, sports, clubs, or groups”]; 3) faith-based [“such as clubs, youth groups, Saturday or Sunday school, prayer groups, youth trips, service or volunteer activities”]; and 4) other-type [“such as dance lessons, piano lessons, karate lessons, or horseback riding lessons”]. Since there were four activity-types, each with responses that ranged from 0 to 3 or more, the total possible range of number of activities spanned from “0” to “12 or more” across all types of activities. For the purposes of this study, we captured extracurricular participation in three distinct ways. First, we created a binary variable (none vs. 1 or more) where adolescents who reported no involvement in any extracurricular activity, regardless of activity type were coded as 0 and adolescents with involvement in one or more activity were coded as 1. Second, to explore the influence of different types of activities on perception of risk of harm from binge drinking, we created a binary variable (none vs. 1 or more) for each type of activity (school, community, faith, other). Lastly, to examine whether the number of activities influence risk perception, we examined the level of involvement in extracurricular activity (0= “None,” 1= “One,” 2= “Two,” or 3= “Three or more.”) in each activity type (school, community, faith, other).

Covariates

Other variables to be considered in the study were selected based on the health belief model and the socio-ecological model, their associations with both the independent and dependent variables, as well as from existing studies of similar scope [43,44]. As such, the following demographic variables were included: age in years (12-13, 14-15, 16-17); sex at birth (female, male); race/ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, other race); total family income (less than $20,000, $20,000-$49,999, $50,000-$74,999, $75,000 and above); and rural/urban residence (large metropolitan, small metropolitan, non-metropolitan). There is strong evidence to suggest that poor mental health is related to an individual’s assessment of risk and their involvement in extracurricular activities [2,30,45]. Therefore, whether the adolescent had reported a major depressive episode in the last year (yes/no) was included as a covariate. Similarly, the overall support provided to adolescents by their parent/guardian has been linked to their participation in extracurriculars along with their perception of risk [46,47]. As such, we created a parental support composite variable based on the responses to four questions assessing various aspects of this construct (e.g., “During the past 12 months, how often did your parents provide help with your homework when you needed it?”). Responses to these questions were “always”, “sometimes”, “seldom”, “never”. Participants who responded, “always or sometimes” to all four questions were coded as “high parental support”; otherwise, they were deemed “medium-to-low parental support.” This parental support construct has been used in a similar paper using the NSDUH survey data [44].

Data analysis

All analyses were conducted in Stata Version 18.0 [48] and survey weights were used to account for the complex sampling design. The final analytical sample consisted of 11,870 eligible adolescents. None of the demographic data had any missing data as we used the imputed variables provided in the NSDUH dataset [35]. Data for depression and parental support were missing for 359 (3.02%) and 731 (6.16%) adolescents, respectively. Data for each extracurricular activity were missing as follows: school (n=106, 0.89%), community (n=165, 1.39%), faith (n=199, 1.68%), other (n=137, 1.15%). Missing data were imputed using Multiple Imputation by Chained Equations [49]. Frequencies and percentages were used to describe the sample across study characteristics. Bivariate regression analyses were used to examine differences between adolescents who perceived weekly binge drinking as harmful compared to adolescents who did not across all covariates. Bivariate and multivariate logistic regression models were run to test the unadjusted and adjusted associations between participation in extracurricular activities and perceived risk of harm from weekly binge drinking. Specifically, we ran separate models for activity participation measured as a binary variable regardless of activity type (none vs. 1 or more), activity participation measured as a binary variable (none vs. 1 or more) separated by activity type (school, community, faith, other), and activity participation measured as a ordinal variable (0, 1, 2, 3+ activities) across each activity-type (school-based, community-based, faith-based, and other-type extracurricular activities). For the latter two analyses examining each activity type, we conducted four different regression models that each contained only one type of activity (school, community, faith or other) to avoid multicollinearity since there is the possibility that adolescents who participate in one type or activity may be likely to participate in other types of extracurricular activity. All multivariate logistic regression models included relevant covariates (demographic factors, depression, and parental support). Both unadjusted (OR) and adjusted odds ratios (aOR) were reported along with their respective 95% confidence intervals (95% CI).

Sample characteristics

Of the 11,870 eligible adolescents, approximately half in this sample identified as male (51.52%), non-Hispanic white (50.52%), and had a household income of $75,000 or more (45.03%) (See Table 1). Thirty-five percent of the sample were aged 12-13 years, with 35% aged 14-15 years and 30% aged 16-17 years. Most adolescents reported participating in at least one activity of any kind (91.69%). Over half of adolescents reported participating in two or more school-based activities (56.59%) while 47.53%, 36.21% and 17.83% participated in two or more community-based, faith-based or other activities respectively. About 4% of adolescents in the sample endorsed perceiving no risk of harm from weekly binge drinking.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographics of adolescents by perception of risk of harm from binge drinking.

https://doi.org/10.1371/journal.pmen.0000278.t001

Table 1 outlines the differences between adolescents who perceived weekly binge drinking as having no risk of harm compared to those who perceived any risk. Adolescents who perceived any risk of harm from weekly binge drinking were more likely to be female (F=14.05, p=0.000), 16-17 years old (F=12.31, p=0.001), Non-Hispanic White (F=9.16, p=0.004), and have a total household income of $75,000 or more (F=4.85, p=0.032). Additionally, adolescents who participated in any extra-curricular activity compared to those that did not participate were significantly more likely to perceive risk of harm from binge drinking weekly (F=30.93, p=0.000). Across school-based, community-based, and faith-based types of extracurricular activities, adolescents who were involved in one or more extracurricular activity were more likely to perceive that weekly binge drinking as risky (School F=21.34, p=0.000; Community F=12.80, p=0.001; Faith F=14.13, p=0.000). Participating in other types of activities was not significantly associated with perception of risk of harm.

Associations between participation in any extracurricular activity and perception of risk of harm from weekly binge drinking

Table 2 presents the results of both the unadjusted and adjusted logistic regressions used to examine the associations between participating in any extracurricular activity regardless of type and perception of harm from binge drinking. Results show that adolescents who participate in at least one type of extracurricular activity have more than twice the odds of reporting that weekly binge drinking is harmful. This association remained significant after controlling for potential confounders (aOR = 1.86, 95% CI: 1.39, 2.49). In addition, being female and older (16-17 years) was associated with perceiving binge drinking as harmful while having a household income of less than $50,000 was associated with reduced likelihood of perceiving binge drinking as harmful (See Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Participation in extracurricular activities and perception of risk of harm from binge drinking.

https://doi.org/10.1371/journal.pmen.0000278.t002

Associations between participation in different types of extracurricular activity and perception of risk of harm from weekly binge drinking

Table 3 shows results of the association between different types of extracurricular activity (school, community, faith and other) and perception of harm from weekly binge drinking. Unadjusted analyses show that compared to adolescents who reported no participation in any extracurricular activity, those who reported one or more activities were significantly more likely to perceive that weekly binge drinking was harmful across all types of activities (Table 3). However, this association was sustained in the adjusted analyses for school, community and faith-based activities (school aOR: 1.82, 95% CI: 1.40, 2.38; community aOR: 1.40, 95% CI: 1.09, 1.79; faith aOR: 1.33, 95% CI: 1.03, 1.72). Like the first model (See Table 2), age, gender and income were associated with perception of risk of harm from binge drinking with older age and female sex being positively associated with higher perception of risk while lower income was associated with lower perception of risk (See Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Association between types of extracurricular activity and perception of risk of harm: unadjusted and adjusted regression.

https://doi.org/10.1371/journal.pmen.0000278.t003

Associations between number of extracurricular activity and perception of risk of harm from weekly binge drinking across different activity types

Unadjusted analyses revealed that participating in any number of school-based or community-based activities (1 activity, 2 activities, or 3+ activities) was significantly associated with having a higher perception of risk of harm from binge drinking, as well as participating in 3 or more faith-based activities or one other activity type (see Table 4). After adjusting the model for relevant covariates, participating in any number of school-based activities remained significant (aOR 1 activity = 1.75, 95% CI: 1.29, 2.37, aOR 2 activities = 1.83, 95% CI:1.32, 2.53, aOR 3+ activities = 1.91, 95% CI:1.35, 2.75), However, this effect was only seen for participating in either one or two community-based activities but not for three activities, although the p-value was 0.06 for three or more activities (aOR 1 activity = 1.35, 95% CI: 1.02, 1.79, aOR 2 activities = 1.47, 95% CI:1.06, 2.02, aOR 3+ activities = 1.41, 95% CI: 0.98, 2.02). Similarly, adjusted analysis was significant for only for participating in three or more faith-based activities, (aOR 3+ activities = 2.07, 95% CI: 1.48, 2.88).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Association between levels of extracurricular activity and risk perception of binge drinking across different activity types.

https://doi.org/10.1371/journal.pmen.0000278.t004

Similar to the first two models, additional significant findings emerged from the multivariate logistic regression model. Across all types of extra-curricular activities, females were more likely to endorse a high perception of risk of harm from binge drinking compared to males (aOR school = 1.58, 95% CI: 1.22, 2.06, aOR community = 1.59, 95% CI: 1.22, 2.09, aOR faith = 1.56, 95% CI: 1.20, 2.03, aOR other = 1.60, 95% CI: 1.22, 2.10). Additionally, there was an effect of age on risk perception across types of extra-curricular activities. Compared to 12–13-year-olds, 16–17-year-olds were more likely to endorse a higher risk perception (aOR school = 1.66, 95% CI: 1.26, 2.20, aOR community = 1.63, 95% CI: 1.22, 2.16, aOR faith = 1.69, 95% CI: 1.27, 2.56, aOR other = 1.67, 95% CI: 1.25, 2.23). Compared to those with a total household income of $75,000 or more, those with less than $20,000 and between $20,000-$49,000 had reduced odds of endorsing binge drinking as risky across all extra-curriculars (see Table 4). Parental support was significantly associated with increased risk perception from binge drinking in the faith-based activities and other activity models but not in the school-based or community-based activities models. Race/ethnicity, metropolitan areas, and depression were not significantly associated with perception of risk of harm from binge drinking across all models.

Limitations and future research

The current study was subject to several limitations. The cross-sectional design of the NSDUH 2019 dataset limited our ability to infer whether participation in extracurricular activities is causally linked to an enhanced risk perception of binge drinking. Prospective experimental designs would be most effective in better understanding the direction of this relationship and to measure the long-term effect of these activities. Second, we cannot be certain that an increased perception of risk of harm from binge drinking translates to reduced alcohol use; however, there is evidence to suggest that a high-risk perception is a deterrent of future binge drinking [1]. Third, the way the NSDUH assesses perceived risk of harm may be confusing to some adolescents, particularly the phrase “harming themselves physically and in other ways” which may result in either an under- or overestimation of risk depending on the adolescent’s interpretation. However, this phrasing has been utilized widely in the field for measuring perceived risk from alcohol and other substances [37,39,40,44,70]. Fourth, the four broad categorization of extracurricular activities available in this dataset prevented additional examination into the potential differential impact of specific types of activities on perception of risk of harm. Further studies should consider including a detailed list of activities to capture the possibility that, even within specific settings (e.g., schools), certain activities offer distinct benefits. Fourth, there are additional confounding variables such as health literacy and cognitive ability that were not assessed in the dataset but are likely associated with participation in extracurricular activities and risk perception. Lastly, although this is a nationally representative sample of US adolescents, these findings lack generalizability to those excluded from the NSUH sample, such as adolescents who are homeless.