The methods for the current study were largely adapted from our previous research (12). Each school's participation in the R4F program is determined by the individual school. In some schools, every student within a grade level will go through the R4F program. In other schools, the program may be within an elective course. Data on students with physical disabilities and special educational needs is not currently collected, but is an area of future interest. However, many schools do report sourcing adaptive bikes to ensure all students are included, and this is an area of future growth and development of the program. The current study lacked a comparison/control group, like our previous study, and instead relied on before and after program participation measurements (12). The program was part of standard educational practice at the school, and parents and legal guardians were sent home letters outlining school participation in the program and surveys as part of the program. Parents or legal guardians who did not wish for their child to participate in the survey signed a form and returned it to the teacher (passive consent). In addition, each student had the option to opt-out of the survey, even if a parent or legal guardian did not opt them out. At the beginning of the surveys, students were invited to only continue if they would like to participate in the surveys, with no impact to their grades. Data were collected via a voluntary anonymous survey that was provided in English as well as Spanish from participants in the Outride R4F scholastic middle school cycling physical education program, further ensuring participant anonymity and minimizing risk. The study was approved by an Institutional Review Board (Advarra), with secondary analysis approved by Loma Linda University IRB. Participants completed online surveys through a Qualtrics platform on classroom or at-home electronic devices before and after the R4F program, which may introduce variability in the survey environment. We do not have information on how much additional instruction was provided to students at each different school or by their parents in the home environment.

A total of 8,639 student responses (4,820 PRE surveys and 3,819 POST surveys) from 44 schools across the United States who participated in the R4F program during the 2021–2022 school year participated in the surveys, during a period of easing COVID-19 lockdown restrictions (48). As the surveys were conducted anonymously as part of a program evaluation and surveillance process, we were unable to verify that every student completed both a pre- and post-program survey, with analysis being performed using unpaired analysis techniques. To achieve the most balanced and unbiased set of PRE and POST responses across schools a series of filters were applied to the dataset. In an initial screen, 13 schools that only participated in PRE surveys or POST surveys but not consistently both sets of surveys were removed from analysis, resulting in 8,007 remaining responses (92.7%) from 31 participating schools. To ensure only students who actively completed the survey during class time were included, we also filtered survey responses based on response time. The mean time for survey completion was 10.56 min, with a standard deviation (SD) of 10.27 min. Surveys completed in fewer than 3 min or not completed within 31.1 min (2 SD longer than the mean completion time) were removed. This method was used to provide a more unbiased approach to eliminate data where we suspect that the participant did not perform the survey with intent, including answering the questions too rapidly, without properly reading the question or answer set or that the individual allowed the survey to sit open for many minutes without being attentive to the survey. The length of the survey for the current study is shorter than other routine surveys or assessments taken during the school day, many of which far exceed 10–15 min. For example, the full Youth Risk Behavioral Survey takes about 35 min for students to complete (49), while the California School Climate, Health, and Learning core module takes middle school students an average of 17 min (50). In addition, we applied a response quality filtration, including only surveys with complete responses to all sub-items of the youth self-report version of the Pediatric Symptom Checklist (PSC-17-Y) (51) and WHO-5 well-being index (52, 53). This step prevented skewed scores due to non-responses to individual items and resulted in the removal of an additional 794 survey responses. After applying these filters, the final dataset consisted of 3,924 pre-program survey responses and 3,289 post-program survey responses across 31 schools. Unlike our previous study (12), most schools were not subject to COVID-19 related lockdowns during this period (54, 55). The included 31 schools for this study represent 14% of the total number of schools that have been awarded the R4F program since its inception. Students were between the ages of 11 and 14, corresponding to grade levels 6 through 8, and are considered adolescents as defined by WHO and UN criteria (11) and by modern viewpoints regarding the transition to adulthood (56). The schools included in this study were from multiple regions throughout the United States, as depicted in Figure 1. The demographics of self-reported participant gender, race and ethnicity along with lifestyle and other factors are provided in Table 1.

Students self-reported their gender, race, and ethnicity. For gender, students self-selected between Male, Female, A gender not listed, and Prefer not to say (Table 1). Due to the small sample sizes of participants not selecting Female or Male, only students who selected Female and Male were included in the comparative analyses. For race and ethnicity, students self-reported as American Indian or Alaska Native, Asian, Black, Hispanic or Latino, Native Hawaiian or Pacific Islander, More than One Race, White, or they did not specify (Table 1). As the data are analyzed using many sub-categories (i.e., days per week physically active), the small sample sizes in many of the reported racial and ethnic minority groups become challenging to interpret. As a result, throughout this manuscript, students from racial and ethnic minority groups (American Indian or Alaska Native, Asian, Black, Hispanic or Latino, Native Hawaiian or Pacific Islander, or students who selected More than One Race) will be aggregated for analysis and referred to as underrepresented groups (UG). While these data were aggregated for statistical power, this grouping may mask distinct experiences across racial and ethnic subgroups.

Students also answered questions about sleep duration, screen time, PA levels, and involvement in clubs and sports teams. They also self-reported their previous participation in a R4F program, and whether they received free or reduced-cost school lunches, which served as a proxy for socioeconomic status (SES) (12, 57). To further understand the SES of the study population, we evaluated the social vulnerability index of each school site, which was 0.41 ± 0.23 (mean ± SD) with a range of 0.10 to 0.92 on a 0–1 scale (58). The schools were in the following SVI ranges, including 8 low SVI, 12 moderate, 2 high, and 4 very high.

To measure psychosocial well-being, students completed assessments that were chosen because they are adolescent centered and designed for a middle school level. Questions were chosen from pre-existing validated instruments for this age group and address both positive and negative psychosocial functioning. Many questions were adopted from CDC's Middle School Youth Risk Behavioral Survey (59). Among the number of metrics available including the Strengths and Difficulties Questionnaire (SDQ) (60), KIDSCREEN (61), the WHO-5 Well-Being Index (52, 53) and the youth self-report version of the PSC-17-Y (51, 62, 63), we chose the WHO-5 and PSC-17-Y metrics based on their psychometric strengths, their brevity, and ease of delivery in school settings. Although these are both strong metrics, it is worth noting concerns regarding the psychometric properties of the WHO-5 and PSC-17-Y measures used in this age group. While they are both established instruments, their validity and sensitivity in younger adolescents (11–14 years) may be limited, as this age range presents distinct developmental characteristics that can influence both self-reflection and reporting accuracy (64, 65). Reliance on self-report alone may also overlook important behavioral or contextual factors; complementing these data with proxy-reports from parents or teachers, as supported by instruments such as the SDQ or KIDSCREEN, could provide a more balanced perspective in future work.

The WHO-5 is a 5-item, self-report questionnaire evaluating positive well-being and depression, validated for use with adolescent respondents ages 11 and older with higher scores indicating greater well-being (66, 67). The PSC-17-Y complements the WHO-5, helping to identify and assess emotional and behavioral problems, with lower scores indicating fewer psychosocial challenges (63, 68). The PSC-17-Y is not meant as a diagnostic tool, rather it is meant to help understand what emotional and behavioral challenges a child may be facing. It covers a range of emotional and behavioral problems and assesses psychosocial functioning, including sub-scores reflecting behavioral conduct and disruptive behavior (externalization), anxiety and depressive symptoms (internalization), and attention difficulties (attention). As both the WHO-5 and PSC-17-Y scores may fluctuate slightly over time, it is important to consider what is considered a meaningful change, as opposed to a statistically significant change. Here, we report WHO-5 scores out of 100, with changes of 10% or more reflecting a meaningful change (69). Each of the 5 items was scored 0 (“At no time”) to 5 (“All of the time”), yielding a raw total of 0–25, which was converted to a percentage score by multiplying by 4. For the PSC each of the17 items were scored 0 (“Never”), 1 (“Sometimes”), or 2 (“Often”) for a total score from 0 to 34. While a difference of 6 or greater for the total score, or differences of 2 or more on the subscales are considered a reliable change for the longer 35-item version (51, 63), reliable change benchmarks have not been published for the short-form version. After completing the R4F program, students took the same survey with additional open and closed questions about the program. The survey was designed to be completed in 10–15 min.

The Riding for Focus (R4F) program curriculum used in the current study was identical to that employed in our previous research (12). Designed to be integrated into the school day, R4F sessions are run by the physical education teachers at the school and typically take place during students' physical education class time. The program is structured to span 6 to 8 weeks, comprising 12 recommended sessions. Throughout these sessions, students acquire various cycling skills while aiming to achieve at least 20 min of moderate-to-vigorous PA per session. The skills are designed to prepare the participants to be safe while riding on the road. This includes how to start and stop the bike, how to balance and control the bike, along with shifting gears, how to scan and avoid obstacles, and how to conduct themselves as a vehicle on the roadway. Activities are conducted on diverse terrains surrounding the school campus. The R4F program is designed to provide 1 h of PA each day, adhering to the Public Health Strategies (PHS) and WHO guidelines for PA (27, 28). The program also and aligns with the standards set by SHAPE America (70), ensuring a comprehensive and well-rounded approach to physical education and cycling instruction.

Data analysis procedures and graph production largely followed the methods outlined in our previous study (12). We used R (71) to create data frames and filter data, streamlining the workflow for the current dataset. GraphPad Prism 10 (La Jolla, CA) was employed for graphing and statistical analyses. Prior to analysis, all datasets were tested for normality, with none showing normal distribution. For comparisons between two groups, we used the Mann–Whitney U-test to identify potential differences in well-being metrics. For analyses involving more than two groups, we performed a Kruskal–Wallis ANOVA with Dunn's multiple comparisons test. Graphs present violin plots showing median, 25th and 75th quartiles, along with minimum and maximum data ranges. Statistical significance was set at P < 0.05, with further distinctions made for P < 0.01, P < 0.001, and P < 0.0001 where appropriate. We calculated Cohen's d test statistic based on Cohen's recommendations (72, 73) to estimate effect size in pre- and post-program comparisons. According to widely accepted benchmarks, a Cohen's d value of 0.2 reflects a small effect size, 0.5 indicates a medium effect, and 0.8 or above signifies a large effect. These thresholds are valuable because they help interpret the real-world significance of observed changes, emphasizing not only whether a difference is statistically significant but also whether it is likely to be meaningful in practice. Consequently, Cohen's d provides additional insight into the practical magnitude of changes in adolescent mental health and well-being scores beyond what traditional p-values alone can offer. To examine the relative clinical risk of developing psychosocial disorders, we used a Fisher Exact Test in combination with Koopman asymptotic scores (74) for PSC-17-Y and WHO-5 metrics, based on established cutoff scores of ≥15 and ≤ 50, respectively (63, 67). Decreases in the PSC-17-Y scores signify reduced anxiety, depression, attention, and behavioral challenges, while increases in the WHO-5 score are reflective of better mental health and well-being. Both of these instruments are used as screening tools, and individuals meeting the established cutoff scores are referred to for professional evaluation and potential treatment for mental health disorders.

We evaluated the impact of various risk factors on WHO-5 and PSC-17-Y scores using specific cutoffs for various risk factors based on current research and recommendations as outlined in our previous work (12). The physical activity cutoff was set at 4 days a week of 60 min of PA, as adolescents who do not meet this threshold have an increased risk of developing psychosocial disorders compared to those who participate in PA more (75). For sleep, we used a cutoff of 8 h each weeknight, considering that adolescents who sleep less than the recommended amount (9 h for 6–12 year olds, 8 h for 13–18 year olds) on school nights are at greater risk of psychosocial disorders (76). Screen time was limited to 2 h a day, in line with the American Academy of Pediatrics' 2016 policy statement on adolescent media use (77). Regarding breakfast consumption, we set the cutoff at 7 days a week, as regular breakfast consumption is associated with maintaining a healthy body weight, greater life satisfaction, and better academic performance (78, 79). For breakfast consumption we only analyzed whether something was eaten, without considering the content or caloric amount of the meal.

We conducted linear regression analyses to examine relationships between psychosocial wellness scores (WHO-5 and PSC-17-Y) and various self-reported factors, including hours of sleep, screen time, the number of days of breakfast consumption, and the number of days of PA that met PHS guidelines. An F-test was used to determine if the slope of each association was significantly different from zero (80). Additionally, we compared the slopes of the pre- and post-program regression analyses using an F-test to assess whether the two datasets shared a common slope or had independent slopes (80). For all slope analyses, a P-value of P < 0.05 was considered statistically significant.

The current study builds upon our previous research demonstrating that participating in the R4F program is associated with better psychosocial well-being in adolescents, as evidenced by increased WHO-5 scores (12). Our findings regarding gender differences in mental health outcomes align with both our earlier work and the broader literature (29, 95, 104). Consistent with previous studies, our results show that males generally report better psychosocial well-being and a reduced risk of potential mental health issues compared to females (12). This gender disparity in mental health outcomes has been well-documented, particularly in the context of the COVID-19 pandemic (2, 104). Research suggests that females report lower mental well-being and are at greater risk for poor mental health outcomes compared to males (105–109). These differences have been attributed to a combination of biological and non-biological factors. These include gender-specific stressors, societal expectations, differences in mood symptom development, disproportionate hormonal transitions, lower life satisfaction, and higher levels of psychological distress among females (2). Additionally, females tend to participate in PA less frequently compared to their male counterparts (110), which may also contribute to these disparities in mental health outcomes (95, 104). However, it's important to note that the relationship between PA and mental health is complex and understudied, with mental health status potentially influencing PA habits and vice versa (111–113). Promoting PA opportunities that are supportive and tailored to the interests of female students is essential for beginning to close the gender gap in participation (110). Epidemiological evidence consistently demonstrates a pronounced female predominance in the prevalence of depression, with women being approximately twice as likely as men to experience depressive symptoms. Although studies suggest a narrowing of this gender disparity, the divergence typically emerges around age 12, peaks during mid-adolescence (around age 16), and subsequently declines, stabilizing in adulthood (105, 106, 108, 109). Our findings regarding the potential prevalence of psychiatric disorders across genders align with the existing literature, which consistently reports higher rates of mood, anxiety, trauma-related, and depressive disorders in females. Our findings regarding the prevalence of specific types of disorders among genders are also consistent with existing literature such as increased mood, anxiety, trauma-related, and depressive disorders seen in females and (105, 106, 108, 109).

Studies often show that adolescents from racial and ethnic minority backgrounds report higher rates of experiencing mental and behavioral health (MBH) conditions-including anxiety, depression, ADHD, impulse control disorder, but are also less likely to access treatment services (114). This lack of care often results in more persistent and more severe MBH conditions. Additionally, racial and ethnic minority adolescents are less frequently offered high-quality, evidence-based care and are more likely to face adverse stressors early in life, such as toxic environments, cultural hardships, and macro-societal factors (115, 116). However, in contrast to this previous work, we did not observe consistent differences in mental well-being outcomes between students from underrepresented groups and white students. This is in line with the most recent Youth Risk Behavior Surveillance System (YRBSS) data from the CDC (88), whereby across most mental health measures, there were not significant differences across racial and ethnic groups. Of course, factors contributing to overall mental health are complex, and several factors may have contributed to this change. For example, our earlier study was conducted during the COVID-19 pandemic—a period that disproportionately affected the mental health of UG students compared to their White counterparts (84, 117). In addition, these results highlight potential differences among the schools that chose to implement the R4F program compared to those that do not, as they are offering increased access to additional programming to their students. Such opportunities, including tailored social-emotional learning curricula, targeted teacher and student training programs, and increased access to after-school programming opportunities are all associated with better mental health outcomes among students (118–121). Ultimately, identifying the precise and complex factors underlying these findings is beyond the scope of our current study, but warrants additional investigation given the potential mental health benefits for students.

We did not observe any significant differences in student well-being across students from low and high SES backgrounds. This result is surprising based on previous research on socioeconomic disparities in adolescent mental health, which typically shows a higher prevalence of externalizing, internalizing, and attention behavior issues among students from low SES households (122–124). Two possible reasons for the lack of difference may be due to the schools that tend to implement the R4F program, as well as reliance on student self-report data. Many students either were not aware of their free or reduced lunch status (39% stated “I don’t know”) or did not want to share it, which could lead to student self-report of free and reduced lunch being less reliable. Secondarily, the wide range of social vulnerability indices among the sampled schools reflects the diversity in socioeconomic context among the sampled participants, which may contribute to the differences in student well-being that could impact study outcomes.

However, students from low SES backgrounds reported higher levels of well-being (WHO-5) after participating in the R4F program compared to before the program. Yet, there was not a statistically significant change in students from high SES backgrounds. These findings contrast with our previous work (12). This difference may reflect the changing influence of COVID-19-related stressors on low SES student mental health, which may have been more prominent in our earlier cohort but less so in this more recent, post-pandemic lockdown group (125).

Lower SES households are likely to have reduced access to sport facilities, related equipment, and other activities known to improve mental health (113, 126), which may have been exacerbated by the pandemic (127, 128). The subsequent school year may have seen a restoration of these opportunities, contributing to the observed improvements (125). This underscores the potential importance of easily accessible, school based, programs for adolescents from low SES backgrounds, such as the R4F program. Schools may be uniquely positioned to intervene in the traditionally bidirectional, positively-reinforcing relationship between low SES and poor mental health by providing such programs (129–131). However, these differences could also be attributed to factors not accounted for in this study, such as geographical and economic barriers, nutritional status, or other variables that interact with socioeconomic status to influence mental health outcomes. Further research is needed to elucidate these potential confounding factors and their impact on the effectiveness of programs like R4F across different socioeconomic groups.

Students involved in extracurricular clubs exhibited higher levels of psychosocial well-being both before and after participating in the R4F program. Numerous studies support this observation, demonstrating that participation in extracurricular activities predicts positive outcomes among adolescents including improved social belonging, reduced depression, better mental health, increased civic engagement, and higher educational achievement (132–135).

However, participating in the R4F program was associated with better WHO-5 scores for students who were not engaged in clubs. This suggests that structured PA programs offered during the school day in a non-competitive setting can provide positive avenues for social interaction, confidence-building, and improved well-being. This finding aligns with research highlighting the psychological, educational, and social health benefits of participation in club-based or team-based sports (103, 136, 137). While the R4F program may not fully replace the positive aspects of extracurricular club involvement, it appears to offer valuable opportunities for enhancing mental health and well-being, particularly for students with limited access to traditional extracurricular activities.

Examining the common elements between clubs and the R4F program could provide valuable insights for shaping school policies. Both clubs and the R4F program offer opportunities for skill development, social interaction, and a sense of belonging, which are crucial factors in promoting adolescent well-being. The quality of experiences in organized activities, rather than mere participation, may be key to understanding their developmental benefits (138). Integrating structured PA programs like R4F into the school day could complement existing extracurricular offerings and provide more equitable access to the mental health benefits associated with organized activities. Future research should focus on identifying the specific mechanisms through which these programs enhance adolescent well-being and how they can be optimally implemented in diverse school settings.

Participation in the R4F program was associated with increased WHO-5 scores, regardless of whether students ate breakfast daily or not. However, students who ate breakfast daily had significantly higher baseline and post-program WHO-5 scores than those who did not eat breakfast daily. This aligns with existing evidence suggesting that breakfast consumption is associated with improved mental health and well-being in adolescents (79, 139, 140).

In addition, we observed a dose-dependent relationship between the number of days students reported eating breakfast and their psychosocial well-being, a finding supported by current literature (79, 141). For instance, a nearly linear relationship has been observed between higher frequency of breakfast consumption and greater life satisfaction in children and adolescents (79). To build on these findings, future research should explore not only the frequency of breakfast consumption but also the quality of the breakfast consumed in relation to PA, given their potentially synergistic impact on student psychosocial well-being and associations with better mental health outcomes in adolescence (139, 141). Indeed, as cycling can be a vigorous activity, ensuring students are properly fueled before going for a bicycle ride can ultimately impact their experience with the activity.

We observed higher WHO-5 scores following participating in the R4F program compared to before participating in the program, regardless of whether the students were regularly active outside of the program. However, both at baseline and after participating in the program, students who were active at least 4 days a week had higher WHO-5 and lower PSC-17-Y scores, aligning with previous work (12, 142). In addition, students who reported being active at least 4 days a week also had lower externalization, internalization, and attention sub scores both pre- and post-program compared to those who were not regularly active. This finding is consistent with existing literature (143, 144), which suggests that PA can be an important tool that helps reduce anxiety, mood disorders, and ADHD symptoms.

We also saw a clear dose-dependent relationship between the number of days of PA reported and the WHO-5 and PSC-17-Y scores. In line with current PA recommendations (27, 88, 95), we observed the highest levels of student well-being, across both measures, in those who reported being active 7 days a week. However, this must be balanced against the observation that those already meeting PA recommendations experienced only modest increases in their mental health and wellness scores. These findings illustrate that further research is needed to resolve the optimal frequency and intensity of PA for maximizing mental health benefits in adolescents (95, 104).

The relationship between PA, screen time, and adolescent mental health is complex and increasingly relevant in our digital era. Our study builds upon previous research and provides new insights into this relationship, particularly regarding the impact of interventions like the R4F program (12). Consistent with existing literature, we found that students adhering to screen time recommendations generally reported better psychosocial well-being (98, 145, 146). In addition, students who exceeded the recommended screen time of 2 h per day also reported significantly higher PSC-17-Y externalization, internalization, and attention sub scores than students who met screen time recommendations. This observation is consistent with studies reporting associations between excessive screen time use and various mental health challenges, including behavioral issues, emotional problems, and attention difficulties (147, 148). However, we also observed that participating in the R4F program was associated with larger changes in mental well-being for students who did not meet screen time recommendations compared to those who met the recommendations. While more research is needed, cycling may be a particularly promising intervention for adolescents with higher screen time use, an important consideration given the increasing time young people spend on electronic devices. However, the persistence of the gap in WHO-5 scores, PSC-17-Y composite scores, and PSC-17-Y sub scores between the two groups indicates that the benefits of limited screen time remain significant even after the intervention.

Our study also identified a dose-response relationship between screen time and mental health outcomes. The negative association between screen time use and WHO-5 scores, coupled with the positive association with both the PSC-17-Y composite and sub scores, suggests that even incremental increases in screen time may have negative impacts on mental health. This relationship persisted after the R4F program, indicating that the impact of screen time on mental health is robust and cannot be fully offset by PA intervention on its own (145, 149–151). Further, while PA interventions like R4F show promise in improving some mental health outcomes (12, 145, 149), they do not comprehensively address all aspects of psychosocial well-being, especially for heavy screen users. The program's limited impact on certain mental health risks associated with excessive screen time warrants further investigation.

The influence of sleep on adolescent psychosocial well-being and the efficacy of the R4F program is a critical area of investigation, given the importance of adequate sleep for mental health (12, 99, 100). Our findings provide compelling evidence for the significant role of sleep in adolescent mental health and its interaction with PA interventions. Both before and after participating in the R4F program, students meeting sleep recommendations demonstrated markedly better psychosocial well-being than those who did not. Additionally, students who slept at least 8 h a night exhibited lower baseline externalization, internalization, and attention sub scores. These results align with existing literature emphasizing the positive impact of adequate sleep on various aspects of mental health and cognitive function in adolescents (152, 153).

Participation in the R4F program was associated with modest improvements in student well-being, regardless of meeting sleep recommendations, consistent with our previous findings (12). However, sleep is an important factor contributing to overall mental well-being, and participating in R4F alone is not enough to overcome poor sleep. Indeed, large gaps in well-being scores persisted between those who did and did not report meeting sleep recommendations.

Our analysis of relative risk for developing psychosocial illness reveals a concerning trend and again supports our earlier work (12). Students not meeting sleep recommendations consistently demonstrated higher rates of poor mental health and an elevated risk for psychosocial disorders, both before and after the R4F program. This finding highlights the critical role of sleep in mental health resilience and suggests that PA interventions alone may not be sufficient to mitigate the risks associated with inadequate sleep.

The observed dose-response relationships between sleep duration and mental health outcomes provide valuable insights into adolescent well-being. Our findings reveal a positive association between sleep duration and WHO-5 scores and a negative association between sleep duration and PSC-17-Y composite and sub scores. Each additional hour of sleep incrementally improved mental health and well-being scores in the adolescents of our study. These results align with current literature (150), which emphasizes the importance of adequate sleep for mental health. Importantly, even modest increases in sleep duration were associated with measurable improvements in mental health outcomes. These findings underscore the independent and enduring role of sleep in adolescent mental health, highlighting the need for sleep-focused interventions alongside PA programs to comprehensively support adolescent well-being.