Athletic and PE program questionnaires were pre-tested for comprehension and face validity with school personnel from five non-study schools and modified based on their feedback. The primary outcome measure for this analysis was sports team participation in high school. We assessed this by asking high school students, signaling pathway “In the past 12 months, on how many sports teams did you play?” Responses were dichotomized into a dummy variable (i.e., 0 = 0 sports teams, 1 = 1 or more sports teams). Sports offered per 100 students. Based on information from the

New Hampshire Interscholastic Athletic Association 24 and school websites, we created a comprehensive list of interscholastic and intramural sports and asked school personnel to indicate which sports their school offered for boys and girls. We used these responses to create a sex-specific index reflecting the number of sports available per 100 students for each school (i.e., by sex: ((number of interscholastic + intramural sports offered)/number of students × 100)). We used this metric to standardize sport availability across schools of different sizes. We dichotomized responses into two groups: less than one sport per 100 students and at least one sport per

100 students. Percent of unrestricted sports. We defined unrestricted sports as sports in which there RG7204 order was no limit to the number of students who could participate. Because intramural sports are unlimited by design, we counted all the intramural sport opportunities as unrestricted. For interscholastic sports, we determined whether or not participation was limited by asking school personnel many to “list any (interscholastic) sport for which participation is limited (i.e., the number of players per sport is capped)” and to specify whether boys and/or girls teams were limited. We then calculated the percent of unrestricted sports separately (i.e., by sex: (number of unrestricted

sports offered/(number of unrestricted + restricted sports offered) × 100)). We categorized responses into three groups of unrestricted sports based on the distribution of the data: less than 85%, 85%–99.9%, and 100%. We measured and adjusted for a number of potential confounders. Adolescent-level covariates measured concurrent with the outcome included sex, grade in school, ethnicity, and overweight/obese status. Overweight/obese status was based on self-reported height and weight and sex-specific Centers for Disease Control and Prevention BMI-for-age growth charts.25 and 26 We also adjusted for sports participation at baseline, when adolescents were in elementary school. This was assessed by asking, “In the past year, did you participate in any of the following things? Team sports (yes/no)?”27 Parent/household level covariates measured through parent surveys included education, income, and single vs. two parent household. School and school-town covariates included school enrollment, town median household income, and school town population.