Two primary data sets, identified henceforth as data set 1 (D1) and data set 2 (D2) were used as source data for simulations. The primary individual-level data and estimations of participation rates were derived from survey data obtained in 4 North Carolina middle schools in an urban school district (D1) (16,19). The D1 survey was administered to 2,582 sixt- through eighth-grade students (response rate, 89.8%) during the 2008–2009 school year. The 4 schools were selected because of their different approaches to providing a sports program. Two schools offered only competitive interscholastic sports, and 2 schools offered intramural sports exclusively in their co-curricular programs. The primary measures of sports participation were students’ self-reports of participation in co-curricular sports at school (school sport) or participation in a sport outside of school (outside sport). For the purposes of this analysis, a community sport was an organized sport within a student’s community (ie, in a recreation program, at a YMCA, at a church, or in a youth sports’ organization team).

Although the D1 data represent the most comprehensive individual-level physical activity data collected from middle school students in North Carolina, they include schools offering only 1 type of school sports program. Some schools in the state offer a combination of intramural and interscholastic sports. The first wave of the National Education Longitudinal Survey (NELS), which includes a national sample of schools with different types of sport programs, was used to calculate differences in sports participation rates when intramurals were offered along with interscholastic sports within the same school. This first NELS was conducted in 1988 by the National Center for Education Statistics and includes data obtained from 27,394 eighth-graders in a random national sample of 1,000 middle schools (20).

To simulate physical activity rates based on sports participation, data were obtained from 1,188 observations conducted from April 2009 through December 2009 at the same schools from D1, by using the System for Observing Play and Leisure in Youth (SOPLAY) (21). Trained observers recorded use, location characteristics, and student physical activity levels in the predetermined physical activity areas that were designated for sports in each of the 4 schools, between 2:30 PM and 4:30 PM. SOPLAY uses momentary time sampling to provide a count of individuals within each designated activity zone and classification of observed activity. Standard Metabolic Equivalents (METs) were assigned to SOPLAY codes as 1.5 = sedentary, 3.0 = walking/moderate, and 6.0 = vigorous. An estimate of kilocalories (kcal) per hour was calculated on the basis of Ridley et al (22) by multiplying the average energy expenditure per sport times 43.1 kg (the mean weight for children aged 11, 12, and 13 years, assuming a 50–50 sex ratio per school).

School-level data were integrated from 2 sources. Information related to school-based sport and physical activity programs was obtained from a self-administered Web-based questionnaire completed by personnel at 325 North Carolina middle schools (response rate, 74.4%) in January 2009 (data set 2, [D2]) (7). Respondents received a $20 incentive for completing the survey. The questionnaire was adapted from the School Health Policies and Programs Study 2006 to measure school-based physical activity environments. School population data (ie, enrollment, percentage female, percentage minority, percentage economically disadvantaged) were obtained from publicly available data at the North Carolina Department of Public Instruction (NCDPI).

Data were analyzed using SPSS 21.0 (IBM Corp, Armonk, New York). Baseline estimates of state participation rates were established on the basis of data from D1. It was assumed that the participation patterns of sixth- through eighth-graders in this sample would be representative of statewide participation rates. We also assumed equilibrium of supply and demand. That is, this sample represented ideal levels of motivation for sports participation (demand) as well as ideal levels of opportunities (supply) both in school and out of school (eg, public recreation departments). On the basis of analysis of school-level data in D2, the sampled schools offered similar numbers and types of activities within their co-curricular programs as school programs statewide.

For participation rate calculations, a stratified approach was used to estimate sports participation (both school- and community-based sport) on the basis of grade level (sixth grade/seventh and eighth grade), race/ethnicity (white/race other than white), economic status (economically disadvantaged/not economically disadvantaged), and sports program type (intramural sports/interscholastic sports). Grade levels were split in this way because a North Carolina policy prevents sixth-graders from participating in interscholastic sports, thus restricting their access to sports participation. The actual rates of sports participation for each population group in D1 were then used to extrapolate participation rates statewide by applying estimated participation rates to the sample of schools in D2. These rates were then extrapolated to the total student population at all North Carolina middle schools. School-level sports participation rates were estimated for each school in D2 on the basis of its enrollment in each of the stratified groups and sports program type.

To estimate physical activity intensity, standard METs were assigned to each available activity within either an intramural or interscholastic program. Because no direct observations of community sports activities were available, the average of observations of participation within this activity was used (ie, the average METs of intramural soccer and interscholastic soccer = community soccer). By using our estimated statewide sports participation rates, we then calculated rates of participation within each activity among sports participants. We were able to estimate participation rates by activity (ie, type of sport), program delivery type (ie, school intramural, school interscholastic, and community), and style of participation (ie, community only; intramural only; interscholastic only; intramural and interscholastic; intramural and community; interscholastic and community; and intramural, interscholastic, and community). Because participation style and activity type differed significantly by economic status, calculations were stratified by economic status. The number of estimated participants within each activity was then multiplied by the standard MET rate for the activity and then converted to kcal per hour. For nonparticipants, METs were held constant at 1.5 (64.65 kcals/h).

To simulate potential changes in statewide rates of adolescent sports participation, a multiple regression model was estimated by using the simulated school-level sample, controlling for percentage of sixth graders, percentage of minority students, and percentage of economically disadvantaged students. Parameter estimates to calculate estimated variation were generated from existing school program conditions. We recalculated both participation rates and related energy expenditure for schools based on 2 hypothetical scenarios: 1) All schools in the D2 sample adopted policies to offer only intramural sports, and 2) All schools in the D2 sample adopted policies to offer both intramurals and interscholastic sports. Because of the expected influence of economic status on sports participation, the potential effects of policy changes were analyzed separately for students who were economically disadvantaged and those who were not.

Fifty-three percent of sampled schools (n = 168) offered no intramurals or noncompetitive activities, and of the 47% of schools (n = 149) that do offer these activities, 6.7% (n = 10) do not allow sixth grade participation. Because of state and school policies restricting sixth grade participation, 32,000 sixth- graders in the sample attended schools that provided no access to co-curricular sports.

On the basis of the imputation of stratified estimated sports participation rates, 59.39% of sixth-graders and 63.4% of seventh and eighth graders were estimated to participate in organized sports either at school or in the community. Overall, an estimated 62.3% of middle school students statewide participated in organized sports. Our estimated figure was similar to estimates based on the aggregate results of the North Carolina Youth Risk Behavior Survey that estimated 61.5% of sixth- through eighth- graders participated in sports at or outside school in 2009 (23). On the basis of NCDPI’s report of 330,062 students attending public middle schools, this estimate suggests that approximately 39% (124,502 students) in North Carolina middle schools may not be participating in sports. The current participation rate for students classified as economically disadvantaged was estimated at 52.3% compared with 70.2% for students not classified as economically disadvantaged. Among all sports participants, 37.2% were estimated to be economically disadvantaged.

In terms of physical activity expenditure, we estimated the energy expenditure of sports participants averaged 117.34 kcal per hour during the after-school period, compared with the assumed average of 64.65 kcal per hour for nonparticipants. We estimated the overall energy expenditure per student at an average of 97.56 kcal per hour and an average of 59,481.02 kcal per hour expended daily per school. Economically disadvantaged students averaged 93.40 kcal per hour compared with 100.86 kcal per hour for students not economically disadvantaged.

The baseline regression model used to estimate the effects of changes to participation rates based on policy changes is presented in Table 1. Controlling for the other variables, offering intramural sports had the strongest positive association with higher sports participation rates (increasing rates from 15% to 17%). The percentage of economically disadvantaged students had the strongest negative association with sports participation rates.

If all middle schools offered only intramural sports, we estimated there would be a 1.1% increase in existing rates of sports participation. It was further estimated that economically disadvantaged students may benefit more from this policy change than might students who are not economically disadvantaged (Table 2). Policy adoption would increase participation among economically disadvantaged students but decrease participation among students who are not economically disadvantaged. In terms of kcals, this policy implementation was estimated to increase physical activity 4.9% for economically disadvantaged students but only 1.2% for those not economically disadvantaged.

If all middle schools offered both intramural and interscholastic sports, we estimated sports participation could increase by 21.2% and by more than 36% for economically disadvantaged students, resulting in 43,504 new sports participants. More than 28,000 new sports participants (64.5%) were estimated to be economically disadvantaged students. Additionally, we estimated that adopting this policy could yield an increase of 7.66% kcal expenditure per hour per student; and it could yield a 12.12% increase in kcal expenditure per hour for economically disadvantaged students (Table 3).