Background
School-based physical activity (PA) interventionists recognize the critical role of school policies and environment (Sallis et al., 1998; 2003), and there is growing interest in applying ecological approaches to examine relations among school policies and children’s PA. Opportunities to be active at school vary widely, and there is little understanding of how school policy relates to PA. School PA policy research is in its infancy, and new research tools that take into account the complex and multi-level nature of school policy need to be developed and tested.

Objectives
(1) To develop and validate a generalizable instrument (School Physical Activity Policy Assessment, S-PAPA) to assess school- and district-level policies and school site characteristics thought to be associated with school PA and, (2) To use the newly developed instrument to assess the relations of policy and environmental variables and PA opportunities provided in elementary schools.

Methods
For Objective One, we assembled focus groups consisting of physical education (PE) policy stakeholders, perused relevant literature, and analyzed related instruments. We used standard psychometric procedures for item development. To assess construct validity, we consulted educational policy makers, PE teachers, and university faculty with expertise in PE and education policy. To examine test-retest reliability, a sample of 31 elementary school PE teachers completed the instrument twice, 14 days apart. Given that responses were generally categorical in nature, categorical analyses (e.g., phi, kappa, chi square) were used to estimate individual item reliability.

For Objective Two, we used the new instrument in 65 elementary schools from 9 states to identify PA policy and environmental variables and then assessed their relations to amounts of PA programming provided during the school day. Key informants in schools completed the S-PAPA and two other validated tools, the PARC (Physical Activity Record for Classes), which provides a daily log of minutes classes spend in PE and recess, and SPAS (Structured Physical Activity Survey), which documents the time and numbers participating in other school PA programs (e.g., running clubs) before, during, and after school. PARC and SPAS were used to collect a 2-week sample, and the resulting data were reduced to provide the mean number of minutes students at each school spent in PE, recess, and other PA programs per week.

We used descriptive statistics to describe PA program minutes and policies, and used correlation to assess the relationship between various district- and school-level policies and environmental variables to student time in PA programs. Variables found to be significant (p<.05) in bivariate correlation analysis were re-coded, and logistic regression was used to assess the odds of a specific school policy or environmental variable being identified in the highest 40% of reported PA program minutes when contrasted with the lowest 40% of schools.

Results
S-PAPA uses open-ended, dichotomous, multichotomous, and checklist formatting and has three distinct sections: (1) Physical Education (47 items); (2) Recess (27 items), and (3) Other Before, During, and After School Programs (15 items). Generally, the reliability of PE and recess items had fair to substantial levels of agreement (Kappas=0.31-0.81), and items about other before, during and after school PA programs had fair to perfect agreement (Kappas=0.31-1.00).

There was substantial variability in PA programs among schools, with the mean weekly PARC score being 209 minutes (SD=59), with 70% (146 min; SD=49.0) accrued from recess and 30% (63 min; SD=30.4) from PE. The mean SPAS score was only 10.4 minutes per week (SD=11.7).

The most prevalent PE policies pertained to the assignment of grades, standards/guidelines for content, and PE minutes per week; the fewest were for program evaluation and maximum student-to-teacher ratio. Recess policies were less common than PE policies, with the most prevalent pertaining to playground facility maintenance, recess minutes per day, and supervisor training. Generally, recess policies were more prevalent at the school than district level.

Policy variables most related to PARC minutes were percentage of students eligible for free/reduced meals (r=-.31, p<.05), having a school written policies regarding PE standards/guidelines (r=.32, p<.05), and requiring a specific number of minutes per week of PE (r=.26, p<.05). Logistic Regression confirmed that having a school-based policy of minutes of PE increased the odds of being in the top 40% of PARC minutes when contrasted with the lowest 40% of PARC minutes (OR=4.43, 95%CI=1.28 to 15.35).

Conclusions
Test-retest reliability results suggest the SPAPA items were reliable and can be useful in assessing school PA policies. Data show that while recess had the greatest contribution to school PA program minutes, few policies govern it. The significant negative correlation found between the percentage of students eligible for free/reduced meals and PARC suggests that low income children have reduced opportunities for PA during PE and recess at schools.

Support/Funding
This project was funded by grant #67113 from Active Living Research, a national program of the Robert Wood Johnson Foundation.