Table 4.

Females: Associations of school obesity prevention policies with school-level prevalence of weight-control behaviors among 9^th and 12^th grade students enrolled at a cohort of Minnesota schools, 2007 to 2010

	Extreme behaviors		Unhealthy behaviors		Healthy behaviors
	Mean prevalence difference (95% CI)	P value	Mean prevalence difference (95% CI)	P value	Mean prevalence difference (95% CI)	P value
School policy score
Limit availability of competitive foods (score range: 0–9)	0.2 (−0.2, 0.6)	0.35	Not estimable b	-	0.4 (−0.5, 1.2)	0.43
Make healthy vending options available (score range: 0–2)	0.2 (−0.9, 1.4)	0.68	0.5 (−2.0, 3.0)	0.68	−0.9 (−3.3, 1.4)	0.44
Promote healthy foods and beverages (score range: 0–5)	−0.7 (−1.4, 0.0)	0.06	−0.1 (−1.7, 1.5)	0.91	−1.2 (−2.6, 0.1)	0.07
Ban advertising for unhealthy foods and beverages (score range: 0–4)	0.5 (0.0, 1.0)	0.04	0.3 (−0.8, 1.5)	0.60	1.3 (0.5, 2.2)	0.003
Provide intramural sports (item range: 0–1)	0.6 (−2.0, 3.3)	0.63	1.8 (−3.9, 7.5)	0.53	2.1 (−3.3, 7.5)	0.44
Overall summary score (score range: 0–22)	0.3 (−0.1, 0.6)	0.12	0.3 (−0.4, 1.0)	0.43	0.3 (−0.4, 0.9)	0.57

^aAll models include a fixed effect for school to adjust for measured (school level, location, percentage minority enrollment, percentage eligible for free/reduced-price meals) and unmeasured school characteristics as well as the percentage of enrolled females who were overweight. Year was included in all models to estimate the change in school-level mean student outcomes over time (from 2007–2010), which can be interpreted as the ‘secular change” in each outcome.

^bThe estimated difference in prevalence is not directly interpretable due to a significant year by policy interaction (P value=0.05); there was a non-significant positive association between the competitive food score and prevalence of unhealthy weight-control behavior in 2007 (mean prevalence difference [95% CI]=0.4 [−0.4, 1.2], p=0.35) and a non-significant negative association between this policy and behavior prevalence in 2010 (mean prevalence difference [95% CI]=−0.6 [−1.6, 0.4], p=0.25).