Sallis 1997.

Study characteristics
Methods	Trial name: Sports, Play, and Active Recreation for Kids (SPARK) Study design: non‐randomised Intervention duration: 2.5 years – reported in paper as 3 academic years (1990–1993) (intervention began in autumn 1990 and ended in spring 1992 for fourth grade teachers and started in autumn 1991 and ended in spring 1993 for fifth grade teachers. Follow‐up observations were at 8 months, and then again at 2 years. Length of follow‐up from baseline: baseline was during autumn at fourth grade, and then follow‐up was during spring at fifth grade. Maintenance effects also studied 1.5 years after the termination of the programme (i.e. 4‐year follow‐up from baseline total). Differences in baseline characteristics: not reported (however, matched by size and ethnic make‐up (% white)). There was a significant difference in age by condition (9.49 vs 9.62 years). Unit of allocation: schools Unit of analysis: classroom
Participants	School type: elementary schools Region: Southern California, USA Demographic/socioeconomic characteristics: schools were situated in a middle‐class suburb of a large city containing 82% European American, 12% Asian/Pacific Islander, 4% Latino, 2% African American with 53% male. Inclusion/exclusion criteria: school level was not reported Student level inclusion ‐ Students were required to complete baseline and final survey and fitness test. Number of schools allocated: of the 7 schools, there were 4 interventions and 3 controls. The 4 intervention groups were further subdivided into 2 groups; 1 teacher arm and 1 specialist arm. The specialist arm was excluded in this review as the in‐school programme was delivered by certified PE specialist. Conversely, the other intervention arm was delivered by school teachers. Consequently, the study reported 5 schools as allocated. Students: 2 consecutive cohort of fourth graders, followed to fifth grade Numbers by trial group n (controls baseline) = 3 n (controls follow‐up) = 3 n (intervention (teachers‐led) baseline) = 2 n (interventions (teachers‐led) follow‐up) = 2 Recruitment Schools:schools agreed to participation in an experimental programme and be randomised to 1 of 3 study conditions. Schools were stratified by percentages of minority student and within those strata; 2 schools were randomly assigned to each condition (PE specialist, teacher‐led or control). The remaining school was added to the control condition. Teachers: not reported Students:2 consecutive cohorts of fourth‐grade students entered the study. All fourth‐grade students were invited to participate. Approximately 98% provided informed consent through a passive consent procedure. Recruitment rate: 12/16 principals consented (consent rate 75%). Resourcing constraints meant that only 7/12 schools were randomised. All fourth‐grade classes in the 7 schools participated.
Interventions	Number of experimental conditions: 3 (2 intervention conditions (PE specialist condition, teacher‐led condition) 1 control condition). PE specialist: credentialed PE specialists were employed and trained by the investigators to ensure full implementation of the intervention. In the trained classroom teacher condition, teachers were trained in the intervention methods. Note: this review only reported the Teacher‐led and control conditions. Policies, practices or programmes targeted by the intervention SPARK PE was designed to be a comprehensive programme for upper elementary students to increase PA. It was deigned to influence the quantity and quality of elementary PE lessons and the amount of PE through: ‐ Number of lessons per week ‐ Minutes of PE per week Implementation strategies EPOC: educational materials ‐ Written curriculum guide identified the programme philosophy and goals and included a yearly plan which was divided into instruction units with activity progressions within each unit. A detailed plan was provided for each PE lesson, which typically had 2 parts: health‐fitness activities and skill‐fitness activities. EPOC: length of consultation ‐ An additional 30 minutes per week was allocated for classroom instruction and practices in self‐management activities and skills. EPOC: other ‐ To support implementation of the curricula, equivalent types of equipment were provided to all 7 schools, including control schools, and replacement equipment was added each year. Trained classroom teacher condition EPOC:educational meetings ‐ Classroom teachers were trained to implement SPARK PE. EPOC:educational outreach visits ‐ On‐site support which was provided during the 3 years ensured the curriculum was followed. A PE specialist provided feedback, encouragement and direct assistance during schools visits. The specialist assisted teachers by leading grade‐level planning meetings, modelling lesson segments, co‐ordinating space and equipment, and giving verbal and written feedback after observing lessons. Theoretical underpinning: not reported Description of control: usual PE was implemented by untrained classroom teachers (usual care).
Outcomes	Outcome relating to the implementation of school policies, practices or programmes ‐ Duration (minutes) per week of PE lessons ‐ Frequency (per week) of PE lessons Data collection method: measured by direct observation by trained assessors for 1 full week twice a year in each school year. Validity of measures used: not reported; however, the measure was objective. Outcome relating to cost: not reported Outcome relating to adverse consequences: not reported Outcome relating to child diet, PA or weight status: PA (MVPA of students in classrooms and out of school PA as well as a fitness test) as well as height and weight. Data collection method MVPA of students in classrooms:SOFIT was used to obtain student activity levels. Codes were used to estimate energy expenditure associated with PA, which had been calibrated using heart rate monitoring and the system had been validated using Caltrac accelerometers. Out of school PA:accelerometer was the primary measure of PA for out of school. Fitness:1‐mile‐run test. Anthropometric measures:height and weight were measured in bare feet. Calf and triceps skinfolds were assessed 3 times using calibrated Lange calipers. Validity of measures used MVPA of students in classrooms: system validated using Caltrac accelerometers. Out of school PA: valid Fitness:objective Anthropometric measures:the interobserver agreement (intraclass correlations) was 0.87 for triceps skinfold and 0.93 for calf skinfold (n = 47). Anthropometry is a valid tool.
Notes	Notes: the SPARK trial included postintervention implementation outcome data only; however, they used random assignment for 6/7 included schools, randomly allocating them to 1 of 3 conditions. The remaining school was allocated by the researchers to the control group. Despite the lack of baseline implementation data, given the use of random assignment, and similarity of other trial sample characteristics, the trial was retained in the review. Research funding: NIH grant HL44467 Conflicts of interest: not reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Randomisation of all schools did not occur resulting in high risk of selection bias. Within each stratum, 1 school was randomly assigned to each of the 3 experimental conditions. To guard against loss of control schools, the remaining school was assigned to the control condition.
Allocation concealment (selection bias)	High risk	Randomisation of all schools did not occur and there was no indication that allocation was concealed and, therefore, were at high risk of selection bias.
Blinding of participants and personnel (performance bias) Implementation outcome	High risk	Outcome group: all. Given the nature of the intervention, participants and study personnel were likely to have been aware of study allocation and, therefore, high risk of performance bias.
Blinding of outcome assessment (detection bias) Implementation outcome	Unclear risk	Outcome group: observations of physical education classes. Implementation of the school physical education programme was assessed by direct observation by trained assessors not part of the intervention team. There was insufficient information about whether these assessors were blinded.
Incomplete outcome data (attrition bias) Implementation outcome	Unclear risk	Outcome group: observations of physical education classes. Insufficient reporting of attrition/exclusions to permit judgement of 'Low risk' or 'High risk' (e.g. no reasons for missing data provided).
Selective reporting (reporting bias)	Unclear risk	No study protocol, therefore, it was unclear if there was selective outcome reporting.
Other bias	Low risk	Appeared free from other bias.
Potential confounding	Unclear risk	There was insufficient information to determine the risk of potential confounders.
Overall risk of bias assessment	Unclear risk	Most domains were at unclear risk of bias.