Oy physical activity may be more amenable to a sport-for-health intervention.

Of the 1393 potentially eligible children at T0, 1143 SB 202190 manufacturer completed baseline measures (92 response rate); 961 children completed assessments at T0, T1, and T2 and were included in the final analyses (84 participation rate). BMC Public Health (2016) 16:Page 6 ofyear follow-up. To determine whether the intervention effect was different for boys and girls, a dichotomous interaction term (labelled `sex') was constructed. Regression coefficients in each model were assessed for significance using the Wald statistic with one degree of freedom. As the prevalence of negative attitudes towards smoking was high in both study arms, odds ratios were converted to relative risks [66] to avoid overestimation of effects and for ease of interpretation of results. Analyses were performed using IBM SPSS Statistics v.22 and MLwiN 2.30 software (Centre for Multi-level Modelling, University of Bristol, UK). Statistical significance was set at p title= journal.pcbi.1005422 groups were transcribed verbatim, imported into NVivo 10 software, and subjected to thematic analysis [67]. This process involved reading and re-reading text and assigning broad thematic codes, some of which werepre-defined from topics covered in the group schedule. Subsequently, broad codes were collapsed into higher and lower order themes and descriptive and interpretive summaries were written based on recursive engagement with the data. A combination of inductive analysis and deductive techniques title= fpsyg.2015.01865 were used to generate codes. To aid the credibility and trustworthiness of the results, analyses and interpretations of the data were discussed amongst three members (CM, JT and LF) of the research team [67].Results Figure 2 shows the flow of schools and participants through the trial. In total, 43 schools participated in the study (27.9 response rate), including 32 (31 ) from Liverpool and 11 (22 ) from Knowsley. Schools that declined to participate provided diverse reasons for not taking part (e.g., too busy, key teacher on sick leave, already inFig. 2 Flow of schools and participants through the study. *Six children were absent at both post-intervention and follow-upMcGee et al. BMC Public Health (2016) 16:Page 7 ofreceipt of external projects). Of the 1393 potentially eligible children at T0, 1143 completed baseline measures (92 response rate); 961 children completed assessments at T0, T1, and T2 and were included in the final analyses (84 participation rate). Participant retention ranged from 80 (T0) to 79 (T2) in the comparison group. The intervention group's retention ranged from 83 at baseline to 68 at T2. However, the withdrawal of two intervention schools due to internal staffing issues excluded 68 children. Had the schools not withdrawn, assuming all children would have continued through the study, the retention at follow-up would have been 74 . Compared with intervention children included in the analyses, a higher proportion of intervention children that were excluded from the analyses had a sibling that smoked (p