Table 3 Effect modification analyses
Baseline Characteristics | LRT p-value for interaction term between baseline characteristic and treatment group | Outcome of subgroup analysis |
|---|---|---|
Proportion of free school meals (school level) | 0.884 | The model with the interaction term does not do a better job of explaining the data than the model without the interaction term |
Median distance of home to school (school level) | 0.371 | |
IMD (school level): weighted IMD | 0.885 | |
IMD (pupil level): IMD quintile | 0.368 | |
Meeting CMO guidelines of an average of at least 60Â min MVPA per day on weekdays (pupil level) | 0.138 | |
Mode of transport from school (pupil level): active travel (walk or cycle/scoot) vs. non-active travel (car or bus/train) | 0.658 | |
Nominated peer supporter | 0.047 | The model with the interaction term does a better job of explaining the data than the model without the interaction term. Treatment effect among those who were nominated peer supporter (in intervention and control schools) is -5.98 (-10.34, -1.61) p < 0.01. Treatment effect among those who were non-nominated peer supporter (in intervention and control schools) is -1.90 (-5.18, 1.38) p = 0.26. Difference in treatment effect between nominated peer supporters and non-nominated supporters is estimated as -4.08 (-8.14, -0.01) p = 0.049 |
Proportion of sedentary time at baseline | < 0.001 | The model with the interaction term does a better job of explaining the data than the model without the interaction term. Proportion of sedentary time at baseline is a continuous measure. The difference in treatment effect is a function of proportion of sedentary time at baseline: -49.34 + 61.32*(Proportion of sedentary time at baseline). |