Effect of the intervention on the behavioural constructs for the clinical behaviours 'managing patients without referral for plain x-ray' and 'advising patients to stay active'.

1No. Practices = 34 and No. GPs = 41, except for variable “Perceived behavioural control (indirect)” where No. Practices = 34 and No. GPs = 40.
2No. Practices = 38 and No. GPs = 42, except for variable “Behavioural intention (performance)” where No. Practices = 37 and No. GPs = 41.
3Only general practitioners who provided both baseline and follow-up measures of the behavioural constructs are included in the calculation of the means.
4Adjusted difference in means estimated from marginal linear regression models using GEEs with an exchangeable correlation structure and robust variance estimation to allow for clustering within general practices. For models where the estimated within cluster correlation was negative (indicated by*), the model was refitted assuming an independent correlation structure. All models adjusted for the design strata (number of GPs per practice, location of practice [metropolitan or rural/remote]) and confounders specified prior to undertaking the analysis (age of GP (years), special interest in LBP, baseline measure of the behavioural construct).
5Adjusted odds ratio estimated from marginal logistic regression models using GEEs with an exchangeable correlation structure and robust variance estimation to allow for clustering within general practices. For models where the estimated within cluster correlation was negative (indicated by*), the model was refitted assuming an independent correlation structure. Models adjusted for the same variables as in footnote 4.