Table 3. Effect of the intervention on adherence as measured by the vignettes.
| Variable | No. practices (no. GPs) | Rx group adherence | Control group adherence | Adj OR1 (95%CI) | Incremental Effect2 (SE)3 | |||
| Rx | Control | No. | % | No. | % | |||
| X-ray 4 | 31 (38) | 36 (40) | 126/152 | (83) | 109/160 | (68) | 1.76* (1.01, 3.05) | 0.099 (0.052) |
| Imaging 4 | 31 (38) | 36 (40) | 119/152 | (78) | 89/160 | (56) | 2.36**(1.48, 3.79) | 0.177**(0.056) |
| Activity 5 | 31 (38) | 36 (40) | 121/152 | (80) | 82/160 | (51) | 4.49**(1.90,10.60) | 0.297**(0.044) |
| Bed rest 6 | 34 (41) | 38 (43) | 163/164 | (99) | 168/171 | (98) | 2.91 (0.30,27.83) | 0.011 (0.012) |
: p<0.05; **: p<0.01.
X-ray adherence defined as GPs not referring for a lumbosacral plain x-ray.
Imaging adherence for vignettes was defined as GPs not referring for any of following three diagnostic tests: lumbosacral plain x-ray, lumbar CT scan, lumbar MRI.
Activity adherence defined as “Advise the patient to continue with their normal daily activities” regardless of other interventions selected (“Paracetamol”, “Non-steroidal anti-inflammatory drugs”, “Advise the patient to do specific back exercises”, “Advise the patient to do general exercises (e.g. walking)”,”Manual therapy”, “Referral to another health care provider”, “Other”).
Bed rest adherence defined as either not recommending “Bed rest”, or recommending “Bed rest” for ≤ 2 days.
Adjusted Odds Ratio (OR) = Estimate of intervention effect adjusted for design strata and potential confounders (specified prior to undertaking the analysis). Adjusted OR estimated from models fitted using xtgee family(binomial) link(logit) vce(robust) yielding semi-robust standard errors.
Incremental effect = change in probability that simulated consult will be adherent to the key messages of the CPG due to exposure to the intervention after controlling for design strata and potential confounders (specified prior to undertaking the analysis). Here, incremental effects derived from model predicted values using method of recycled predictions [24].
Standard errors derived from bootstrap using bsmultiv.do [24].
Models adjusted for the following design strata and pre-specified potential confounders: GP age (years), years since GP graduated, self-reported special interest in LBP, number of GPs per practice, practice method of billing, rural/metro practice.
Models adjusted for the following design strata and pre-specified potential confounders: GP age (years), years since GP graduated, self-reported special interest in LBP, number of GPs per practice, practice method of billing, rural/metro practice, baseline measure of fear-avoidance beliefs.
No adjustment for stratification variables or potential confounders because of limited events of non-adherence.