Measuring practitioner/therapist effects in randomised trials of low back pain and neck pain interventions in primary care settings

In psychological health treatment studies it has been shown that differences between therapists account for some of the non-specific effect of treatment but this phenomenon has not so far systematically been investigated in musculoskeletal disorders. In this study we evaluated and compared the size and potential influence of the 'practitioner effect' (or 'therapist effect') in three randomised treatment trials of low back pain and neck pain patients in primary care. We calculated the proportion of variance in outcomes attributable to differences across practitioners, i.e. the practitioner-variance partition coefficient (p-vpc). As measures of outcome, we focused on self-reported disability as the primary outcome, but we also investigated assessed psychological outcomes. The p-vpc for the disability measures ranged from 2.6% to 7.1% across trials and time points (post treatment and follow up). Estimates differed between treatment subgroups within trials; being highest in treatment subgroups assigned to psychosocial-based interventions. A 'practitioner effect' does exist and is more pronounced in treatments involving greater psychosocial emphasis. This has implications for both practice and research in this clinical area. It highlights the importance of patient-practitioner interactions, and the need to address practitioner effects in designing and analysing outcome studies in low back pain and neck pain in primary care.