Margit Kooijman

General discussion | 123 Influence of the therapist In an attempt to unravel some of the mystery, we investigated first whether a therapist effect exists in physiotherapists treating patients with shoulder pain and second if personality traits of the physiotherapist influence patients outcome. It was found that the physiotherapist-effect explained 12% of variance and that being more extravert was associated with better treatment results (chapter 6) . Literature on therapist effects in primary care settings is very scarce and on the influence of intrinsic factors non-existing. One study on patients with low back and neck pain compared three RCTs in which a wide variation of treatment types was applied. It found that less than 2% of the total variance in outcome was explained by treatment across trials whereas therapist effect was found to account for 3-7% of the variance 33 . The only study on the influence of personality traits of the physiotherapist on outcome (in patients with chronic complaints) confirmed inter-therapist variation of around 7% 34 . In addition, their findings carefully indicated that less neurotic therapists seemed to influence outcome positively. It is also known that a therapist effect is more pronounced in delivering psychosocial and behavioural treatment strategies than in, for example, manual therapy 33 . Psychosocial and behavioural treatments strategies are particularly required in patients presenting with long-lasting complaints, such as most patients with shoulder pain. Limitations As a consequence of being explorative, the methodology of this research is not flawless. By nature, personality traits are complex constructs; hence there are many ways to measure them. Combined with a lack of research on therapists’ effect on outcome, the specific findings of present study on the influence of being extravert should be interpreted with care and mainly considered an incentive for further research. Furthermore, part of the findings is based on previously gathered data with the disadvantage of not being able to influence its design. For the current study in particular, this shortcoming concerned that therapists determined treatment outcome based on input by the patient while no direct assessment of the outcome by patients themselves was registered. So where do we go from here? It seems that as clinicians we arrived at a time and place where we want to recognize we not only make interventions, but are part of the intervention as well. Although it is largely unknown which context factors are of importance, preliminary research in primary care settings point in the direction that context factors at the level of the therapist are operational, as they are known to be in other professions. Also, multi-level analysis is increasingly used to correct for such therapist effects 35 . And it makes sense that for persisting musculoskeletal pain such as shoulder pain, besides generic prognostic factors (common across pain sites) also factors common across therapies impact clinical outcome. Acknowledging this complexity suits the inclusive goal of positive health and