Margit Kooijman

122 diagnosis, when using musculoskeletal ultrasound, findings should at least always be placed in the clinical context based on physical examination and especially history taking. In patients without a clear biomedical diagnosis, GPs and physiotherapists should be very aware that concomitant neck pain, long duration at the start of treatment and high pain intensity are important drivers of (non-)recovery. More specifically, presence of (one of) these factors requires extra attention on keep/refer decision-making and prioritizing of patient education. There seems to be promising consensus on generic prognostic factors. Still the majority of studies on this topic concentrated on patients with low back pain. For this reason, there is a need to enlarge the body of evidence for prognostic factors in other anatomical pain sites to assess true generality. Prognostic models appear to match our broader aim towards health and vitality, yet its usefulness for clinical care is largely unknown. Evidence for such new categorisations is indispensible, as is information on appropriate measurement intervals. What is more, not only clinicians but also patients are used to treatment based on diagnosis. In my opinion, it would be worthwhile to investigate how they feel about refraining from a diagnosis and targeting treatment based on prognosis; ‘clinical usefulness’ is also about the patients’ trust in and mandate for the proposed treatment. III Context factors “In the absence of effective treatment, clinicians have always understood that prognosis can be highly variable in persons with a particular diagnosis” 23 . This mystery ‘variable’ is likely endorsed by many GPs and physiotherapists in clinical practice. It concerns factors other than specific treatment effect including characteristics related to the patients, practitioners, setting and the way treatments are designed and delivered 26,28,30 . Non-specific or context factors have been investigated within the field of psychotherapy for a longer time and shown to account for between 0 to 18% of variability in patient outcomes 31 . More recently they have started to get recognized in research on patients with musculoskeletal complaints too. Context factors might explain the heterogeneity between responses such as expectations regarding treatment 27 . At the same time, these factors are an explanation why across large numbers of trials, conditions and treatment strategies a similar pattern of improvement emerges. The size of the influence of context factors is largely unknown but it is suggested that it dominates any individual response to treatment 26,30,32 .