Methods and analysis - This research comprises a prospective questionnaire-based cohort study with a nested mixed-methods study. A minimum of 144 practitioners will be recruited from public and private sector settings (48 physiotherapists, 48 osteopaths and 48 acupuncturists). Practitioners are asked to recruit 10–30 patients each, by handing out invitation packs to adult patients presenting with a new episode of LBP. The planned multilevel analysis requires a final sample size of 690 patients to detect correlations between predictors, hypothesised mediators and the primary outcome (self-reported back-related disability on the Roland-Morris Disability Questionnaire). Practitioners and patients complete questionnaires measuring non-specific treatment components, mediators and outcomes at: baseline (time 1: after the first consultation for a new episode of LBP), during treatment (time 2: 2 weeks post-baseline) and short-term outcome (time 3: 3 months post-baseline). A randomly selected subsample of participants in the questionnaire study will be invited to take part in a nested mixed-methods study of patient–practitioner interactions. In the nested study, 63 consultations (21/therapy) will be audio-recorded and analysed quantitatively and qualitatively, to identify communication practices associated with patient outcomes.
Ethics and dissemination - The protocol is approved by the host institution's ethics committee and the NHS Health Research Authority Research Ethics Committee. Results will be disseminated via peer-reviewed journal articles, conferences and a stakeholder workshop.