Background: Non-pharmacological interventions such as education, exercise, and weight loss (if necessary) are core to the management of Osteoarthritis (OA). The role of nurses in managing symptomatic knee OA has been advocated but whether nurses can deliver such interventions as a complex package of care is unknown. The overall aim of this research was to develop and test the feasibility of a nurse-led complex intervention for knee pain comprising non-pharmacological and pharmacological components. The specific objectives of this thesis were to: 1) Systematically review the literature evaluating complex interventions for knee pain due to OA, 2) Evaluate fidelity of delivery of a nurse-led non-pharmacological complex intervention for knee pain, 3) Assess the acceptability of the non-pharmacological component of the intervention, issues faced in delivery, and resolve possible challenges. Methods: Systematic review and meta-analysis of complex interventions for knee pain due to OA: A systematic literature search was conducted on MEDLINE, EMBASE, AMED, PsycINFO, and CINAHL up until September 29th, 2020. Randomised Controlled Trials (RCTs) comprising at least patient education, exercise, and weight loss interventions were searched. Data were extracted by a single reviewer and cross-checked by two others. Standardised mean differences (SMD) and 95% confidence intervals (CI) were calculated using the random-effects model. The risk of bias was assessed with the Revised Cochrane risk-of-bias tool, and intervention reporting with the template for intervention description and replication (TIDieR) checklist. The primary outcome of interest was knee pain. Package development phase: 18 participants with knee pain (five with mild severity, eight with moderate, and five with severe) participated in a single-arm study. The fidelity and acceptability of a nurse-led non-pharmacological intervention comprising assessment, education, exercise, use of hot/cold treatments, footwear modification, walking aids, and weight-loss advice (if required), delivered in 4 sessions over 5 weeks were evaluated. Fidelity of delivery of intervention: Each intervention session with every participant was video recorded and formed part of the fidelity assessment. Self-reported fidelity checklists were completed by the research nurse after each session and by an independent researcher, after viewing the video recordings blinded to nurse ratings. Fidelity scores (%), percentage agreement, and 95% Confidence Intervals (CI) were calculated. Two semi-structured interviews were conducted with the research nurse. Acceptability assessment of the non-pharmacological components: Eighteen adults with chronic knee pain (defined as pain for longer than three months) were recruited from the community. The intervention comprised holistic assessment, education, exercise, weight- loss advice (where appropriate), and advice on adjunctive treatments such as hot/cold treatments, footwear modification, and walking aids. Participants had one-to-one semi- structured interviews at the end of the intervention. The nurse was interviewed after the last visit of the last participant. These were audio-recorded and transcribed verbatim. Themes were identified by one author (PAN) using framework analysis of the transcripts and cross-checked by another (AF). Results: Systematic review ad meta-analysis of complex interventions: We reviewed 2,649 titles and abstracts in the systematic search. The screening process identified twenty RCTs recruiting 3,069 participants with knee OA. Twelve RCTs were included in the meta- analysis. More than half of the studies were judged to be of high quality. The completeness of intervention reporting was poor. Complex interventions for OA produced moderate benefit for pain relief (-0.47, 95% CI -0.77, -0.16) and physical function (-0.49, 95% CI -0.72, -0.25). However, studies delivering non-pharmacological interventions for knee OA rarely reported both fidelity of delivery and acceptability of non-pharmacological interventions. Fidelity of delivery of intervention: Fourteen participants completed all visits. 62 treatment sessions took place. Nurse self-report and assessor video rating scores for all 62 treatment sessions were included in the fidelity assessment. Overall fidelity was higher on nurse self- report (97.7%) than on objective video-rating (84.2%). The percentage agreement between nurse self-report and video-rating was 73.3% (95% CI: 71.3 - 75.3). Fidelity was lowest for advice on footwear and walking aids. The nurse reported difficulty advising on thermal treatments, footwear, and walking aids, and did not feel confident negotiating achievable and realistic goals with participants. The nurse found the discussion of goal setting to be challenging. Acceptability assessment of the non-pharmacological components: Most participants found the advice from the nurse easy to follow and were satisfied with the package, though some felt that too much information was provided too soon. The intervention changed their perception of managing knee pain, learning that it can be improved with self-management. However, participants thought that the most challenging part of the intervention was fitting the exercise regime into their daily routine. Conclusion: A non-pharmacological package of care comprising patient education, exercise, and weight loss advice is more beneficial than usual care or any other single non- pharmacological component. A trained research nurse could deliver such a non- pharmacological package of care with high fidelity and acceptability for the participants and the nurse delivering the intervention. Future research should consider measuring the fidelity of delivery of intervention and acceptability in a real-world primary-care setting before evaluating it further in a multicentre RCT. Measuring the extent to which components are delivered as intended across different settings and populations, fidelity research may assist to understand which intervention components are effective and in which situations.