This multicentre, three-arm sham-controlled randomised trial will be conducted in 43 individual Physiotherapy Centres that provide services for Primary Care Physicians located in 21 NHS Trusts situated in the Midlands and Cheshire regions of the UK. Multi-centre ethical approval has been obtained from the West Midlands Multicentre Research Ethics Committee and local approval was given by 12 ethics committees (Southern Derbyshire, Shropshire, Worcestershire, Warwickshire, Mid Staffordshire, South Staffordshire, Sollihull, North Birmingham, South Birmingham, West Birmingham, East Birmingham, East Cheshire). The trial was designed by a steering group with expert input from physiotherapists, an acupuncture specialist and trial methodologists. Information will be collected from the individual participating physiotherapists (demographics, current training and use of acupuncture, attitudes and beliefs about knee pain, and beliefs and expectations of the three treatment packages being compared in the trial) prior to the commencement of the trial and after the trial has been completed (Table 1).

Participants include patients with knee pain aged 50 years and over referred to physiotherapy centres by their general practitioner. Participants will be randomised to one of three groups: (i) advice and exercise alone, (ii) advice and exercise plus true acupuncture, (iii) advice and exercise plus sham acupuncture. Follow up will be at 2 weeks (by telephone), 6 weeks, 6 months and 12 months after randomisation, by postal questionnaire. Non-responders will be followed up.

Eligible patients are male and female subjects aged 50 years and above with pain (with or without stiffness) in one or both knees presenting to primary care. They must be naïve to acupuncture treatment (i.e. have never experienced acupuncture before for their present or any past complaints), and considered suitable for referral to a physiotherapy outpatients department by their general practitioner. Participants must be able to read and write English, be willing to consent to participation, and able give full informed consent. They must also be available for telephone contact.

Patients with potentially serious pathology (e.g. inflammatory arthritis, malignancy etc) on the basis of general practice or physiotherapy diagnosis or from past medical history, those who have had a knee or hip replacement on the affected side(s), are already on a surgical waiting list for total knee replacement, or for whom the trial interventions are contraindicated are excluded from the trial. Those who have received an exercise programme, from a physiotherapist, for their knee problem within the last 3 months (normal recreational involvement in sport or exercise will not be an exclusion) or an intra-articular injection to the knee in the last 6 months are also excluded.

Eligible patients referred by their GP to the physiotherapy departments will be invited to take part. Recruitment will take place over 18 months and will operate in one of two ways (See Figure 1):

A minimum of 48 hours after receiving the information leaflet, patients will be telephoned by the nominated local assessor/trial nurse (within 10 working days) to further screen eligibility. Information will be recorded on a standard proforma. For those patients either not eligible or not willing to be recruited, the proforma will be used to detail the reason for ineligibility, or reason for decline. Where patients are willing, two additional questions are asked to those that decline to participate in the trial to capture their treatment preference and expectations with respect to acupuncture and an advice & exercise treatment package. This information will be anonymised.

For patients willing to be recruited to the trial, an appointment is arranged for a research assessment at the patient's local physiotherapy department. After gaining verbal consent, the patient will be posted the baseline questionnaire to complete prior to their research appointment. At the research assessment visit the local physiotherapy assessor/trial nurse will perform a more detailed eligibility screen, explain the study, gain informed written consent to randomisation and conduct a baseline research interview and examination. Following consent to the study, the participant will be registered with the Research Centre by fax and allocated a unique trial number. The baseline assessment will be carried out blind to subsequent treatment allocation. An appointment will be made for the treating physiotherapist to begin treatment within 10 working days of the research assessment. Consent to treatment will be gained from each treating physiotherapist prior to commencing treatment, as is current physiotherapy practice. All participants will have an initial clinical physiotherapy assessment and treatment session of up to 40 minutes duration. During this session the physiotherapist will identify and record potential acupuncture points to be used should the participant be randomised to receive acupuncture (true or sham). A minimum of 6 and maximum of 10 points will be selected, based upon the participant's presentation and the clinical opinion of the physiotherapist. This will be carried out as part of the overall physical examination of the knee – the therapist will not draw the participant's attention to the localisation of acupuncture points to avoid raising their expectations about the possibility of receiving acupuncture. The advice and exercise package will then be started during this initial treatment visit.

Randomisation will take place after this initial physiotherapy session. The treating physiotherapist will telephone the Research Centre at Keele University, during normal working hours. This methodology ensures that the initial physiotherapy assessment and advice and exercise package provided is performed blind to subsequent treatment allocation. The specific trial interventions will commence during the participant's second treatment visit. During the randomisation telephone call the physiotherapist will be asked to identify the selected acupuncture sites to check that the participant has received their first pre-randomisation treatment session. The physiotherapists will also be asked questions about their own expectations of the individual participant's likely clinical outcome and their beliefs about which treatment they would like the participant to receive. Participants recruited to the trial will then be randomised to one of the three trial interventions in a 1:1:1 ratio based on their unique trial number. Computerised third-party randomisation will be performed using random permuted blocks of 12 (blocked by treatment centre).

The interventions will be delivered within 10 working days of randomisation by experienced physiotherapists, trained in acupuncture to at least the minimum standard for basic membership of the Acupuncture Association of Chartered Physiotherapists (AACP) (35 hrs of training). The participant's GP will be contacted at the time of randomisation and asked to avoid co-interventions for the period of the trial wherever possible, but especially until the 6-week follow up has been completed. However, if the GP feels that symptoms are sufficiently troublesome to need further treatment this will be at the GPs discretion. Information about co-interventions will be collected in the follow-up questionnaires and by review of a sample of participants' clinical records.

Using a standard proforma, the physiotherapists record the number and duration of treatment sessions each participant receives, plus details about the advice and exercises prescribed, the location and number of acupuncture points (where applicable) and any adverse reactions. The sensation that needling (true or placebo) evokes has been shown to be a significant correlate of acupuncture-induced analgesia (this has been a finding from both clinical and experimental studies). Hence, the sensation evoked from each treatment in the acupuncture groups will also be recorded. Acceptability and credibility of the interventions will be evaluated using a telephone follow-up at 2 weeks from the beginning of treatment and a questionnaire administered at 6 weeks 60.

Participants who give verbal consent to participate in the trial are posted a baseline questionnaire which they are asked to complete and bring with them to the research interview and examination. Information collected on this questionnaire is detailed in Table 2. These variables will be used to describe the study sample and as baseline measures of outcomes. Immediately following written informed consent, participants undergo a research interview and examination which follows a previously published schedule 66.

Outcome measures will be performed at 2 weeks (Table 3), 6 weeks, 6 months and 12 months (Table 4). Follow-up assessments will be performed using a telephone call at 2 weeks after the first treatment and self-completed postal questionnaires at all other time points. Non-responders will be telephoned 2 weeks after mailing the follow-up questionnaire on up to 2 occasions and posted a replacement questionnaire with a reminder letter if there is still no response at 4 weeks.

The primary outcome measure for this trial is the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain sub-scale 61. We have defined overall success as a 20% difference in the WOMAC pain sub-scale between true acupuncture and advice and exercise alone at 6 months. For this comparison, a minimum of 90 participants is needed in each group to reject the null hypothesis with 80% power and at a 5% significance level (two-tailed) 68. As our trial will compare sham acupuncture with advice and exercise alone we have three groups and so need 270 participants. Allowing for a 30% drop out rate in those recruited to the trial, the total number of participants required to be randomised is 350.

Collection of data and statistical analysis will be performed blinded to treatment allocation. Analysis will be performed on an intention to treat basis and the primary outcome will also be analysed on a "per protocol basis".

Univariate analysis will be performed using t-tests to analyse numerical data and chi-square tests for categorical data. The clinical and demographic data collected at baseline will be inspected and if there are any important differences between the trial groups, these factors will be used as covariates. These analyses will be performed using ANOVA and logistic regression as appropriate.

The analysis of the secondary outcomes will be exploratory. A univariate analysis with respect to the different treatment will be performed for the WOMAC pain sub-scale at 6-weeks and 12-months and for the WOMAC functioning sub-scale at 6-weeks, 6- and 12-months. The global outcome assessment with be analysed both as an ordinal and a dichotomous variable (categories 1–3 defined as "success"). Moreover, area under the curve slopes will be calculated for each treatment group over the whole treatment period and compared.

Statistical significance will be set at the 5% level (two-tailed). Statistical analysis will be performed using Stata 7.0. The trial will be monitored by an independent Data Monitoring and Ethics Committee. No interim analysis of the primary or secondary outcomes will be undertaken during the trial period.