Study design

The study is a pragmatic RCT, with participants randomised individually to one of two parallel groups, incorporating economic evaluation and nested qualitative research. The comparison is between patients randomised either to be offered a service based on initial telephone assessment and advice from PhysioDirect, followed by face-to-face treatment when necessary, or to be offered usual care consisting of allocation of patients to a waiting list for face-to-face care.

The study was designed to assess equivalence between the two treatment groups in the primary clinical outcome. If equivalence in clinical outcomes is established, differences in the costs of providing care and in the secondary outcomes (particularly waiting times for treatment, time lost from work and usual activities, patient satisfaction, patient preference) become particularly important and relevant to future provision of services.

The equivalence design was chosen over a conventional superiority design because there was no hypothetical justification to suggest that the PhysioDirect service would produce better clinical outcomes than usual care by the main follow-up time point of 6 months. However, because it was anticipated that patients randomised to PhysioDirect would receive advice more quickly than those randomised to usual care it was conceivable that they would achieve improved outcomes at the first time point of 6 weeks, but this was a secondary outcome. Furthermore, although we wanted to test a hypothesis of equivalence in the primary outcome we also wanted to test for difference in some of the secondary outcomes. For these reasons the study was designed as an equivalence study rather than a non-inferiority study, as the former would require a larger sample size and would provide power to establish differences in the secondary outcomes.

It is arguable that the most important outcome for this study would be improved cost-effectiveness – for example, the PhysioDirect service might be slightly less effective but substantially less costly and therefore might be cost-effective. However, the different sources of uncertainty in cost-effectiveness analyses make it very difficult to design a study with cost-effectiveness as the primary outcome and in particular to estimate the appropriate sample size for such a study. Furthermore, the interpretation of findings that a service is less effective and less costly than usual care is contentious.⁶⁴

Setting

PhysioDirect services were newly established for the purposes of the trial in four areas of England. Each physiotherapy service provided care for patients from a defined group of general practices within one of the following PCTs: Bristol, Somerset, Stoke-on-Trent, and Central and Eastern Cheshire. The total population covered across all four PCTs was approximately 625,000 people.

Patients from 94 general practices participated in this study. Participating GP practices were typical of NHS general practices in England, representing a broad mix of practice sizes, the smallest serving a population of 2121 and the largest 28,599 people. These practices covered a wide range of types of area, including inner-city, suburban, market towns and rural areas. Although detailed information about the population of each participating practice was not collected, between them the practices provide care for patients of all age groups. The areas differed in terms of their ethnic mix, but none had a high proportion of patients from non-white ethnic backgrounds. Although 12% of the population of Bristol is from black and minority ethnic (BME) groups, the BME population in the city is concentrated in particular inner-city wards. By contrast, only 1.2% of the population of Somerset is from BME groups. About 5% of the population of Stoke is from BME groups (predominantly Pakistani and Bangladeshi). About 3% of the Cheshire population is from a range of different minority ethnic communities.

Between them, in 2008 the existing physiotherapy services in these PCTs received approximately 18,300 referrals from primary care professionals in the general practices participating in the trial. This equates to 29.3 referrals per 1000 patients per annum, which is slightly higher than the rate of 24.4 reported in the most recent available figures for England.¹³ However, these national data were last reported in 2004–5 and it was noted then that referral rates were rising by about 4% per annum.¹³

The physiotherapy services that participated in the trial were typical of NHS primary care-based physiotherapy in the UK. NHS physiotherapists in the UK accept referrals from primary and secondary care physicians, other health-care practitioners (such as nurse practitioners and allied health professionals) and, in some cases, offer direct access or self-referral pathways to physiotherapy. Treatment is delivered following a clinical assessment of the patient by the physiotherapist, and common interventions for MSK conditions include advice and education, specific and general exercise, manual therapy and pain-relieving modalities, such as electrotherapy and acupuncture.

Participants

The inclusion criteria were deliberately as broad as possible in order to maximise generalisability and to reflect the ‘real-world’ operation of PhysioDirect services. Inclusion criteria were adults (aged ≥ 18 years) who were referred by GPs or other members of the primary health care team or who referred themselves (self referred) for physiotherapy for a MSK problem.

Exclusion criteria were children (< 18 years); patients referred to physiotherapy by a hospital consultant, emergency department or primary/secondary care interface service; those needing domiciliary physiotherapy; those needing postoperative physiotherapy; those needing physiotherapy for non-MSK problems; very urgent referrals; those who did not confirm that they wanted physiotherapy; second or subsequent referrals for physiotherapy for an individual during the trial period; people unable to communicate by telephone in English; and those with hearing difficulties who may have had difficulty communicating by telephone.

It was necessary to exclude participants who had problems that were so urgent that it was not safe to delay assessment for the length of time it would take to gain consent to the trial. Frequently, referrals were designated as urgent by the referring clinician. It was important not to exclude most of these referrals, as one of the potential advantages of a PhysioDirect service is that it allows an early telephone assessment by a physiotherapist to prioritise those who need urgent care. Therefore, all referrals to each participating service were screened by a senior physiotherapist on receipt (see below) and only those which were deemed very urgent were excluded from the study (regardless of whether or not the referring clinician had marked the referral as urgent). The excluded patients were contacted and offered physiotherapy in the usual way for urgent cases.

Recruitment of participants

General practitioners or health-care professionals in the relevant practices referred patients to physiotherapy in the usual way, although patients could refer themselves in some areas of Stoke, as that service already had developed a self-referral pathway for patients for some GP practices. Referrals to each physiotherapy service were screened by a senior physiotherapist within one working day of receipt to confirm that the patient appeared to be eligible for the study. These potentially eligible patients were sent information about the trial by post from participating physiotherapy centres, along with a consent form and a baseline questionnaire.

The consent form asked potential participants to choose from one of three options: to consent to participation in the trial; to state that they wanted physiotherapy but did not want to participate in the trial; or to state that they no longer wanted physiotherapy.

Those who agreed to participate were randomised into the trial; those who declined participation but wanted physiotherapy were put on the waiting list for usual care; and those who did not want physiotherapy were discharged. All referrals were logged in order of when the referral form was originally received, so the process of gaining consent did not affect the patient's position on the waiting list for face-to-face physiotherapy if they ultimately received usual care.

If people did not reply to the information letter, they were sent a reminder mailing after 2 weeks to encourage them to respond. The reminder informed them that if they did not respond in any way within 2 weeks of the date the reminder letter was sent then they would be discharged and taken off the physiotherapy waiting list. This is standard practice, in that most physiotherapy services and many hospital outpatient services send patients a ‘partial booking’ letter inviting them to respond to request an appointment, and patients who do not respond are discharged and removed from the waiting list.

Randomisation

Patients who gave consent to participate and had completed the baseline questionnaire were randomised in a 2 : 1 ratio to PhysioDirect or usual care. This allocation ratio was chosen to ensure that sufficient patients were randomised to PhysioDirect to make this new service viable, given that all non-consenting and excluded patients continued to receive usual care, as well as those randomised to usual care. Randomisation was undertaken using web or telephone access to a secure remote automated allocation system maintained by the UK Clinical Research Collaboration (UKCRC)-registered Bristol Randomised Trials Collaboration. Allocation was made at the level of the individual, stratifying by physiotherapy site and minimising by sex, patient age group, and site of the presenting MSK complaint. Allocation was therefore fully concealed. Periodic checks were made during the study of the allocation ratios according to the minimisation factors, to ensure that the randomisation schedule was performing as intended.

Following randomisation, patients were sent a letter either inviting them to contact PhysioDirect and explaining how to do so (intervention arm), or explaining that they were on a waiting list for a face-to-face physiotherapy appointment and would be contacted by the service when the next appointment became available (control arm).

The number of patients excluded from or not participating in the trial for different reasons was recorded. The age, sex and postcode of all patients were also recorded in anonymised form, to make it possible to compare participating and non-participating populations.

Description of intervention and control arms

Intervention: ‘PhysioDirect’

The intervention is defined as the patient being offered treatment within the PhysioDirect treatment pathway. PhysioDirect is the provision of an easily accessible telephone assessment and advice service from an experienced physiotherapist, supported by a computerised assessment algorithm. Following the telephone assessment, patients are usually given exercise advice and then invited to telephone back to report progress. They can then be invited for a face-to-face appointment if necessary, or this can be offered following the initial assessment if appropriate.

The rationale for PhysioDirect is that all patients will have access to earlier assessment and advice about their problem from a physiotherapist. Furthermore, those patients most likely to benefit from face-to-face physiotherapy should be able to receive it more quickly (by filtering out the patients who do not need face-to-face care), hopefully leading to a faster clinical improvement and a quicker return to work and/or usual activities.

The term ‘PhysioDirect’ is used variably by different services in the UK, but the model used in this trial was modelled closely on the system developed by the Huntingdonshire PCT in 2001. This service was runner-up in the 2003 Health and Social Care Awards. Currently, 350,000 people in Cambridgeshire are served by this PhysioDirect system. The established nature of the service, the structured format of the system and the experience of the Huntingdon physiotherapy staff both in using and training other physiotherapists to use this system all made the system particularly suitable for use in the PhysioDirect trial.

Setting up the service

In order to establish a PhysioDirect service, each participating site needed to train sufficient physiotherapists to provide telephone assessments using the assessment software and to provide a suitable office. Apart from the training and set-up costs, no extra physiotherapist resources were used within the study – the time of the existing staff was reallocated within existing resources.

Training the physiotherapists

All physiotherapists operating the PhysioDirect telephone services undertook a structured training programme led by the physiotherapy service in Huntingdon which involved attending a 2-day course of teaching, demonstrations and observation of live calls. Each PCT trained about eight physiotherapists in order to provide sufficient capacity across the week (with some contingency in case of absence) and without any one physiotherapist having to spend more than half of their time working on the PhysioDirect service.

The PhysioDirect software (see below) was installed at each of the PCTs in the PhysioDirect trial prior to the trial physiotherapists attending the Huntingdon training so that following the training session the physiotherapists could practise and hone their skills. The Huntingdon PhysioDirect trainers were available by telephone throughout this period to provide advice if problems were encountered.

A visit to each participating PCT was undertaken by a PhysioDirect trainer approximately 2 weeks after completion of the training programme. The trainer observed calls and facilitated a problem-solving session. The trainer then listened to individual calls and, using a structured format, assessed each physiotherapist's competency to utilise the system safely and effectively. If physiotherapists did not reach the required competency at the first site visit a further visit to recheck competency was carried out approximately 6 weeks after the initial training. The competency check consisted of the trainer assessing 53 aspects of the telephone assessment process and completing a checklist indicating whether or not each aspect was performed to a satisfactory level. Each section was evaluated on a yes/no basis, with overall comments about performance, issues to be addressed and an agreed action plan if required. All physiotherapists had to be certified as competent to undertake PhysioDirect before they assessed patients in the trial.

Office facilities

Each site required a quiet office with at least two computers with installed software, and a telephone system with headsets for the physiotherapists, a fax machine and an answering machine. Ideally (but not necessarily), the receptionist would work close by in an adjacent office.

The PhysioDirect software

Shortly after its inception, the Huntingdon PhysioDirect system was translated into a computer-based system to assist the safe, efficient and effective delivery of the telephone assessment and advice service. The challenge during the development of the computerised system was to maintain sufficient structure to guide an effective and efficient assessment by prompting physiotherapists to cover all key aspects of the patient assessment, while being flexible enough to be responsive to presentations of individual patients and taking account of the lack of visual clues provided by a patient during a face-to-face consultation. The system had to be as simple as possible as physiotherapists undertaking telephone assessment are required to perform a number of skills simultaneously, including asking questions, visualising the patient and their presentation, analysing the patient's responses, formulating the next question and typing responses into the system. The solution was to include mandatory fields with dropdown menus and tick boxes for key aspects of the assessment, with text boxes provided to allow the physiotherapists to record responses to supplementary questions eliciting further information or clarification from the patient.

The software is a based on a program developed using Microsoft Access 2003 (Microsoft Corporation, Redmond, WA, USA), which first allows the patient's demographic details to be checked, then brief details regarding the patient's occupation, hobbies/activities and main problem for which they have been referred. When using the software the physiotherapist selects an appropriate assessment framework, depending on the area of the body affected. This ensures any relevant ‘special questions’ are asked that relate to that area. For example, if the patient is complaining of low back pain, questions regarding symptoms suggestive of cauda equina compression are included by the computer software in the assessment algorithm. This assessment, prompted by the software, includes the investigation and recording of the patient's presenting, medical and drug history, details of aggravating, easing and diurnal patterns and assists the physiotherapist in reaching a clinical diagnosis. An example screen from the system is shown overleaf.

Further information about the PhysioDirect software used in this trial can be obtained from Mrs Jill Gamlin, Physiotherapy Department, Hinchingbrooke Hospital, Hinchingbrooke Park, Huntingdon PE29 6NT, UK (e-mail: jill.gamlin@nhs.net).

Standardisation of PhysioDirect across the sites

All four physiotherapy services participating in this trial set up a PhysioDirect service following the same model of organisation and using the same assessment software as developed in Huntingdonshire (Figure 1).

FIGURE 1

Example screen shot from PhysioDirect assessment software. Reproduced with permission from Cambridgeshire Community Services NHS Trust.

As far as possible, the PhysioDirect service was implemented in the same way in each of the four sites. The physiotherapists received the same training and were accredited in the same way, they used the same PhysioDirect software, and they used the same suite of advice and exercise leaflets. Furthermore, all recruitment and other research procedures were standardised in all sites. In this way, a high degree of consistency in how the intervention was delivered in each local service was ensured.

After the physiotherapists had been trained and assessed as competent to assess and advise patients via PhysioDirect, each physiotherapy site carried out a run-in period of at least 1 month to ensure smooth running of the service before patients to be included in the main trial were recruited. During this run-in period, all trial procedures of recruitment and randomisation were followed, making it possible to fully pilot the research procedures in parallel with the physiotherapy services gaining experience of the new way of working. Follow-up data collection for most outcome measures was also undertaken, although with less intensive use of reminders than in the subsequent main trial because of resource limitations.

Staffing the service

Based on advice from Huntingdon, it was decided that only experienced physiotherapists would be involved in providing the PhysioDirect telephone advice. Each service therefore trained several of their more senior staff who were on Agenda for Change Band 6 or above to conduct the PhysioDirect sessions. Details of the training and experience of these physiotherapists were collected at the training sessions. These are described in Table 1.

TABLE 1

Sex, training and experience of physiotherapists providing PhysioDirect

Because (in the context of a trial) only a minority of patients were randomised to PhysioDirect, the number of patients did not justify having a physiotherapist available to answer the telephone at all times in the week. Each site was asked to provide telephone sessions spread across a range of days of the week and at different times of day, to allow patients some choice about when to telephone. These opening times were included in the letter inviting patients to telephone the service. If patients telephoned at other times, an answerphone message asked them to telephone again when the service was open.

Process of the encounter

As soon as possible after consent to participate in the trial was received, patients were sent a letter inviting them to telephone an experienced physiotherapist for initial assessment and advice and explaining the times during which the PhysioDirect service was available each week. When a patient called the service, if a physiotherapist was not engaged in a call, they would be the first person the patient spoke to and the assessment would start immediately. If the staff were all engaged with other calls then the first person they would speak to would be a receptionist. The receptionist would take the patient's details and place them on a ‘call-back’ list along with information regarding when would be most convenient for this to happen. The physiotherapists would either field any calls arriving if they were free, or consult the call-back list for available patients.

The physiotherapist responding to the telephone call followed the computer-assisted assessment system to assess the patient and record the findings.

Process of care following the encounter

There were several possible outcomes following the initial telephone call:

• In some cases, at the end of the call the physiotherapist posted a relevant advice leaflet about self-management and exercises to the patient, inviting them to telephone back to report progress after about 2–4 weeks, if appropriate. If the patient telephoned back they could be given further advice or be booked for a face-to-face consultation if necessary.
• In some cases, the initial telephone call established that face-to-face assessment and/or care was needed, in which case this was arranged either by putting the patient on the PhysioDirect waiting list for face-to-face care or by arranging an urgent appointment if necessary.
• In other cases, the initial assessment established that physiotherapy was unlikely to be effective and the patient was given appropriate advice or referred to another service, and discharged from physiotherapy.
• In a few cases the initial assessment revealed that urgent medical intervention was required and these patients were sent to accident and emergency (A&E) with an accompanying fax detailing the need for attendance. If the medical intervention was not considered ‘an emergency’ then patients were asked to attend their GP, for example to obtain a radiograph.

The assessment process allocates patients to one of five categories. These categories and the subsequent management pathways are summarised in Table 2.

TABLE 2

Patient categories and subsequent management pathways resulting from PhysioDirect assessment

Providing PhysioDirect in the context of a randomised controlled trial

Separate waiting lists were held for patients randomised to PhysioDirect and to usual care, so that any impact of the PhysioDirect pathway on how long people waited for a face-to-face appointment could be detected. For the same reason, we sought to allocate physiotherapist time in proportion to the number of people who were randomised to the PhysioDirect arm of the trial. Based on experience in the pilot study it was possible to estimate the proportion of patients referred for physiotherapy who would be eligible for the trial and the proportion of those who would consent to take part. Using these data, and also taking into account that one in three of the eligible, consenting patients would be randomised to usual care, it was possible to estimate that approximately 20% of all patients would be allocated to PhysioDirect rather than usual care. Each site was therefore asked to allocate 20% of their total physiotherapist staff resources to patients allocated to the PhysioDirect arm of the trial, and within this proportion to allocate about half the time to telephone sessions and the remainder to face-to-face care for people in the PhysioDirect arm.

Patients randomised to the PhysioDirect arm could state at any time that they did not want to discuss their problem on the telephone but instead wanted to wait for a face-to-face appointment. This was allowed within the PhysioDirect pathway and did not mean that the patient needed to withdraw from the trial. It is important to note that the PhysioDirect ‘intervention’ was based on offering patients an initial telephone assessment with a physiotherapist, with face-to-face care available if necessary. The ‘intervention’ is not the telephone call, and a face-to-face consultation should not be equated with ‘usual care’. The PhysioDirect intervention is the care pathway, which can encompass both telephone advice and face-to-face care.

Outcome assessment

Outcomes were assessed at baseline, and at 6 weeks and 6 months after randomisation.

The primary outcome was clinical outcome at 6 months, assessed using the Physical Component Score (PCS) measure from the Short Form questionnaire-36 items, version 2 (SF-36v2) questionnaire.⁶⁵ The SF-36v2 PCS is a well-recognised generic measure of health status. It was particularly suitable for this trial because, unlike disease-specific measures, it is applicable to the wide range of MSK problems referred to physiotherapy.

Although there could be a concern that generic measures may be less responsive than disease-specific measures, the physical functioning and bodily pain scales of the SF-36 (which contribute most of the variation in the PCS summary measure) compare reasonably well with disease-specific measures in patients with MSK problems.^66–70

Several further measures of clinical outcome were also used. The first was the Measure Yourself Medical Outcomes Profile (MYMOP)⁷¹ questionnaire (version 2), which is a patient-generated measure. It allowed patients to specify up to two symptoms and one functional limitation for which they had been referred to physiotherapy, then follow-up questionnaires assessed change in those specific symptoms/limitations. This individualised and validated measure has also been used to assess patients with a wide range of problems.⁷¹ The MYMOP was included as, by focusing on the patient's main problem, it might be more sensitive to change than the PCS. The MYMOP was designated an important secondary outcome.

Second, a single question was included as a global measure of individual rating of overall change, based on a seven-point Likert scale from ‘very much worse’ to ‘very much better’.

Third, we created a composite measure of response to treatment using the approach recommended by the Outcomes Measures in Rheumatology Clinical Trials – Osteoarthritis Research Society International (OMERACT-OARSI) initiative.⁷² This combined measures of physical function (the SF-36 physical function scale), pain (SF-36v2 bodily pain scale) and overall perception of the main problem (question 1 from the MYMOP questionnaire).

Other secondary outcomes were:

• costs (described in more detail later)
• quality of life [measured using the EQ-5D measure⁷³ (European Quality of Life-5 Dimensions; EuroQol health utility measure)]
• the individual scales and the mental component summary measure from the SF-36
• waiting times for treatment, based on data collected from routine physiotherapy records
• time lost from work and usual activities, based on patient questionnaires
• satisfaction with care provided (development of the measure is described in Appendix 2)
• preference for physiotherapy or usual care if they needed physiotherapy in future, based on a single question in the patient questionnaire.

Collection of data

Baseline data about patient characteristics, and the nature of the problem for which patients were referred to physiotherapy, were collected from the referral forms at each participating physiotherapy centre.

Outcome and process data were collected from questionnaires administered to patients at baseline, 6 weeks and 6 months after randomisation, from electronic data downloaded from the PhysioDirect software, from routine records of consultations collected by each physiotherapy service, and from general practice records.

The baseline patient questionnaire collected data about patient characteristics and about the outcome measures. Data about outcomes at follow-up were collected from similar questionnaires sent by post at 6 weeks and 6 months after randomisation. Non-responding patients were sent a first reminder questionnaire by post after 2 weeks, and a second after a further 2 weeks, if needed. If patients did not respond to this second reminder within 2 weeks, attempts were made for up to a further 2 weeks to collect outcome data by telephone. In this situation, priority was given to obtaining data about the primary outcome (the SF-36v2 PCS). Similarly, if patients did complete postal questionnaires but key questions were omitted then telephone contact was attempted in order to collect missing question items that were necessary to calculate the primary outcome.

Some ‘screening’ questions were included to identify patients who had accessed rarely used but potentially expensive resources (e.g. hospital admissions) because of the problem for which they were referred for physiotherapy. Patients who responded positively to these questions were then telephoned to obtain more detailed information.

Data about telephone consultations with the PhysioDirect service were collected by download from the assessment software used in this service. Data about face-to-face consultations were obtained from routine records maintained by each physiotherapy service. Data about primary care consultations were obtained from general practice notes, extracted by research staff. Details of consultations with NHS walk-in centres, emergency departments, hospital outpatient appointments and admissions were obtained from the patient questionnaire, as were details of non-NHS consultations.

Details of the physiotherapists participating in the PhysioDirect service were obtained from questionnaires administered before these physiotherapists undertook training to use the PhysioDirect software. Details of the grades of the other physiotherapists providing face-to-face consultations were obtained from the managers of each service.

Table 3 lists the various outcome measures, the timing of data collection and the source of the data.

TABLE 3

Outcome measures, timing of data collection and source of data

Data entry

All questionnaire data were entered into a bespoke database in Microsoft Access v2000. A 1-in-10 sample of each questionnaire type was independently checked to assess the accuracy of data entry. Of 58,615 fields checked, there were just 73 errors (0.125% error rate).

Blinding

In pragmatic trials of this type, as in the real world, it is not possible to blind participants or physiotherapists, whether they are in the intervention or control arm of the trial.

Most of the outcome data were obtained from patient questionnaires, which were therefore not blind to treatment allocation. However, all data entry from these questionnaires was conducted blind to allocation. The collection of data about PhysioDirect calls was by electronic download. The collection of data about face-to-face physiotherapy consultations from administrative records and from general practice records was conducted by research staff not involved in providing the intervention, as far as possible blind to allocation, although in some cases (7%; 160/2228) this was not possible because the allocation was recorded in the records.

Sample size and power

This study was powered to establish clinical equivalence using the PCS scale from the SF-36. At the time of planning the study, previous studies in a range of populations and conditions had suggested that a minimum clinically important difference on the SF-36v2 PCS was about 4 points [0.4 standard deviation (SD)],^74–77 although the latest version of the SF-36v2 manual now suggests that ‘a minimum important difference of 2–3 points is reasonable’.⁷⁸

In this study, a difference of no greater than 2 points was conservatively specified as demonstrating equivalence. This is equivalent to an effect size of 0.2 SDs, which is considered a small effect size.⁷⁷ Sample sizes for analysis of 976 and 488 in the PhysioDirect and usual-care groups, respectively, would yield 95% power to reject a null hypothesis of non-equivalence (i.e. the difference in means, μ_PhysioDirect − μ_{usual care}, is not less than 0.2 SDs) with an overall two-sided alpha of 0.05 alpha if the observed difference in means is zero. The same sample size would yield 80% power to reject the null of non-equivalence if the observed difference between the groups is 0.046 SDs rather than zero.

The target sample size for patients completing the final 6-month follow-up questionnaire was 1000 patients in the PhysioDirect arm and 500 patients in the usual-care arm. Assuming 20% non-collection of the primary outcomes, it was necessary to recruit 1250 and 625 patients in the PhysioDirect and usual-care arms, respectively, or 1875 patients in total.

Statistical methods

The main hypotheses were that care via the PhysioDirect service would be clinically equivalent to usual care and more cost-effective at 6 months after randomisation.

Economic analysis and qualitative research

The methods for the economic analysis and for the qualitative research are described in detail in subsequent chapters, after the main quantitative results from the trial.

Investigation of adverse events and serious adverse events

General practitioners from participating practices and physiotherapists treating trial patients were asked to report any death, hospitalisation, significant disability or incapacity, life-threatening circumstance or other medically significant occurrence that he/she considered may be potentially related to physiotherapy, or to the trial procedures, to the Chief Investigator within 5 days of occurrence. The Chief Investigator would investigate these using a predetermined set of criteria and report any related and unexpected serious adverse events (SAEs) to the Main Research Ethics Committee, within 15 days of becoming aware of the event, using the SAE report form for non-CTIMPs (Clinical Trial of an Investigational Medicinal Product) taken from the National Research Ethics Service website. A log would be kept of all AEs or SAEs reported to the Chief Investigator; his/her assessment of the intensity, causality, expectedness and seriousness of the AE; the reasons for those decisions; and (when appropriate) the dates on which the SAE was reported to the main Research Ethics Committee, Sponsor, and Data Monitoring and Ethics Committee.

Ethics and research governance approval

Multisite research ethics approval was obtained from Southmead Research Ethics Committee, Reference 08/H0102/95. All necessary research governance approvals were also obtained for each physiotherapy site.

Trial registration

Current Controlled Trials ISRCTN55666618.

UK Clinical Research Network (UKCRN) 4778.

The trial protocol has been published.⁸⁴

Summary of changes to the project protocol

The following changes were made to the original protocol after it was funded. The first five changes were all made before the trial started, the sixth was included a priori in the trial analysis plan, and the seventh was made during recruitment. All were approved by the Trial Steering Committee.

1. Patients who were referred by a primary care health professional or self-referred to physiotherapy were also included, rather than only those referred by a GP.
2. Needing physiotherapy for non-MSK problems was added as an exclusion criterion.
3. Sex was included as a minimisation factor in the randomisation procedure.
4. Analysis of health status in order to calculate quality-adjusted life-years (QALYs) was based on the EQ-5D measure rather than the Short Form questionnaire-6 Dimensions (SF-6D) measure derived from the SF-36v2.
5. The primary outcome was originally described as the clinical outcome assessed using two measures: the SF-36v2 PCS and the MYMOP. However, after discussion with the TSC it was agreed that having two primary outcomes could lead to problems in interpretation of findings. Therefore, the SF-36v2 (as the better established measure) was designated as the primary outcome and the MYMOP as an important secondary outcome. Wording was changed to clarify that the equivalence limit for the PCS was based on 2 points rather than 0.2 SDs.
6. The recruitment rate was defined as the percentage of patients consenting to participate from those responding to the initial letter to confirm that they did want physiotherapy. Patients who did not respond to the initial invitation or reminder letters from the physiotherapy service, despite reminders, were taken off the physiotherapy waiting lists and were defined as not eligible for the study.
7. Because of lower than anticipated follow-up rates at 6 months in patients recruited in the run-in phase it was agreed to continue inviting patients to the main trial until 2000 had given consent to participate, recognising that the final total recruited would be about 2143 patients because of the lag between invitation and consent. This sample size would provide the target sample size of 1500 patients completing questionnaires at 6 months, even if only 70% were followed up successfully.