Stratified versus usual care for the management of primary care patients with sciatica: the SCOPiC RCT

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 12/201/09. The contractual start date was in November 2014. The draft report began editorial review in March 2019 and was accepted for publication in November 2019. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Sue Jowett is (from 2016 to present) a member of the National Institute for Health Research Health Technology Assessment Clinical Trials Committee and reports personal fees as an independent advisor at the Pfizer (Pfizer Inc., New York, NY, USA) chronic pain advisory board meeting in November 2018, outside the submitted work. Kate M Dunn reports grants from the Wellcome Trust (London, UK), during the conduct of the study.

Disclaimer

This report contains transcripts of interviews conducted in the course of the research and contains language that may offend some readers.

Copyright statement

© Queen’s Printer and Controller of HMSO 2020. This work was produced by Foster et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2020 Queen’s Printer and Controller of HMSO

Background

The term sciatica is commonly used to describe symptoms of pain radiating from the lower back to the leg, often extending to the foot. 1,2 Patients may also have other leg symptoms such as pins and needles, numbness or leg muscle weakness. 1,2 Occasionally, there may not be any lower-back pain (LBP), and, commonly, the leg pain is worse than the LBP. 1,2 The most common reason for sciatica is compression or irritation of a lumbar spinal nerve root by a prolapsed or bulging disc; less often, it is due to tightening of the central or lateral spinal canal (spinal stenosis). 1,2 Some clinicians prefer to use only the terms lumbar nerve root pain, lumbar radicular pain or radiculopathy, to distinguish it from non-specific, referred leg pain, which may arise from spinal structures such as muscles, ligaments or facet joints, rather than compression of a nerve root. 3 However, the term sciatica is still used widely to indicate symptoms arising from lumbar nerve root compression, mainly from a disc prolapse,4 and this is the term used throughout this report.

There is significant variation in sciatica prevalence estimates in published studies, from 1.2% to 43%,5 depending on each study’s diagnostic criteria and sampling methods. 5–7 In clinically evaluated general population cohorts, the point prevalence of sciatica was estimated to be 4.8%. 8,9 In a 2015 primary care cohort in the UK, presenting with back and leg pain and attending a clinic for assessment, 74% of participants were suspected to have sciatic pain. 10 Although imaging tests were not used to inform the initial clinical diagnosis in the UK cohort, the prevalence estimate shows that sciatica is commonly encountered or suspected in primary care. 10

Research on characteristics potentially associated with outcome in sciatica has identified a limited number of prognostic factors independently associated with outcome, mainly in studies of secondary care cohorts. 11–14 Only pain and condition-specific disability are consistently associated with having spinal surgery, which, in this secondary care context, is taken as a proxy of poor outcome for natural course and conservative management. 13 In the UK primary care cohort of patients with suspected sciatica previously mentioned, the high impact of sciatic pain on patients and their expectation of non-improvement over time, were independently associated with non-improvement. 14

It is generally believed that, for most patients with sciatica, symptoms improve considerably within 12 weeks of onset, usually without any active treatment. 1,2 Results from both clinical trials and cohort studies involving sciatica populations and utilising a number of different outcomes and definitions of improvement show that the percentage of patients reporting improvement varies between 32% and 65%. 14–18 Therefore, at least one-third of patients continue to suffer with pain for ≥ 1 year. The personal, social and economic burdens are significant,19,20 with UK annual costs estimated at £268M in direct medical costs and £1.9B in indirect costs. 21

Treatments for, and current clinical management of, sciatica

The vast majority of patients with sciatica in the UK are assessed and managed in primary care. Current evidence,22–24 the updated UK National Institute for Health and Care Excellence (NICE) guidance (2016)25 for the treatment of LBP and sciatica in those aged > 16 years and the National Low Back and Radicular Pain Pathway 201726 all support mainly a stepped-care management approach for most patients with sciatica, in the absence of suspected serious pathology. NICE25 recommends considering the use of stratification tools when assessing patients with sciatica to decide on the intensity of conservative management. Current management options may include advice about staying active; non-steroidal anti-inflammatory drugs (NSAIDs); weak opioid or neuropathic pain medications; physiotherapy interventions, such as exercise, with additional options of manual therapy techniques; or psychological therapies. Most sciatica patients will improve over time and will not require specialist care or invasive management options. However, for patients with severe and/or non-resolving sciatica, steroid spinal epidural injections and/or surgery are recommended for pain relief, in the presence of concordant imaging findings, for example from magnetic resonance imaging (MRI). Of all the available treatments for sciatica, surgery has the most robust evidence from research studies; it provides rapid and substantial relief of pain, but outcomes for those having surgery and those managed conservatively have been shown to be similar in the longer term, with approximately one-fifth of patients reporting ongoing complaints, which fluctuate over time, irrespective of which treatment is received. 27 On average, although patients receiving surgical intervention improve rapidly and substantially, they still seem to report mild to moderate pain and disability 5 years after the surgery;28 however, evidence also shows that increased pre-treatment symptom duration is related to worse outcomes following both conservative and surgical management than receiving treatment earlier in a patient’s symptom presentation. 29

In practical terms, stepped care may include an initial ‘wait-and-see’ period in primary care, with advice and pain medication, with subsequent referral to physiotherapy or similar treatments/services (e.g. osteopathy, chiropractic), and, for those not improving or with severe and incapacitating pain, referral to specialist spinal services at the interface and/or the secondary care setting for investigations and further management. 26 In the UK NHS, it is in these specialist settings where treatment options such as spinal injections and/or surgery are considered. 26 However, an immediate referral of all sciatica patients to injections or surgery is not considered a cost-effective model of care,22,23,30 and it is unlikely that it is needed for all patients. Currently, only patients with possible serious spinal pathology are fast-tracked to specialist services. 25,26 For all other sciatica patients, there is variation in clinical practice in terms of referrals from general practice to physiotherapy and specialist services. The UK Spinal Taskforce31 highlighted problems in the management of sciatica caused by variation in clinical practice, specifically delays in the treatment of patients with severe pain, believed to be mainly caused by delays in referral to specialist services. The Spinal Taskforce31 highlighted the need for good-quality trial evidence on the clinical effectiveness and cost-effectiveness of early referral of patients with severe symptoms for consideration of treatments such as surgery or spinal epidural injections. As early referral to surgical services for all sciatica patients cannot be recommended, better, and earlier, identification of patient subgroups in primary care for early matched treatment pathways [stratified care (SC)], including early referral to spinal specialist services for some sciatica patients, may be the key to improving outcomes for sciatica patients. The challenge for primary care clinicians [e.g. general practitioners (GPs) and physiotherapists] is how to identify, early in the presentation, patients who are likely to do well with conservative management in primary care and patients who may need early, fast-track referral to spinal specialist opinion for consideration of more invasive management options, such as spinal injections and surgery.

Stratified care model and rationale for the SCOPiC trial

In the field of non-specific LBP, a SC approach comprising two components – first, subgrouping according to risk of persistent back pain-related disability and, second, matching each subgroup of patients to treatment – demonstrated better clinical and health economic outcomes than non-SC. 32,33 In summary, the approach uses a brief self-report questionnaire, the Subgroups for Targeted Treatment (StarT) Back Screening Tool,34 developed for and validated with primary care patients with non-specific LBP (with and without leg pain), and allocates LBP patients to subgroups of low, medium or high risk of persistent back pain-related disability. The Subgroups for Targeted Treatment (STarT) Back Screening Tool has nine items: four focus on physical constructs and five focus on psychological constructs. A score of < 3 indicates that the patient is, most likely, at a low risk of future persistent back pain-related disability; a score of ≥ 4 of the five psychological items indicates that the patient is, most likely, at a high risk of persistent disability. 34 Any other score identifies patients as being at a medium risk of persistent disability. 34 By estimating the risk of persistent back pain-related disability, the STarT Back Screening Tool supports early clinical decision-making about conservative treatments for patients with non-specific LBP (such as GP care and physiotherapy management). Patients with sciatica have been shown to have more severe symptoms than those with non-specific LBP;20 management options such as injections and spinal surgery may be applicable to sciatica, but are not applicable to non-specific LBP. A SC approach for patients consulting with suspected sciatica in primary care may result in improved outcomes, but evidence is lacking.

Prior to the Sciatica Outcomes in Primary Care (SCOPiC) trial, no tool has yet been developed and tested with which to stratify care specifically for patients with sciatica early in the presentation of symptoms. In particular, it was not possible to predict which patients might benefit from early consideration for invasive treatment such as surgery. 35 In a cohort of sciatica patients with at least 6 weeks’ duration of symptoms, all of whom were surgical candidates, only high levels of pain and disability were associated with having surgery at some point. 12 The lack of clear and consistent factors predicting poor outcomes in patients with sciatica11,13,14 has made it challenging to design prognostic models that can guide early treatment decision-making. 13

In the absence of stratification approaches to help with clinical decision-making in early primary care management of patients with sciatica, we used information about the characteristics of patients referred to specialist spinal services from the only available UK primary care cohort of sciatica patients. 10 Using this, and working closely with clinicians, we developed an algorithm combining information about patients’ risk of persistent pain-related disability (using the STarT Back Screening Tool) and key findings from the clinical assessment that were associated with referral to spinal specialist services to allocate patients to matched care pathways, including an early, fast-track referral to MRI and specialist spinal opinion. 21 The stratification algorithm is described in Chapter 2.

Responding to the National Institute for Health Research (NIHR) Health Technology Assessment programme commissioning call (12/201, February 2013) about research to improve the management and outcomes of patients with sciatica, the aim of this study was to test, in a definitive randomised trial, whether or not the use of a new stratification algorithm and the matched care pathways lead to better outcomes than non-stratified, usual care (UC) for patients consulting with suspected sciatica in primary care.

The SCOPiC trial aims and objectives

The overall aim of this trial was to investigate whether or not a new SC approach for sciatica that combines information on the risk of persistent disability with information about sciatica clinical severity to guide decision-making about matched care pathways results in better clinical and cost outcomes for patients consulting their GP with symptoms of suspected sciatica.

The two main aims were to investigate the:

1. clinical effectiveness of SC versus non-SC for patients consulting with suspected sciatica in primary care, in terms of patient-reported time to symptom resolution

2. cost-effectiveness of SC versus non-SC over a 12-month period.

Further objectives included investigating the clinical effectiveness of SC versus non-SC on sciatica-related disability, LBP, leg pain, quality of life, time lost from work, health-care use and patient satisfaction with care received, and the results of care. In addition, we investigated the impact of SC on service delivery, in terms of referrals to physiotherapy services and spinal specialist services, and the acceptability of this SC approach from the perspective of patients and clinicians.

Design and setting

The SCOPiC trial was a multicentre, pragmatic, assessor-blinded, randomised controlled trial (RCT), with an internal pilot health economic evaluation and linked qualitative interviews. The trial’s setting included NHS primary care services (general practices), community physiotherapy services and spinal interface services between primary and secondary care.

An economic evaluation was conducted alongside the RCT to evaluate the cost-effectiveness of the intervention over 12 months (see Chapter 4). A nested qualitative study was conducted to explore patients’ and clinicians’ views of the fast-track care pathway associated with the SC model tested in this trial (see Chapter 5).

General practices

We recruited patients from 42 general practices in the areas of North Staffordshire, North Shropshire/Wales and Cheshire. These areas were the three recruiting centres. We initially started with 30 general practices and then added a further 12 practices over the course of the trial, to recruit the number of participants required in the trial.

Physiotherapy sites

Five community NHS physiotherapy services were involved in the trial: two in Staffordshire, one in Shropshire, one in Wales and one in Cheshire. We initially started with three physiotherapy services (those in Staffordshire and Shropshire), then added a further two (in Wales and Cheshire) over the course of the trial.

Spinal specialist sites

Patients recruited to the fast-track pathway were seen by spinal specialists in either Staffordshire (Staffordshire and Stoke-on-Trent Partnership NHS Trust), Shropshire (The Robert Jones and Agnes Hunt Orthopaedic Hospital NHS Trust) or Cheshire (Leighton Hospital, Mid Cheshire Hospitals NHS Trust).

Ethics approval

The trial was registered on the International Standard Randomised Controlled Trial Number (ISRCTN) registry (ISRCTN75449581). Ethics approval was granted by the National Research Ethics Service West Midlands – Solihull (reference number 15/WM/0078). Recruitment commenced in June 2015 and was completed in July 2017. The Trial Steering Committee (TSC) and Data Monitoring Committee (DMC) provided independent oversight of the trial.

Participants

We recruited adults consulting in general practice with symptoms of sciatica of any duration and severity.

Interventions

Trial procedures

Baseline assessment, eligibility screening and informed consent

The SCOPiC trial sciatica clinics were operated as integrated research and NHS service clinics. At these clinics, trial physiotherapists explained the purpose of the clinic and answered questions about the clinic and the trial. Potential participants expressing interest in participating in the trial proceeded to have a standardised clinical assessment by the physiotherapist to confirm the clinical diagnosis of sciatica. The assessment was documented on a standard pro forma. In the context of the trial and according to literature37,38 and clinical guidelines,25 sciatica is, in most cases, indicative of nerve root entrapment due to a disc prolapse, and, less frequently, due to spinal stenosis. 1 Therefore, as well as patients with symptoms thought to be due to disc prolapse, we also included patients with symptoms thought to be due to spinal claudication. Eligibility for the trial was based on the assessing physiotherapist being ≥ 70% confident in their diagnosis of sciatica. We also asked the assessing physiotherapists to record a specific diagnosis of disc prolapse or stenosis for the sciatic symptoms. At least one of the following signs and symptoms contributing to the clinical diagnosis of sciatica7,39,40 had to be present: leg pain approximating a dermatomal distribution; leg pain worse than, or as bad as, back pain; leg pain worse with coughing/sneezing/straining; subjective sensory changes approximating a dermatomal distribution; objective neurological deficits indicative of nerve root compression; positive neural tension test; and (specifically for spinal claudication/spinal stenosis) leg pain worse with weight-bearing activities and better with sitting. Patients were excluded if the assessing physiotherapist thought that the leg pain was a result of causes other than sciatica (e.g. referred leg pain, hip pathology, vascular claudication), if diagnostic certainty was < 70% or if potentially serious pathology was suspected (e.g. malignancy, cauda equina syndrome). All reasons for exclusion were documented.

For those patients who were eligible and interested in taking part in the trial, the trial physiotherapist explained the trial in detail, answered any questions and took written informed consent. The information required from self-report and clinical examination to allocate patients to one of the three groups based on the stratification algorithm was recorded by the assessing physiotherapist on a standard pro forma. Patients had completed and brought with them the baseline questionnaire, or completed the questionnaire during the research clinic visit, along with a second baseline questionnaire. The clinic administrator checked the baseline questionnaires for missing data while the patient was in the clinic.

Patients who did not wish to participate or who were ineligible received advice and education from the physiotherapist in the clinic, with the option to be referred for further treatment as appropriate, outside the trial. GPs received information (by a standard letter generated by the trial management database) of the outcome of their patients’ attendance to the SCOPiC trial research clinic, and their care plan. Figure 1 outlines the SCOPiC trial recruitment procedures.

FIGURE 1.

The SCOPiC trial recruitment procedures.

Randomisation

Eligible patients who consented to take part were randomised to one of the two trial arms, in a 1 : 1 ratio, using a web-based randomisation service from Keele Clinical Trials Unit (CTU), which ensured allocation concealment, operated by the clinic administrator. There was a back-up process to telephone the CTU randomisation service in the event of web access failure. The assessing physiotherapist gave the clinic administrator the completed standard pro forma, which included the sciatica group the participant was allocated to according to the stratification algorithm. In each recruiting centre, the clinic administrator used the randomisation service and individual participants were randomised, stratified by centre (North Staffordshire, North Shropshire/Wales and Cheshire) and group allocation according to the stratification algorithm, using random permuted blocks of varying size (two, four and six), to either SC or UC. With three centres and three groups equating to nine stratified cells, participants were randomised in block sequences with random selection from the following blocks: AB, BA, AABB, ABAB, ABBA, BAAB, BABA, BBAA, AAABBB, AABABB, AABBAB, AABBBA, ABAABB, ABABAB, ABABBA, ABBAAB, ABBABA, ABBBAA, BAAABB, BAABAB, BAABBA, BABAAB, BABABA, BABBAA, BBAAAB, BBAABA, BBABAA, BBBAAA.

If a participant was randomised to the SC arm of the trial, the administrator informed the assessing trial physiotherapist, who saw the participant and commenced the care matched to each group. If a participant was randomised to the UC arm of the trial, the administrator informed the physiotherapist delivering care for the UC arm, who saw the patient and initiated treatment. Different physiotherapists delivered treatment to participants in each trial arm, at the research clinic and during subsequent appointments. Information about which arm of the trial a participant was allocated to was not disclosed to either the participant or their GP. A participant’s GP was informed in writing that the individual was participating in the trial, but GPs were not informed about which arm of the trial the participant had been randomised to. Usual clinician-to-clinician correspondence continued as per usual practice; the research did not interfere with this. Each participating general practice had a site file containing contact details of the trial team.

Blinding

The main issue in the SCOPiC trial was the concealment of the stratification algorithm and its use in matching participants to care pathways. Participants were told that the trial was comparing two primary care approaches for the treatment of sciatica: one based on matching patients to treatment using a simple tool that helps to decide on the treatment pathway most likely to help them and one based on the treatment needed as discussed and agreed by the physiotherapist and themselves. To further help with allocation concealment, participants randomised to the UC arm of the trial were seen by physiotherapists who had not carried out their detailed clinical assessment and eligibility screen for the trial. Therefore, physiotherapists in the UC arm remained blinded to both the details of the stratification algorithm and the individual patient’s sciatica group status, hence avoiding contamination between the two trial arms. It was not possible for physiotherapists treating participants in the SC arm to remain blinded. To further protect against contamination, no physiotherapists treating participants in the SC arm were involved in the treatment of participants in the UC arm. Research nurses blinded to treatment allocation conducted the minimum data collection (over the telephone) at 4 and 12 months’ follow-up for participants who did not respond to questionnaires. The risk of contamination from GPs knowing to which arm a participant was randomised was deemed to be very low. In qualitative interviews with the GPs (as part of the nested qualitative interviews), we collected data on whether or not involvement in the trial contributed to changes in GPs’ referral decisions for patients with sciatica (see Chapter 5). Initially, we planned to collect information about GP referral patterns before and during the trial, using anonymised general practice electronic records, to check whether or not trial participation influenced referral patterns for sciatica patients, but this was not technically possible to do.

Statisticians and health economists were blinded to treatment arm during the development of the statistical analysis plan (SAP) (the SAP was agreed in advance with the TSC members) and through monitoring phases of the trial, including the interim data analysis specified in the internal pilot phase of the trial. Furthermore, two statisticians performed the primary analysis of the main trial independently and both were blinded to treatment arm allocation. The trial statistician carrying out further key analyses of primary and secondary outcomes was blinded up to the point of conducting the per-protocol analysis (a sensitivity analysis). Treatment allocation was stored in a secure computer server accessible only by authorised CTU staff, for whom blinding was not relevant. For the quantitative data analysis, throughout the monitoring period, the treatment arm variable was blind dummy-coded until the key primary and secondary data had been analysed.

Follow-up

For the primary outcome, time to resolution of sciatica symptoms, data collection via text messages occurred weekly for the first 4 months for all participants, starting on the first Sunday following the participant’s randomisation at the SCOPiC trial research clinic. Then, between 4 and 12 months’ follow-up, the text message data collection changed to once every 4 weeks, or until ‘stable resolution’ of symptoms (stable resolution was defined as 2 consecutive months’ responses of ‘completely recovered’ or ‘much better’; see Outcome measures and assessments for details on the primary outcome). Once stable resolution occurred, data collection for the primary outcome via text message, beyond the first 4 months, ceased. Non-responders to the first week’s text message received a reminder text 48 hours later, and those who still did not respond were mailed a postcard the next day. Non-responders to the second week’s text message received a reminder message 48 hours later, and those who still did not respond received a telephone call from a research nurse after at least a further 24 hours. For subsequent non-response reminders, the processes described above were repeated. For those participants not providing any response using text messages, there was an option to transfer to data collection by brief telephone call with a research nurse. A small number of participants chose to receive only phone calls for collection of the primary outcome (n = 49).

The secondary clinical outcomes and health economic outcomes were collected at 4 and 12 months using participant self-completed postal questionnaires. Participants who did not respond to the postal questionnaires were sent a reminder postcard after 2 weeks of non-response. After a further 4 weeks, participants received a repeat full postal questionnaire and a short version if they had not responded by 6 weeks. If a total of 8 weeks had passed and no data had been received, the participant was contacted by a research nurse blinded to treatment allocation to collect a minimum data set over the telephone.

Quality assurance

For quality assurance purposes, we implemented monitoring procedures to ensure that all personnel involved in the trial adhered to the trial’s protocol and procedures. All physiotherapists involved in taking participant consent were observed at least once by a senior member of the research team. All physiotherapists involved in eligibility assessment were observed at least once in the research clinic. Administrators involved in the preliminary eligibility criteria checking over the telephone were also observed by a senior member of the research team. Standardised pro formas were used for all trial documentation, to minimise variation in trial procedures. We did not encounter significant problems during quality control monitoring.

Outcome measures and assessments

Clinical outcomes

Process outcomes

Process outcomes were collected to investigate the impact of SC on service delivery for both physiotherapy and specialist spinal services. With the use of case report forms (CRFs), data were collected, in each arm of the trial, on the number of participants referred to physiotherapy services, the number of physiotherapy sessions received, treatments provided and the number of participants referred to specialist spinal services and/or secondary care settings. The timing of referral and treatment were also captured, when possible. In addition to CRFs, we captured information on health-care use from participant questionnaires and hospital record reviews from participating NHS specialist services.

Description of the interventions

Stratification algorithm

Figure 2 shows the stratification algorithm used to direct participants’ care in the stratified arm of the trial. The algorithm utilises information on the risk of a poor prognosis from the STarT Back Screening Tool and information on factors associated with referral to spinal specialists to allocate participants with clinically diagnosed sciatica to one of three sciatica groups, each matched to a care pathway, with one of the care pathways being fast-track referral for MRI and to a consultation with a spinal specialist for opinion. We briefly describe here the decisions and methods utilised to derive the stratification algorithm. Full details of the development and internal validation of the stratification algorithm for primary care patients with sciatica are available in a separate peer-reviewed paper. 50

FIGURE 2.

Stratification algorithm for allocating patients to groups and matched care pathways.

To identify the group of sciatica participants likely to need fast-track referral to imaging tests and a spinal specialist assessment and opinion, we used data on patients with a sciatica clinical diagnosis from the Assessment and Treatment of Leg pain Associated with the Spine (ATLAS) study,10 a prospective, treatment cohort of primary care patients with back and leg pain. The ATLAS cohort included patients with symptoms of any duration and severity, and excluded patients with potential serious pathology. All patients without contraindications to MRI had the scan after their baseline clinical assessment. In the development of the algorithm, ATLAS study participants clinically diagnosed with sciatica with diagnostic confidence of ≥ 70% (as reported by the assessors) were included. The outcome definition was referral to NHS spinal specialist services (yes/no) at some point over the course of 12 months’ follow-up in the ATLAS study. Potential predictors of referral to specialist services that were informed from the literature more broadly and that were available in the ATLAS study data set comprised self-report and clinical examination findings. The selected factors were categorised into five domains: impact of condition, pain levels/symptoms, psychological factors, symptom behaviour and presentation (e.g. increased leg pain with coughing, reporting of numbness in the leg), and clinical examination findings. Logistic regression (univariable and multivariable) was used to determine the association between each factor and the outcome of referral to spinal specialist services. The factors derived from the statistical models as being most strongly associated with referral to specialist services, within each domain and overall, were discussed with clinical and research experts (epidemiologists, triallists, statisticians, spinal surgeons, spinal physiotherapy specialists, rheumatologists, pain specialists, GPs). The final list of factors thought to be most relevant to the decision to refer to spinal specialists was agreed by all clinical stakeholders, and included the following: effect of lower-back and/or leg pain (sciatica) on ability to do one’s job or ability to do jobs around the house [Numerical Rating Scale (NRS) of 0–10, with a cut-off point of > 6], current leg pain intensity (NRS of 0–10, with a cut-off point of > 6), sensory deficits in a dermatomal distribution recorded during the clinical examination (yes/no) and pain below the knee (yes/no). The first three factors were derived from the statistical analyses. The binary cut-off points (NRS scores of > 6) for impact and pain intensity were considered to be reasonable thresholds for pain and functional limitations, and had face validity when considering early referral to specialists. Presence of pain below the knee was subsequently added to the list as it is considered the best proxy indicator of leg pain due to nerve root involvement;37 the clinical experts considered this to be important to combine with the other factors to guide referral decisions for patients with sciatica.

Data from the ATLAS study cohort showed that only one participant (out of 52) with a STarT Back Screening Tool score of up to 3 (out of a total of 9), indicative of low risk of poor prognosis, was referred to spinal specialist services. On the basis of that, patients at a low risk of poor prognosis (STarT Back Screening Tool score of up to 3), irrespective of clinical characteristics, were not considered for referral to specialists or for a full course of physiotherapy management. For the remaining participants, we considered a number of possibilities in terms of combinations of factors from the clinical assessment and information on risk of poor prognosis, using the STarT Back Screening Tool score, for identifying which participants with sciatica should be fast-tracked to spinal specialist services. For each combination of factors, we considered the implications for sensitivity in terms of identifying observed referrals, and feasibility and practicality in relation to use in clinical practice. For all scenarios, sensitivity/specificity, positive/negative predictive values and percentage of the sample fast-tracked were calculated. Based on the results of this analysis, participants at a medium risk of poor prognosis who had all four clinical characteristics described above, and participants at a high risk of poor prognosis with any three of the clinical characteristics, were allocated to group 3 and matched to the care pathway of fast-track referral to MRI and specialist opinion. The remaining participants were allocated to group 2 and matched with a course of physiotherapy management of up to six sessions. Sensitivity, specificity and positive predictive value (PPV) of the algorithm for participant allocation to the fast-track pathway (group 3) were 51%, 73% and 22%, respectively. The algorithm, applied to the ATLAS study cohort, allocated 12% of participants to group 1, 57% to group 2 and 31% to group 3.

Training and auditing

All personnel (i.e. administrators, trial physiotherapists for intervention and control arms, research nurses) involved in the trial received training in the trial’s procedures and documentation. For audit purposes, treatments delivered to participants in the intervention and control arms of the trial were recorded in a standardised format on CRFs. The information recorded included dates of start and completion of treatment, number of treatments received and types of interventions (i.e. exercise, advice, manual therapy) and the physiotherapist’s clinical grade (NHS Agenda for Change banding). Protocol deviations in both trial arms were recorded and reported. Physiotherapy record reviews were conducted in the cases of missing or incomplete CRFs for both SC and UC participants.

Details of the care of participants in group 3, the fast-track pathway (e.g. referrals for physiotherapy or surgical or injection procedures), were collected from the clinical letters generated in the specialist clinics for each participant and recorded on CRFs. Following the 12-month follow-up, data on treatments actually received for sciatica and the time frame of any interventions delivered were supplemented by reviews of the hospital records at the participating NHS trusts whenever possible. Given that participants could choose to have treatments in other NHS sites, we anticipated that we would not be able to retrieve hospital record data for all participants.

Adverse events

It was considered unlikely that participants in the SCOPiC trial would be at risk of serious adverse events (SAEs). The trial investigated the approach of matching appropriate care pathways to three groups of sciatica patients. The treatments themselves, in each of the three care pathways, were all part of routine clinical practice. SAEs were defined as those that resulted in death, unscheduled hospitalisation or significant disability as a result of the trial’s interventions or procedures. Any SAEs were documented in CRFs and reported immediately to the trial team. Any SAEs that were considered to be related to the trial procedures or interventions were reported to the Research Ethics Committee by the chief investigator within 15 days of the chief investigator becoming aware of the event, and to the TSC and DMC. We also documented SAEs during hospital records reviews, carried out as part of the trial’s data collection procedures.

Possible potential AEs included worsening of symptoms, for example as a result of an exercise programme prescribed by the physiotherapist. Physiotherapists treating participants in the SCOPiC trial were asked to report potential AEs on the CRFs or to the trial team. Participants were also asked about AEs in the follow-up questionnaires.

Sample size justification

A total sample size of 470 participants was required to test for superiority of SC compared with UC, to detect a hazard ratio (HR) of between 1.4 and 1.5 for time to resolution of symptoms (primary outcome) with 80–90% power (given a two-tailed significance level of 5%), assuming an event (resolution) rate of ≥ 60%, a 20% drop-out rate and intraclass correlation (ICC) for clustering by physiotherapist at the level of 0.01 and allowing for a coefficient of variation in physiotherapist cluster size of 0.65 (Eldridge et al. 51). The sample size allows for a least conservative HR of 1.4 in median survival times with 90% power (if all participants in the trial are recovered by the 12-month follow-up and ICC for physiotherapist effect is < 0.001), and a most conservative HR of 1.5 in median survival times with 80% power (if 60–65% of participants in the trial are recovered by the 12-month follow-up and ICC for physiotherapist effect is 0.01, given an average cluster size of around 12–15).

In this context, a HR of > 1 (denoting a higher ‘successful’ event rate) is a positive outcome.

The sample size of 470 would also provide > 80% power to detect a ‘small’ to ‘moderate’ standardised mean difference (effect size) of 0.3552 between the two trial arms in respect of sciatica-related physical disability [measured using the Roland–Morris Disability Questionnaire (RMDQ), a key secondary outcome], at the 12-month follow-up, allowing for physiotherapist effect and 20% attrition. The trial was not powered to detect differences between the sciatica groups between the trial arms.

Internal pilot phase

The internal pilot phase of the trial was designed to assess participant recruitment and follow-up rates over the first 8 months of recruitment; the success of general practice recruitment and retention; the success of physiotherapy site recruitment, including training and engagement; adherence to the treatment protocols; the proportion of participants allocated to each of the three groups according to the stratification algorithm; the time to MRI and specialist opinion for those in the fast-track pathway (group 3); the event rate of the primary outcome; and rate of missing data (text messages) for the primary outcome up to the 4-month follow-up for all participants recruited during the 8 months of the pilot trial phase.

The following were set as criteria for progression to the main trial from the internal pilot phase, and were agreed with the DMC and TSC:

1. recruitment rate of ≥ 70% of that anticipated; specifically, ≥ 90 patients would need to be recruited over the first 8 months of recruitment (where 130 would have been expected across the centres, taking into account staggered general practice recruitment and set-up)

2. overall loss to follow-up including non-response and dropouts (e.g. withdrawals, deaths, departures) in the primary outcome measure (resolution of symptoms) not exceeding 25% (based on 4 months’ follow-up of participants recruited during the first 7 months).

In the first success criterion, the expected number of 130 participants recruited in the trial for the duration of the internal pilot phase was based on the total number of 470 adults who were to be recruited from approximately 30 general practices over a 22-month recruitment period (or approximately 0.7 participants recruited per practice per month). Recruitment of centres was staggered; active recruitment was expected to start with the first centre (North Staffordshire), from five general practices, during the first month of recruitment, then increase at the first centre with an additional 10 practices by end of the second month. Recruitment would then start at the second centre (North Shropshire) from five practices by end of the third month, and increase at the second centre with an additional 10 practices by end of month 4, with full patient recruitment thereafter. Therefore, expected patient recruitment was (practices × months recruiting in first 8 months × recruitment rate):

The internal pilot did not involve formal interim analysis of between-group effects on the primary or any other outcomes. A review of the internal pilot phase data was undertaken by the Trial Management Group (TMG) and shared with the DMC, which reported their recommendations on progression to the TSC and funder.

Data management

Statistical analysis

All analyses were conducted and reported following the Consolidated Standards of Reporting Trials (CONSORT) guidelines. 54–56 The primary analysis was by ‘intention to treat’ (ITT), with analysis being carried out as per randomised allocation.

Process outcomes analysis

Descriptive statistics were used to examine health-care utilisation by participants in the trial arms and by sciatica groups (1, 2 and 3) in both arms. Numbers (and percentages) of participants referred for physiotherapy treatment or to spinal specialists, and numbers (and percentages) of participants receiving spinal surgery and/or spinal injections, were described. For participants receiving physiotherapy treatment, the numbers (median and IQR) of physiotherapy sessions received and summaries of types of treatments received were also described. Time to physiotherapy treatment and appointments to spinal interface clinics, spinal orthopaedic clinics and pain clinics for participants in both trial arms and by sciatica groups (1, 2 and 3) were also evaluated (median and IQR) when possible, taking the randomisation date as the starting point. Statistical testing was carried out through non-parametric tests of significance, the Mann–Whitney U-test was used for between-arm testing of numerical data, and chi-squared tests were used for between-arm testing of frequency (count) data.

Patient and public involvement

Patient and public involvement (PPI) was supported by the PPI infrastructure within the Arthritis Research UK Primary Care Centre at Keele University, which includes a research user group (RUG), made up of 75 members aged from 32 to 82 years, all with experience of living with a long-term condition.

Previous interviews that we conducted with sciatica patients emphasised the impact that it has on them and the requirement for clearer information on treatment and prognosis. 60 Prior to the SCOPiC trial taking place, we held a meeting with three PPI members who were currently experiencing, or previously had experienced, sciatica. Outcomes from this meeting highlighted that prompt pain relief is key, given the severity of the pain. This informed the choice of time to symptoms resolution as the primary outcome of the trial, which emphasised the significance of early assessment and the requirement to match treatments to patients more efficiently, and the need for this trial. Furthermore, it was thought that collecting pain data via regular, brief texts or telephone calls was acceptable.

Further meetings with PPI members were held throughout the trial to seek advice on trial documents, the language of the text messaging and to discuss the findings from the qualitative interviews with patients in the fast-track pathway of the SC arm. A final PPI meeting was held to discuss the overall trial results and agree the key messages for patients and the public.

Two PPI members were part of the TSC and attended each TSC meeting, providing valuable input throughout the trial from a patient perspective.

Trial oversight

The trial was supported by Keele CTU, a UK Clinical Research Collaboration fully registered CTU, and was monitored by the CTU operations group. The conduct of the trial was designed and managed in line with the Keele University Health and Social Care research quality management system and adhered to Keele CTU standard operating procedures.

The trial was overseen by the TMG, which included all key personnel involved in the design and operational management of the trial and its processes. The TMG met on a monthly basis to discuss and monitor progress, including recruitment and follow-up, in line with the agreed trial management plans; to identify solutions to challenges; and to agree actions in the event of any protocol non-compliance issues that might arise. The TMG worked closely with the Clinical Research Network (CRN) West Midlands, CRN Cheshire and CRN Wales in monitoring and facilitating recruitment via general practices and the clinical teams in the participating NHS sites, delivering the care pathways in the SCOPiC trial.

Participant flow into the trial

Trial invitation mailing to potentially eligible patients commenced on 28 May 2015; the first participant was randomised on 4 June 2015 and the final participant was randomised on 18 July 2017. The last mailing of follow-up was on 18 July 2018 and the last participant follow-up response was received on 30 August 2018. Three centres participated in the trial: North Staffordshire and North Shropshire were planned centres from the start of the trial, and Cheshire was added as a centre on 1 April 2016 following the internal pilot phase review to ensure continued successful recruitment to the trial. The North Shropshire centre’s general practice numbers were increased to include four general practices from the Shropshire/Welsh border area. By the end of the recruitment phase, 45 general practices had agreed to take part in the trial and SCOPiC trial participants from 42 of those practices consented and were randomised in the trial (range of the number of participants consenting and randomised per practice 1–34). Among those 42 practices, the total number of participants recruited as a percentage of the total number invited to the SCOPiC trial clinics to be screened for eligibility ranged from 4.8% to 35.7%. The observed recruitment rate over the recruitment period was approximately 18 people per month, which was just under the expected number of approximately 20 per month.

Trial population

In total, 2719 potential participants with suspected sciatica were mailed invitations to attend the SCOPiC trial research clinics, 1718 (63%) of whom contacted Keele CTU and were screened for preliminary eligibility over the telephone (median 4.5 days later, IQR 2–11 days later). Of these, 268 were found to be ineligible (the most common reasons being ‘no leg pain’ and ‘ongoing/recent physiotherapist/specialist care’) and 97 declined to attend the research clinics. A further five were screened but had no appointment booked. Hence, 1348 potential participants were booked into the SCOPiC trial research clinics and 1269 (46.7% of invitees) attended (median 14 days after invitation, IQR 9–21 days after invitation). Of these, 552 (43% of attendees) were eligible to take part in the trial and 717 were ineligible. The most common reasons for ineligibility were ‘physiotherapist’s low confidence in sciatica diagnosis’, ‘referred leg pain’ and ‘no symptoms in the leg’. Of the 552 participants attending clinic who were eligible to take part in the trial, 76 declined participation and 476 provided written, informed consent and were randomised to one of the two trial arms: 238 to SC and 238 to non-stratified UC. The flow diagram of participants into and through the trial is shown in Figure 3.

FIGURE 3.

The SCOPiC trial participant flow diagram. a, More than one reason for ineligibility was possible; b, one did not provide any data and nine (five) had experienced the event [resolution (stable resolution)] by the time of withdrawal; c, one did not provide any data and 10 (seven) had experienced the event [resolution (stable resolution)] by the time of withdrawal. FQ, full questionnaire response; MDC, minimal data collection response.

Stratified block randomisation ensured that the number of participants in each arm of the trial was balanced in terms of centre (North Staffordshire/North Shropshire and Wales/Cheshire) and sciatica group (1, 2 and 3). More participants were recruited from North Staffordshire [40.3% of the total sample (192/476)] and North Shropshire [47.5% (226/476)] than from Cheshire [12.2% (58/476)], as the Cheshire centre commenced recruitment after the internal pilot phase. More participants were in sciatica group 2 [44.3% (211/476)] than in group 1 [22.5% (107/476)] or group 3 [33.2% (158/476)] – these proportions were similar in the two trial arms. These observed proportions were somewhat different from our projected estimates before the trial of 12% in group 1, 57% in group 2 and 31% in group 3.

The mean ages of participants who were randomised into the trial (n = 476) and those who were invited but were not randomised (n = 2243) were similar: 52.0 (SD 14.1) years and 52.3 (SD 16.1) years, respectively. There was a similar female-to-male ratio among trial participants (262 : 214, ratio 1.22) and non-participants (1260 : 983, ratio 1.28). Participants had a higher median Index of Multiple Deprivation (IMD) rank value (designating lower social deprivation) than non-participants: 15,304 (IQR 8021–21,790) and 13,353 (IQR 4613–21,294), respectively. There was a difference in participant uptake (i.e. consent rate of those invited) between centres: North Staffordshire [14.9% (192/1291)], North Shropshire/Wales [19.8% (226/1143)] and Cheshire [20.4% (58/285)]. General practice computer templates were completed to capture participating GPs’ preferred treatment choices for patients at that point of consultation. The proportions selecting ‘keep under GP care’, ‘refer to physiotherapy’ or ‘refer to specialist spinal services’ were similar for trial participants and non-participants (30.3% and 32.7% for ‘keep under GP care’, 63.2% and 58.4% for ‘refer to physiotherapy’, and 6.5% and 8.9% for ‘refer to specialist spinal services’, respectively).

Results of the internal pilot phase of the trial

Review of the internal pilot progression criteria took place on two occasions, first on 4 February 2016 (to review progression in relation to participant recruitment) and then on 15 July 2016 (to review progression in relation to follow-up rates for the primary outcome).

Recruitment during the internal pilot phase of the trial (the first 8 months of the recruitment period) was in line with anticipated figures: 129 participants were recruited, compared with an anticipated number of 130 (124 participants within 7.5 months by the time of DMC/TSC reporting). This 99% recruitment rate (against the expected figure) exceeded the threshold criterion for progression to the main trial, which had been set at 70% (i.e. 90/130), thereby fulfilling the first of the two progression criteria. In total, 28 general practices were participating at this stage, only two short of the 30 that had been anticipated by this point (and further practices were ready to join the trial). There was a difference in the observed and expected proportions of participants in sciatica groups 1 and 2, but fidelity checks of research clinic CRFs showed no evidence that suggested that these observed proportions were inaccurate.

It was noted on review of follow-up data that (1) the event rate of the primary outcome (percentage of participants reporting resolution of sciatica symptoms according to the primary definition) exceeded 60%, which satisfied the assumption of the original sample size calculation, and (2) the follow-up rate for the primary outcome data up to 4 months was 88% (1706 responses from a potential total of 1938). Hence, the loss to follow-up was 12%, which was lower than the 25% threshold (thereby fulfilling the second of the two progression criteria). The follow-up rate for the 4-month postal questionnaire (including minimum data collection) was 77%. There was little difference in withdrawal numbers between trial arms (13 participants had withdrawn from one arm at the 4-month follow up, and nine from the other arm).

The DMC and TSC agreed that the internal pilot phase progression criteria had been met and that the trial could proceed to the main trial phase.

Baseline characteristics of randomised participants

A summary of baseline characteristics for the 476 trial participants, overall and by trial arm, is given in Table 2. The mean age was 52.1 years (50.8 years in the SC arm and 53.3 years in the UC arm). The overall percentage of females was 55.0% (262/476): 55.5% and 54.6% in the SC and UC arms, respectively. Data were available from 475 baseline questionnaires, because one participant was withdrawn immediately after randomisation (owing to error, as the participant was ineligible).

The majority of the trial participants were employed (68.4%), of whom 72.1% were full-time workers; the majority of these had taken time off work in the previous 12 months because of their sciatica. Around half of the trial participants had acute pain (< 6 weeks’ duration, 51.5%); among the rest, 21.8% and 26.5% had subacute (6–12 weeks’ duration) and chronic (> 12 weeks’ duration) symptoms, respectively. The mean sciatica-related physical disability score (RMDQ score) was 11.2 (SD 5.3): 11.1 for the SC arm and 11.3 for the UC arm. The majority of participants classified their general health as ‘good’, around two-thirds of participants were experiencing significant sleep disturbance due to the sciatic pain, around half were classified as having possible/probable anxiety and one-third were classified as having possible/probable depression at baseline (using the Hospital Anxiety and Depression Scale). For 398 participants, the assessing physiotherapists diagnosed disc prolapse as the reason for the sciatic symptoms; for 63 participants, the recorded clinical diagnosis was spinal stenosis. In approximately 64.3% of the disc-related sciatica cases, the S1 nerve root was suspected of being compressed. In about 85% of the cases clinically diagnosed with stenosis, leg symptoms were unilateral.

The baseline characteristics for each trial arm, stratified by sciatica groups (1, 2 and 3), are summarised in Table 3. Age was similar across all strata (the mean age ranged between 50.1 and 55.3 years). There was a slight trend towards an increasing proportion of females from group 1 (≈50%) through group 2 (≈55%) to group 3 (≈60%). The centre in North Shropshire/Wales had a higher proportion of group 1 participants (28%) than the centres in North Staffordshire and Cheshire (17% each). Furthermore, there was a trend towards a lower proportion of employed participants from group 1 (74%) through group 2 (71%) to group 3 (61%). Trends of greater proportions of time off work and lower work performance levels (indicated by higher Work Performance Index scores) were reported from group 1 through group 2 to group 3. Table 2 shows that the expected trend of worsening health status for all baseline measures of pain levels, disability and mental and general health, from group 1 through group 2 to group 3, was observed in the trial population.

Trial follow-up

The flow diagram in Figure 3 also summarises the follow-up rate of participants. For the primary outcome (collected through text messages), 9467 responses were received from 10,601 potential texts (or telephone calls) (i.e. completeness, 89.3%). The response rate was similar between the two arms: 88.3% (4688/5310) in the SC arm and 90.3% (4779/5291) in the UC arm. A total of 83.6% (398/476) of participants responded to at least 80% of potential texts/telephone calls, 75.4% (359/476) responded to at least 90% and 60.1% (286/476) responded to all potential texts/calls [138 (58.0%) in the SC arm and 148 (62.2%) in the UC arm]. For the 4-month follow-up questionnaire, the overall response rate was 82.6% (393/476): 80.7% (192/238) in the SC arm and 84.5% (201/238) in the UC arm. For the 12-month follow-up questionnaire, the response rate was 75.4% (359/476) overall: 74.4% (177/238) in the SC arm and 76.5% (182/238) in the UC arm. The summary breakdown of responses as full questionnaire response or minimal data collection by trial arm is shown in the flow diagram (see Figure 3); the proportion of responders who completed the full questionnaires was 80.4% (316/393) at 4 months and 72.4% (260/359) at 12 months. Those who did not complete the 4- and 12-months questionnaires were younger, lived in significantly more deprived neighbourhoods (lower average IMD rank value) and had slightly worse baseline health status than those who completed the questionnaires. Summary baseline characteristics comparing responders to follow-up questionnaires with non-responders are shown in Table 4.

Loss to follow-up was mainly through non-response; however, there were 38 participant withdrawals over the 12-month follow-up period: 18 in the SC arm and 20 in the UC arm. A summary of the withdrawal reasons (when known), by trial arm, is provided in Table 5 and shows that, as well as the total number, the reasons for withdrawal were similar between the trial arms. As detailed in the legend of the flow diagram in Figure 3, half (i.e. 19) of participants had experienced the primary event (resolution of sciatica symptoms) by the time of withdrawal: nine (half of withdrawals) in the SC arm and 10 (half of withdrawals) in the UC arm. One participant (in the SC arm) was withdrawn immediately post randomisation and was not followed up, as they had been randomised in error (the patient did not fulfil the eligibility criteria).

Primary outcome

The summary time-to-event data for the primary end point of first resolution of sciatica symptoms are shown in Figure 4. The event rate was rapid over the first few weeks, with 25% of events (sciatica symptom resolution) occurring by week 4 in the SC arm and by week 5 in the UC arm. In addition, inspection of the percentiles presented shows significant data skewness, which explains the difference between the mean and median values. The median time-to-event time (weeks) was slightly shorter in the SC arm than in the UC arm: 10.0 and 12.0 weeks, respectively. Comparing each trial arm, the figure for resolved symptoms (‘completely recovered’ or ‘much better’) by 16 weeks was 137 out of 238 (57.6%) in both arms; by 48 weeks, it was 344 out of 476 (72.3%): 176 out of 238 (73.9%) in the SC arm and 168 out of 238 (70.6%) in the UC arm. Table 6 summarises the Kaplan–Meier event-rate probabilities and event numbers across the time intervals of data collection. For the ratio of the SC arm to the UC arm, the estimate for the primary end point time-to-event HR was 1.14 (95% CI 0.89 to 1.46; p = 0.288) (adjusted for centre, baseline stratification sciatica group and symptom duration, and accounting for clustering by named physiotherapist). There was limited overlap in hazards across time and the statistical interaction term for trial arm by time within the model assessment of proportional hazards was not statistically significant [either when looking at time (p = 0.430) or log_etime (p = 0.377)]. Given a total sample size of 476 participants, a follow-up of 89.3%, a primary event (first resolution of sciatica symptoms) rate of 72.3% and a design effect/inflation factor of 1.32 (based on the following: an ICC for clustering by physiotherapist of 2.6% for the cumulated occurrence, or not, of an event by week 48; average cluster size of 4 with coefficient of variation of 1.5), the post hoc power for detecting a HR of between 1.4 and 1.5 ranged from 0.73 to 0.87.

FIGURE 4.

Kaplan–Meier time-to-event analysis of the primary end point (time to first resolution of sciatica symptoms). a, Estimation is limited to the largest event-free time if it is censored. Cumulative proportion of resolved cases by week 48: 0.754 (all): 0.780 (SC) and 0.8729 (UC).

Sensitivity analyses for the primary outcome measure

Figures 5–7 show the time-to-event data for the secondary ‘event’ definitions. In all cases, there was no clear graphical indication against proportional hazards and no statistically significant interaction of trial arm by time (hence, proportional hazards were assumed).

• A total of 286 out of 476 (60.1%) participants had experienced the event ‘stable resolution of symptoms’ (Figure 5) by week 48: 146 out of 238 (61.3%) in the SC arm and 140 out of 238 (58.8%) in the UC arm. The median event times were 24.0 weeks for SC and 24.0 weeks for UC. The HR for SC relative to UC was 1.11 (95% CI 0.87 to 1.43; p = 0.389).

• A total of 437 out of 476 (91.8%) participants had experienced the event ‘first improvement of symptoms’ (Figure 6) by week 48: 219 out of 238 (92.0%) in the SC arm and 218 out of 238 (91.6%) in the UC arm. The median event times were 2.0 weeks for SC and 2.0 weeks for UC. The HR for SC relative to UC was 0.95 (95% CI 0.79 to 1.15; p = 0.585).

• A total of 396 out of 476 (83.2%) had experienced the event ‘stable improvement of symptoms’ (Figure 7) by week 48: 198 out of 238 (83.2%) in each trial arm. The median event times were 6.0 weeks for SC and 5.0 weeks for UC. The HR for SC relative to UC was 1.05 (95% CI 0.86 to 1.29; p = 0.606).

FIGURE 5.

Kaplan–Meier time-to-event analysis of the secondary event classification (time to stable resolution of symptoms). a, Estimation is limited to the largest event-free time if it is censored. Cumulative proportion of resolved cases by week 48: 0.636 (all), 0.662 (SC) and 0.613 (UC).

FIGURE 6.

Kaplan–Meier time-to-event analysis of the secondary event classification (time to first improvement of symptoms). a, Estimation is limited to the largest event-free time if it is censored. Cumulative proportion of resolved cases by week 48: 0.935 (all), 0.941 (SC) and 0.929 (UC).

FIGURE 7.

Kaplan–Meier time-to-event analysis of the secondary event classification (time to stable improvement of symptoms). a, Estimation is limited to the largest event-free time if it is censored. Cumulative proportion of resolved cases by week 48: 0.865 (all), 0.875 (SC) and 0.855 (UC).

Table 7 summarises the hypothesis test results for the time-to-event sensitivity analyses. Estimates for the HR (SC relative to UC) were close to 1.0 (range 0.86–1.22) in all cases and mostly in the range of 1.0 to 1.2, indicating a slightly faster time to event in the SC arm than in the UC arm. Similar to the primary end point evaluation, all tests showed no statistically significant differences between the trial arms (p > 0.05). Sensitivity analyses addressing different patterns for the missingness of data showed similar HR estimates (and close to 1) across the different scenarios.

TABLE 7 - Sensitivity analyses of the primary outcome measure

A total of 286 out of 476 (60.1%) participants had experienced stable resolution of symptoms by week 48 (SC, 61.3%; UC, 58.8%).

A total of 437 out of 476 (91.8%) participants had experienced a first improvement of symptoms by week 48 (SC, 92.0%; UC, 91.6%).

A total of 396 out of 476 (83.2%) participants had experienced stable improvement of symptoms by week 48 (SC, 83.2%; UC, 83.2%).

Analysis of 286 complete responders.

	HR (95% CI); p-value
Stable resolutiona	1.11 (0.87 to 1.43); 0.389
Improvementb	0.95 (0.79 to 1.15); 0.585
Stable improvementc	1.05 (0.86 to 1.29); 0.606
Resolution: missing
Mean imputation	1.15 (0.90 to 1.47); 0.272
Resolved imputation	1.12 (0.89 to 1.41); 0.324
Stable resolution: missing
Mean imputation	1.12 (0.87 to 1.43); 0.382
Resolved imputation	1.11 (0.89 to 1.39); 0.337
Improvement: missing
Mean imputation	0.95 (0.79 to 1.15); 0.620
Resolved imputation	0.92 (0.77 to 1.11); 0.386
Stable improvement: missing
Mean imputation	1.06 (0.86 to 1.29); 0.595
Resolved imputation	1.07 (0.88 to 1.29); 0.506
Parametric
Weibull model	1.16 (0.91 to 1.50); 0.237
Exponential model	1.19 (0.89 to 1.59); 0.249
Log-normal model	0.88 (0.65 to 1.18); 0.380
Interval-censoring
Weibull	1.17 (0.91 to 1.49); 0.219
Exponential	1.20 (0.89 to 1.62); 0.237
Log-normal	0.86 (0.62 to 1.19); 0.361
Log-rank (non-parametric) test	p = 0.368
Breslow (non-parametric) test	p = 0.489
Tarone–Ware (non-parametric) test	p = 0.432
Complete-case analysisd	1.22 (0.89 to 1.66); 0.212

Per-protocol analysis

A total of 51 participants were judged to not have had care according to the SC intervention protocol and were removed from the per-protocol analysis. The main reason for protocol deviation was participants in sciatica group 2 in the SC arm of the trial receiving fewer than three physiotherapy sessions; this was due to either patients not completing their physiotherapy course of treatment or being discharged by their treating physiotherapist. A further 15 patients, one in sciatica group 1 and 14 in group 2, were eventually referred to spinal specialist services by their treating physiotherapist (for details of protocol deviation reasons, see Appendix 2, Table 25). A total of 138 out of 187 (73.8%) had experienced the event (symptom resolution) by week 48 in the SC arm [compared with 168/238 (70.6%) in the UC arm]. The median event times were 10.0 weeks for the SC and 12.0 weeks in the UC. The HR for SC relative to UC was 1.10 (95% CI 0.84 to 1.45; p = 0.491). Similar to the primary ITT analysis, there was no statistically significant difference between SC and UC.

Subgroup analyses

By sciatica group (1, 2 and 3)

Figure 8 shows the time-to-event data for the primary outcome stratified according to sciatica group (groups 1, 2 and 3) and by trial arm. For group 1, 68 out of 107 (63.6%) participants and 82 out of 107 (76.6%) participants reported that their sciatica symptoms had resolved by weeks 16 and 48, respectively: 31 out of 53 (58.5%) and 38 out of 53 (71.7%), respectively, in the SC arm, and 37 out of 54 (68.5%) and 44 out of 54 (81.5%), respectively, in the UC arm. For group 2, 127 out of 211 (60.2%) and 157 out of 211 (74.4%) participants reported resolved symptoms by weeks 16 and 48, respectively: 66 out of 105 (62.9%) and 81 out of 105 (77.1%), respectively, in the SC arm, and 61 out of 106 (57.5%) and 76 out of 106 (71.7%), respectively, in the UC arm. For group 3, 79 out of 158 (50.0%) and 105 out of 158 (66.5%) participants had resolved symptoms by weeks 16 and 48, respectively: 40 out of 80 (50.0%) and 57 out of 80 (71.3%), respectively, in the SC arm, and 39 out of 78 (50.0%) and 48 out of 78 (61.5%), respectively, in the UC arm. The median event times are displayed in Figure 8 and show that participants in sciatica groups 1 and 3 in the SC arm took slightly longer to reach resolution than participants in sciatica groups 1 and 3 in the UC arm. Conversely, participants in group 2 in the SC arm achieved resolution of sciatica symptoms slightly faster than participants in group 2 in the UC arm. The adjusted HR for group 2 relative to group 1 (ratio SC-to-UC) was 1.46 (95% CI 0.84 to 2.54; p = 0.182), and for group 3 relative to group 1 (ratio SC-to-UC) it was 1.43 (95% CI 0.78 to 2.62; p = 0.243). For the secondary event definitions, the adjusted HRs were, respectively: (i) 1.70 (95% CI 0.94 to 3.09; p = 0.081) for group 2 relative to group 1 and 1.52 (95% CI 0.79 to 2.90; p = 0.209) for group 3 relative to group 1 for ‘stable resolution’; (ii) 1.17 (95% CI 0.72 to 1.90; p = 0.526) for group 2 relative to group 1 and 1.05 (95% CI 0.63 to 1.77; p = 0.849) for group 3 relative to group 1 for first ‘improvement’; (iii) 1.11 (95% CI 0.67 to 1.84; p = 0.689) for group 2 relative to group 1 and 0.96 (95% CI 0.56 to 1.64; p = 0.891) for group 3 relative to group 1 for ‘stable improvement’.

FIGURE 8.

Kaplan–Meier time-to-event (time to first resolution of sciatica symptoms) in each of the sciatica groups (1, 2, and 3).

By sciatica clinical diagnostic classification

Figure 9 shows the time-to-event data for the primary outcome stratified according to the clinical diagnosis of either disc-related sciatica or spinal stenosis. Information on clinical diagnosis of sciatica was extracted from the SCOPiC trial clinic physiotherapists’ standardised clinical assessment forms. Physiotherapists recorded a response for this for 461 participants: for 398 out of 461 (86.3%) participants, physiotherapists recorded a clinical diagnosis of disc-related sciatica, and for the remaining 63 (13.7%) participants, a clinical diagnosis of sciatica due to stenosis was recorded. Figure 9 displays the mean and median event times. It shows similar average event times for the SC and UC arms within the subgroup of participants with a clinical diagnosis of disc-related sciatica, but faster resolution for those with clinically suspected stenosis in the SC arm than for those with clinically suspected stenosis in the UC arm. The adjusted HR for the clinically diagnosed spinal stenosis subgroup relative to the group with disc-related sciatica (ratio SC-to-UC) was 1.92 (95% CI 1.01 to 3.65; p = 0.043). For the secondary event definitions, the adjusted HRs were 1.56 (95% CI 0.78 to 3.12; p = 0.213) for ‘stable resolution’, 1.05 95% CI (0.61 to 1.81; p = 0.868) for first ‘improvement’ and 1.18 (95% CI 0.67 to 2.09; p = 0.563) for ‘stable improvement’.

FIGURE 9.

Kaplan–Meier time-to-event (time to first resolution of sciatica symptoms) for those with a clinical diagnosis of disc-related sciatica and those with clinical diagnosis of spinal stenosis.

Table 8 displays summary baseline characteristics by clinically diagnosed sciatica due to either disc prolapse or spinal stenosis.

TABLE 8 - Baseline characteristics by sciatica diagnostic label (and trial arm)

HADS-A, Hospital Anxiety and Depression Scale – Anxiety subscale; HADS-D, Hospital Anxiety and Depression Scale – Depression subscale; SBI, Sciatica Bothersome Index; S-LANSS, Self-report Leeds Assessment of Neuropathic Symptoms and Signs; TSK, Tampa Scale of Kinesiophobia; WPI, Work Performance Index.

Scored 1–32,844; higher scores indicate lower levels of deprivation.

Assessed in clinic (clinic questionnaire); otherwise, self-reported in the patient baseline questionnaire.

Leg pain interference was considered present if a participant scored ≥ 6 on the 0–10 NRS.

Scored 0–23; higher scores indicate greater disability.

A 0–24 composite score; higher scores indicate worse symptoms.

Possible range: 0–24; a score of ≥ 12 indicates possible neuropathic pain.

Scored 17–64; higher scores indicate a greater fear of movement.

Scored 0–21; higher scores indicate greater levels of anxiety/depressive symptoms. A score of ≥ 11 is considered indicative of a ‘probable case’ of depression/anxiety.

Note

Frequency counts do not always add to total column numbers due to some missing data.

Characteristic	Clinically diagnosed disc-related sciatica			Clinically diagnosed spinal stenosis
	All (N = 398)	SC (N = 203)	UC (N = 195)	All (N = 63)	SC (N = 30)	UC (N = 33)
Age (years), mean (SD)	49.8 (13.2)	48.4 (13.6)	51.3 (12.7)	66.5 (11.2)	66.8 (11.2)	66.2 (11.3)
Sex, n (%)
Females	215 (54)	108 (53)	107 (55)	38 (60)	21 (70)	17 (52)
Males	183 (46)	95 (47)	88 (45)	25 (40)	9 (29)	16 (48)
IMD rank value,a median (IQR)	14,228 (7193–21,626)	14,228 (7084–21,790)	14,604 (7193–21,790)	20,215 (10,569–22,887)	18,046 (6767–24,210)	20,841 (13,447–23,491)
Employed, n (%)	295 (75)	155 (77)	140 (72)	27 (43)	12 (40)	15 (45)
Time off work because of sciatica, in the previous 12 months, n (%)	164 (56)	78 (51)	86 (61)	9 (32)	3 (23)	6 (40)
WPI score, mean (SD)	6.0 (2.8)	5.9 (2.8)	6.1 (2.7)	5.2 (3.4)	4.5 (3.3)	5.8 (3.5)
Prescription,b n (%)	315 (81)	168 (84)	147 (77)	49 (79)	21 (70)	28 (88)
Sciatica group, n (%)
1	93 (23)	47 (23)	46 (24)	11 (17)	4 (13)	7 (21)
2	173 (43)	87 (43)	86 (44)	28 (44)	16 (53)	12 (36)
3	132 (33)	69 (34)	63 (32)	24 (38)	10 (33)	14 (42)
Usual back pain score, mean (SD)	5.9 (2.8)	5.9 (2.7)	5.8 (2.8)	5.7 (3.1)	5.7 (2.9)	5.6 (3.4)
Usual leg pain score, mean (SD)	6.8 (2.3)	6.7 (2.3)	6.8 (2.3)	7.6 (1.7)	7.5 (1.6)	7.7 (1.8)
Leg pain interferencec (yes/no), n (%)	302 (76)	153 (76)	149 (76)	52 (83)	26 (87)	26 (79)
Leg pain interference score, mean (SD)	6.5 (2.4)	6.3 (2.5)	6.6 (2.4)	6.4 (2.6)	6.1 (2.8)	6.6 (2.5)
Symptom duration, n (%)
< 2 weeks	41 (10)	13 (6)	28 (14)	6 (10)	2 (7)	4 (12)
2–6 weeks	170 (43)	89 (44)	81 (42)	18 (29)	6 (20)	12 (36)
6–12 weeks	85 (21)	50 (25)	35 (18)	15 (24)	8 (27)	7 (21)
3–6 months	50 (13)	26 (13)	24 (12)	10 (16)	5 (17)	5 (15)
6–12 months	15 (4)	7 (3)	8 (4)	5 (8)	3 (10)	2 (6)
> 12 months	36 (9)	17 (8)	19 (10)	9 (14)	6 (20)	3 (9)
RMDQ score,d mean (SD)	11.1 (5.4)	11.1 (5.3)	11.2 (5.5)	11.5 (4.9)	11.6 (5.2)	11.4 (4.7)
SBI score,e mean (SD)	14.5 (5.0)	14.5 (5.0)	14.4 (5.1)	14.8 (4.6)	15.2 (4.8)	14.5 (4.5)
S-LANSS score,f n (%)
< 12	214 (57)	109 (58)	105 (56)	38 (69)	15 (60)	23 (77)
≥ 12	164 (43)	80 (42)	84 (44)	17 (31)	10 (40)	7 (23)
TSK score,g mean (SD)	40.7 (6.0)	40.2 (6.1)	41.1 (6.0)	39.8 (6.8)	41.1 (6.5)	38.7 (7.0)
HADS-A score,h mean (SD)	8.0 (4.1)	7.7 (4.1)	8.3 (4.0)	7.2 (3.9)	8.5 (3.6)	5.9 (3.7)
HADS-A, n (%)
Normal (0–7)	192 (48)	102 (51)	90 (46)	38 (60)	12 (40)	26 (79)
Possible (8–10)	99 (25)	53 (26)	46 (24)	13 (21)	11 (37)	2 (6)
Probable (≥ 11)	106 (27)	47 (23)	59 (30)	12 (19)	7 (23)	5 (15)
HADS-D score,h mean (SD)	6.3 (4.0)	6.4 (4.0)	6.3 (4.0)	6.6 (3.8)	6.6 (3.4)	6.6 (4.2)
HADS-D, n (%)
Normal (0–7)	259 (65)	133 (66)	126 (65)	40 (64)	20 (67)	20 (61)
Possible (8–10)	71 (18)	33 (16)	38 (19)	12 (19)	6 (20)	6 (18)
Probable (≥ 11)	67 (17)	36 (18)	31 (16)	11 (17)	4 (13)	7 (21)
Sleep problem, n (%)	260 (65)	127 (63)	133 (68)	41 (65)	19 (63)	22 (67)
General health, n (%)
Excellent	21 (5)	10 (5)	11 (6)	1 (2)	0 (0)	1 (3)
Very good	87 (22)	47 (23)	40 (21)	12 (19)	4 (13)	8 (24)
Good	173 (44)	89 (44)	84 (43)	32 (51)	17 (57)	15 (46)
Fair	88 (22)	41 (20)	47 (24)	16 (25)	7 (23)	9 (27)
Poor	28 (7)	15 (7)	13 (7)	2 (3)	2 (7)	0 (0)
STarT Back Screening Tool risk subgroup, n (%)
Low	93 (23)	49 (24)	44 (23)	12 (19)	5 (17)	7 (21)
Medium	218 (55)	114 (56)	104 (54)	34 (54)	16 (53)	18 (55)
High	85 (21)	39 (19)	46 (24)	17 (27)	9 (30)	8 (24)

Secondary outcomes

A summary of overall Global Perceived Change (GPC) ratings from the 4- and 12-month questionnaires is shown in Figure 10. There was no statistically significant difference in GPC ratings between trial arms at either follow-up time point, although, at 12 months, slightly more participants in the SC arm than in the UC arm reported greater GPC ratings. In total, 78 out of 188 (41.5%) in the SC arm reported being ‘completely recovered’ or ‘much better’ at 4 months, compared with 85 out of 194 (43.8%) in the UC arm; at 12 months, these numbers were 97 out of 174 (55.8%) and 88 out of 176 (50.0%), respectively. The summary results for all secondary outcome measures are presented in Table 9. Usual back pain and usual leg pain intensity reduced from 5.8 and 6.9, respectively, at baseline to 3.6 and 3.2, respectively, at 4 months and to 2.9 and 2.8, respectively, at 12 months. The mean reduction in usual back pain was 3.7 at 4 months and 4.1 at 12 months for participants who reported being ‘completely recovered’ or ‘much better’, compared with 0.8 and 0.9 at 4 and 12 months, respectively, for participants who did not report these levels of improvement. There was a similar differential for usual leg pain, which was reduced by 5.6 and 5.8 at 4 and 12 months, respectively, for those who reported being ‘completely recovered’ or ‘much better’, and by 2.2 and 1.8, respectively, for those who were not. As these contrasts in average scale scores between those patients who were completely recovered or much better and those who were not were similar across back and leg pain scales, this would suggest that participants’ rating of GPC of symptoms was similarly driven by both back and leg pain intensity (and not only one of these).

FIGURE 10.

Secondary outcomes: GPC by trial arm at 4 and 12 months. (a) GPC at 4 months; and (b) GPC at 12 months. The ORs are the proportional odds of a favourable outcome relative to not being better, derived through ordinal logistic regression.

Sciatica-related physical disability levels (RMDQ mean scores) also reduced, from 11.2 at baseline to 6.3 at 4 months and 5.3 at 12 months. As shown in Table 9, similar improvements, on average, were reported for all outcomes in both trial arms. There were no statistically significant differences in the secondary outcomes between trial arms. Absolute effect sizes for numerical secondary outcomes (standardised mean difference relative to baseline SD) were nearly all lower than the preclassified cut-off point for a ‘small’ effect size (i.e. ≤ 0.2).

Table 10 summarises the data for the secondary outcome measures between trial arms for sciatica groups 1, 2 and 3. There was little difference in health outcomes between interventions arms in each of the three sciatica groups. A trend of worsening health status values was recorded from group 1 through group 2 to group 3. For example, for disability, the 4-month mean value was lowest, at 3.1, in group 1 and highest, at 9.2, in group 3 (similar trends were observed for other measures).

Process outcomes

Process outcomes comprised data on referrals generated at the SCOPiC trial research clinic (randomisation point) and care delivery data in physiotherapy and specialist spinal settings, including secondary care referrals and treatments. Data were captured with CRFs (at the SCOPiC trial research clinic and at physiotherapy services), patient self-reports and from hospital records review. Waiting times for treatments (other than physiotherapy) and/or tests were collected/calculated from hospital records only.

Referral patterns from SCOPiC trial research clinics

Table 12 presents the referral patterns for participants attending the SCOPiC trial research clinic and randomised to the trial. The stratification algorithm for allocating SC participants to the matched care pathways was followed in all but four cases. In the UC arm, physiotherapists provided initial assessment and advice and directed patients’ care according to clinical judgement, with the majority of participants referred to physiotherapy treatment; only 28 out of 238 (12%) participants were discharged back to GP care from the research clinic. Appendix 3, Table 26, summarises the GPs’ preferred care pathways for UC participants versus care pathways decided by the physiotherapists delivering trial UC at the point of randomisation at the SCOPiC trial research clinic; it shows that, in 62% of cases, GPs said that they would have referred to physiotherapy and, in 32% of cases, they would have continued with GP care.

Summary of physiotherapy care

Many trial participants were referred for physiotherapy treatment. Data were collected with the use of CRFs, for 214 (89.9%) SC participants and 235 (98.7%) UC participants. In terms of sciatica groups, completed CRFs in the SC arm were obtained for 51 out of 53 (96.2%) participants in group 1, 104 out of 105 (99.2%) participants in group 2 and 59 out of 80 (73.8%) participants in group 3. For the UC arm, these proportions were 54 out of 54 (100%) participants in group 1, 105 out of 106 (99.1%) participants in group 2 and 76 out of 78 (97.4%) participants in group 3.

The overall median number of total treatment sessions was similar for participants in the SC and UC arms [2 (IQR 1–4) and 2 (IQR 0–3), respectively]. Participants in group 2 who were allocated to SC had, on average, more treatment sessions than similar participants in the UC arm. The median time to the first physiotherapy appointment (for those who were referred to physiotherapy) was 9 days (IQR 6–15 days) for SC participants and 21.5 days (IQR 11–46 days) for UC participants. Physiotherapy treatment was delivered over a shorter time frame in the SC arm than in the UC arm [median 38 days (IQR 12.5–70 days) and 66 days (IQR 29–97 days), respectively]. The differences in time to the first physiotherapy appointment and in treatment delivery were driven by SC groups 1 and 2. Appendix 4, Table 27 and Figures 17 and 18, presents information from CRFs that captured physiotherapy treatment data.

Data from spinal interface clinics

Table 13 summarises data on appointment numbers at the spinal interface clinics, and time to appointment, over the trial follow-up period, and data on referral decisions made at the spinal interface clinics. Information was derived from available hospital records review only; therefore data are missing for a number of participants. It should also be noted that, if any participants had their NHS appointment or treatments carried out in a private setting, these are not included in a patients’ NHS records and would not be captured by the hospital records review. Hospital data were available for 427 out of 476 (89.7%) participants in the trial: 215 out of 238 (90.3%) in the SC arm and 212 out of 238 (89.1%) in the UC arm. By arm and sciatica group, data were available for the following numbers of participants: SC group 1, 49 out of 53 (92.5%); SC group 2, 94 out of 105 (89.5%); SC group 3, 72 out of 80 (90.0%); UC group 1, 47 out of 54 (87.0%); UC group 2, 95 out of 106 (89.6%); and UC group 3, 70 out of 78 (89.7%). In line with the trial protocol, a significantly larger proportion of group 3 participants in the SC arm than in the UC arm were referred to and attended the spinal specialist interface clinic (82.0% vs. 58.3%, respectively). The overall time taken to be seen in the interface clinic from the date of randomisation was significantly shorter for the SC arm (median 12 days) than for the UC arm (115 days). The total number of visits to secondary care spinal orthopaedic services (referrals made from spinal interface clinics plus any other referral route) was 32 for the SC arm and 23 for the UC arm. There were five visits to pain clinics (two in the SC arm and three in the UC arm). Overall, there was not much difference in time to appointment in orthopaedics between SC and UC participants, but, as expected in the fast-track pathway, sciatica group 3 participants in the SC arm were seen more quickly than similar participants in the UC arm [87 days from randomisation (IQR 58–105 days), compared with 120 days (IQR 42–151 days), respectively]. In addition, SC participants listed for spinal epidural injections had the procedure more quickly than similar participants in the UC arm [60 days (IQR 41–93 days), compared with 161 days (IQR 113–253 days), respectively]. For spinal surgery, the time to procedure was very similar between SC and UC arms for the small numbers of patients having surgery [236 days (IQR 99–372 days) and 216 days (IQR 102–305 days), respectively]. Appendix 5, Table 28, summarises available data on time to procedures, as captured by hospital records review.

Secondary care interventions

Data from self-report and hospital records review were combined; Table 14 summarises the numbers of participants who underwent MRI, spinal surgery or spinal epidural injection, per trial arm and per sciatica group within each arm. Overall, based on the available data, few participants had spinal surgery and more SC than UC participants had spinal epidural injections. Information about the specifics of injections or surgery was available for a limited number of participants from all those who received these treatments. Seventeen caudal epidural injections and six transforaminal injections were recorded, along with nine discectomies and two decompressions (laminectomies).

Adverse events

There were no SAEs or AEs in either arm of the trial.

Summary

The SC participants reported improvement a median of 2 weeks earlier than UC participants; however, this difference was small and not statistically significant (HR 1.14, 95% CI 0.89 to 1.46). The trial did not find convincing evidence that this model of SC led to faster improvement or better clinical results than UC, for patients with clinically diagnosed sciatica. The results are discussed in detail in Chapter 6.

Aim

The overall aim of the health economic evaluation was to compare the resource use, costs and benefits of the SC intervention with UC for participants presenting to primary care with sciatica, and to estimate the cost-effectiveness of SC.

Methods

The primary evaluation was a cost–utility analysis; the base-case analysis was performed from the UK NHS and Personal Social Services (PSS) perspective, with additional analyses for the health-care provider perspective and the societal perspective incorporating productivity costs. The quality-adjusted-life-year (QALY) was used as the measure of health outcome, using data from the EuroQol-5 Dimensions, five-level version (EQ-5D-5L), questionnaire. Participant-level intervention and sciatica-related costs over the 12-month follow-up period were included, with costs reported in 2017 prices. The analysis of cost-effectiveness focused on estimation of the incremental cost per QALY gained.

Costs and outcomes were not discounted, as the trial was limited to a period of 12 months’ follow-up. Trial findings are reported in accordance with the Consolidated Health Economic Evaluation Reporting Standards. 61

Measurement and analysis of outcomes

The EQ-5D-5L questionnaire data were collected at baseline and at 4 and 12 months. Responses were converted to index scores using the interim cross-walk value set for mapping from the EQ-5D-5L to the EuroQol-5 Dimensions, three-level version. 62 QALYs were generated for each participant using the area under the utility curve approach assuming linear interpolation between the three utility measurements. 63 To avoid bias, adjustment for differences between the groups in baseline EQ-5D-5L scores was undertaken using a regression-based adjustment. 64 Health utility values and QALYs accrued over the 12-month follow-up were summarised by trial arm and assessment time point and presented as means and SDs, with between-arm differences compared using bootstrapped mean differences. A univariate and multivariate regression model was performed to estimate QALY differences between the SC and UC arms. In the univariate model, QALY differences over the 12-month follow-up period were regressed against the trial arm variable. In the multivariate model, the QALY differences were regressed against the baseline EQ-5D-5L scores, age and sex variables.

Resource use and costs

Analysis

Results

Resource use and costs

Work-related outcomes

Results regarding paid employment, work status, sciatica-related work absence and reduced productivity are reported in Table 19. Overall, sciatica-related work absence costs were slightly lower, and presenteeism costs were slightly higher, in the SC arm than in the UC arm.

Incremental costs

Table 20 summarises the total costs per participant for the imputed and complete-case analysis for the NHS, health-care and societal perspectives. The costs per participant are also shown, including additional hospital records data on health-care use, from the sensitivity analysis. The results show that, although the SC and UC costs to the NHS were very similar, the cost of SC was slightly higher, overall, than the UC cost (difference £46.21, 95% CI –£110.60 to £187.06).

Cost-effectiveness analysis

Table 20 provides a summary of the cost-effectiveness analysis using data from the base-case NHS perspective. Results from the health-care and societal perspective are also reported. In the base-case analysis, SC was associated with an adjusted mean increase in costs of £46.21 (95% CI –£110.60 to £187.06) and an adjusted mean reduction in QALYs of –0.011 (95% CI –0.035 to 0.013) per participant, compared with UC, over the 12 month follow-up. The SC intervention was therefore dominated by UC; the net monetary benefit (NMB) was –£275 if society’s willingness to pay for a QALY (λ) is valued at £20,000. Uncertainty around the value is illustrated in Figure 11 (NHS perspective) and Figure 12 (health-care perspective). The majority of the points are in the north-west quadrant, showing that SC was more costly and less effective than UC and is unlikely to be cost-effective at the £20,000–30,000 per QALY threshold range, which is currently used by NICE to determine the cost-effectiveness of health-care interventions for the NHS. The CEACs show the probability that the SC intervention is cost-effective at different levels of willingness to pay for a QALY. For the NHS perspective, at £20,000 and £30,000 per QALY, the probability that the SC intervention is cost-effective was 0.18 and 0.19, respectively.

FIGURE 11.

Cost-effectiveness from the NHS perspective. (a) Cost-effectiveness plane comparing the SC intervention with UC, showing 5000 bootstrapped replicates of the ICER; and (b) CEAC for SC (intervention) compared with UC.

FIGURE 12.

Cost-effectiveness from the health-care perspective. (a) Cost-effectiveness plane comparing the SC intervention with UC; and (b) CEAC for SC (intervention) compared with UC.

Sensitivity analyses

Complete-case analysis

Under this scenario, from an NHS perspective, SC was slightly less costly (–£14.03) and less effective (–0.025 QALYs) when the analysis was restricted to participants with complete cost and outcome data (see Table 20). The ICER was therefore lower, at £550.10, and the NMB was –£461 (λ = £20,000). The probability of SC being cost-effective at λ = £20,000 was 0.15; at £30,000, it was 0.14. This is illustrated in Figure 13.

FIGURE 13.

Cost-effectiveness for the complete-case analysis, from the NHS perspective. (a) Cost-effectiveness plane comparing the SC intervention with UC; and (b) CEAC for SC (intervention) compared with UC.

Impact of stratified care on health-related quality of life by sciatica group

Quality-of-life data (EQ-5D-5L and QALY scores) at each time point by trial arm and sciatica group are provided in Table 21. At baseline and at the end of 12 months’ follow-up, participants in the SC arm had slightly lower EQ-5D-5L scores than participants in the UC arm, in all sciatica groups, with the exception of group 1. At 12 months, QALY outcomes in all sciatica groups were similar (very slightly in favour of UC: –0.0004, –0.0198 and –0.008 for groups 1, 2 and 3, respectively), with all CIs crossing zero.

Incremental costs

Point estimates of incremental costs for each sciatica group are reported in Table 22. Results for NHS costs showed that SC was slightly cheaper for participants in sciatica groups 1 and 2, but more expensive for participants in sciatica group 3, because of the higher number of spinal injections and costs associated with the fast-track pathway (MRI and specialist consultation). Similar results were observed from the societal perspective, with SC being less expensive than UC for participants in sciatica groups 1 and 2, because of lower work absence-related costs, but more expensive for participants in sciatica group 3 because of the costs associated with the fast-track pathway.

Cost-effectiveness analysis by sciatica group

Table 22 reports the point estimates of incremental costs, incremental QALYs and ICERs by each sciatica group. These analyses revealed considerable uncertainty around the central estimates of the incremental costs and QALYs because of the reduced sample size in each group, but SC remained unlikely to be cost-effective. Overall, the results demonstrated that SC generated slightly fewer QALYs, on average, and was slightly cheaper than UC for sciatica groups 1 and 2, but not for group 3. Based on conventional willingness-to-pay thresholds for additional QALYs (approximately £20,000 per QALY),75 SC was dominated by UC over 12 months for those in group 3, that is a slightly lower mean health benefit (0.00841 additional QALYs) and a higher mean NHS cost (£209.56). For participants in groups 1 and 2, SC was associated with a slightly lower mean NHS cost (–£9.45 and –£54.45 for groups 1 and 2, respectively) and a lower mean health benefit (–0.00041 and –0.01981 QALYs for groups 1 and 2, respectively). Figures 14–16 present the uncertainty around the cost-effectiveness scenarios, along with the cost-effectiveness planes for the analysis per sciatica group. The location of the ‘swarm’ of bootstrapped cost–utility pairs explains the nature of the uncertainty in the generated incremental cost and QALY estimates. The CEACs in Figures 14–16 show the probability that SC is cost-effective compared with UC for a range of cost-effectiveness thresholds. In sciatica group 1, minimal cost savings were realised, but with considerable uncertainty regarding both costs and health benefits, although the ICER (in the south-west quadrant) was > £20,000 per QALY. Similarly, for group 2, cost savings were realised, with the majority of bootstrapped cost-effectiveness mean differences falling in the south-west quadrant of the £30,000 and £50,000 per QALY cost-effectiveness threshold lines. For sciatica group 3, the majority of the replicates of the ICERs were located in the north-west quadrant, with SC dominated by UC. The results suggest that SC is unlikely to be cost-effective at the £20,000–30,000 per QALY threshold currently used by NICE to determine the cost-effectiveness of health technologies. 75 The CEACs in Figures 14–16 show that the probability that SC is cost-effective compared with UC is 0.51, 0.19 and 0.26 for sciatica groups 1, 2 and 3, respectively.

FIGURE 14.

Sciatica group 1: comparing SC with UC. (a) Cost-effectiveness plane; and (b) CEAC.

FIGURE 15.

Sciatica group 2: comparing SC with UC. (a) Cost-effectiveness plane; and (b) CEAC.

FIGURE 16.

Sciatica group 3: comparing SC with UC. (a) Cost-effectiveness plane; and (b) CEAC.

Summary

In the context of the economic analysis results, the SC model tested in the trial is unlikely to be a cost-effective option for the NHS, using commonly applied willingness-to-pay threshold values of £20,000–30,000 per QALY gained.

Aim and objectives

The aim of the nested qualitative research was to determine, and understand, the acceptability of the fast-track care pathway to patients and the clinicians who either were directly involved in patient care in this pathway or would be if this were to become incorporated into current clinical management in future, that is GPs, spinal specialist physiotherapists and spinal surgeons. As outlined in the Chapter 2, GPs identified patients with suspected sciatica for invitation to the SCOPiC trial research clinics for assessment of their eligibility to participate in the RCT. Spinal specialist physiotherapists were directly involved in the management of patients in the fast-track pathway: they provided specialist assessment and opinion at spinal interface clinics about further management, including consideration for more invasive treatments. Spinal surgeons were not involved in assessing patients as part of the fast-track pathway following the MRI scan, but they would assess patients who were referred for surgical opinion following spinal specialist opinion in interface clinics. Spinal surgeons were, however, not aware of which patients had been referred as a result of the trial’s fast-track pathway.

Although previous research has explored the views of clinicians and patients on the clinical management of sciatica,60,77,78 it is the fast-track pathway to MRI and spinal specialist assessment and opinion that is the novel addition in the SCOPiC trial. Gaining an understanding of how the fast-track pathway is perceived by patients and clinicians can provide important insights into this element of the SC model tested in the trial. It is for this reason that the fast-track pathway is the focus of the qualitative research.

This aim was addressed through the following specific objectives:

• to explore and understand patients’ –

  • views of their clinical care and its appropriateness and timeliness for addressing their problem

  • views of the time taken for assessment, treatment and symptom resolution

  • experiences of sciatica symptoms and their impact, to provide context in terms of the treatments received.

• to explore and understand clinicians’ views of –

  • the suitability of patients on the fast-track pathway

  • how the fast-track pathway compared with usual care

  • the added value of the fast-track pathway for this group of patients with sciatica

  • the limitations of the fast-track pathway for this group of patients with sciatica.

In addition, interviews with GPs explored perceptions of whether or not their involvement in the trial contributed to changes in their referral decisions for patients with sciatica.

Theoretical frameworks

The introduction of the fast-track care pathway clearly represents a change to usual care for a group of patients with sciatica. Changing health-care practice brings diverse challenges,79,80 including whether or not an intervention is considered meaningful and relevant to clinicians and patients, and the degree to which the intervention is seen to ‘fit’ within established ways of working. 81 Examining such issues using a theoretically underpinned approach can extend the scope of purely descriptive approaches, enabling a more cogent and coherent explanation of the issues identified in the data. Therefore, we drew on two theoretical frameworks to guide the qualitative research. The first of these is normalisation process theory (NPT),82 a theoretical framework that has been widely used to explore the introduction of new health-care interventions across a diverse range of settings, for example cancer care,83 hypertension84 and mental health. 85 NPT provides a framework for understanding why some health-care interventions are accepted and successfully embedded in routine practice, yet others are less successful in their uptake. 82 NPT aims to identify and explain ‘factors that promote and inhibit the routine incorporation of complex interventions into everyday practice’,86 through highlighting the ‘ways in which work must be reconfigured both individually and collectively by multiple stakeholders involved in the work of implementation’,87 thus taking account of both the individual-level and broader organisation-level factors. This aim is reflected in the framework’s four main components, as outlined in Murray et al:86 coherence (or sense-making), cognitive participation (or engagement), collective action (work done to enable the intervention to happen) and reflexive monitoring (formal and informal appraisal of the benefits and costs of the intervention).

Normalisation process theory aims to address both the initial introduction of a new intervention to the clinical setting, namely implementation, as well as its subsequent routinisation, namely its embedding, and the sustaining and reproduction of behaviours that allow for the longer-term integration of the intervention in everyday clinical work. Given that the aims of the nested qualitative research were to explore and understand patients’ and clinicians’ views about the acceptability of the fast-track pathway in the context of the trial, our focus was on the implementation stage of the normalisation process; we could not assess the degree to which the fast-track pathway was embedded or integrated into routine working at this early stage. Of the four key components of NPT outlined in the previous paragraph, it is the concept of ‘coherence’ that we drew on as a lens through which to interpret the interview findings. Coherence relates to the degree to which a new intervention is seen to be ‘meaningful’ to key stakeholders, and whether it ‘makes sense’ in the context of existing ways of working. The meaning attached to a clinical intervention and whether or not clinicians and patients ‘buy in’ to a new way of working is key to its early adoption and implementation. The concept of coherence thus aligns closely with our aims, as the degree to which clinicians and patients perceive the fast-track pathway as making sense in the context of the existing care pathway (i.e. ‘stepped’ care) for sciatica is key to understanding its overall acceptability.

The second theoretical framework that we drew on was Allen’s88 conceptualisation of care pathways as ‘boundary objects’. Allen88 defines a ‘boundary object’ as a loose concept, but with strong cohesive power, which enables the bringing together of the interests of different groups, while still allowing these groups to maintain their respective social identities. Allen88 argues that care pathways can be considered as ‘boundary objects’ in that they have the ability to ‘align clinical, management and service user interests around a healthcare quality agenda’. 88 She also points out, however, that they can, at the same time, give rise to conflicting agendas among different stakeholders (e.g. clinicians, patients, managers), which results in challenges when trying to meet the needs of multiple groups in the design and operationalisation of new care pathways. 88 This ‘boundary object’ conceptualisation was also drawn on as a lens through which to interpret the findings, enabling the investigation of how the fast-track pathway addressed the shared goals of patients and the three clinician groups (GPs, spinal specialist physiotherapists and spinal surgeons), aspects in which the fast-track pathway did not align with the priorities of these respective groups and where competing agendas were present in terms of the care pathway’s goals, components and operationalisation.

Methods

Individual semistructured interviews were conducted with patients in the SC care arm of the trial who were allocated to group 3’s matched fast-track pathway (n = 20) and with participating clinicians (n = 20: seven spinal physiotherapists, nine GPs and four spinal surgeons) between January 2016 and February 2018.

Sampling

Patients were purposively sampled to capture diverse characteristics including the treatment centre attended and participant demographics such as age, sex, leg pain intensity, treatments received (including patients who had received invasive treatments such as spinal injections or surgery) and response to treatment. Clinicians were sampled for variation in geographical location across the different practice areas from which patients were recruited to the trial (North Staffordshire, North Shropshire/Wales and Cheshire). GPs were further purposively sampled based on the number of patients during the trial for whom a sciatica-related diagnostic Read code had been entered, prompting the electronic ‘pop-up’ on the GP’s online system indicating that the patient may be eligible for invitation to the SCOPiC trial clinics. It was felt that interviewing those GPs who had had the most opportunity to engage with the trial, in terms of identifying patients to be invited to the SCOPiC research clinics, might yield rich interview data. Sampling based on opportunity to engage with the trial was not relevant for the other clinician groups: the spinal physiotherapists had, by definition, been involved in the management of this group of patients in terms of providing specialist opinion and assessment as part of the fast-track pathway, and spinal surgeons would not have been aware of which patients they had seen that were part of the fast-track pathway.

Recruitment

Patients were initially sent, via post, an interview invitation letter along with a participant information leaflet about the qualitative study, having consented to contact as part of their participation in the RCT. A member of the research team then telephoned these patients to ask if they would be interested in taking part in an interview. For those who were willing to take part, a convenient time and location for the interview was arranged and a confirmation letter was sent via post.

Clinicians were initially contacted by e-mail or, in the case of some GPs, via their practice manager and were subsequently posted an invitation letter and information leaflet about the qualitative study. Clinicians who agreed to take part were contacted again to arrange the interview.

Semistructured interviews

Interviews with patients, spinal physiotherapists and spinal surgeons were conducted after the 4-month follow-up time point of the RCT. This was to allow patients to reflect on their experiences of the fast-track pathway up to that point, and for clinicians to have had the opportunity to see patients as part of the fast-track pathway. Interviews with GPs were conducted once recruitment to the trial had finished (November 2017 onwards), to avoid the possibility of influencing their normal referral patterns of patients with sciatica.

Of the 20 patient interviews, 13 took place at participants’ homes, two took place at the university and five were conducted via telephone, in line with participants’ preferences. The duration of the patient interviews ranged between 21 minutes and 1 hour 15 minutes (average 48 minutes). Of the 20 clinician interviews, eight were carried out at the clinician’s practice or hospital, four took place at the university and eight were conducted via telephone. The duration of clinician interviews ranged between 19 minutes and 32 minutes in length (average 26 minutes).

Interviews were audio-recorded, with the exception of one GP interview as the GP consented to take part in an interview but not to the interview being audio-recorded. This GP reported that, although they were willing to have their views included in the research, they felt uncomfortable being audio-recorded. Instead, the interviewer took detailed notes during and immediately after the interview to capture the GP’s views. All participants provided written, informed consent prior to the start of interviews, or audio-recorded consent in the case of telephone interviews. Consent was reaffirmed verbally at the end of each interview.

Topic guides

Topic guides were used in interviews, which covered a range of areas relevant to the qualitative study aims. Separate topic guides were used for patient and clinician interviews. The clinician topic guide was adapted for the three clinician groups to reflect the differing roles of these clinicians in patient assessment and management and in the trial’s fast-track care pathway (as outlined in Aim and objectives in this Chapter). For instance, interviews with spinal surgeons placed more emphasis on exploring the suitability of patients in the fast-track pathway for surgical opinion, whereas interviews with GPs explored the impact of their participation in the trial on their management of patients with sciatica in general practice. Spinal physiotherapist topic guides placed more emphasis on asking the participants to reflect on concrete experiences of managing specific patients in the trial, as a result of their role providing specialist spinal opinion as part of the fast-track care pathway.

Topic guides were used primarily as an aide-memoire; the interviewer retained flexibility to follow up on any unexpected findings emerging during the interviews, allowing aspects of the interviews to be participant led. Early findings informed subsequent interviews, with the topic guides iteratively revised throughout the data collection process. An example of this was in early GP interviews when the GP participants highlighted that, as part of usual care, they could often identify the more ‘severe’ patients with sciatica who they felt were particularly in need of early investigation and consideration for more invasive treatments. Based on those early findings, the topic guide was revised to prompt the interviewer to explore in more detail, in subsequent interviews, the characteristics that GPs felt make patients’ presentation more ‘severe’ and, thus, their condition more suitable for early investigation, as well as the potential implications of the fast-track pathway for managing this group of patients.

Data analyses

Audio-recordings of interviews were transcribed in full and anonymised by replacing names with pseudonyms and removing other potentially identifiable information. A two-stage analysis framework was adopted that incorporated an inductive thematic analysis followed by mapping the identified themes onto the two theoretical frameworks: the ‘coherence’ construct in NPT and the conceptualisation of care pathways as ‘boundary objects’. Adopting this two-stage framework reflected a desire to avoid imposing meaning on the data, or potentially missing important findings, which can be risked when pre-established theories are applied in a deductive manner. Analysis was an iterative process and data collection continued until saturation was judged to have been reached, defined as ‘informational redundancy’: the point at which additional data no longer offers new insights. 89

Anonymised transcripts were first systematically coded on a line-by-line basis by one of the authors (BS) with the aid of the software program NVivo 10 (QSR International, Warrington, UK), to identify recurrent concepts inductively. Coding was, at first, largely descriptive, and later became more conceptual as interpretations of the data moved towards a higher level of theoretical abstraction. Coding was reflexive and recursive, with codes being revisited in the light of the findings of subsequent data collection. A random sample of six patient transcripts and six clinician transcripts was independently coded by three other members of the team to check for inter-coder reliability. Coders brought different disciplinary perspectives to the data (BS: medical sociology; BB: social gerontology; MA: clinical academic and general practice expertise; and KK: clinical academic physiotherapy and spinal specialist expertise). The aim of independent coding was also, therefore, to understand cross-disciplinary perspectives on the data and, through discussion, to come to an agreement on shared meanings and interpretations. Early findings from the patient interview data, along with a sample of three interview transcripts, were also shared with four members from the Research Institute’s PPI group. A meeting was held in which the researchers looked through these transcripts with the PPI members and explored whether or not their interpretations of the data aligned with the emergent findings. All four PPI members were broadly in agreement with the early interpretations of this interview data.

Data were analysed thematically using the constant comparison method. 90 This involved looking for connections within and across interviews, and across codes, highlighting data consistencies and variations. Initially, patient and clinician data were coded separately; these data sets were then mapped onto one another, looking at how each of the main themes that was identified did or did not manifest across both the clinician and patient interviews. Three of the four main themes that were identified could be neatly mapped across both data sets; one theme identified in the patient interviews did not emerge in the clinician data: the impact of sciatica symptoms on everyday life. Although clinicians demonstrated understanding and awareness of the impact of sciatica on patients’ lives, they discussed this mainly in relation to how the impact of symptoms on patients informed their clinical management. Whereas the other three themes relate directly to the experiences and perceptions of the fast-track pathway, this theme, as it emerged in relation to the patient data, is broader in scope and focused primarily on patients’ experiences outside clinical/health-care settings. For this reason, this theme is reported on only briefly here to provide context for understanding patients’ views and experiences of treatments received during the trial; the theme is explored in greater detail in Saunders et al. 91

The second stage of the analysis involved mapping the three themes identified in both the patient and clinician interviews onto the ‘coherence’ construct in NPT and the conceptualisation of care pathways as ‘boundary objects’. We explored the degree to which the identified themes could be seen to ‘fit’ within these frameworks, and how the theoretical constructs manifested in relation to these themes. Exploring the findings in the context of these theoretical frameworks allowed us to move beyond a purely descriptive analysis, developing insights at a higher level of abstraction and, thus, allowing for the development of a more robust understanding of the identified issues. In what follows, we outline the characteristics of the sample, before reporting on the findings in relation to each of the key themes: in brief, the one theme identified only in the patient data, followed by the three key themes identified in both the patient and clinician interviews, in turn.

Results

Main themes

The four main themes identified through the thematic analysis were:

1. the impact of sciatica symptoms on everyday life (identified in patient data only)

2. the acceptability of the fast-track care pathway

3. the perceived benefits of the fast-track care pathway

4. the waiting times for onward treatment following fast-track.

Discussion

The nested qualitative research showed that patients and all three clinician groups experienced benefit from the fast-track pathway, particularly in providing reassurance based on MRI scans. This reassurance related primarily to patients feeling assured that there was no serious underlying pathology related to their pain, such as cancer. They also reported feeling reassured that they had been given an explanation for the cause of their pain, which they could see on the scan, and, as a result, they felt more hopeful that their condition was treatable, a view supported by the availability of viable treatment options discussed with clinicians. This aligns with Pincus et al. ’s 92 classification of ‘cognitive reassurance’, relating to reassurance provided through the giving of information about aetiology and diagnosis, and discussion of suitable treatment options. The finding contrasts with recent arguments by Wheeler et al. 93 that imaging patients with LBP can lead to patient distress if degenerative changes or aberrant findings are identified. However, the difference in our findings may be explained by the clear biomedical cause that is more often present with sciatica than with non-specific LBP, which can lead to patients feeling a greater degree of certainty about their condition and, thus, stronger cognitive reassurance. These findings align with those of Ryan and Roberts,94 who found that patients with sciatica experienced ‘relief’ when investigations led to the identification of a cause of their pain. However, although the MRI findings led to reassurance, for four patients in our study, the speed with which they underwent MRI and specialist assessment led them to believe that their condition must be particularly severe and, therefore, had the opposite effect of leading to concerns.

There were differences in views identified regarding when to offer the fast-track pathway in terms of sciatica symptom duration. For patients, early MRI investigation and assessment by spinal specialists was viewed very positively compared with stepped care, which shows similarity with previous research findings on the importance patients with sciatica place on undergoing early MRI. 60,94 Among some clinicians, however, there was a concern that those with short symptom duration (i.e. < 6 weeks) might be inappropriately ‘fast-tracked’, as their symptoms may well resolve without the need for imaging and specialist assessment. Spinal surgeons and some GPs felt that it was acceptable for patients with short symptom duration to be ‘fast-tracked’, but only to provide early reassurance rather than to direct treatment. In general, there was a reluctance among participating clinicians to consider more invasive treatments for patients with symptom durations of < 6 weeks, or even 2–3 months in the view of spinal surgeons. This suggests that the overall expectation of a favourable natural course for most patients with sciatica, and the embeddedness of the current ‘stepped’-care approach in routine clinical care, makes clinicians reluctant to consider most patients with short symptom duration for more invasive treatment options early on. This shows similarity with findings in other health-care contexts: for instance, in the Dutch health-care context, Hofstede et al. 78 also observed a reluctance on the part of clinicians to refer patients with short symptom duration for investigations and specialist opinion before a period of conservative management has first been tried, and failed.

Both patients and clinicians also highlighted limitations in the fast-track pathway in not being able to influence waiting times for further treatment.

Summary

• Most patients and all three clinician groups saw potential added value in the fast-track care pathway, particularly for providing early reassurance to patients who had no serious underlying pathology and who had a clear explanation for their symptoms. However, conversely, for some patients, the speed of undergoing MRI and spinal specialist assessment led to concerns that their condition was particularly severe.

• Patients reported positive views on the short time frame for fast-track; however, there was variation in views among clinicians about the suitability for fast-track of patients with short symptom duration (i.e. those with pain lasting < 6 weeks), and some clinicians expressed concern that such patients may be inappropriately ‘fast-tracked’.

• Clinicians reported a reluctance to move from the usual ‘stepped’-care model for patients with short symptom duration, expressing reservations about considering ‘acute’ patients for invasive treatments. Some patients chose to opt for conservative management even when offered invasive treatment options. This reluctance on the part of clinicians and patients, which is based on the hopeful expectation of natural resolution, is evidenced in the fact that most of the patients interviewed (16/20) were initially referred for physiotherapy as a first treatment option following fast-track.

• Patients and clinicians expressed concerns that the ‘fast-track’ pathway did not extend to shortening the time that patients waited for further treatments after their MRI and initial specialist assessment at the spinal interface clinics.

Overview of the SCOPiC trial aim, justification and key findings

Summary of the key findings

We did not find convincing evidence, overall, that the SC model tested in this RCT led to better outcomes for patients than UC. On average, participants in both arms of the trial improved substantially from baseline, and to a similar degree, over time. In relation to care pathways, at the SCOPiC trial research clinics and at the point of randomisation, all trial participants received a one-off physiotherapy session, including assessment and advice tailored to each individual patient. Following this, the management of participants (n = 237) in the SC arm was directed by the use of the stratification algorithm (see Chapter 2): 153 participants were referred for physiotherapy treatment (49 participants were scheduled to receive up to two physiotherapy sessions to help with self-management and 104 were scheduled to receive up to six physiotherapy sessions), four patients were discharged back to their GP (as it was deemed that they did not require any further treatment) and 80 were fast-tracked to MRI and an appointment with a spinal specialist physiotherapist at the spinal interface clinics, for further assessment and management. In the UC arm (n = 238), 200 patients were referred for further physiotherapy treatment, 28 were discharged back to their GP and 10 were referred to the spinal interface clinics. Data on the care received by participants in both trial arms showed that, during the 12-month follow-up period, 22 participants in the SC arm went on to receive spinal epidural injections and five underwent spinal surgery (two from group 3), whereas 13 participants in the UC arm received spinal injections and eight underwent surgery (five from group 3). On average, participants in the SC arm received their treatments sooner than participants in the UC arm, including both primary and secondary care interventions.

In the primary trial data analysis of time to first self-reported resolution of symptoms, participants in both the SC and UC arms reported improvements in a similar time frame. In the SC arm, the median time to first resolution of symptoms was slightly shorter (2 weeks), with a non-statistically significant ‘time to event’ HR of 1.14 (95% CI 0.89 to 1.46). Approximately one-quarter of participants in the SC and UC arms reported ‘resolution of symptoms’ by week 4 and week 5 from randomisation, respectively. By 12 weeks, approximately 50% of participants in both arms had reported first ‘resolution of symptoms’. By the end of the follow-up period at 12 months, 74% of participants in the SC arm and 71% in the UC arm had reported resolution of symptoms. Adopting a more inclusive definition of improvement, to include ‘better’ in addition to ‘completely recovered’ and ‘much better’, just over 90% of participants in both trial arms had reported improvement by the end of the follow-up period.

Analysis of the secondary outcomes also showed no statistically significant differences between participants in the two trial arms; this included the key secondary outcome of sciatica-related physical disability (mean difference in RMDQ score at 12 months –0.53 points, 95% CI –1.84 to 0.78). The number of days lost from work due to sciatica over the 12-month follow-up was also similar in both arms (SC arm, mean 5.48 days and UC arm, means 5.67 days).

There was a trend for larger numbers of participants in the SC arm to report being satisfied with care, and with the results of care, at the 4-month follow-up, yet this was reversed at the 12-month follow-up. However, any between-arm differences were small and statistically non-significant, and likely to be driven by a few individuals. At 12 months, fewer than 20 participants in each arm reported not being satisfied with the care they had received or the results of the care received.

Subgroup analyses, by sciatica group (1, 2 and 3), showed similar improvements in both trial arms, in terms of time to symptom resolution, with HRs for all definitions of improvement showing no statistically significant differences between SC and UC. There were statistically non-significant trends in the data: the median time to ‘symptom resolution’ for groups 1 and 3 in the SC arm was slightly longer than for those in the UC arm, whereas the median time to ‘symptom resolution’ for group 2 was slightly faster than for those in the UC arm. Although prespecified, these subgroup analyses were exploratory as the trial was not powered to detect differences at the level of the sciatica groups (1, 2 and 3).

Similarly, exploratory subgroup analysis by the clinical diagnosis of disc-related sciatica/spinal stenosis showed that the between-arm effect for SC over UC was significantly higher for the subgroup of patients with a clinical diagnosis of stenosis than for those with a clinical diagnosis of disc prolapse. SC participants with a clinical diagnosis of spinal stenosis reported faster resolution of symptoms than UC participants with a clinical diagnosis of spinal stenosis (HR 1.92, 95% CI 1.01 to 3.65). However, this was an exploratory analysis, the reasons for the difference are unclear, and given the small sample size, it may be a chance finding.

The primary health economic analysis found that the adjusted mean difference between SC and UC in health-related quality of life (measured by QALYs) at 12 months was small, at –0.011 (95% CI –0.036 to 0.013), in favour of UC. The mean difference in NHS and PSS costs at 12 months was also small: £46.21 (95% CI –£110.60 to £187.06) in favour of UC. Compared with UC, the SC model tested in the SCOPiC trial was marginally more costly and associated with slightly lower health-related quality-of-life outcomes over 12 months. For both QALYs and costs, CIs were wide and crossed zero, highlighting the uncertainty in the cost-effectiveness estimate. Given these results, SC is unlikely to be a cost-effective option using commonly applied willingness-to-pay threshold values of £20,000–30,000 per QALY gained. Extrapolation beyond 12 months’ follow-up was not required as the health outcomes and costs for SC and UC were very similar at 12 months, and unlikely to change over a longer time period.

The qualitative investigation of the acceptability of the fast-track care pathway in the SC arm showed that both patients and clinicians reported some benefits, particularly in providing reassurance early in the presentation of symptoms, based on MRI results, and also in considering more invasive treatment options, if appropriate and desired, for patients with persistent symptoms. However, for patients with a short duration of symptoms, there was collective clinical reluctance to consider tests or invasive treatments early on, with spinal specialists reporting that they would choose to wait at least 6 weeks, or even as long as 2–3 months before considering surgical options. Some patients, even when offered treatments such as injection or surgery, also chose to continue with conservative treatment options or to ‘wait and see’ whether or not their pain would improve over time. Data from the ‘time-to-event’ analysis showed that approximately 25% of participants in sciatica group 3 had reported improvement by 4–5 weeks (in both trials arms). So, in that respect, although ‘fast-tracking’ some sciatica patients (mainly those with more severe symptoms) to MRI and specialist opinion was considered by most as a positive action, it was not consistently helpful to clinicians in their discussions with their patients about management options when the duration of symptoms was short.

Discussion of trial findings

A number of reasons may be considered in explaining the trial results and the key finding of no difference, overall, in clinical outcomes between the SC and UC arms of the trial. The first is the stratification algorithm we used to subgroup patients and allocate them to one of three matched care pathways. The sensitivity, specificity and PPV of the algorithm for patient allocation to group 3, namely the fast-track pathway, were 51%, 73% and 22%, respectively. 50

In the absence of factors consistently associated with outcomes specifically in sciatica patients, the stratification algorithm combined prognostic data from a previously developed tool for non-specific LBP patients (the STarT Back Screening Tool34) with data associated with referral to specialist spinal services, in order to identify sciatica patients for the fast-track pathway (group 3). Those with a good prognosis using the STarT Back Tool (group 1) and, irrespective of clinical characteristics, had a low likelihood of being referred to spinal specialists and were matched with a brief intervention comprising two sessions with a physiotherapist. All other patients (group 2) were matched with a course of up to six physiotherapy sessions. Participants in both trial arms still had access to GP care. Participants in group 3 (the fast-track) scored consistently higher on all measures than participants in groups 1 and 2; it is clear that group 3 is at the worse end of the symptoms spectrum, based on the specific algorithm used in this trial. In terms of further health-care resource utilisation, only small numbers of participants from groups 1 and 2 had spinal injections and/or spinal surgery. Nevertheless, the algorithm’s ability to correctly identify those patients most likely to need or benefit from referral to a spinal specialist may be limited (PPV was 22%),50 and a number of participants in group 3 either improved by the time they attended the spinal interface clinic or had no evidence of nerve root compression on their MRI scans. Data from other studies of sciatica in primary and secondary care settings show that, in about 20–40% of clinically diagnosed sciatica cases, MRI scans do not show evidence of nerve root compression;4,10,96,97 therefore, invasive management options are deemed inappropriate. Similarly, the algorithm’s sensitivity of 51%50 indicates that it is possible that a number of patients who were likely to have been appropriate for the fast-track pathway were not identified using the algorithm. Future research may result in a different algorithm for directing care for patients with sciatica; however, we believe that the approach taken here, which combined prognostic information and sciatica severity characteristics identified from the clinical assessment, is a reasonable approach to have developed and tested for this group of patients.

Another consideration is the effectiveness of the UC intervention in this trial. All UC participants received treatment in the SCOPiC trial. Participants consulted their GP and then attended the SCOPiC trial research clinic where they were assessed by a physiotherapist, and all had advice tailored to them at this visit. From the research clinic, 84% (n = 238) in the UC arm were referred for further physiotherapy; only 4% were referred to specialist spinal services and only 12% were discharged back to their GP (indicating that the assessing physiotherapist felt that these participants did not require any further treatment). This compares to data on GP-suggested referral decisions (for UC participants) at the point of identifying potential participants for the trial, where 32% said they would recommend that the patient continue with GP care, 62% to be referred to physiotherapy and 6% to be referred to spinal specialists. Thus, in the operationalisation of the trial from SCOPiC trial research clinics, where physiotherapists assessed patients and made decisions about clinical management, more participants received physiotherapy intervention than would have been the case had GPs made decisions about usual care. However, it is important to point out that these decisions by GPs and physiotherapists about onward treatment were made at different time points, with the difference in time ranging from a few days to a couple of weeks from when each patient consulted their GP to when decisions were made about onward care by physiotherapists in the SCOPiC trial clinics. It is possible that symptoms changed between the two assessment points and this may be reflected in the different decisions about management in 36% of UC participants. Overall, as regards primary care management, more participants in the UC arm received treatment in addition to GP care than would have been observed in normal, usual GP-led primary care practice. The consequence is that the UC intervention in the SCOPiC trial may have been more effective than the care usually received in true usual primary care practice. Participants in the UC arm, on average, waited longer before being referred to specialist services than those in the SC arm; this was expected, taking into account the inclusion of the fast-track pathway for participants in group 3 of the SC arm. The time to be seen in physiotherapy services was also somewhat longer for participants in the UC arm than for those in the SC arm, but, overall, the SCOPiC trial UC arm may have provided an enhanced level of care compared with true usual GP-led primary care, and that may help to explain the findings of the trial. However, this finding in itself is important, as it indicates that good clinical care for this population leads to good results overall, and that the SC model we tested does not lead to better outcomes for patients.

From the qualitative research findings, it was evident that, for patients who might be potential candidates for injection and/or surgery but have symptoms for a short time only, both clinicians and patients preferred to try conservative management while giving time for improvement of symptoms. Whether or not this had any bearing on the trial results is unclear and, on balance, unlikely, as the numbers that proceeded to have invasive procedures were small.

Comparison with existing studies

We are not aware of any studies that have investigated a SC model for sciatica management, and, generally, very few sciatica studies have been successfully conducted in primary care. Comparisons are not straightforward between studies owing to differences in characteristics, for example levels of disability and/or pain, duration of symptoms, variation in condition case definitions and availability of similar outcome measures. We could not identify any primary care studies that used ‘time to improvement’ as their outcome. However, we can make some comparisons in terms of improvements in pain intensity and disability levels between the SCOPiC trial participants and the participants of a recent primary care study, as the latter study has a broadly similar population, including in terms of the duration of symptoms. Mathieson et al. 99 investigated the effectiveness of pregabalin versus placebo medication for primary care sciatica patients; similar improvements were seen in both trial arms, with approximately two-thirds of participants using additional health services, including physiotherapy treatment by approximately one-third. The mean leg pain intensity score reduced from 6.3 and 6.1 at baseline to 3.4 (percentage change: 46%) and 3.0 (50.9% change) at 1 year for the active drug and placebo groups, respectively. 99 In the SCOPiC trial, the mean change was slightly greater: in the SC arm, the leg pain intensity score reduced from 6.8 to 2.9 (57.3% change) and, in the UC arm, it reduced from 6.9 to 2.8 (59.4% change). Similar changes were observed in disability scores: 54.9% and 51.3% change in the SCOPiC trial SC and UC arms, respectively, and 44.6% and 52.3% change in the active drug and placebo arms of the Mathieson et al. 99 trial, respectively. Data from the ATLAS study,14 a clinical cohort that included sciatica patients presenting to primary care, all seeing a physiotherapist and mostly treated according to a stepped-care model, showed a 38.8% change for disability score and a 55.3% change for leg pain intensity score at 1 year, which is slightly less change than the SCOPiC trial or the Mathieson et al. 99 trial. However, a larger number of patients in the ATLAS study had longer pain duration; this may explain the difference in disability results between this population and the trial populations described earlier in this section. These comparisons with other similar sciatica studies set in primary care provide some evidence that the UC arm in the SCOPiC trial can be considered a best-care comparison, which, without the use of any stratification tools, seems to have worked well for approximately 75% of participants. The percentage change in disability score in both arms of the SCOPiC trial (SC arm, 54.9%; UC arm, 51.3%) was considerably higher than that achieved in a study of SC for non-specific LBP (SC, 43.9%; best-care control, 34%). 32

Strengths and limitations

The CONSORT guidelines were followed both in the conduct and in the reporting of the trial. To protect against bias, randomisation was carried out by means of a computer-generated code to ensure concealment and balance of allocation, participants (and their GPs) were not told of their allocated trial arm and validated outcome measures were used to reduce measurement error. However, the design of the trial makes it impossible to conceal allocation of trial arm for the physiotherapists delivering the interventions. We protected against contamination by having different physiotherapists deliver treatment for the SC and UC arms for the duration of the trial. The main analyses were conducted according to the ITT principle and the overall statistical and health economics analyses plan was agreed in advance with the TSC.

The SC model tested, with its two components, the stratification algorithm and the matched care pathways for each of the three sciatica groups, which were developed prior to the trial and were agreed with input from all stakeholders in the management of patients with sciatica (i.e. GPs, spinal surgeons, spinal physiotherapists, rheumatologists, patients and researchers). The primary outcome (time to symptom resolution) was chosen in consultation with patients with experience of either current or previous sciatica. For patients, it was important to gain pain relief as swiftly as possible, and clinicians were comfortable with the characteristics of patients who would then be fast-tracked to MRI and specialist assessment. As part of this, clinicians involved in developing the matched interventions for sciatica groups in our SC model agreed that symptom duration should not be a key factor in a model of SC. However, it is known that short symptom duration reduces the predictive ability of stratification tools more generally. 103

A further strength is the high follow-up rate (88.3% for the SC arm and 90.3% for the UC arm) for the primary outcome in the SCOPiC trial, obtained via text messages. A sensitivity analysis assessing the impact of missing data on the primary outcome did not affect the trial results.

One of the limitations is that the stratification algorithm that we developed and internally validated prior to the trial, to match patients to the trial care pathways, was not externally validated in an independent primary care sciatica cohort before being tested in the trial. 50 We were not aware of any similar primary care data set that we could use to do this that was available at the time before starting this trial. We acknowledge that the algorithm’s sensitivity, specificity and PPV were not optimal; however, in the absence of any other methods to guide matching sciatica patients to care pathways early on, including a referral for MRI and specialist assessment, this stratification algorithm was the first step in investigating the effectiveness of SC for this population. It is acknowledged that the design of the trial does not allow differentiation between the effect of the stratification algorithm (the subgrouping) and the matched care pathways (matched treatments). Any effects observed can be attributed only to the combination of using the algorithm and the matched treatment pathways.

A further limitation is that health-care resource use information was based on self-reported data, which can potentially be affected by under-reporting. In addition, the number of missing data for the health economic analysis (up to 50% for cost data) was a weakness. A sensitivity analysis, however, showed comparable results. We also acknowledge that the trial was not powered to detect differences at the level of each sciatica group (1, 2 and 3) or in the suspected specific clinical diagnosis of disc-related sciatica or spinal stenosis between the trial arms. Therefore, caution is needed in the interpretation of the results of the subgroup analyses, and we recommend that these are seen as very exploratory findings.

Implications for current practice

Contrary to the evidence for superior clinical effectiveness and cost-effectiveness of a prognostic model of SC for patients with non-specific LBP,32,33 the new SC model specifically developed for patients with sciatica and tested in the SCOPiC trial (comprising both prognosis and factors associated with referral to spinal specialists) was not superior to non-stratified UC. Overall, SC participants reported symptom improvement a median of 2 weeks earlier than UC participants, and, although this is important for individual patients, on a population level, this difference is small and most likely to be of little consequence in terms of overall improvement over time. The UC arm involved all patients having a session with a physiotherapist for assessment and advice, and most were referred for further physiotherapy input; therefore, the improvements observed in the UC arm could be attributed to the physiotherapy intervention received. However, it is difficult to disentangle the treatment effect from the improvement due to the natural history of sciatica. A few more participants in the SC arm than in the UC arm received epidural injections, but this did not lead to better outcomes compared with UC.

In contrast to patients with non-specific LBP for whom invasive treatments are not appropriate, for sciatica patients, these are potential management options, depending on MRI findings and a patient’s overall health and preferences. However, lessons learned from the crossover levels observed in sciatica trials comparing surgery with conservative management4,104 clearly indicate that a number of patients improve over time without treatment and others decide against the risk of invasive procedures if there is chance of improvement by other means. It may be, therefore, that, for patients with suspected sciatica, a SC model in which all care pathways are considered early on will not be appropriate for most patients, because some of the available treatments carry risks (and are expensive) and both patients and clinicians prefer not to consider these options early in the presentation of symptoms, as was evident from the qualitative research results. Taking into account the significant burden sciatica has on patients and society, trying to identify which patients need earlier or more intensive or invasive treatment makes intuitive sense, but this trial shows that, in reality, this is difficult to achieve. It could be that our SC algorithm has potential but needs to have different cut-off points for allocation to sciatica groups and matched care pathways, or it could be that different approaches to subgrouping and targeting treatment for patients with sciatica need to be developed and tested, other than the approach we used in this trial. Further research should attempt to identify factors consistently and differentially associated with outcome or treatment effect in the sciatica population, to facilitate construction of better predictive models for use in clinical decision-making. In the meantime, efforts to systematise clinical care delivery for patients with sciatica may help reduce unhelpful practice variation.

Suggestions for further research

As the common prognostic factors shown to be relevant in non-specific LBP are not consistent prognostic factors of pain persistence in sciatica,11,13 avenues for further research in prognostic factors in sciatica may include investigation of sensory profiles based on the hypothesis that differences in these profiles are associated with different pathophysiological mechanisms of the involved nerve fibres, and, therefore, differential prognoses or response to treatments may follow. 105–107 There is some evidence from the treatment of other neuropathic conditions that pharmacological treatment stratification according to sensory profiling has a differential treatment effect. 108

Further research that develops and tests new SC models, for example symptom-based and pathophysiological mechanisms stratification using sensory profiling, might have potential to better inform treatment decisions early on.

Contributions of authors

Nadine E Foster (https://orcid.org/0000-0003-4429-9756) (NIHR Research Professor of Musculoskeletal Health in Primary Care, NIHR Senior Investigator and Director of Keele CTU) was responsible for the original proposal, securing funding for the trial and drafting the original proposal. As chief investigator, she had overall responsibility for the trial and was a member of the TMG. She also critically revised all chapters of the report and edited several versions of the report.

Kika Konstantinou (https://orcid.org/0000-0003-4228-9386) (Senior Clinical Lecturer and Consultant Spinal Physiotherapist) contributed to the original funding proposal. She was a member of the TMG and led the trial on a day-to-day basis as principal investigator, provided training for the trial physiotherapists and oversaw quality control for the trial. She drafted the final report and led the critical revision and editing of the final report.

Martyn Lewis (https://orcid.org/0000-0002-3667-132X) (Reader in Biostatistics) contributed to the original funding proposal, was a member of the TMG, wrote the SAP, carried out data cleaning, was responsible for the statistical analysis and reporting, drafted the trial’s clinical outcomes results chapter and contributed to the methods chapter.

Reuben Ogollah (https://orcid.org/0000-0002-5777-4117) (Research Fellow in Biostatistics) wrote the SAP, was a member of the TMG (until September 2017), carried out data cleaning, was responsible for the statistical analysis and reporting, and contributed to the trial’s methods and clinical outcomes results chapters.

Benjamin Saunders (https://orcid.org/0000-0002-0856-1596) (Research Associate: Health Services Research) was a member of the TMG, conducted the qualitative interviews, co-led the qualitative data analysis and drafted the qualitative chapter.

Jesse Kigozi (https://orcid.org/0000-0001-7608-4923) (Research Fellow in Health Economics) was a member of the TMG, conducted the cleaning of the health economic data, worked on the health economic analyses and data interpretation and drafted the health economics chapter.

Sue Jowett (https://orcid.org/0000-0001-8936-3745) (Reader in Health Economics) was a member of the TMG, contributed to the original funding proposal, developed the economic data collection and analysis plan, and oversaw the health economic analysis and the drafting of the health economics chapter.

Bernadette Bartlam (https://orcid.org/0000-0002-8557-6222) (Senior Lecturer, Applied Health Research) was the senior qualitative research team member, was responsible for the data collection and analyses and was a member of the TMG (until February 2018). She conducted a number of the qualitative interviews and contributed to the drafting of the qualitative chapter.

Majid Artus (https://orcid.org/0000-0001-5319-9015) (NIHR Clinical Lecturer in General Practice) contributed to the original funding proposal, provided general practice expertise, was a member of the TMG and contributed to the conduct and analysis of the qualitative interview research and the drafting of the qualitative chapter.

Jonathan C Hill (https://orcid.org/0000-0001-6246-1409) (Senior Lecturer in Physiotherapy) contributed to the original funding proposal and was a member of the TMG.

Gemma Hughes (https://orcid.org/0000-0003-0056-9291) (Trial Manager) was responsible for overseeing the day-day-running of the trial, was a member of the TMG, contributed to drafting the report and formatted the final report.

Christian D Mallen (https://orcid.org/0000-0002-2677-1028) (Director, Institute for Primary Care and Health Sciences and NIHR Research Professor in General Practice) contributed to the original funding proposal and provided general practice expertise.

Elaine M Hay (https://orcid.org/0000-0002-9545-4296) (Professor of Community Rheumatology) contributed to the original funding proposal.

Danielle A van der Windt (https://orcid.org/0000-0002-7248-6703) (Professor of Primary Care Epidemiology) contributed to the original funding proposal.

Michelle Robinson (https://orcid.org/0000-0002-2266-8250) (Study Co-ordinator) was responsible for the day-to-day co-ordination of the trial and was a member of the TMG. She also carried out data cleaning and provided reports.

Kate M Dunn (https://orcid.org/0000-0002-6202-2606) (Professor of Epidemiology) contributed to the original funding proposal, was a member of the TMG and contributed to the drafting of the methods chapter.

All authors contributed to the final report and approved the final version.

Publications

Konstantinou K, Lewis M, Dunn KM, Ogollah R, Artus M, Hill JC, et al. Stratified care versus usual care for management of patients presenting with sciatica in primary care (SCOPiC): a randomised controlled trial. Lancet Rheumatol 2020;2:e401–11.

Saunders B, Konstantinou K, Artus M, Foster NE, Bartlam B. Patients’ and clinicians’ perspectives on a ‘fast-track’ pathway for patients with sciatica in primary care: qualitative findings from the SCOPiC stratified care trial. BMC Musculoskelet Disord 2020;21:469.

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Access to available anonymised data may be granted following review.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care.