Summary of lessons learnt
This study has several important findings with major implications for developing evidence, within a UK setting, for challenging, expensive interventions with the potential for rare but significant side effects. We found that treatment pathways for acute sciatica varied across research sites and changed during the study period. This necessitated a flexible and heterogeneous approach to study recruitment, matching local treatment pathways. It was possible to introduce a novel treatment approach (biologic therapy) requiring specialist services (rheumatology), not normally a part of existing treatment pathways, within the context of a clinical trial. However, delivering this RCT was challenging, involving multiple professional groups across different health care organisations. In the future, additional feasibility studies, more efficient site set-up, improved and pilot-tested recruitment methods and longer recruitment periods might be appropriate.
There were four main factors that led to delays and early trial closure: contracting issues, securing the payment of ETCs, site withdrawal due to concerns about reactivating TB in a highly prevalent area, and a complex trial recruitment process that did not always match local care pathways. There were long delays agreeing and exchanging subcontracts with participating centres and sites, and contractual discussions with one site were never concluded. Earlier agreement between sponsor, university centres and NHS sites might have been facilitated by the use of model agreements such as the Brunswick research collaboration agreement [
40]; site feasibility questionnaires; or research infrastructure that could facilitate the contracting process in multi-site research, such as the National Institute of Health Research Translational Research Partnership [
41].
Negotiations for the ETCs were protracted in England, where responsibility for these costs had to be negotiated with different NHS organisations with competing cost pressures; new arrangements are needed [
42]. One site withdrew from the RCT before starting recruitment because of a change in the PI’s perception of acceptable risk in the local population, fuelled by recent high-profile media cases. Further discussions between the trial management group and the local PI around potential risks, related concerns and the degree of equipoise might have prevented site withdrawal.
During trial set-up new national management guidance was published [
6], as well as a new national back pain and radicular pain pathway [
7]. In one site the local sciatica management pathway changed around the time that it opened to recruitment. This site initially relied only on referrals to its secondary care musculoskeletal service but later involved the primary care research network, which was starting to identify participants just before trial closure. In the other open site there were operational issues with identifying the research physiotherapist resource and fitting the trial around the clinical commitments of participating clinicians.
The main method for identifying participants was retrospective GP record review, but this had a low rate of response, with only a small proportion seen at the screening assessment. We had modelled the numbers of eligible participants for our recruitment projections on the previous Assessment and Treatment of Leg pain Associated with the Spine (ATLAS) cohort study, which identified patients in real time as they were consulting, with retrospective record review used only as a backup [
43]. Although we identified large numbers of potential participants, only small numbers returned reply slips indicating a willingness to participate. It is not known why potentially eligible participants did not wish to participate. Informal feedback suggested that some patients might have been much improved by the time they were contacted about the trial; some might have found the trial procedures too burdensome, such as the complex two-stage recruitment process; whilst others might not have wished to participate in an RCT, especially in a clinical trial of an investigational medicinal product involving a medication with significant potential adverse effects. Greater patient and public involvement could offer insights into how to explore this. Two of the clinical sites were going to recruit participants using the same methods as the ATLAS cohort, which have been used successfully in another RCT of primary care-delivered treatment for sciatica [
44]. However, these two sites signed their contracts just prior to trial closure. Although potential participants had started to be identified, there was insufficient time to recruit them.
The current management of RCTs within the United Kingdom has emphasised recruitment efficiency and delivery of outcomes within short timelines [
45]. This remains appropriate for treatments that fit within existing treatment pathways; when they do not, a new pathway must be developed specifically for the trial. In the current study we introduced medical screening and biologic therapy administration delivered through experienced secondary care rheumatology services. The heterogeneity of existing clinical pathways for sciatica (in primary and in secondary care) necessitated a multifaceted approach, with different solutions for different sites, requiring flexibility when pathways at single sites changed between the planning and execution of the trial.
Comparison with previous literature
The previous systematic review of biological agents for sciatica found a small number of RCTs and other studies with small numbers recruited [
17]. Many of these studies also had poor rates of recruitment, both in the UK NHS [
46] and elsewhere in Europe [
47].
Slow or inadequate recruitment to publicly funded multicentre RCTs is still a common problem [
48]. Systematic reviews of RCTs that compared methods to increase trial recruitment found that effective interventions included telephone or text reminders 2 weeks after receiving the letter of invitation, the use of lay advocates who were already involved in the study, monetary incentives, and non-blinding of trial participants. The evaluation of recruitment strategies within RCTs was advocated [
49,
50]. Results of a systematic review concerning the recruitment activity of clinicians in RCTs include the use of qualitative research to identify and overcome recruitment barriers, reduction of clinical workload associated with participation in RCTs, extra training and protected research time [
51].
Implications for future research
We make a number of recommendations for future researchers (Table
3). A number of these are pertinent to all RCTs conducted in the United Kingdom. For example, we would recommend full discussions between the sponsor’s contracting department and all university centres and NHS sites to obtain early agreement about what the contracts need to include and how the contracting process should be arranged, so that the university centres and the NHS sites have a clear understanding of their delegated roles and tasks. This may involve model contracts such as the Brunswick research collaboration agreements, which have been designed to be suitable for the majority of cases where two or more universities receive a joint research grant [
40]. Early discussions about site requirements, perhaps using a site feasibility questionnaire, early dialogue with sites’ research and development departments, and the early appointment of research staff in each site would facilitate trial set-up.
Table 3
Ten lessons learnt for consideration in a future trial
Contracts | 1 | Early agreement between sponsor, NHS sites and university centres about how the contracting process should be arranged with model research collaboration agreements |
Site set-up | 2 | Early discussions about site requirements using a site feasibility questionnaire |
3 | Recruitment of a dedicated research physiotherapist (or other personnel) at each site |
Treatment acceptability | 4 | Establish if the proposed treatment is acceptable to all principal investigators |
| 5 | Determine if the proposed treatment is acceptable to sciatica patients, using further qualitative research |
Recruitment | 6 | Simplify two-stage recruitment process |
7 | Use telephone or text reminders two weeks after patients receive letter of invitation |
8 | Use of lay advocates already recruited into the study |
9 | Recruitment during real-time GP consultations |
Feasibility study | 10 | Feasibility study testing several key recruitment methods |
We also make recommendations pertinent to the circumstances of this particular RCT. The impact of research staff shortages, in this case a research physiotherapist at one site, could be avoided by having dedicated research staff. In addition, involvement of the research staff during the initial planning stage would have been useful for planning the recruitment strategy.
Further qualitative research is needed to identify reasons for low recruitment rates, using methods such as the QuinteT Recruitment Intervention, which uses a combination of standard and innovative qualitative research methods, with some simple quantification, to understand recruitment and identify sources of difficulty [
52]. Possible reasons for poor recruitment include concerns about the nature of the trial intervention and its side effects, perceived burden of trial participation, natural history of recovery of severe sciatica, perceptions about the nature of sciatica itself, and whether the treatment under study is consistent with these. People who believe that their sciatica will resolve quickly (either spontaneously or with treatment) are unlikely to commit to a trial of medical intervention with long follow-up, particularly if they perceive that it would not provide (and might delay) definitive treatment. Such beliefs in the study population might not be well-founded in evidence, and pre-recruitment education might be necessary to help potential participants appreciate the possible benefits that might be achieved from novel interventions that are being investigated.
Patient recruitment from ‘real-time’ GP consultations may have reduced the delays associated with retrospective checks of GP consultations and from referrals to physiotherapy and secondary care settings. Unfortunately, because of delays in agreeing ETCs and finalising contracts, there was insufficient time to recruit any participants using this method before trial closure.