Enhancing Recruitment Using Teleconference and Commitment Contract (ERUTECC): study protocol for a randomised, stepped-wedge cluster trial within the EFFECTS trial

Many randomised controlled trials (RCTs) fail to meet their recruitment goals in time [1, 2]. A study of 114 multicentre trials in the UK showed that 45% failed to reach 80% of their recruitment goal. Less than one third of the trials recruited their original target number of participants within the time originally specified, and around one third had to be extended in terms of time and resources [3], and it has been identified as the highest priority to find methods to enhance recruitment to the RCT [4]. There are a few trials that have sought to evaluate different strategies for recruiting patients. However, these studies are small, and some are hypothetical, making the interpretation unclear [5]. Qualitative interventions within trials to improve recruitment have been developed in the UK [6], and in addition, online resources [7, 8] are available. In spite of that, we need to know more about the barrier and success factors for trials in recruiting patients and especially in RCTs. Given these facts, trialists are advised to include study recruitment strategies within their trials [5].

One alternative is to embed trials of recruitment interventions within host trials. An embedded recruitment trial is defined as [9]:

‘…a RCT in which an intervention (or several interventions) to enhance recruitment outcomes are tested in the context of another RCT (or several RCTs) known as the host RCT(s)’.

The EFFECTS (Efficacy oF Fluoxetine – a randomisEd Controlled Trial in Stroke, Clinical Trials number NCT02683213) study seeks to investigate whether 20 mg fluoxetine daily compared with placebo for 6 months after acute stroke improves patients’ functional outcome [10]. The EFFECTS study is a multicentre trial aiming to recruit 1500 patients in Sweden. The study collaborates with two other investigator-led studies, FOCUS (UK) and AFFINITY (Australia/New Zealand/Vietnam). Each trial is funded independently and intends to report its own results [10].

The first patient in EFFECTS was included on 20 October 2014. Our primary recruitment goal was that each centre should randomise at least two patients per month. In reality, there are huge differences between centres regarding recruitment. Looking back between June 2016 and May 2017, for instance, only three centres have achieved the recruitment goal: Danderyd Hospital, Karolinska University Hospital Solna and Sundsvall Hospital, while another three (Mora General Hospital, Alingsås General Hospital and Skaraborg Hospital Skövde) came close. In fact, these six centres have so far included half of the individuals. This pattern – that a few centres have included the majority of individuals – has been consistent since the start, irrespective of different strategies. The recruitment rate per month and centre is updated in real time and is available in the public domain [11].

Currently, the EFFECTS trial includes 30 patients per month [11], and recruitment projections would take the trial beyond its current funding. Thus, a new strategy is needed.

After talking to the principal investigators (PIs) at the centres, and to trialists in the UK with similar studies, we are confident that five of the top recruiters (Danderyd Hospital, Karolinska University Hospital Solna, Sundsvall Hospital, Mora General Hospital and Skaraborg Hospital Skövde) have reached their full potential, and little would be gained from the planned intervention.

We have asked the centres via a survey, and have also discussed this in person at investigator meetings, what they consider to be the most important barrier to recruitment for trials. They have said that the most important barrier is the lack of time for the physician responsible to identify the right patients and to carry out the study-specific procedures at baseline. The second most important factor is the lack of time for all other study personnel. They have a high clinical workload at their clinic and do not have any time specifically dedicated to working with clinical trials. It is the head of department (verksamhetschef) who is responsible for finances and personnel. Without the support of the head of department there will hardly be any change in recruitment. We will, therefore, invite the head of department (verksamhetschef) at the local centre to the teleconference in addition to the PI and the research nurse. Hopefully, we can work together to identify what could be done to provide more opportunities at the clinic to do research for both physicians and nurses within working hours. We have chosen a teleconference because it is time effective and less expensive than a face-to-face meeting. In addition, we hypothesised that a commitment contract would make it more personal for the local centre.

We have followed the 2013 SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) Checklist [12] in conjugation with the 2013 SPIRIT explanation and elaboration guidance for protocols of clinical trials [13] (Additional file 1).

Does a structured teleconference re-visit with the study personnel (PI, at least one research nurse, and the head of department) at the centres, accompanied by a commitment contract, enhance recruitment by 20% during 60 days post intervention, compared with 60 days pre-intervention in the EFFECTS study?

We have chosen the stepped-wedge cluster randomised study design [14] for three reasons. First, it is not possible for us to carry out the intervention at 14–15 centres at the same time. Second, we believe that all medium and low recruiting centres could gain from the intervention, and in a stepped-wedge cluster design all centres are exposed. Every step provides before and after observations, and every step switches from control to become exposed to the intervention. Third, we have noticed a seasonal variation in recruitment. During the Christmas and Easter holidays, and especially during the summer, recruitment is falling. This intervention gives a realistic view of recruitment throughout the whole year. To the best of our knowledge, only one randomised stepped-wedge study in the field of stroke has been done [15].

EFFECTS has been running for almost 3 years, and we know the recruitment rate per centre and month. Initially we had 35 centres. Three centres – Karolinska Hospital Huddinge, Visby General Hospital and Högsbo Rehabilitation Hospital – have been closed due to low recruitment mainly because of a lack of PI. Of the 32 centres, we have identified five centres that have achieved our goal, that is recruiting two or more patients per month. They are the top recruiters: Danderyd Hospital, Karolinska University Hospital Solna, Skaraborg Hospital Skövde, Mora General Hospital and Sundsvall Hospital. We will not intervene at these centres because we believe that they have reached their full potential and the intervention is too weak for them. Not including the top recruiters, this leaves us with 27 centres in this study.

Figure 1 shows a Consolidated Standards of Reporting Trials (CONSORT) flow diagram [16] of the study. Hence, ERUTECC is a randomised, stepped-wedge cluster design [14] study. We made some minor changes to the CONSORT flow diagram layout.

We will use a randomised, stepped-wedge cluster design, where every step provides data before and after intervention, but not at the same point in time.

Figure 2 illustrates the stepped-wedge cluster study in our study. As the name indicates, the intervention has 11 different steps, and the schema takes the form of a wedge. Figure 3 is a schematic diagram of the time schedule of enrolment, interventions (including any run-ins and washouts) and assessments for the participating centre.

In ERUTECC, all centres have a 60-day running-in period and a 60-day post-intervention period. Our plan is to start in September 2017 and finish in October 2018.

The 27 sites will be divided into two categories: medium and low recruiters. The rationale for this is that we do not want to risk all medium recruiters falling into the same step, for example the summer period, which usually a low-recruiting period. The centre will be randomised in terms of the order in which intervention will be performed, leading to (at least) one medium and (at least) one low recruiter in each of the 11 steps (Fig. 2).

All centres in the EFFECTS study, their classification, type and numbers of patients are listed in Table 1.

This is what we will do before the intervention (teleconference):

This is a teleconference between the chief investigator (CI) plus the trial manager and the study personnel. The meeting requires that at least the PI, one research nurse and the head of department (verksamhetschef) must attend the meeting. If they are not all able to attend, the meeting will be rescheduled. Ideally, as many members as possible listed on the delegation list will attend the meeting.

The agenda

If a centre refuses to participate in the study, the reason for this is noted. We will include the centre as ‘intention-to-treat’.

First, we excluded the top recruiting centres. Second, the remaining centres were categorised as low or medium recruiters. Third, in each category (low and medium) the order of the intervention was randomised using a computer-generated allocation sequence using SAS version 9:4 by our statistician, PN. Neither the intervention nor the order of the intervention is communicated in advance. The intervention in known by the Steering Committee, ACL (author of this manuscript) and the Regional Ethical Committee, and is not communicated to the centres. The order of the allocation is known by three persons (PN, EI and EL), and is kept in an Excel file behind a secure firewall at Stockholm County Council. The centre will be contacted by the trial manager (EI) 3–5 weeks before the planned intervention. For obvious reasons, the intervention is not blinded.

The study has been running for almost 3 years and we know the recruitment rate per centre and month. Among the 32 centres in EFFECTS, we have identified five centres that have achieved our goal, i.e. recruiting two or more patients per month. These are the top recruiters (Danderyd, Karolinska Solna, Skövde, Sundsvall and Mora). We will not intervene with these centres because we believe that they have reached their full potential and the intervention is too weak for them. Not including the top recruiters leaves us with 27 centres in this study.

The participants in the EFFECTS study are not aware of the embedded study. We will not mention the intervention or in what order we will do the intervention outside the group responsible for ERUTECC. For obvious reasons, there is no blinding of the intervention. The centres will not be informed that we are measuring numbers of randomised patients before and after the teleconference, but they are fully aware that we want to enhance recruitment. The exact numbers of recruitments per centre has been available in the public domain through a link that has been updated in real time since the start of the EFFECTS study.

We will compare the numbers of included individuals 60 days before the teleconference with the numbers of subjects 60 days post intervention. The null hypothesis is that there will be no difference before and after. We consider a 20% increase in recruitment rate as clinical relevant, although it is arbitrarily chosen.

Statistical example:

In EFFECTS, we have a screening list in which we note all eligible individuals who have been considered as candidates for EFFECTS. If these patients are not part of the study, the reason is noted. Usually the reasons are administrative and non-willingness. We will compare the numbers of randomised patients with the eligible patients on the screening list.

The outcome of the study is the number of included patients per centre. Inclusion is available through our electronic randomisation system, and we will keep a separate log for every teleconference. If a centre refuses to be part of ERUTECC, we will still include the centre in the analysis (intention-to-treat). In addition, we will also do a ‘per-protocol analysis’.

We aim to start this study in September 2017. At that time, we will have 27 active centres, not including the five top recruiters.

It is hard to anticipate any harm from ERUTECC. The host trial, EFFECTS, has specific monitoring performed by the Karolinska Trial Alliance. The monitor is independent and the Karolinska Trial Alliance follows a specific monitor plan, which, for example, checks every patient consent. The participants in ERUTECC will received no extra benefits for being part of the study and there will be no compensation for those who claim that they have suffered harm from trial participation.

Anticipated problems.