Recruiting patients with persistent physical symptoms and no specific diagnosis for clinical trials in primary care: a qualitative study of four scenarios

Abstract

Aim:

To investigate how to recruit patients rigosrously in general practice for trials based on symptoms as predefined inclusion criteria rather than a well-defined diagnosis. We used persistent physical symptoms (PPS) as a target condition.

Background:

Conducting randomized controlled trials (RCTs) in general practice is crucial for evidence-based decision-making and treatment in frontline healthcare. However, recruitment is often challenged in general practice because many patients have vaguely defined conditions, such as PPS.

Methods:

We used qualitative semi-structured interviews to explore the perspectives of general practitioners (GPs) on different recruitment scenarios, while paying particular attention to the logistics and feasibility in routine care. A total of 11 GPs from seven clinics were strategically recruited. The GPs were presented with one recruitment scenario (pre-consultation screening) and encouraged to suggest other scenarios. We conducted a thematic framework analysis of the interview material.

Findings:

To ensure valid intervention research, one predefined scenario was discussed, and three additional scenarios were co-created during the interviews: (1) opportunistic screening, (2) pre-consultation screening, (3) audit (retrospective and prospective), and (4) random sample screening of the practice population. These scenarios differ with respect to logistical complexity, comparability to daily clinical practice, and entailed selection bias. Every scenario requires individual adaptation and implementation support to be feasible in routine care in general practice. The present study provides approaches for rigorous primary care research based on RCTs for frequent and vaguely defined conditions, but it also highlights a need to develop research methods better suited to frontline healthcare interventions.

Introduction

Randomization is the state-of-the-art method in a clinical trial when investigating the effects of a new clinical intervention. Hence, conducting randomized controlled trials (RCTs) in general practice is crucial to establish evidence-based clinical decision-making and treatment in frontline health care. However, recruiting general practitioners (GPs) and general practice patients for such trials is often challenging (Hetherton, 2004; Mason et al., 2007; Johnston et al., 2010; Ngune et al., 2012; Foster et al., 2015; Riordan et al., 2021). A recent systematic review on recruitment to RCTs in general practice identified a wide range of factors associated with recruitment rates (Moffat et al., 2023). Such factors relate to patients, practices, and practitioners, and they are characterized by a high degree of complexity with many potential factors to consider and mitigate, including demographic, medical, practical factors, beliefs, and attitudes toward trial organization. The review suggests that clinical relevance is crucial. In line with other studies (Bower et al., 2014; Andersen and Vedsted, 2015; Foster et al., 2015), it highlights competing demands and time pressure as notable barriers. Another review reporting on 34 general practice RCTs in the UK suggested that only 30% of trials reach recruitment targets within the planned time frame (Bower et al., 2014). These rates increase when researchers (not GPs) are responsible for securing patient consent (Bower et al., 2014). Notably, when considering patients with persistent physical symptoms (PPS) often without a specific diagnosis, the task of recruitment becomes even more daunting, which may delay RCTs or even cause discontinuation (van der Gaag et al., 2017; Schreijenberg et al., 2018; Tan et al., 2019; Doe et al., 2022; Elzinga et al., 2023).

Symptoms that cannot be attributed to a specific diagnosis constitute a large proportion of patient contacts in general practice. A large cross-sectional Danish study of consecutive face-to-face contacts found that 27% of adult patients visiting their GP received no diagnosis; i.e. they had symptoms only (Rosendal et al., 2015). Efforts of early intervention may include PPS (not functional disorder) or conditions such as alcohol consumption (not alcohol abuse), stress, worry (not anxiety disorder), and affected mood (not depression). Research within these areas is needed to enhance the evidence base for primary health care and to identify opportunities for early intervention and prevention.

Besides the distinct challenges of identifying the target population, general practice also faces barriers related to the rigorous logistics of research projects (Hetherton, 2004; Mason et al., 2007; Goodyear-Smith et al., 2009; Page et al., 2011; Fletcher et al., 2012; Foster et al., 2015) and the individual organization of practices (Goodyear-Smith et al., 2009; Foster et al., 2015). Therefore, GPs are well positioned to offer guidance on how to design RCTs targeting general practice, including how to identify and recruit less distinct patient groups (Hetherton, 2004; Fletcher et al., 2012; Riis et al., 2016).

The present study aimed to investigate how to ensure rigorous patient recruitment (based on symptom assessment as predefined inclusion criteria rather than a well-defined diagnosis) to a trial in general practice. The challenge is to strike a balance between scientific inclusion standards and feasibility in general practice. We wanted to explore GP perspectives on different recruitment scenarios with a special focus on logistics and feasibility in routine care. PPS was used as a target condition for vaguely defined conditions to explore patient recruitment to RCTs in general practice.

PPS represents a broad spectrum of severity ranging from mild and self-limiting symptoms to persistent conditions with severe functional impairment, such as functional disorders (Löwe et al., 2024). No strict demarcation of categories exists, but the risk of symptom persistence increases with the number of symptoms (Rosendal et al., 2017). Up to 17% of adult patients attending general practice fulfil the criteria for PPS (Budtz-Lilly et al., 2015b). Thus, patients with PPS are a heterogeneous group, and it can be difficult to establish whether PPS is caused by a physical disease or other reasons. A rigorous evaluation of treatment interventions requires attention to two key factors: (1) choice of intervention and (2) consecutive and uniform enrolment. No unique diagnosis exists for PPS (Schaefert et al., 2010). Therefore, primary care studies based on registry data may end up applying proxy data (such as frequent attendance) since public registers and medical records are based on diagnoses or use of healthcare services. Consequently, the true target population cannot be included using this method. A previous study found substantial variations in the subjective GP assessment of PPS (Rosendal et al., 2003). This calls for more knowledge on how to conduct consecutive, criteria-based patient enrolment.

Methods

Study setting

General practice forms the foundation of primary healthcare in Denmark. Most Danish GPs are self-employed but work under a collective agreement with the public healthcare system, ensuring free access to GP services for 98% of all Danish citizens. GPs are central to delivering public health initiatives, preventive care, chronic care, and diagnostic assessment of symptoms. Each GP is responsible for a patient list of app. 1,600 individuals, serving as both their primary healthcare provider and the gatekeeper for access to specialist and hospital care. Most patient issues are managed within general practice, with only a small proportion of cases being referred to secondary care.

Recruitment and participant characteristics

We conducted a qualitative study based on interviews with GPs working in seven different clinics. All GPs were recruited from three Danish regions: North Denmark Region, Central Denmark Region, and Region of Southern Denmark. A total of 11 GPs participated in the study: There was one GP from each of three clinics, i.e. 3 GP, and two GPs from each of four clinics, i.e. 8 GPs. GPs working in the same clinic were interviewed together. The GPs were recruited through the professional network of one of the authors (MR) from 15 December 2020 to 7 July 2021, and they were purposively selected based on their interest in interventions and the development of primary care. Variation was sought in relation to gender, age, clinic type, seniority, and geographical location. The participants had 2–29 years of experience as a GP, and six were women. Three of the clinics were based in cities (urban area), while four clinics were based in small towns (rural area). All GPs owned their general practice clinic (partnership practice or solo practice being part of a formal practice partnership). Two of the GPs were also senior researchers in primary care.

Interviews

The qualitative interviews each lasted for approximately one hour and took place in the clinic of the individual GPs, except for one interview that took place online. The interviews were semi-structured (Kvale, 2015), and the interview guide was inspired by insights from the existing literature on recruitment for RCTs in general practice (see Box 1). The interviews were conducted by MR (GP and researcher) and JHA (anthropologist). The interviewers introduced the case project, i.e. a self-help programme for patients with PPS (Rask et al., 2021), outlined the conceptualization of PPS, and presented ‘pre-consultation screening’ as a possible scenario to set the scene for discussion (see Figure 1 and Box 2). In this scenario, all patients aged 21–65 years would be screened immediately before their consultation in accordance with the inclusion criteria. The three-step screening process involved the 25-item BDS checklist, which has previously been validated in general practice (Budtz-Lilly et al., 2015a) (see Appendix 1). Second, the screening comprised questions about functioning and sick leave. Third, the GP evaluated whether any exclusion criteria applied, or the patient could be included in the intervention. The interviewers stressed that all comments on this scenario were welcome and that the intention was to discuss if (and how) this procedure could be conducted in a feasible manner. This information was followed by open questions about other possible scenarios for the case project as seen from the GPs’ perspectives. Interviews and analyses were done in an iterative process; emerging themes and new scenarios described by the GPs were concurrently incorporated in the interview guide (and presented in subsequent interviews).

Box 1.

Themes in the interview guide

• Background information about the GPs and the organization of their clinic

• Brief description of scenario 1 + inclusion criteria (presented by interviewers)

• Patients with PPS

  • Conceptualizations of patients with PPS
  • Discussion of GPs’ own patient cases
• Recruitment scenario 1: Screening before GP consultation

  • Description of the pre-consultation recruitment scenario (presented by interviewers)
  • How could this scenario work in the clinic (in terms of approaching the patients, initiating the intervention, manpower, financial restraints, and time resources)?
  • Other comments from the GPs to scenario 1?

• Which other opportunities exist for recruitment of patients with PPS according to GPs?

Figure 1.

Proposal for patient recruitment in a pre-consultation screening scenario in case of persistent physical symptoms.

Box 2.

Criteria for patient recruitment in case project

The aim of the case project was to deliver high-standard PPS treatment in general practice to prevent patients from developing chronic symptom disorders.

Inclusion criteria

• Listed with participating practice

• Age 21–65 years

• Giving informed consent to participate in the study

• Consulting a GP or clinic staff (supervised by GP)

• Multiple symptoms, i.e. four or more symptoms within the past four weeks according to the validated 25-item BDS checklist (1).

Exclusion criteria (assessed by the GP)

• Not consulting for symptoms

• Symptoms explained by previously diagnosed physical disease or mental disorder (including addiction/substance abuse)

• Acute severe disease

• Not eligible due to language deficiency or cognitive problems

• Other (stated by GP), for example:

  • Recent onset of all symptoms, and I see no need for intervention at this time.
  • I believe that the patient will not benefit from participating in the case project at this time.

• GP recommends the case project, but the patient declines.

• Ongoing diagnostic assessment is not a reason for exclusion.

Intervention in case project

• A GP-prescribed, internet-based programme for patients with PPS, named ‘My Symptoms’ https://www.mine-symptomer.dk/My_Symptoms.

• The programme is based on self-help, i.e. not supported.

• The programme is inspired by cognitive behavioural therapy.

• The programme provides psychoeducation about PPS by offering modules on the impact on lifestyle, stress and strain, thoughts, feelings, values, and self-care.

GPs were remunerated (one hour) for their participation in the interview in accordance with the national remuneration agreement for GPs in Denmark.

Analysis

Interviews were audio-recorded, transcribed, and analysed through the use of framework analysis (Ritchie, 1994; Gale et al., 2013). The coding and analysis were primarily done by JHA, but each step was discussed with the co-authors. The first step in framework analysis is familiarization. In this step, JHA listened to audio recordings, read the entire material, and listed key ideas and recurring themes. The co-authors read a selected number of interviews, and possible themes were discussed in the author group. Important themes were included in a thematic framework, allowing the material to be sorted and organized. For this purpose, several codes were developed for each theme to designate different issues under this theme. These codes were then applied to a few transcripts and refined. The entire material was coded in NVivo 12 Pro, thereby indexing how different parts of the transcripts belonged to different issues and themes. Some themes and codes were predefined, e.g. the theme ‘recruitment scenario 1’ and the associated code ‘who does the screening’, while others emerged from the material, e.g. the theme ‘how do GPs conceptualize and handle patients with PPS’ and the associated code ‘when do GPs start to think that symptoms might not be related to physical disease’. Framework analysis allows for this combination of inductive and deductive analysis (Ritchie, 1994; Gale et al., 2013). After having coded the entire material, JHA made charts for each theme and summarized participant statements on each theme. MR reviewed the charts. The final analysis explored the content of each theme, both within and across the interviews (Ritchie, 1994). Two overarching themes were identified: conceptualization of patients with PPS and recruitment scenarios.

Key quotes were translated from Danish into English by two translators individually (one with clinical expertise and one with linguistic expertise), and a final consensus version was made.

Results

Theme 1: Conceptualization of patients with PPS in the intervention

During the interviews, we informed the GPs about the target patient population of the case project. We also presented the inclusion and exclusion criteria. This ignited a discussion about their understanding of patients with PPS and the circumstances required for offering the intervention. As also demonstrated in a previous qualitative study (Rask et al., 2021), the understandings of PPS varied according to the spectrum of symptom severity. Some GPs found that a large number of patients present symptoms with no biomedical explanation, whereas others would confine PPS to severely affected individuals only, i.e. functional disorders. One GP described these patients with terms like ‘hiking trophies’, ‘regular guests’, ‘problem patients’, and ‘heart-sink patients’. She did not know of any patients with PPS who were still relatively well-functioning. Another GP knew of well-functioning patients with PPS, but these patients were likely to go unrecognized:

I think that this is a new observation for general practice (…). Those staying under the radar are the almost well-functioning [patients], who just keep on returning again and again and are somewhat health anxious and feel themselves a little too much. (GP G)

Several GPs mentioned that it took some time before being assured that PPS in a patient were not signs of an underlying disease. One GP said that she would often have done multiple consultations before suspecting PPS. Others explained that − before voicing any hypotheses about PPS to the patients − they had to be quite sure that the patient did not have a physical disease:

Often you need to just sweep in front of the somatic door before excluding a psychiatric or functional diagnosis. You must be quite sure that you haven’t missed anything because they show up with something that might raise a slightly red flag. But then along the way, I think you can begin to explain that it’s your nervous system that is overloaded, it is your body that makes noise, but you have been carefully examined. (GP G)

In line with this, the GPs said that they would usually conduct a medical assessment of the patient (possibly involving specialized care) before recommending an intervention addressing PPS. They chose this approach for two reasons. First, the GP had to be sure that a physical disease was not present. Second, the GP wanted to maintain a good relationship with the patient. Some GPs explained that conducting physical examinations could also serve to make the patient feel taken seriously.

Theme 2: Recruitment scenarios

The recruitment scenarios discussed in the interviews varied in terms of their degree of integration into routine care in general practice. The same inclusion and exclusion criteria for PPS were used for all scenarios, but the role of the GP and the logistics differed. Detailed descriptions of the relevant logistics can be found in Table 1. In the following, we present four scenarios, with the first being most comparable to routine care.

Table 1.

Details on recruitment scenarios informed by case project

		Administrative work in the clinic	In the clinic before consultation	In the consultation*	Subsequent consultations	Research team	Pros	Cons
Scenario 1: Opportunistic recruitment (numbers 1–3 refer to the order of events)
	a)	No	No	1) The GP assesses patient eligibility according to predefined inclusion and exclusion criteria. The GP briefly introduces the project to relevant patients. The GP hands out written information and informed consent. The GP initiates the intervention.	No	2) Researchers may subsequently exclude patients who do not fulfil the inclusion criteria based on Patient questionnaires (PQs) or telephone interviews.	Routine clinical practice is almost undisturbed, and the procedure is easy to integrate into daily practice.	Inconsistent patient enrolment between different GPs and consequently between randomized groups. High risk of selection bias.
	b)	No	No	1) The GP assesses patient eligibility according to predefined inclusion and exclusion criteria. The GP briefly introduces the project to relevant patients and refers them to the research team.	3) Yes. If the researchers assess that the patient can participate, a subsequent consultation is booked with the GP, who initiates the intervention.	2) The research team will screen patients with regard to the inclusion criteria using PQs or telephone interviews.	Routine clinical practice is almost undisturbed, and the procedure is easy to integrate into daily practice.	Inconsistent patient enrolment between different GPs and consequently between randomized groups. High risk of selection bias.
Scenario 2: Screening before consultation (numbers 1–5 refer to the order of events)
	a)	1) Research assistants located in the practice will mark patients in the target age group (21–65 years in the present case) in the time schedule of the clinic. The day before an appointment, they will call the patients and briefly inform about the project. If patients accept further information and possible participation, they will be asked to arrive 10 minutes early in the clinic. An alternative option is to screen the patients by telephone interview or offer the patient to answer the PQs online via a link sent by email. If the appointment is scheduled on the same day, the patient will be asked to arrive 10 minutes early when the booking is made.	2) There will be a notice at the arrival point asking patients to go to the front desk, where the research assistants provide oral and written information about the project. If patients accept participation, they will answer a brief questionnaire and provide informed consent on a tablet or on paper. PQs will be decoded by the assistants. If fulfilling the inclusion criteria, the patient is given a notification to hand over to the GP.	3) For patients presenting the notification, the GP assesses whether any of the exclusion criteria apply. If not, the intervention can be initiated.	4) Possibly. The GP may not have time to introduce the intervention in the present consultation.	Supervision of assistants in clinics.	The recruitment of patients is consecutive and consistent across randomized groups. Low risk of selection bias.	High workload and interference with routine clinical practice. May result in inopportune timing of intervention.
	b)	1) Administrative work as described in a), except that patients fulfilling the inclusion criteria are not revealed to the GP. 4) Instead, results from patient questionnaires are compared to GP registrations of exclusion criteria on all patients after clinical hours. Eligible patients are contacted by the assistants by phone or email from the clinic and offered an extra consultation with their GP.	2) As in a). However, the PQ will be returned to the assistants in a closed envelope and decoded after clinical hours.	3) For all patients in the target age group, the GP assesses whether any of the exclusion criteria apply.	5) Yes. A supplementary consultation is offered to eligible patients by phone or email. The GP re-assesses whether any of the exclusion criteria apply. If not, the intervention is initiated.	Supervision of assistants in practices	The recruitment of patients is consecutive and consistent across randomized groups, but dropout rates may be higher due to the extra consultation. The GP’s contribution is easy to integrate into clinical practice than in a). Low to medium risk of selection bias.	High workload, but little disruption of routine clinical care compared to a)
Scenario 3: Audit – prospective or retrospective (numbers 1–5 refer to the order of events)
	a)	1) For each participating GP in the clinic, one staff member may transfer name and birthday from the present-day time schedule to an audit questionnaire for patients in the target age group (21–65 years in the present case). The GP will be able to enter additional patients, e.g. in case of acute patients. 4) Staff in the clinic will contact patients to make an extra appointment with their GP.	No	2) The GP will prospectively register one line of information, including exclusion criteria, for all patients in the target group seen in face-to-face consultations. Patients visiting due to health problems will be briefly informed about the project by the GP. Patients will be given written information and a link to the project website to fill in the PQ online at home. Alternatively, they may be given a phone number to a research assistant, who can help them fill in the PQ via a telephone interview.	5) Yes. A supplementary consultation is offered to eligible patients by phone or email.	3) The research staff will decode the PQ for inclusion criteria and the GPs’ audit registrations for exclusion criteria and inform clinics twice a week about patients who can be enrolled.	The prospective registration of all patients provides consecutive and consistent patient recruitment across randomized groups. Routine clinical practice remains relatively undisturbed.	Patients may choose not to answer the PQ subsequently. Medium risk of selection bias.
	b)	1) The GP reviews the appointments of the past weeks to identify target patients (21–65 years in the present case) who have consulted with symptoms. Identified patients are registered, and the GP looks up the patient’s electronic medical record to explore whether any exclusion criteria apply. Potential eligible patients are contacted by the assistants by phone or email from the clinic and are informed about the project. If the patients accept participation, they are asked to answer the PQ online or by telephone interview. 3) Clinic staff will contact patients to make an extra appointment with their GP.	No	No	4) Yes. Patients fulfilling the inclusion criteria will be offered a supplementary consultation by phone or email. The GP re-assesses whether any of the exclusion criteria apply. If not, the intervention is initiated.	2) The research staff will decode the PQ for inclusion criteria and inform clinics twice a week about patients who can be enrolled.	Routine clinical practice is undisturbed.	The retrospective registration increases the risk of GP-induced selection bias, and patients may choose not to answer the PQ. The GPs must spend administrative time on the audit process. Medium to high risk of selection bias.
Scenario 4: Population screening (numbers 1–3 refer to the order of events)
		2) The research team sends a list of potentially eligible patients to the GPs for assessment of exclusion criteria according to existing information in their electronic medical records. Eligible patients will be approached by their GP by email and will be asked whether they would be interested in participating in the project.	No	No	3) Yes. Patients accepting participation will be offered a supplementary consultation by phone or email. The GP re-assesses whether any of the exclusion criteria apply. If not, the intervention is initiated.	1) The research team selects a random sample of patients registered with the clinics and in the target age group (21–65 years in the present case). Patients are contacted by letter or email. Patients will receive written information and a link to the project website to fill in the PQ online at home. Alternatively, they may be given a phone number to a research assistant, who can help them fill in the PQ via a telephone interview. PQs are decoded, and a list of patients fulfilling the inclusion criteria is generated and sent to participating clinics.	The recruitment of patients is consecutive and consistent across randomized groups. However, there is a high risk that potential eligible patients may refuse participation. Low to medium risk of selection bias.	Contact to patients may be clinically irrelevant. Recruitment and intervention occur outside of the clinical context.

* GP may be replaced by a nurse under GP supervision (in this column).

PQ: Patient questionnaire, including the BDS checklist and questions on functioning and sick leave.

Scenario 1: Opportunistic recruitment

Opportunistic recruitment requires the GP to include patients during the consultation in accordance with predefined criteria (Box 2). This scenario relies exclusively on the GP’s interpretation and application of the criteria, even if fulfilment is subsequently checked by a researcher (Table 1, scenario 1b).

GP D argued that this scenario was preferable as it would be easy to implement in daily practice; it would depart from the symptoms that patients want to discuss and in the GP’s assessment of intervention relevance. In this way, the intervention will be delivered with perfect timing. However, two GPs from another clinic were sceptical about it. They argued that this type of recruitment requires GPs to know their patients very well, and this could complicate recruitment for younger doctors. Moreover, on a busy day, the assessment of eligibility for enrolment may easily be skipped:

I believe that if you must do it, [you need to] consider it on the day when they [the patients] visit. But if you are busy, there are clearly some [cases] where you think, ‘ah, well’, and then you move on […] It needs to be some of the five [patients] that we, ehh, already suspect to have functional issues. (GP J)

The GPs’ conceptualizations of patients with PPS suggest that the problem with picking ‘the five’ is that these patients are likely to be severely affected by their symptoms. Thus, only a few patients from the target group will be included. Additionally, according to GP K, patients might not be recruited in this scenario due to time pressure, as most GPs do not inquire into multiple symptoms. Instead, they tend to focus on one symptom, which holds a risk of missing eligible patients.

Scenario 2: Pre-consultation screening

This scenario involves screening of the patients before their consultation with their GP, after which the GP will evaluate whether the patient may be recruited for the intervention. Hence, this scenario is positioned somewhere in between daily practice and an efficacy evaluation of the intervention alone, as routine care would be modified by the pre-consultation screening of all patients.

All participating GPs emphasized that they did not have sufficient manpower to conduct the initial screening; everyone, including the regular practice staff, was too busy doing their jobs. These time concerns are also related to obtaining consent and answering questions from patients about the project. However, most GPs would agree to have project assistants conduct the screening and answer the patients’ questions. Another barrier foreseen by some GPs was space limitations, as the screening needed to be carried out isolated from other patients in case personal issues came up. Not all clinics would be able to provide this extra space.

Based on the GPs’ concerns, the interviewers (MR and JHA) developed two different models for pre-consultation screening (see Table 1). These general recruitment models required adaptation to the needs of each participating clinic. In our study, the GPs came from very diverse clinics in terms of the number of GPs, staff, and patients. They also differed in which clinical tasks were carried out by nurses. These organizational factors are likely to influence recruitment logistics, e.g. how many research assistants are needed, and whether patients with a nurse appointment are possible candidates for enrolment. Despite its flexibility, some GPs perceived this scenario as complex and overwhelming:

It’s a highly relevant subject and all, but it [the recruitment procedure] seems very comprehensive, I must say. (GP B)

Besides the logistics involved in the pre-consultation screening, another issue was the possible effect of patient screening. All GPs said that answering the BDS checklist might raise additional questions and concerns in patients, which the GPs would then have to deal with. It might even lead to extra examinations:

When I consider these symptoms and while talking about them, then – because of the way we are oriented these days − if I took three to four of these symptoms, I would think ‘attention cancer’. (GP A)

In contrast, another GP explained that becoming aware of the number of symptoms on the checklist might help him think along important new lines rather than solely searching for a physical disease.

One GP was so critical of the screening approach that she would refuse to follow this scenario. She argued that screening patients might create new health concerns and thereby disturb the consultation:

Some of my patients…they don’t need to be reminded of all the conditions that you could possibly suffer from. They need to focus on what they actually experience in their body and in their life and then talk to the GP about it, and if they are presented with this [BDS checklist], then their mind will go somewhere else, and we don’t get the conversation we need. I believe it would harm the consultation process. (GP D)

Scenario 3: Audit

Recruitment through an audit is more systematic than pure opportunistic enrolment, but it requires the research team to decode patient questionnaires outside the clinical context. Audit scenarios in the present context entail booking an extra consultation for enrolled patients, which some GPs might find problematic due to their busy schedules. Audits may be conducted prospectively or retrospectively.

The prospective audit scenario came up in the last interview. For all patients in the relevant age group who were seen in a face-to-face consultation, the GP would prospectively register on a paper form whether the exclusion criteria applied. Patients visiting for health issues (including any symptom) who did not fulfil the exclusion criteria would receive written information with a link to the project website and be encouraged to fill in the BDS checklist online at home. Patients fulfilling all inclusion criteria would be asked to make a new appointment with the referring GP to initiate the intervention.

The retrospective audit requires the GPs to inspect the appointments in their schedule for the past week to check if any of their patients consulted for symptoms without fulfilling the exclusion criteria. Relevant patients would be contacted and informed about the project. If willing to participate, they would be asked to fill in the BDS checklist. If fulfilling the inclusion criteria, they would be invited for a new consultation and recruited for the intervention.

GP I found this to be an easier task than the prospective audit:

Intuitively, I think that it would actually be easier for me to look back than ahead. (GP I)

And GP J agreed:

Yes, because I also believe that if you know it’s simple, then you think ‘okay, the past two weeks, I can do that’. (GP J)

Likewise, GP K found this scenario to be operational and systematic.

However, the GPs also identified some challenges. GP K said that the scenario required dedicated GPs who were willing to spend time going through the appointment list. GPs J and I pointed out that GPs do not necessarily write everything in the medical records; sometimes, only the patient’s main reason for consulting is stated, whereas other symptoms presented are not noted and cannot be identified.

Scenario 4: Population screening

This scenario entails that project researchers send a screening questionnaire to a randomly selected part of the patient population in the relevant age group who are listed at a participating clinic. Hence, it is completely taken out of the clinical context. After the screening questionnaire, this scenario has three steps. First, the questionnaires will generate a list of patients fulfilling the inclusion criteria. Second, the GP must assess whether the exclusion criteria apply to patients on the list. Third, eligible patients will be contacted by their GP and asked whether they would be interested in participating in the project and (if so) to book an appointment with their GP to finalize their inclusion and initiate the intervention.

The GPs suggesting this scenario said that it might be too much work for some GPs to go through the patient lists. They also acknowledged that there might be a risk to include patients who had not thought of contacting their GP due to the symptoms. GPs G and H were not supportive of the scenario; they believed it would create a lot of extra work by stirring a need for consultations, which did not exist before the questionnaires were sent out:

This would induce some sort of need that might not have been there already […], and this would cause a lot of extra activities in the clinic. (GP H)

In line with GPs G and H, GP K found this scenario to be extremely remote from routine clinical practice because it holds a risk of including patients solely on the basis of their symptoms, although they may have no intention of seeking healthcare.

In another clinic, GPs I and J supported this point made by GPs G and H. However, GP J added that many of the patients to be offered a consultation would be patients attending relatively often anyway:

I just think that very many of the patients that you wish to find come here relatively often anyway. So, if they obtained a different understanding of how to handle their symptoms, then perhaps they wouldn’t need to come here so often. (GP J)

Thus, GP J believed that if the patients agreed to accept the intervention, they might learn to manage their symptoms better, and their need for GP consultations might thus decrease accordingly.

Discussion

Using the case of PPS, we explored GP perspectives on how to enable rigorous patient recruitment for an RCT in a general practice setting. We introduced one scenario (pre-consultation screening) in qualitative interviews, and the GPs suggested refinements that resulted in the co-creation of another three scenarios. The scenarios differed in logistical complexity, and all scenarios were found to require substantial individual adaptation and implementation support to be feasible in routine care.

Strengths and limitations

Several of the recruited GPs were part of MR’s professional network in primary care research. They might have been more open to research than the average GP and hence not representative. On the other hand, they had more expertise as clinician-researchers, which suggests a high degree of relevance and applicability of the study results. The broad representation of GP age, geographical location, and clinic organization ensured breadth and variety in the represented perspectives.

In terms of method triangulation, the interviews included open-ended questions and techniques to enhance discussions about rigorous research in general practice. In the open part of the interviews, the GPs were encouraged to articulate their own views and to voice both positive and negative perceptions of the scenario presented by the interviewers (MR and JHA). They were also encouraged to come up with suggestions for alternatives. Nevertheless, a more thorough discussion of the uncovered scenarios could have been obtained by arranging subsequent focus groups with a new group of GPs. However, the aim of the present study was to explore a variety of possibilities rather than obtain complete descriptions of each scenario.

Comparison with existing literature

To our knowledge, this is the first study to investigate GP views on the recruitment of patients with a vaguely defined condition, such as PPS, for an RCT. Previous studies on RCTs for patients with PPS have varied considerably with respect to inclusion criteria and recruitment strategies (Rosendal et al., 2013). Some studies combine screening of patients in the waiting room with GP assessment and diagnostic interviews to reduce the risk of selection bias; this approach is similar to our pre-consultation screening scenario. This mainly entails the risk of bias caused by overrepresentation of frequent attenders, but bias may also occur if a large number of patients decline participation (Larisch et al., 2004; Morriss et al., 2006; Morriss et al., 2007). In a study by Rief et al. (Rief et al., 2006), GPs were instructed to identify patients based on inclusion criteria and screening scales. This procedure was demanded by the ethics committee, but it resulted in a non-intended group of patients who were both severely affected by their symptoms and frequent attenders in the clinic. Our study shows that GPs often think of patients with PPS as severely affected frequent attenders. In line with this, Rask et al. have shown that it is difficult for GPs to identify patients with mild-to-moderate PPS (Rask et al., 2014). This is a significant limitation of what we have called opportunistic recruitment, and it may potentially undermine the evaluation of the intervention itself in an RCT.

More efficacy-oriented recruitment procedures are often complex, which can also be a barrier for GPs. Furthermore, some studies have shown that concerns about burdening patients with research projects can make GPs decline participation (Hetherton, 2004; Mason et al., 2007). In all our interviews, we found three central concerns: How much time did the GPs need to spend on the project? How much would it disturb the clinical routines? Would extra consultations be needed for intervention patients? Similar concerns have been reported in several studies on barriers to recruiting GPs and thereby their patients for RCTs; a complex and time-consuming design can make GPs decline participation in a research project, and those who agree to participate might fail to recruit patients (Hetherton, 2004; Goodyear-Smith et al., 2009; Page et al., 2011; Ngune et al., 2012; Fletcher et al., 2012; Reed et al., 2013; Foster et al., 2015). In our pre-consultation screening scenario, we wanted to relieve the GPs of the screening burden by involving research assistants, but it remained difficult to avoid disturbances in everyday routines. In this scenario, it becomes extremely important to adapt the procedures to the individual clinic. Audits and random sample screening are less complex. However, they require the GP to spend time on patient lists for recruitment after clinical working hours and subsequent time on extra consultations for relevant patients.

Implications

This study shows that researchers need to strike a difficult balance between scientific standards for recruitment (consecutive, uniform, and unbiased recruitment), practical feasibility in the clinic, and possible burdens placed on GPs and patients when selecting a recruitment procedure for trials aimed at vaguely defined conditions, such as PPS. Opportunistic screening has the advantage of being relatively simple, requiring little time, and being easily adaptable to routine clinical care. However, opportunistic screening holds a high risk of selection bias, with high inclusion variations induced by the GP through individual choices of inclusion/exclusion. Pre-consultation screening and audit scenarios are more labour intensive and logistically complex, but they involve more systematic recruitment procedures, which increase the chance of enrolling the targeted patients. The random sample of population screening is very systematic and not logistically complex for GPs but is conceptually very remote from clinical practice and holds a risk of recruiting patients who had not planned to consult their GP. Consequently, patients recruited in scenario 1 (opportunistic screening) and scenario 4 (population screening) might differ substantially from the original target population in need of early intervention.

An important concern, which was highlighted in the discussions of pre-consultation screening and random sample screening, is the possible negative effects that screening for inclusion criteria might have on patients in terms of inducing new health worries. If the screening is done just prior to a planned consultation, this consultation might be derailed from its original purpose. The strict criteria of randomized efficacy trials often conflict with the need to generate evidence about how new interventions perform in routine care settings, where GPs have the freedom to apply interventions to patients they consider suitable. In these real-world scenarios, numerous factors related to both practice and patient characteristics influence outcomes, making it difficult to attribute effects solely to the intervention itself. Our study highlights the need to develop or refine research methods that can effectively evaluate frontline healthcare interventions under such conditions. For example, the UK Medical Research Council has addressed this challenge by introducing guidelines that emphasize iterative research and process evaluation rather than focusing exclusively on quantitative measures of efficacy (Skivington et al., 2021).

Conclusion

We aimed to investigate rigorous patient recruitment in general practice for trials based on symptom assessment as predefined inclusion criteria rather than a well-defined diagnosis. Our study contributes new insight into how to produce valid research in general practice settings. Based on interviews with experienced GPs, the study provides an overview of scenarios for recruitment, including detailed descriptions and implied challenges of each scenario. Specifically, the study shows that RCTs may still be challenged even after adaptation to individual clinics and provision of substantial support (both financially and in manpower) when uniform labels, such as diagnoses, cannot easily be used to identify the targeted patients. Such challenges include declining GPs, inopportune timing of patient intervention, and poor transferability to routine care. If left unaddressed, this may leave a major part of general practice out of clinical intervention research. Therefore, we need to continue discussions and searches for the development of better ways to study interventions for conditions such as PPS in general practice.

Ethics

All participants received written and verbal information about the project and gave informed consent. The data are securely stored, and participants have been anonymized in the reporting data. The project was approved by the Danish Data Protection Agency (file no. 1-16-02-16-19).

Supporting information

Frumer et al. supplementary material

Frumer et al. supplementary material

Supplementary material

The supplementary material for this article can be found at https://doi.org/10.1017/S1463423625100583.

Data availability statement

The manuscript includes relevant data; however, to protect the privacy of study participants, the raw empirical data cannot be shared.

Authorship contribution statement (CRediT Roles)

Michal Frumer: Formal analysis, Writing – original draft and preparation, Writing – review and editing. Julie Høgsgaard Andersen: Conceptualization, Formal analysis, Investigation, Methodology, Project administration, Visualization, Writing – original draft and preparation, Writing – review and editing. Mette Trøllund Rask: Conceptualization, Formal analysis, Funding acquisition, Visualization, Writing – review and editing. Lisbeth Frostholm: Conceptualization, Formal analysis, Funding acquisition, Writing – review and editing. Marianne Rosendal: Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Visualization, Writing – original draft and preparation, Writing – review and editing.

Funding statement

The study was supported by grants from Innovation Fund Denmark and the Danish foundation TrygFonden. The funding sources had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests

The authors declare to have no conflicts of interest.