Introduction
In The Netherlands general practitioners (GPs) have a crucial role as gatekeepers. Despite the specific complaint a patient has, GPs decide whether referral towards secondary (or more specialized tertiary) care is considered necessary or if the ailment can be controlled within primary care. Currently there is a dilemma present within the referral strategy for inflammatory rheumatic diseases (IRD). Since IRD is difficult to recognize by GPs, a lengthy referral delay leads to late diagnosis and ultimately decreased quality of life on the one hand [
1,
2]. On the other hand, promoting early referral leads to a high percentage of inadequate referrals [
3,
4].
One of the most potent ways to increase appropriateness of referrals within the increasing number of patients with musculoskeletal complaints [
5] is a structured referral sheet (e.g. algorithm) [
6]. By using structured referral sheets, the appropriateness of referral improves significantly, alongside it showing good potential to improve cost-effectiveness. For IRD several validated referral sheets have been developed, like the CaFaSpA for axial spondyloarthritis [
7], the PEST for psoriatic arthritis [
8] and the CARE for early arthritis [
9]. Recently, we have validated that the CARE is suitable to be used for all forms of IRD [
10], and we also published a composite algorithm for all forms of IRD [
11]. The actual implementation of such a (digital) healthcare innovation is however a complex process [
12] and needs preceding research. The main constraint to an optimal implementation process is lack of patient and provider involvement [
13‐
19]. To our knowledge, both patient and provider involvement are lacking so far when considering implementation of a digital referral algorithm to identify IRD patients. For successful implementation, it is essential to identify the barriers and facilitators that might affect the adaptability and implementation [
20,
21], however these are considered not to be influenceable. Therefore it is also important to identify factors that can be actively addressed to facilitate implementation, e.g. implementation factors. Feedback and clarity from patients and healthcare providers on awareness, support base and usability are expected to facilitate the development and implementation of a valid strategy [
22].
Qualitative research has been proven an effective methodology to draw upon attitudes, experiences and reactions from end users for successful implementation [
23]. A constructivist grounded theory approach to qualitative research is proficient to understand both patients’ and caregivers’ perspectives on our referral strategy. Focus groups are considered a well suited method for the purpose of understanding the perspective of patients with musculoskeletal complaints because of the dynamic interaction within the group leading to fuller and deeper discussion and triggering of new ideas [
24,
25]. A survey is considered useful and feasible to understand the caregivers’ perspective, since they are often hard to engage due to their scarce time [
26].
Therefore, in this qualitative study, we aimed to explore whether there is a support base for a digital IRD referral algorithm among patients, general practitioners and rheumatologists. Subsequently we investigated the barriers and facilitators from a patient and a caregiver perspective regarding awareness, usability, accessibility and acceptability of the referral algorithm.
Discussion
In this qualitative study we established that there is indeed a need for a digital referral algorithm indicating the risk of IRD and need for referral. This is supported by patients, GPs and rheumatologists, perceiving it as a helpful innovation in order to increase the appropriateness of referrals to the rheumatologist. The fact that the added value is seen by and for all included stakeholders indicates that a support base is present which can be a facilitator in the adoption phase of this healthcare innovation. In fact, perceived benefit of an innovation by the stakeholders involved is considered to be the most powerful facilitator [
35]. On top of that, from our results it was shown that all stakeholders felt a need for change resonating with the proposed strategy, which is also a known important facilitator [
36].
Both patients and caregivers labeled the narrowness of the outcome of the proposed algorithm as a major barrier. Important information for the patients and GPs is missing in the current outcome, especially when the outcome of the algorithm is that there is no risk of an IRD. To increase usability, the algorithm should be able to advice on possible treatment options in primary care when providing a negative referral advice. For many non-inflammatory joint complaints, elaborate guidelines for management within primary care already exist that can be incorporated into the outcome of our referral algorithm [
37,
38].
A remarkable finding is the discrepancy between expectations of patients and rheumatologists at some points. For example, participating rheumatologists claimed to know for sure that all patients with musculoskeletal complaints always want to have a face-to-face consultation with a rheumatologist, no matter the complaint. This was mentioned as one of the main barriers by the one third of rheumatologists who did not support the idea of implementation. On the contrary, patients state they just want to receive the right care at the right place, no matter what the right place is. On top of that, we have proven that quality of life significantly increases if non-IRD patients are being withheld from unnecessary outpatient care [
39]. This highlights the importance of organising healthcare towards more personalised healthcare [
40]. Patients within the present study are ready for this change, and very willing to take more control over their own patient journey. Next to patients, also GPs acknowledge the added value of shared decision making. Shared decision making is increasingly advocated as an ideal model of decision making during the medical encounter, as it has shown to improve patients outcomes and increase benefits for both clinicians and the healthcare system [
41,
42]. Therefore, to convince the hesitant one third of rheumatologists, personalised healthcare and shared decision making should be given a more prominent role within healthcare policies.
Our initial idea was to implement this algorithm for patients to use at home. The qualitative analysis indicated that this approach should be reconsidered. GPs themselves admit to have doubts on the necessity of referral in about a quarter of all patients with musculoskeletal complaints. Therefore use of this referral algorithm by GPs might be convenient in supporting GPs in making a decision on referral.
Previously, referral tools for use by patients have been considered effective in increasing appropriateness of referrals, for example in patients with axial spondyloarthritis, one of the most common forms of IRD [
43]. However, it has been noted that referral tools for use by patients should be applied supplemental to a referral tool for use by the GP. The referral algorithm for use by the GP should typically be implemented first [
43]. In order to do so, it is essential that the algorithm is incorporated into GP information systems already in use in daily primary care practice to increase awareness and accessibility. Participating GPs often offer that it should be integrated into their GP information system.
The main strength of this study is the high number of responses gathered on the caregivers’ surveys. Since qualitative research is a method with high information richness of the data, mostly 75 responses to open ended surveys per group are sufficient to provide reliable results [
44], where we collected over 100 responses per group. Especially noteworthy is the fact that almost 40% of all Dutch rheumatologists participated in our survey. Rheumatologists and GPs were selected completely at random, without any prior selection on age, seniority and affinity with digital innovations. Therefore, this gives a good reflection of the overall field of caregivers dealing with patients with musculoskeletal complaints, the population in which our algorithm is ought to be used.
Next to the caregivers’ perspective, we were also able to include a broad patient perspective. It is agreed upon that patient participation within healthcare is beginning to play a more important role [
45]. A review of the literature reveals that participation of patients with musculoskeletal complaints in healthcare is associated with improved treatment outcomes and higher rating of the quality of care [
45,
46]. Since patients were included with no prior selection on age, diagnosis and affinity with digital innovation, they are a good representation of the overall population of people with musculoskeletal complaints.
The qualitative methodology allows for several other strengths. The designed open-ended questions from both qualitative research methods were specific enough to yield coherent responses across respondents, yet broad enough to invite a spectrum of answers [
47]. The combination of open and closed questions in the surveys allowed for efficient collection of valid data within closed questions, with the possibility of immediately reassuring that all relevant issues have been covered using the open ended questions [
48]. The chosen questions within the survey were focussed on implementing our algorithm for patients to use at home, whereas in hindsight it may have been more useful to include questions on use of this algorithm by caregivers.
The iterative design of this study facilitated a more comprehensive insight in the perspectives of the stakeholders with regard to the proposed algorithm. Through verification of stakeholder claims with other stakeholders, validity and data saturation were enhanced. The rigorous thematic analysis approach has allowed to produce trustworthy and insightful findings [
49]. Transparency of the results is ensured by providing direct quotes from participants.
What might be considered a limitation, is that two of the three focus group discussions had to take place online. Online focus groups might allow for less interaction and less perception of non-verbal communication [
50,
51]. However, recently it is shown that online focus groups also have an interactive character and bear a strong resemblance to offline focus groups [
52]. Since online focus groups are not bound by geographical location, it allows patients to participate from the comfort of their own location of preference, which may lower the threshold for patients with musculoskeletal complaints to participate. Online focus groups are well suited to gather a first reaction of respondents to a new healthcare innovation [
53]. We believe that the weaknesses and benefits of online focus group discussions outweigh each other and the fact that no differences were found between the different focus group settings in the present study underlines that.
Another limitation might be that the results from our study are only transferable to settings with a similar healthcare system; in which primary care consultants act as gatekeepers to refer to secondary care. Since practice patterns for musculoskeletal and rheumatic diseases are distinct from other specialists [
19], it is questionable whether the current results are directly transferable to other fields of healthcare. Lastly, the fact that researchers study a tool they have developed themselves may also be considered a limitation. However, during the focus groups one of the moderators present had not been involved during development of the algorithm, and is therefore considered to be independent.
Based on this study, we have some recommendations for clinical implication. Since the analysed digital algorithm is considered not suitable for patients to use at home, the focus shifts towards sufficiently supporting GPs in the decision-making on referral for patients with musculoskeletal complaints. Conveniently, the content of the algorithm as presented in this study is appropriate for use by primary care clinicians. The outcome of the algorithm however needs to be expanded with information on treatment options for non-inflammatory complaints within primary care. To do so we recommend to incorporate information already available, for example at the Dutch College of General Practitioners [
38]. When these recommendations can be achieved, we encourage rheumatologists to place greater trust in the GP making the decision not to refer when sufficiently supported by the digital advice.
Another clinical implication is that IT facilities have to become supporting in realising sustainable, effective and patient centred healthcare at the right time and the right place. Until now, the multitude of systems that is available in primary care prevents working with a digital algorithm in real time. In order to modernize the systems towards patient-centred care, a sense of urgency must rise within primary care consultants to make changes happen [
54]. Within this study, we have shown that GPs underline the need for change, and are very willing to work with such an innovation on the condition that it can be integrated within their current systems.
While the proposed algorithm and its included referral sheets have been validated [
10], future research may be conducted focussing on validation of this referral algorithm in an independent, unselected primary care population, when in fact used by general practitioners. Within this validation, both sensitivity and specificity of this algorithm when used by GPs need to be established. Next to that, the cost-effectiveness of using the proposed referral algorithm should be investigated and compared to current usual care.
Concluding, patients, rheumatologists and general practitioners express the need for a distinctive referral algorithm indicating the risk of IRD in patients with musculoskeletal complaints. Our algorithm is expected to increase appropriateness of rheumatology referrals when aimed at use within primary care and sufficiently validated. To ensure successful implementation the barriers and facilitators for implementation need to be considered and identified from the perspective of patients and caregivers as in the present study. The majority of identified barriers and facilitators was overlapping between stakeholders. The current findings can reliably inform further decision strategies for implementation of the digital referral algorithm for inflammatory rheumatic diseases.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.