Primary care physician perspectives on screening for axial spondyloarthritis: A qualitative study

Background Many patients with axial spondylarthritis (axSpA) experience lengthy diagnostic delays upwards of 14 years. (5–14 years). Screening tools for axSpA have been proposed for use in primary care settings, but whether this approach could be implemented into busy primary care settings remains unknown. Objective To solicit feedback from primary care physicians regarding questions from the Inflammatory Back Pain Assessment: the Assessment of Spondyloarthritis International Society (ASAS) Expert Criteria and gain insight about barriers and facilitators for implementing axSpA screening in primary care. Methods Guided by Consolidated Criteria for reporting Qualitative Research (COREQ-criteria), we recorded, transcribed, and analyzed in-depth interviews with eight family medicine physicians and ten internists (purposeful sampling) using immersion/crystallization techniques. Results Few physicians reported awareness of existing classification criteria for axSpA, and many reported a lack of confidence in their ability to distinguish between inflammatory and mechanical back pain. From three domains, 10 subthemes emerged: 1) typical work-up of axSpA patients in primary care, with subthemes including the clues involved in work-up and role of clinical examinations for axSpA; 2) feedback on questions from the Inflammatory Back Pain Assessment: ASAS Expert Criteria, with subthemes to evaluate contents/questions of a potential screening tool for axSpA; and 3) implementation of the screening tool in primary care settings, with subthemes of perceived barriers including awareness, time, other conditions to screen, rare disease, and lack of structured questionnaire for back pain and perceived facilitators including workflow issues and awareness. Conclusions Primary care physicians believed that an improved screening instrument and a strong evidence-base to support the need for screening for axSpA are required. The implementation of axSpA screening into a busy primary care practice requires integration into the practice workflow, with use of technology suggested as a possible way to improve efficiency.


Introduction
Axial spondyloarthritis (axSpA) is an inflammatory disease characterized by chronic back pain [1]. Prevalence estimates vary widely [2]. In the US,~1% of the adult population is estimated to have axSpA [3], whereas in Canada the prevalence has been estimated to be 213 per 100,000 [4]. The diagnostic journey experienced by patients with axSpA is often lengthy, lasting upwards of 14 years in the United States [5][6][7], and delayed diagnosis can result in greater functional impairment, higher healthcare costs, and worse quality of life [8]. Early initiation of treatment can reduce symptoms [9], and may delay disease progression [9] and prevent disability [10]. Primary care providers have described reasons for delayed diagnosis of axSpA including disease characteristics (e.g., back pain is common, whereas axSpA is relatively uncommon); patient perception (e.g., sharing back pain at end of appointment); provider unfamiliarity (e.g., lack of awareness about axSpA); and healthcare system-related issues (e.g., brevity of primary care appointments) [11]. Although rheumatologists play a pivotal role in caring for patients with axSpA, most patients often consult with primary care providers (e.g., primary care physicians or internalist) seeking symptom relief when they initially have their symptoms (i.e., back pain or spinal pain) and are diagnosed by primary care providers other than rheumatologists [5]. Several sets of classification criteria exist for axSpA, including the modified New York criteria [12], the Berlin criteria [13], the Amor tool [14], the European Spondylarthropathy Study Group (ESSG) criteria [15], and the Assessment of Spondyloarthritis International Society (ASAS) criteria [16]. Yet, it is not known whether healthcare providers, other than those in rheumatology, are familiar with these criteria [17].
Several approaches to assist in identifying patients with axSpA have been explored in primary care settings, including the use of screening questionnaires [18,19], early referral tools that combine clinical criteria and laboratory and imaging test results [20], and automated referral algorithms using electronic medical record (EMR) data [21]. However, few studies [22,23] have engaged primary care providers to gather their opinions about screening tools for axSpA in clinical practice.
As part of a larger qualitative research study [SpodyloArthritis Screening and Early Detection (SpA-SED) Study], the primary objective of this study was to evaluate primary care physician perspectives and views about the use of a screening tool for axSpA. This analysis focuses on statements made by primary care physicians regarding their baseline practices on typical work-up of axSpA patients, feedback on questions from the Inflammatory Back Pain Assessment: the Assessment of Spondyloarthritis International Society (ASAS) Expert Criteria [19], and barriers and facilitators for implementation of axSpA screening in the primary care setting. The results of this study will provide foundational knowledge for the development and implementation of screening tools for axSpA.

Materials and methods
The University of Massachusetts Medical School Institutional Review Board approved this study. All participants provided written informed consent.

Study sample and setting
We recruited 18 primary care physicians to achieve saturation, with gender (8 women) and specialty balance (8 Family Medicine and 10 Internal Medicine) from January to May 2019. To be eligible, a physician must be a primary care physician in clinical practice and were able and willing to participate in an interview. Physicians were ineligible if they were: 1) unable to participate in a discussion lasting no more than 60 minutes; 2) unwilling to be audio recorded; 3) unable to consent; and 4) non-English-speaking subjects. Using purposeful sampling, we reached out to primary care physicians in Massachusetts and Rhode Island who were known to the authors and to regional professional societies (e.g., Rhode Island Academy of Family Physicians) to identify potential participants [24]. We identified 34 potential participants and completed 18 60-minute, audio-recorded interviews (3 in person, 15 on the phone, conducted by either KL, DS with SL or DS taking notes, no other people were present). No repeat interviews were conducted. One participant had conducted research (unrelated to axSpA) with the interviewer >5 years previously. Participants were compensated for their time with a $300 cash card.

Study protocol
A multidisciplinary team developed the interview protocol guided by the consolidated criteria for reporting of qualitative research (COREQ) [25] and informed by a literature review and insights from the research team, which included two rheumatologists (S1 Table). We collected the qualitative data using in-depth interviews with a semi-structured interview outline. Each interview was conducted on a one-to-one basis with an observer from the research team taking notes. We began the interview with physicians' baseline practices on typical work-up of axSpA patients. We then solicited feedback during a question-by-question review of the Inflammatory Back Pain Assessment: ASAS Expert Criteria to understand their perspectives on those questions to screen axSpA patients [19]. The interviews were then completed by asking questions about implementation of axSpA screening tools in primary care settings. To understand the characteristics of physician participants in the study, we lastly administered a structured questionnaire at the end of interviews. It included items about socio-demographics, practice characteristics, awareness of the features of inflammatory back pain according to several different sets of classification criteria (i.e., Calin [26], Berlin [27], and ASAS [19]) and "any other criteria" (open-ended), and the following question: "How confident are you in distinguishing inflammatory back pain from mechanical back pain?" (extremely, very, somewhat, or not confident). Interviewers were members from the research team who had experience with in-depth interview and trained in qualitative research as well as for this study protocol. We also conducted pilot testing of the interview protocol with a primary care physician not participating in the study and did not include the data in this analysis. Interviewers reported no explicit bias related to the topic under study but believed that primary care physicians may lack awareness about axSpA.

Qualitative data analysis
We recorded each in-depth interview and discussion with the participants. We used a content analysis approach to completely transcribe and code the data collected. Transcription was carried out verbatim. We did not ask participants to review the transcripts of their interviews for comment or correction. We developed a hierarchical coding structure and handbook based on review of the first few transcripts and expanded throughout data analysis. Two research assistants performed line-by-line coding of the transcripts with NViVo qualitative software. Qualitative review of the dual coding confirmed that passages often appeared twice, suggestive of effective coding. The data were analyzed as a group using immersion / crystallization techniques [28]. We requested feedback from participants wiling to review results. This manuscript focuses on three themes in our coding structure: 1) baseline practices on typical work-up of axSpA patients; 2) feedback on the Inflammatory Back Pain Assessment: ASAS Expert Criteria questions; and 3) perspectives and views about implementation of axSpA screening tools in primary care settings.

Results
On average, the study participants had been practicing in the field for a mean of 15.7 (±13.0) years (Standard Deviation) ( Table 1). Most physicians were Non-Hispanic White, trained at US allopathic schools, and were affiliated with academic institutions. No family medicine physicians and only 30% of internal medicine physicians had heard of the ASAS classification criteria. No internists and 2 of 8 family medicine doctors reported feeling "extremely confident" and 3 of 8 family medicine and 3 of 10 of internists reported that they felt "very confident" in distinguishing inflammatory back pain from mechanical back pain.
In line with the interview protocols, 3 main themes along with 10 subthemes were established (Table 2).

Typical baseline work-up of axSpA patients
When asking of the typical baseline work-up of axSpA patients (Table 3), physicians noted several clinical observations potentially pointing to axSpA, including: 1) young people presenting with back symptoms without any antecedent injury; 2) the presence of comorbid autoimmune conditions; 3) peripheral joint involvement, and other systemic manifestations (e.g., iritis). When speaking of the role of clinical examinations and tests, most physicians stressed that a thorough medical history would be essential for diagnosing axSpA. Some also expressed concern about having dissatisfied patients when costly tests are performed that may not yield definitive results.

Feedback on questions from the Inflammatory Back Pain Assessment: ASAS Expert Criteria
With respect to questions in ASAS Expert Criteria to screen patients with AxSpA (Table 4), physicians were concerned that some questions were not specific (e.g., Have you suffered from back pain for more than 3 months?) or sensitive (e.g., Did your back pain develop gradually?). They were also concerned that patients with intermittent back pain might be missed. Physicians suggested including questions about decreased range of motion or stiffness, heel pain and other symptoms of enthesitis, psoriasis, and Crohn's disease. Most did not like the  suggestion to immediately refer patients with 4 or 5 affirmative answers to a rheumatologist because of concerns about the shortage of rheumatologists; however, they thought that it would be helpful for specific tests to be suggested as next steps to pursue in evaluating these patients.

Implementation of the screening tool in primary care settings
Physicians considered axSpA to be "uncommon" and perceived this as a barrier to implementation of a screening tool (Table 5). They suggested using such an instrument only in specific clinics (e.g., pain clinics), if a patient presented with recurrent back pain, or before ordering "expensive" tests. All physicians discussed lack of time as a barrier and pointed out that there are many other common conditions for which they need to screen. A few indicated that they would never use such a screening tool. Table 6 summarizes primary care physicians' perceptions of facilitators to implementation of a screening tool for axial spondyloarthritis in primary care. Physicians expressed concern that patients using an online screening tool might request unnecessary referrals. Suggestions included administering the screening tool in the waiting room or by telephone before the visit, during the appointment when axSpA is suspected, or by using a clickable link in the EMR (Table 6). Physicians also suggested using machine learning to trigger when to initiate screening, embedding a screening tool in UpToDate (an online resource designed to provide physicians access to current clinical information), and using smart phrases in EMR software to record responses to the screening tool. Physicians want evidence to support the use of an axSpA screening tool in practice (e.g., sensitivity and specificity; US Preventive Services Task Force endorsement).

Discussion
In our study, many primary care physicians lacked awareness of classification criteria for axSpA and most were not "extremely confident" in their ability to distinguish inflammatory

Major comments Representative quotes
Have you suffered from back pain for more than 3 months?
• Thought it was a reasonable question because mechanical back pain typically gets better in 6 weeks • Thought the question lacked specificity • Recognized that patients may experience flare-ups or intermittent pain and misinterpret the question • D32: Yeah, I think that's in general we're taught six to 12 weeks is kind of what people should get, kind of between that acute and more chronic phase, so I think three months is an appropriate time.
• D7: One is the lack of specificity in the question, you know, because there's some people who, you know, will say I've had back pain for ten years and that means it's come and gone. And so whether that's a follow-up question or whether you put in the preface, you know, continued back pain or chronic back pain or back pain that hasn't gone away for the last three months.
Did your back pain start when you were aged 40 or under?
• Agreed that age is clinically relevant • Thought that this question would be easy for patients to understand and answer • Others thought that the question was "non-specific", "too broad", and "sensitive not specific" • Questioned the value of this information because many people experience back pain under the age of 40 years and there are other contributors to pain (e.g., overweight)

Does your back pain improve with exercise?
• Agreed that with inflammatory issues, improvement with exercise expected • Expressed concern question was not sensitive or specific • Did not like the term "exercise" as they felt the term was generic and vague. Not all patients "exercise" (because "exercise is in the eye of the beholder"); could be interpreted as "going to the gym" • Concerned question may make patients feel "guilty and bad" • Suggested alternative wording: "activity" or "move around" • D34:'Cause exercise connotates I go to the gym and put on gym shorts and I pump iron. That's how I might have-that's what I think people might perceive, and it makes them feel guilty and bad, when you ask, "Do you exercise?" and they say, "No." They're reluctant to answer that. "Are you physically active?" "Yeah, I love going outdoors and playing with the kids, and we bike around." I get a lot more out of, "Are you physically active," versus, "Do you exercise?" I actually don't like the word "exercise."

Do you find there is no improvement in your back pain when you rest?
• Thought the question was good because it is something doctors don't think about • Like that it gets at mechanisms of symptoms (e.g., stiffness") • Thought that the "double negative wording" may be difficult for patients to answer • Some thought this question could be combined with the question on exercise • Suggested rewording: "When you rest, does your back get better?" • D28: If they find there is no improvement in back when you rest, okay, that's something I haven't thought about, so I guess would be a good one.
• D29: It may make more sense, like, "Does your back pain improve with rest?" Because that's, like, not a negative question, right?

Do you suffer from back pain at night which improves upon getting up?
• Thought question could be answered easily by patients • Addresses something doctors don't think about • Ruled out osteoarthritis and was associated with morning stiffness • Thought the question needed a more specific time frame: "better after 30 minutes into your day"; "more in the morning, better towards to evening" • Thought the question was not specific or sensitive. They thought that chronic low back pain can be attributed to many things (e.g., disks when lying flat, bad mattress)

PLOS ONE
from mechanical back pain. Many of the physicians interviewed emphasized the importance of taking a good medical history. Primary care physicians indicated that they would value a screening tool that provided guidance on appropriate tests to order prior to a referral, rather than an immediate referral to a rheumatologist, because of the difficulty in obtaining a timely rheumatology consultation appointment. Barriers believed to impede implementation of screening for axSpA in primary care included lack of time (because other more common conditions require screening) and lack of awareness about axSpA. All physicians interviewed emphasized the importance of integrating any proposed screening strategy into their clinical workflow.
Physicians are concerned about overuse of testing and low-value healthcare given that dissatisfied patients when costly tests are performed that may not yield definitive results [29]. Primary care physicians have difficulty discriminating inflammatory back pain from other kinds of back pain and are often unaware of other SpA features that are important for the differential diagnosis. Add to this fact that the radiological proof is sometimes a late feature of the disease and the result is the diagnostic delay. As such, primary care providers in our study thought that an axSpA screening tool may be helpful if it could make use of visit time and health care Table 5. Primary care physician perceptions of perceived barriers to implement the screening tool for axial spondyloarthritis.

Subtheme
Representative quotes Awareness D07: Awareness is probably the biggest thing. I'm not aware of guidelines for screening, using a specific screening tool. If you don't know that it exists and it's evidence-based, that's probably the biggest. resources more efficient. Pressure from national campaigns to reduce ordering of imaging (e.g., Choosing Wisely [30]), patient concerns about the high costs of laboratory tests that are not fully covered by insurance [31], and the workforce shortage in rheumatology [32] all underscore the value of using screening tools to identify patients in need of appropriate additional testing and referral to evaluate for axSpA. Given the brief amount of time available to spend with patients in primary care appointments [33], physicians reinforced the notion that implementation of an axSpA screening tool must not require additional physician time. Because the prevalence of axSpA in the general population is low, a screening approach would need to target patients with chronic back pain and rely on patient to complete a questionnaire. Participants suggested working through the US Preventive Task Force before making recommendations for screening in primary care practices [34]. This process begins with nominating a topic. If selected for further consideration, the process would involve development of a research plan and review of published evidence, culminating in a final recommendation statement. In our study, physicians remarked that back pain tends to be a 'doorknob' question-one that is left to the end of the visit when time already has run out. Short appointment times impede a physician's ability to fully explore the underlying reasons for back pain [30]. Patients experiencing waxing and waning symptoms may be misidentified as experiencing new onset back pain.
Participants in our study recommended various approaches using technology to assist with axSpA screening. Incorporating axSpA screening questions into EMRs has been suggested as being useful to improve efficiency in physicians' offices [35]. Such an approach has encouraged physician adherence to following recommended screening guidelines and has been successful in some [36,37], but not all, case studies [38]. Approaches that engage non-clinician support staff [39] or employ automated tools improved screening in primary care settings [40].
Patient reported screening approaches (e.g., via online screening or an app) offer an alternative to EMR-based algorithms. Use of a non-invasive, algorithm to find cases of axSpA, based on information contained in the EMR, has been studied using a randomized controlled cluster trial design [41]. In the intervention arm, the diagnosis of axSpA was ultimately confirmed in 8% of subjects but this approach did not have a short term impact on physical function, as average disability scores were similar in both arms at 4 months [41].

Strengths/Limitations
The qualitative study design is a strength, as this approach often provides insights that cannot be obtained by asking more closed-ended quantitative research questions. Our protocol adhered to best practices for qualitative research. Despite that a convenience sample was used [42] and some participants recruited were known to the researchers practicing medicine in Massachusetts and Rhode Island, the primary interviewer (KLL) was not someone with a prior relationship with the participants and therefore the interpretation bias could be reduced. Despite that none of the physicians recalled ever having diagnosed a patient with axSpA, most were confident in distinguishing inflammatory versus mechanical back pain. In addition, the average years in practice was 15.7 years and the majority were affiliated with academic institutions. Although primary care practices in the Northeast may differ from those in other regions of the United States, the findings do not appear to be overly optimistic with respect to the approach of screening for early detection of axSpA. None recalled ever having diagnosed a patient with axSpA, although several reported having treated patients with ankylosing spondylitis that was diagnosed before coming under their care.

Conclusions
Primary care physicians believed that the delay in diagnosis of axSpA is too long. With respect to the ASAS screening questions, physicians agreed that these questions need improvement and noted that some questions were neither sensitive nor specific. Primary care physicians preferred a screening tool that recommends additional testing, rather than one that directs referral to a rheumatologist. They believed that there may be a role for use of such a screening tool in the primary care setting but requested evidence to support its implementation, since they already must follow many other recommendations to screen for conditions more common than axSpA. Strategies to implement axSpA screening must be mindful of practice workflow issues and should be effective in reducing delay in diagnosis of axSpA.