“Gleaning a lot from the history and physical exam,” and “reasonably confident without imaging”: a qualitative study of primary care clinicians' management of patients with low back pain

Abstract

Background

Routine seeking of imaging for patients with low back pain is not concordant with the evidence-based recommendation that imaging is rarely of diagnostic value. Inappropriate imaging is a waste of resources and can lead to undesirable downstream effects for individuals and health systems. To develop effective strategies to reduce unwarranted referrals for imaging in primary care, we must understand the drivers for, and barriers to, guideline-adherent practice. We explored clinicians’ views to identify the dominant influences on clinicians as they choose to pursue, or avoid, imaging for their patients with low back pain.

Methods

We interviewed a purposeful sample of 47 primary care clinicians (14 physiotherapists, 18 chiropractors, 15 physicians) throughout Ontario, Canada, with a guide based on the Theoretical Domains Framework (TDF). We investigated clinicians’ views about their use of imaging in the management of low back pain. Interviews were recorded and transcribed verbatim. We analysed transcripts, then identified themes within TDF domains.

Results

Most clinicians reported that, for most clinical encounters, they adhered to guideline recommendations about imaging. Many clinicians across disciplines expressed the following themes: (1) imaging may result in an incidental finding or otherwise cause harm to patients, and drive up health system costs (TDF domain Beliefs about consequences); (2) clinicians were confident in their abilities to diagnose, to explain to patients the rationale for not recommending imaging, and to respond to their needs (domains Beliefs about capabilities; Skills). Many clinicians identified that patients occasionally want the validation that imaging provides (domain Social influences). Some clinicians described the value of imaging to corroborate a diagnosis (domain Beliefs about consequences).

Conclusions

This study is the first to examine influences on imaging behaviours of a large interprofessional sample of primary care clinicians in Canada. Even among knowledgeable, skilled, confident clinicians who reported mostly adhering to guideline recommendations, there are potential influences on deviating from guideline-adherent care.

Introduction

Non-specific low back pain, wherein the pathoanatomical cause of pain cannot be determined, represents a significant public health challenge in Canada and worldwide [1]. Every year, one in five Canadians experiences this type of low back pain, making it the second most prevalent symptom encountered in family practice [2]. Treatment for low back pain costs CAD$6—12 billion per year and is associated with significant societal costs related to loss of worker productivity and associated disability payments [3, 4]. In Canada, patients experiencing low back pain have access to several primary care clinicians, including publicly funded family physicians and mostly fee-for-service physiotherapists or chiropractors.

Clinical practice guidelines for low back pain from Canada, the United States, the United Kingdom and the World Health Organization [5] provide consistent recommendations for managing patients in primary care with low back pain, focusing on advice, reassurance, and self-management [6]. Imaging for patients with low back pain has limited diagnostic value and is related to potential harm [7]. Despite recommendations to avoid imaging in the absence of suspicion of pathology, clinical practice is frequently not aligned with recommendations about imaging [8, 9]. Several clinician-focused interventions have been developed to reduce imaging for low back pain, including educational interventions for clinicians (Clinically Oriented Relevant Exam (CORE) back tool) [10], clinical decision support tools such as the Choosing Wisely Canada campaign [11], and providing clinician support by way of rapid access clinics [12]. However, a systematic review of qualitative studies identified that clinicians lacked content knowledge of low back pain guidelines [13]. To inform the development of effective knowledge translation interventions relevant to the Canadian context, we need to determine clinician beliefs about imaging for this patient population.

The Theoretical Domains Framework (TDF) is a comprehensive framework used to understand behaviour change among healthcare providers. The TDF identifies 12 domains that influence behaviour, including knowledge, skills, beliefs, and social influences, among others [14]. While we know that some clinicians believe there is value in imaging [13], there is limited insight into how well they are informed about the guidelines and contextual factors which may influence their decision-making processes. The Back ON study aims to: 1) determine the rate of and factors associated with inappropriate lumbar spine imaging (x-ray, MRI, CT) for people with non-specific low back pain presenting to primary care in Ontario; and 2) determine the barriers and enablers to reducing inappropriate imaging for low back pain in primary care settings. We report here on results related to Aim 2, based on using the TDF to gather the perspectives of primary care clinicians across Ontario, Canada, regarding imaging for acute non-specific low back pain. We used the TDF to design the interview guide and analyze the resulting data because the TDF is designed to provide insight about a comprehensive set of potential influences on behaviours. The TDF has been used extensively to understand healthcare behaviours, including in the analysis of clinicians’ use of imaging in the management of low back pain [13, 15,16,17].

Materials and methods

Study design

This qualitative exploratory study used semi-structured interviews to examine contextual factors which may influence the decision-making processes regarding the ordering or referral for imaging for patients presenting to primary care with low back pain.

Participant selection, recruitment and data collection methods were restricted to those of the larger prospective cohort study, the Back ON study [18]. The current study adhered to the consolidated criteria for reporting qualitative research (COREQ) guideline [19].

Participants & recruitment

Our target group comprised community-based primary care physicians, chiropractors, and physiotherapists in Ontario who provided care to patients with low back pain. In Ontario, all family physicians can directly refer patients for imaging; most chiropractors and physiotherapists cannot, but they can influence imaging through referral to a patient's physician, and some chiropractors have plain x-ray imaging equipment in their own practices. We excluded physicians working in academic settings or exclusively urgent care settings.

For the Back ON study, clinicians were randomly selected for recruitment from registered clinicians in the College of Physicians and Surgeons of Ontario, the College of Physiotherapists of Ontario, and the College of Chiropractors of Ontario. We also used snowball sampling by asking participants to identify colleagues and other clinicians [18]. We used a purposive sampling approach for the interview phase to target a subset of Back ON participants.

Study clinicians were interviewed after they completed recruitment of patient participants in the Back ON study. We planned that only clinicians who recruited patients to the larger study would be interviewed and recruited our target sample of chiropractors who met this criterion. We made exceptions to this protocol for physicians and physiotherapists, as many clinicians experienced challenges in recruiting patients during the COVID pandemic. After the initial interviews with chiropractor participants, and to ensure variation, we sought additional chiropractor participants who either possessed or recently had access to their own imaging equipment.

Participating clinicians were given an information letter and consent form. Participants completed a demographic survey about their age, sex, training location, experience, and proximity to an imaging facility. Participants also completed a questionnaire that included familiarity with four low back pain guidelines [20,21,22,23], with an “other: please specify” option. They were asked to rate their level of agreement on statements about the value of imaging for non-specific low back pain (Appendix 1).

This research was approved by the Health Sciences and Affiliated Teaching Hospitals Research Ethics Board at Queen's University (REH-736–18) was conducted in accordance with the principles of the Declaration of Helsinki.

Interview guide

We developed three interview guides (physician / physiotherapists / chiropractors), modifying questions to reflect whether clinicians ordered imaging or recommended that the patient seek imaging from another provider. Interview guides were iteratively developed using the TDF framework [24] and results from a systematic review [13]. The guide (Table 1) included two or three questions from each of the 14 TDF domains, aiming to elicit: clinician knowledge; typical behaviours about imaging and/or referring for imaging; and barriers to, or enablers of, following guideline recommendations.

Table 1 TDF domains, definitions, and an example interview question used for each domain

After completing three pilot interviews, one from each clinician group, we reviewed the transcripts and revised the interview guide. Participants had the option of conducting the interview over telephone or via Zoom. Participants were compensated with an honorarium upon the completion of the interview.

Data analysis

The analysis team comprised a mix of researchers (nursing, physiotherapy, chiropractic, health psychology) with qualitative expertise. We transcribed all interviews verbatim using NVivo® transcription software, with K.N. checking for accuracy. The TDF framework was used to analyze and code the narratives [24]. L.H–L., K.E.N., and I.M.T. met to review transcripts and develop a coding strategy. The coding strategy was initially based on a deductive approach to content analysis using the TDF domains alongside the ability of the coder to identify if an utterance appeared to be outside the domains.

Following the initial coding using NVivo®, the team met to ensure optimal categorization of utterances within domains and identified that utterances difficult to code within the TDF themes often reflected avoidance of the behaviour (e.g., participants talked about how they would avoid imaging). The team came to consensus about how to assign these utterances to TDF domains based on inductively determined themes. Intercoder reliability was carried out in accordance with the process described by O’Connor, et al. [25], with two coders (L.H–L., I.M.T.) independently coding 1 of every 5 interviews (27%) to ensure consistent application of code definitions and interpretations. Additionally, L.H–L., K.E.N., and I.M.T. met to review excerpts within each domain and themes to verify coding aligned with the definitions in consultation with a TDF expert (A.M.P.). Group discussions and consensus resolved any coding discrepancies. A codebook summary was shared with the authorship team to ensure accurate interpretation. Finally, L.H–L., K.E.N., S.F., A.M.P., and I.M.T. reviewed the findings and identified TDF domains as relevant if they had a high number of themes, or conflicting statements amongst respondents, or those with clinical significance towards changing practice behaviour [24].

Results

Participant characteristics

We invited 19 family physicians, 20 chiropractors, and 16 physiotherapists to participate; in total, 47 participants (15 physicians, 18 chiropractors, 14 physiotherapists) completed interviews. Interviews took place from June to October 2021. The audio-recorded interviews ranged from 29 to 104 min (mean 59 min). No repeat interviews were carried out. The participants had an average of 13.7 (range 1–59) years of experience and saw a median of 10 (range 1–100) low back pain patients per month (Table 2).

Table 2 Clinician demographic information

Key themes identified within relevant domains

Table 3 presents the TDF domains of greatest relevance with sample quotes. We identified six TDF domains that aligned to participants’ attitudes about ordering imaging or referring patients to their physician for imaging: (1) Knowledge, (2) Skills, (3) Beliefs about capabilities, (4) Beliefs about consequences, (5) Social influences and (6) Behavioural regulation. The Cohen’s kappa coefficient for inter-rater reliability in applying the codebook was 0.73 [26].

Table 3 Key themes identified within relevant domains

Knowledge

In the pre-interview questionnaire, 24 (51%) of the participants either indicated “Never heard of it” or "Likely read it at some point but can't recall" for all listed guidelines (Fig. 1). However, in interviews, all participants stated that they knew there were guidelines for imaging in low back pain (Knowledge), with some recalling that they were taught the guidelines during formal education. Some participants referred to published sources of best practices, such as the Choosing Wisely Canada recommendations [11] (mentioned by 8 physicians), or the CORE Back tool [10] (2 chiropractors and 5 physicians), with the majority of participants admitting that they were unable to identify the name of a guideline that they followed in practice.

Fig. 1

Familiarity with low back pain guidelines

When probed about what they could remember about imaging guidelines, many provided examples of algorithms or methods utilized when assessing patients with low back pain: e.g., an algorithm published by the provincial health authority (1 chiropractor and 5 physicians); the McKenzie method of Mechanical Diagnosis and Therapy [27] (1 chiropractor and 11 physiotherapists). Although participants recognized they were not able to recall all components of any guidelines, almost all participants went on to articulate the importance of imaging only in the presence of red flags and were able to describe what these red flags were (Knowledge). They also described strategies they incorporated for managing low back pain, such as patient education and reassurance, lifestyle modifications, active recovery, and pain management (Partial knowledge of important elements without knowing guidelines). Many clinicians perceived imaging as potentially appropriate after a specified duration of time or if symptoms persisted beyond a certain number of weeks (Partial knowledge of important elements without knowing guidelines).

When asked about how they keep up to date on guidelines, participants identified two categories of methods – active and passive. Active methods included seeking out information themselves, which included peer-reviewed sources such as journals and Uptodate® (3 physicians), internet searches (1 chiropractor, 3 physicians, 1 physiotherapist), continuing education modules (3 chiropractors, 4 physicians, 4 physiotherapists), or participating in groups such as journal rounds (3 physicians; 3 physiotherapists). Knowledge was also obtained passively via communication bulletins sent to members by professional associations, identified by 17 participants (Sources of knowledge). Four participants admitted to not keeping up to date with recommendations (1 chiropractor, 1 physician, 2 physiotherapists).

Skills and beliefs about capabilities

Participants expressed confidence in their ability to educate and reassure patients about low back pain management without imaging (Beliefs about capabilities). Participants attributed their proficiency and accuracy in diagnosing patients without imaging to adopting a consistent assessment approach (Skills), which instilled confidence in their ability to manage a patient’s imaging expectations. Participants also identified that years of clinical experience plays an important role in arriving at a diagnosis without imaging (Skills). Conversely, some participants reflected that a lack of experience in newer professionals may cause a reliance on imaging to arrive at a diagnosis.

Participants provided scenarios they found to be ‘tricky’ or challenging, wherein they are more likely to image (Beliefs about capabilities). Participants described difficulty in deciding about the need for imaging in the presence of multiple co-morbidities, older age, or an ambiguous presentation. This view was more prevalent among chiropractors.

Beliefs about consequences and behavioural regulation

As imaging avoidance in patients with no red flags was the desired behaviour, we felt it was prudent to capture how participants incorporated low back pain management strategies into their practice to avoid imaging (Behavioural regulation – alternate behaviours / managing without imaging). Some recognized pain management as a priority goal and articulated the importance of exercise and referrals to physiotherapy (for physician participants) as an adjunct to pain management. Additionally, many participants expressed that imaging results are frequently incidental and would not provide any additional information beyond what they had already obtained from their history and physical assessment findings (Beliefs about consequences).

Many participants expressed concerns about potential harm to patients from imaging, such as radiation exposure and out-of-pocket expenses (Beliefs about consequences). Furthermore, participants acknowledged that incidental findings could cause patients emotional distress, as the imaging might reveal a finding irrelevant to their symptoms. Participants also described being hesitant to order imaging due to reducing access to imaging for cases that warrant it or diverting funding from other areas of healthcare. Some also noted that imaging often fails to alter pain management strategies and offers no tangible benefits to patients. A few participants described why they value imaging to arrive at a diagnosis (Beliefs about consequence – value of imaging).

When completing the pre-interview questionnaire, 45 (96%) participants ‘strongly disagreed’ or ‘disagreed’ with the statement “I am likely to refer low back pain patients for lumbar spine imaging (x-rays, CT or MRI) because patients often expect me to do so” (Fig. 2). Nevertheless, in interviews, many participants expressed that they perceived that the patient was expecting imaging or sometimes felt pressured to accommodate a patient who wanted imaging (Social influence). Participants believed that patients wanted imaging to assign a diagnostic label to the source of the pain, which in turn helps to validate their degree of impairment (Beliefs about consequences, Social influence). Other participants highlighted that their ability to reassure the patient was a determining factor. As such, they felt that their ability to assess and promptly address a patient’s concerns was a crucial skill set to prevent a patient’s request for imaging (Beliefs about capabilities). Participants did admit to occasionally ‘giving in’ to imaging requests, particularly if the patient was insistent (Social influence). When faced with a patient wanting imaging, participants said they would often adopt negotiation strategies, such as communication and education on outcomes to avoid imaging (Behavioural regulation). Participants stated that they would offer patients imaging at a time in the future should the proposed treatment not be effective. This redirection allowed for additional time to pass, with the expectation of resolution of low back pain symptoms. Several participants expressed the importance of establishing rapport with patients, which fostered a trusting relationship and, in turn, increased the patient’s confidence in the participant’s ability to address their needs without the reliance on imaging (Beliefs about capabilities). However, participants identified that in some cases, patient communication and education proved to be challenging for them (Skills), especially with patients who needed additional psychosocial support or ongoing reinforcement of adherence to the proposed treatment.

Fig. 2

Agreement with lumbar spine imaging usefulness in confirming a diagnosis

Domains of lesser relevance

Our analysis revealed that the TDF domains of 1) Social professional role and identity, 2) Optimism, 3) Reinforcement, 4) Intention, 5) Goals, 6) Memory, attention and decision processes, 7) Environmental Context and Resources, and 8) Emotions were not as influential on most participants’ decisions to recommend or order imaging (Table 4). Participants were knowledgeable about both their professional limitations and the interprofessional expectations regarding referrals for imaging (Social professional role and identity). Many participants were not optimistic that imaging would be of any value; however, a few participants – principally chiropractors – found value in imaging (Optimism). Participants stated there was no financial incentive or disincentives to order imaging (Reinforcement). When asked about their intent to ‘image the next 10 patients’, there was a range of answers, with 38 (80%) of participants responding that they would only seek an image for 0 to 2 patients of the next 10; three (6%) responded the next 3 of 10; three (6%) responded with 4 to 5 of 10; and two (4%) said 5 to 7. One participant (a chiropractor) stated they would seek an image for all of the next 10 patients (Intention). Within the TDF domain Goals, participants often shared their priorities when interacting with a patient with low back pain. Most participants articulated a clear concept of criteria for deciding when to image (Memory, attention and decision processes – criteria that would prompt an image). Finally, most participants did not express emotional responses to patients seeking care elsewhere if they did not image (Emotion). Nevertheless, 15 (32%) participants articulated that they feared they would miss a critical finding if they did not image (Emotion – fear or doubt of missing something important).

Table 4 Domains of lesser relevance

Discussion

In this qualitative study that included a diverse group of primary care clinicians, we explored the perspectives of clinician beliefs about the use of imaging for patients experiencing low back pain. The majority of participants reported adherence to guideline recommendations by avoiding imaging in this population except when patients presented with ‘red flags’. Our results indicate that participants' confidence appeared to be a key enabler in avoiding inappropriate imaging for patients with low back pain. The confidence was most described in relation to having a standardized assessment that included screening for red flags, and the ability to explain to patients why imaging was not needed to guide diagnosis or treatment in most cases. Our findings are very similar to the findings of a systematic review that practitioners’ knowledge, skills, beliefs about consequences and social influences are all highly relevant TDF domains associated with why physicians may use imaging in the management of patients with low back pain [15].

In contrast, our findings are distinctly different from previous meta-syntheses by Slade, et al. [28] and Sharma, et al. [13], where they reported that clinicians felt imaging was a useful tool to explain the source of the pain and relieve a patient’s anxiety. When we reflect on this contrast, it appears that, at least in the Ontario context, clinicians’ perspectives about imaging for low back pain may have changed in recent years. This change may be partially due to the success of campaigns and tool development. For example, the Choosing Wisely Canada campaign [11] appears to have been a success: 67% of the physician group identified familiarity with the messaging. Also, many participants were familiar with the CORE back tool [10], identifying it as a reliable source for information. However, we note that both resources reference ‘six weeks’—participants recalled this wording, yet not always in the appropriate context. For example, the CORE back tool indicates that “if your symptoms persist for > 6 weeks, schedule a follow-up appointment”; it does not direct clinicians to image at that time. Some participants interpreted these statements as conveying that imaging is required if pain persists after six weeks.

Our study has findings similar to that of Slade, et al. [28], wherein they identified that clinicians commonly aligned with an ‘accepted’ practice among peers rather than guideline recommendations. This finding is unsurprising, as professionals learn from various sources, contributing to the development of ‘mindlines’. Mindlines [29] refers to the internalization of knowledge within guidelines formed by traditional written material, yet also recognizing that knowledge is formed over a lifetime, from personal experience as well as the shared experience of peers and mentors. These sources were all noted as influencing our participants' knowledge; one-third of participants identified that they were kept abreast of clinical guidelines by communication sent out by their professional bodies. Almost two-thirds of participants perceived that their peers had a similar approach to imaging (Social influences). Thus, it is not surprising that participants expressed challenges and less confidence with managing ‘tricky’ patients (e.g., older adults, patients with more co-morbidities or psychosocial issues, insistent patients), given that this ‘atypical’ patient would challenge content within existing mindlines [30]. Mindlines would likely be formed around the ‘prototype’ of a patient with low back pain, with an awareness of the content of guidelines (e.g., red flags) with knowledge limitations about managing the atypical low back pain presentation. An opportunity for future interventions would be to embrace this concept of how clinicians learn, and target both informal sources of knowledge (e.g., peer tutors) [31] in addition to formal sources (e.g., professional dissemination of updates), focusing specifically on the management of tricky cases. Another opportunity would exist with developing channels for clinicians to consult with peers before resorting to imaging (e.g., peer / expert case review; health systems to design rapid referrals).

There were very few participants who found high value in imaging, with chiropractors being the most prominent group of these few participants, which is a similar finding to that of other studies [13, 16]. Most of our study participants, including most chiropractors, believed that imaging would result in an incidental finding in most patients rather than aiding diagnosis, contrary to the findings within the previous meta-analyses [13, 28]. Based on a meta-analysis, Sharma, et al. [13] reported that clinicians may order imaging to reduce the risk of litigation due to missed diagnosis. In contrast, we found no evidence of this influence in our participants. There are differences between Canadian malpractice coverage and that elsewhere, which results in a significantly lower risk of a successful claim than in many other countries, particularly the USA [23]. In our data, concerns about missed diagnosis were primarily about poor patient outcomes rather than litigation.

Limitations

Since participants volunteered for the study, the study generalizability is limited as participants may have been more likely than most Ontario primary care clinicians to keep up with research findings and thus have been aware of the guidelines for imaging in non-specific low-back pain. We also recognize that low back pain is often managed in the context of a longitudinal relationship between the patient and clinician; the pros and cons of carrying out discussions to forgo imaging in the context of patient desire for imaging may vary from one visit to another.

Conclusion

This study reports on the findings from the first Canadian large interprofessional sample of primary care clinicians examining influences on low back pain imaging behaviours. Clinicians identified several opportunities for ongoing dissemination of guidelines and practice areas that can potentially reduce imaging in patients with non-specific low back pain.