Introduction

For patients and primary care physicians (PCPs) alike, persistent non specific low back pain (nsLBP) is well recognised as a “heart sink” [1] and the label nsLBP covers up our almost complete ignorance of underlying pain mechanisms. Critically, the management of patients with nsLBP is unsatisfactory and the resultant burden of disease at both individual and societal levels has become significant [2], [3], [4]. Models of care, service delivery and individual practitioner practices in the assessment and management of nsLBP also vary [5], [6] further impacting negatively upon patient outcomes [7]. Barriers to the implementation of current evidence-informed best practice in primary care have been identified [8], [9], [10] and, in parallel, patients continue to seek a pathway towards less pain and disability [11].

Despite the significant and escalating costs incurred to health systems from the use of various investigational and imaging studies [12] related to diagnosis and management [13], [14] and the costs associated with co-morbid pain [15], clinical outcomes for patients have not improved [16]. Additional contributors to poor patient outcomes include a lack of adherence to guidelines [17], [18] or lack of knowledge by practitioners [19] of best practice guidelines [20], [21] and the variable translation of evidence into practice in the primary care setting [22]. Heterogeneity in nsLBP cohorts [23], [24], [25] compounds these issues, although novel protocols to subgroup patients are emerging [17], [26]. Different systems of health care delivery complicate data comparisons and interpretation of trial outcomes is dogged by a lack of standardised pain assessment measures [27]. System inefficiencies exist due to inadequate use of validated screening criteria designed to match patient complexity with an appropriate level of resource allocation [28]. Furthermore, as there is no simple biomarker for pain and the lived experience cannot be inferred from imaging studies of the spine, including computed and functional magnetic resonance tomography, consistent adoption of evidence-based practice appears even more critical in achieving positive health and economic outcomes [22].

By contrast, the adherence to guidelines by practitioners and the adoption of appropriate self management practices by patients are currently considered rational strategies to reduce the burden of spinal pain [28], [29], [30] and to deliver cost effective patient outcomes [18], [22]. Previously we have demonstrated reduced wait-times and costs at a public pain medicine unit and increased use of active pain management strategies [30] following a system redesign from a traditional model to one that delivers interprofessional patient group education sessions prior to individual appointments [28]. Interprofessional care, that is care arising from the provision of comprehensive health services to patients by multiple health caregivers who work collaboratively in order to deliver quality care evidence-based care within and across settings, is associated with a reduction in health care utilisation and improved function [30], [31], [32]. Integrating biomedical pain management strategies with cognitive behavioural approaches in a health service delivery model such as ours perforce uses an interprofessional platform [28], [30], that combines evidence-based guidelines [33] pharmacological [34], [35] and interventional procedural options [36] with self-management training for patients [30], [37], education in the neurophysiology of pain [38], [39], active strategies such as pacing (time-contingent graded activity [39]), moderating fear avoidance behaviours [40] and active movement-based strategies [41], [42]. In primary care practice, adding advice, education and exercise, or exercise and behavioural counselling to usual physician care, is also more cost-effective than usual physician care alone [32]. However, most physicians have received limited training to cope with the multidimensional nature of complex pain. The implementation of such interprofessional evidence-based models would appear to require a continual cycle of education coupled with practical skills, delivered both to consumers and health providers as described in the Western Australian, Department of Heath, Spinal Pain Model of Care [43]. Therefore, to better align policy and practice [28], [30] in Western Australia, we developed, implemented and evaluated the effectiveness of such an interprofessional educational program designed to enhance the knowledge and skills of primary care physicians (PCPs) managing patients with nsLBP.

Methods

Subjects

Study design/population.

Using a prospective cohort study design, PCPs from metropolitan Perth, Western Australia were invited to attend the General Practitioner Pain Education Program (gPEP). This program was designed to upskill PCPs with practical evidence-based management of patients with nsLBP. Inclusion criteria required that PCPs were registered and practising in primary care. Exclusion criteria included PCPs or specialists already working in a multidisciplinary team that treated patients with acute and chronic musculoskeletal pain.

Ethics statement.

The study was approved by the local South Metropolitan Area Health Service, Fremantle Hospital and Health Service, Department of Health (Government of Western Australia), Human Research Ethics Committee (no. 08/371) and adhered to the Code of Ethics of the Declaration of Helsinki. Written informed consent was obtained from all participants. The Royal Australian College of General Practitioners (RACGP) accredited gPEP as a Continuing Professional Development activity attracting a Category 1 rating with the maximum of 40 points awarded to practitioners post-completion of required activities.

Participation, consent and anonymity.

PCPs were invited through two nominated metropolitan General Practitioner Networks (GPNs) to register and attend one of five workshops, which were run consecutively over a five month period in 2009. One hundred and twenty six PCPs were registered for gPEP (Figure 1). Of this number, eight were excluded: two did not meet the inclusion criteria (not currently practising); six met the inclusion and attended gPEP but did not consent to use of the data for the purposes of the study. Of the one hundred and eighteen remaining registered PCPs, ninety one attended and participated, with the twenty seven remaining registrants not attending on the day. All registrations were accepted by the two geographically separate, metropolitan GPNs. These GPNs allocated a pre-selected unique study identification code to each of the registered PCPs. A set of these unique study codes was pre-allocated by the research team to each GPN. This unique code appeared on all registrants' subsequent data sets and on the associated GPN databases allowing for the efficient cross-matching of both responders and non-responders throughout the study, while also maintaining participant confidentiality. One researcher (HS) was also a member of the educational team and was aware of the unique study identification codes. This researcher was not involved in the data entry or analysis. The remainder of the educational team was blind to the data collection, entry and analysis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. This flow chart indicates the study recruitment process.

Note that some physicians registered and attended but were not included in the analyses as they did not consent to their data being used. These participants were still eligible for their maximum continuing education points if they submitted their pre and post course questionnaires.

https://doi.org/10.1371/journal.pone.0038037.g001

On the day and immediately prior to the intervention, written consent forms were completed by attending participants and collected. Participants were then instructed to open an envelope containing their uniquely-coded data set. This data set contained a battery of questionnaires, as fully outlined below. Once completed, each participant sealed their baseline data in the individual envelopes provided. Data sets were immediately collected by the research team and attending GPN representatives. Following each of the educational programs, one investigator (HS) cross-matched the registration list with the final attendance list and with the signed consent forms. This enabled identification of participants who registered but did not attend, or who did attend but did not consent to their data being used for the purposes of this study. For those participants (n = 6) who did attend the program, but did not consent to the study, no data were included in any analysis. However, for these participants, data were still collected as completing the battery of questionnaires was a requirement for awarding of continuing education points. At 2 months post-intervention, each GPN mailed out to their allocated participants, a code-matched post-course data set with instructions to complete and to send back in the pre-paid envelope or to fax back to the GPNs. Each participant data set was then matched and logged against their baseline data set. In this way pre- and post-intervention data could be matched at data entry. Non-responders were contacted only by the GPNs, who requested completion of the post-course questionnaires to satisfy RACGP post course activity requirements.

Intervention: Educational Team and Educational Materials.

The interprofessional educational team were all actively engaged in either tertiary facilities (hospitals providing specialised consultative health care, and requiring a referral from a primary or secondary health care facility) and/or private practice non-hospital primary care and hospital-based pain management facilities and/or university facilities and who worked together in various clinical and research combinations across these facilities. The team included four pain medicine specialists (one of whom was also a rheumatologist), one senior occupational therapist, four senior postgraduate-qualified musculoskeletal physiotherapists and two clinical psychologists. To ensure the educational content was closely aligned with and relevant to a primary care setting, external feedback was also sought from and provided by PCPs including representatives of the RACGP and the GP networks, and a clinical academic PCP.

The implementation framework for this intervention is summarised in Figure 2 and was based on ‘The Western Australian Spinal Pain Model of Care’ [43] with a focus on key recommendations 1–4 (p. 8 and p. 33 of the Model of Care; http://www.healthnetworks.health.wa.gov.au/modelsofcare/docs/Spinal_Pain_Model_of_Care.pdf). Five members of the educational team, 3 of whom also co-authored this study (HS, SD, JQ), contributed to the development of this evidence-based Model of Care which was designed to ensure consumers with spinal pain receive the ‘right’ care, at the ‘right’ time, from the ‘right’ team and in the ‘right’ place.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. The framework involved in developing and implementing the gPEP intervention is described in this schematic.

The Western Australian Spinal Pain Model of Care (MOC) is a policy document which describes a framework on which evidence can be implemented into policy and practice to ensure consumers with spinal pain receive the ‘right’ care, at the ‘right’ time, from the ‘right’ team and in the ‘right’ place. The Spinal Pain MOC addresses key gaps in policy and practice. Starting with the Model of Care, key stakeholders interested in spinal pain collaborated to apply for grant funding. Key recommendations from the MOC informed the focus of the educational content for gPEP, and the content was peer reviewed (including GP network engagement) and accredited through the Royal Australian College of General Practitioners (RACGP) for Continuing Professional Development points. The intervention was then implemented and the effectiveness evaluated.

https://doi.org/10.1371/journal.pone.0038037.g002

The educational team collaboratively developed and delivered the 6.5 hour single day gPEP program (Figure 3) which included the following modules: (1) Making sense of pain: a missing component of care; (2) Clinical guidelines and evidence-informed best practice for the assessment and management of patients with nsLBP; (3) Movement, activity and pain; pacing activity and goal setting: helping patients with nsLBP map a meaningful course through every-day life; (4) Response to pain: psychological and behavioural factors in managing patients with nsLBP; and (5) Pharmacologic and procedural approaches to the management of patients with nsLBP.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. The interprofessional model of low back pain education for primary care physicians, is shown.

Physicians' evidence base knowledge and skills and clinical practice behaviours were measured at baseline (upper dotted line) and at 8 weeks post intervention (lower dotted line). Five modules were presented over a single day. Each of 5 modules was presented with a short evidence based lecture of 15–20 minutes duration and was accompanied by a related case study integrating and applying the relevant clinical knowledge and skills. Each case study was designed to facilitate interprofessional engagement between both PCPs and the educational team, so participant groups were limited in size (typically n≤12) with each comprising a micro-interprofessional team (pain medicine specialist, clinical psychologist, physiotherapist and occupational therapist). The horizontal arrows indicate the integration of evidence base between and across all modules. Case studies targeted clinical practice related to each module, but also included other modular information, as appropriate.

https://doi.org/10.1371/journal.pone.0038037.g003

Each module comprised a 15–20≤ minute evidence-based, guideline-informed lecture which was followed by an action-learning [44], interactive, ‘know-do’ case study of 45–60≤ minutes duration. Each case study was presented as a patient vignette with clinically relevant questions and interactive discussions relating to the implementation of evidence into clinical practice, specifically targeting appropriate clinical practice behaviours and focusing on practical patient-oriented active self management strategies and co-care. These case-based studies were focused on a matching of resources and management approaches to the level of a consumer's pain and disability (that is, low pain and disability were approached using less complex approaches and high pain and disability were approached using more complex, in-parallel multimodal approaches). Screening tools (for example, Orebro [45], Depression, Anxiety and Stress Scale [46], painDETECT [47]) were presented and their application and scoring demonstrated with reference to these case studies, with 2 specific aims: (i) to provide busy PCPs with a time-efficient diagnostic triage system for screening patients with LBP; and (ii) to enable PCPs to match their clinical findings with scores from tools designed to reflect the multidimensional aspects of pain.

The clinical guidelines which informed the development of the educational materials for this study included: (i) the Australian Evidence-Based Management of Acute Musculoskeletal Pain: A guide for clinicians [48] (ii) the New Zealand Clinical Group Guidelines for the Assessment and Management of Acute Non-Specific Low Back Pain [49], [50]; (iii) the European Low Back Pain guidelines [51], [52], [53], [54] (iv) Diagnosis and treatment of low back pain: a joint clinical practice guideline from the American College of Physicians and the American Pain Society [55]. The updated National Institute for Health and Clinical Excellence guidelines for the management and assessment of nsLBP [56] were not available at the time of the program development, but once available were subsequently reviewed in 2009 to ensure there were no significant evidence omissions.

Program participants were provided with a hard copy of a workbook containing all lecture materials and case studies, a summary table of evidence for the management of patients with acute and chronic nsLBP and a CD with the clinical guidelines listed above and any guideline-associated patient information sheets. All participants were also offered access to an online, not-for-profit database (myLibrary) used as a sustainable repository for all course materials and updates. Two free, optional “web-labs” were also provided to up-skill participants in web-based evidence searches and storage of information using a previously developed evidence-based storage database (http://www.mylibrary.net.au/).

Results

Of the one hundred and eighteen PCPs registered, ninety one attended and participated (attendance rate of 72%). Demographic and clinical data are shown in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. The demographic and clinical practice characteristics of primary care physicians (PCPs) participating in gPEP^†.

https://doi.org/10.1371/journal.pone.0038037.t001

Attitudes and Beliefs (Modified HC Pairs Questionnaire)

Mean (SD) scores are shown pre- and post-intervention, the 95% confidence interval for the difference in means and the p-value for each item of HC-PAIRS (Table 2). In addition, a total HC-PAIRS score was generated. The HC-PAIRS score difference (n  =  mean change  = −5.6±8.2, p<0.0001; 95% confidence interval: −7.6 to −3.6) demonstrated a clinically significant move towards disagreement with the questions, that is, against the suggestion that management of lower back pain should involve rest rather than activity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Primary care physicians' beliefs regarding low back pain and associated disability.

https://doi.org/10.1371/journal.pone.0038037.t002

Self-reported knowledge and skills (gPEP questionnaire)

The pre- to post-intervention change in responses for self-reported knowledge and skills, is shown in Table 3. The general trend for each question is clear from the unmatched data. Some of the questions were left blank by some respondents, which is why the totals vary a little. While approximately 20–35% of responses were inadequate prior to the intervention, only a very small number of people rated their knowledge as inadequate following the intervention (indicating a clear impact of intervention). The paired t-tests showed highly statistically significant movement towards greater knowledge (p<0.0001 for all questions).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Comparison data for primary care physicians' evidence-based self-reported knowledge and skills.

https://doi.org/10.1371/journal.pone.0038037.t003

The pre- to post-intervention change in the frequency of recommendations to patients with nsLBP, is shown in Table 4. While this is a little hard to interpret as it is related to PCPs' workload, if the categories are grouped into ‘Never’ and ‘at least once per week’, it becomes evident that there is a general move for questions 13–16 away from the ‘Never’ category. The trend in all of these questions is towards giving this advice or assistance at least some of the time. The kappa statistic for each item was generally low, indicating that the intervention had influenced responses. For most items, the majority of people responding ‘Never’ at the pre-intervention stage responded ‘at least once’ at the final survey.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Comparison data for primary care physicians' frequency of recommendations for low back pain management.

https://doi.org/10.1371/journal.pone.0038037.t004

The three most important things PCPs obtained from gPEP included the following themes:

1. Management strategies and education (71.5%). Items listed included: a better understanding of pain; increased confidence with managing LBP; improved knowledge of evidence-based assessment and management; making a management plan; pacing advice; importance of belief systems to patient outcome.
2. Importance of an interdisciplinary team approach (51.9%). Items listed included: a multidisciplinary team approach, e.g.; physiotherapy, psychology, exercise program; psychological aspects, including the use of a psychologist; importance of physiotherapy; not all physiotherapists know how to treat LBP.
3. The limitations of imaging (39.5%) and the appropriate use of pharmacological options available/appropriate use of meds (29.6%)

Clinical practice behaviour (patient vignette)

The pre-post comparison showing PCPs' recommendations for activity, work and bed rest, regarding the acute nsLBP patient vignette, is shown in Table 5. The movement for Q1 was marginally against the recommended guidelines, while the other questions showed movement towards guideline consistent behaviour. Q2 showed a large movement towards guideline consistency. The kappa statistic showed poor agreement (low values), confirming that the responses had changed, and Table 5 shows the direction of the change for each question. While over 85% of respondents who were initially guideline consistent for each item remained consistent, the majority of respondents who were guideline inconsistent for questions 2 and 3 gave guideline consistent responses at post-intervention (82% and 62% respectively). For question 1 (exercise recommendation) the results were less conclusive, with a smaller proportion (39%) of respondents changing from inconsistent to consistent.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Comparison of primary care physicians' recommendations for acute non specific low back pain management.

https://doi.org/10.1371/journal.pone.0038037.t005

The baseline means of the HC-PAIRS questionnaire (total of all questions) were compared between participants classified as ‘guideline consistent’ and ‘guideline non-consistent’ according to the 3 questions on the patient vignette (Table 6). The means for the non-consistent group for each question were significantly higher (p<0.003 for each question) than for the guideline-consistent group.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Comparison of HC-PAIRS total scores (pre-intervention), between participants classified as ‘guideline consistent’ and ‘guideline inconsistent’.

https://doi.org/10.1371/journal.pone.0038037.t006

Discussion

We demonstrate promising findings in this before/after interprofessional pain education program set within a framework that aligned health policy and practice. Volunteer PCPs were encouraged to adopt more self-reported evidence-based beliefs, attitudes and clinical behaviours for managing their patients with nsLBP. Notwithstanding the lack of control for confounding factors imposed by a cohort design, our data clearly demonstrate a positive impact of the gPEP intervention, with strong evidence for movement of PCPs towards improved beliefs and attitudes regarding LBP, greater self-reported use of evidence based knowledge and clinical skills, and more guideline consistent recommendations for work and activity. Active self management strategies were likely to be recommended more frequently post-intervention. This intervention used an implementation framework which, in our view, may help to bridge a significant gap in the effective translation of evidence into policy and practice.

The significant decrease in HC-PAIRS scores at post-intervention is consistent with improved PCP beliefs and attitudes about pain and impairment in relation to LBP and of a greater magnitude than previously demonstrated for health professionals following an intervention using printed materials only [61]. While the change in total score can be interpreted as clinically important (≥−4.5 points as proposed by Domenech et al [62]), it was also clear that responses to some questions changed more than others. The two questions which changed the most (on average) were: Q2: An increase in pain is an indicator that a low back pain patient should stop what they are doing until the pain decreases; and Q8: Low back pain patients have to be careful not to do anything that might make their pain worse. This shift of almost 1 point on the 7-point Likert scale towards disagreement may reflect the emphasis of gPEP which focused on delivering simple, evidence-based, guideline consistent messages that PCPs could readily convey to their patients. These key messages were consistently repeated throughout the case studies and also linked with specific evidence-based approaches to clinical management, thereby showing the ways in which guidelines could be implemented in a flexible patient-centred manner.

So that PCPs could more easily align their work practices and recognise the potential benefit for patients [63], the case studies were designed to resonate with the guidelines, and improve the coherence with existing PCP consultation practices. In this context, the following guideline consistent recommendations relevant to HC-PAIRS questions 2 and 8, were strongly integrated into our educational material: hurt does not equal harm and some pain is to be expected as you recover from an episode of nsLBP; stay active if possible, moving helps reduce pain; a time-contingent approach to pacing activity can assist in functional recovery from nsLBP; maintain your usual activities; stay at work if possible; avoid prolonged bed rest. Using simple language to convey key messages closely aligns with the approach used in a population-based strategy designed to shift societal views about back pain and which had a sustained effect on PCP beliefs and stated practice behaviour 4.5 years after its cessation [64]. Linking PCP beliefs explicitly to clinical practice behaviours through the use of interactive real clinical cases appears to be a powerful strategy to facilitate adherence to guidelines, potentially more so than printed materials alone [61]. In our case studies, the use of catastrophizing language in patient interactions regarding activity was strongly discouraged, emphasizing the negative influence fear of movement (for example, associated with the catastrophizing or irrational beliefs) on predicted self-reported disability and poor behavioural performance [65]. Other questions from HC-PAIRS (Q3, Q4 and Q9) relating to pain and impairment and aspects of lifestyle moved smaller amounts (approximately 0.5 of a point), and the remaining questions showed no change (Q1, Q5, Q6, Q7, Q10, Q11, Q12 and Q13), probably reflecting the focus of the gPEP intervention towards the key knowledge and skills outlined above.

Scores on the HC-PAIRS have been reported as the only significant predictor of recommendations for work and physical activity (based on a patient vignette) when controlling for possible confounders including gender, years of experience in the treatment of back pain, judgments of severity of symptoms, and judgments of severity of pathology [57]. Our patient vignette data appear consistent with this finding. At baseline, subjects giving guideline inconsistent responses to the vignette questions appeared to gain significantly higher scores on the HC-PAIRS questionnaire (i.e.; more unhelpful beliefs in relation to current evidence regarding LBP and disability). Also, the majority of PCPs who were guideline-inconsistent for work and bed rest recommendations at pre-intervention (40% and 26% respectively), gave guideline-consistent responses at post-intervention, aligning with the shift in HC-PAIRS score. In contrast, the movement for the exercise recommendation was marginally against the recommended guidelines, with a proportion of PCPs changing from guideline consistent to inconsistent. This response may relate to PCPs' interpretation of ‘pacing’, which in gPEP was focused as a time-contingent approach [39] to exercise rather than a pain-contingent approach [66]. Based on the case studies, PCPs were encouraged to adjust activity levels if their patient used a pain-contingent ‘boom-bust’ or ‘overdo-underdo’ approach to exercise and activity (i.e. behavioral modification) and to monitor these changes, especially given that the relationship between fear of movement and function is moderated when pain persists beyond one year [67]. In this regard, a more conservative initial approach to pacing activity in patients with acute nsLBP might be expected, although a graduated increase in exercise and activity would be appropriate in the mid to longer term.

PCPs baseline self-rated knowledge and skills regarding evidence-based aspects of nsLBP varied across the questionnaire items, with the greatest percentage of clinically adequate responses documented for the question which related to the use of pharmacological options for people with nsLBP, as expected for these domain-specific components of medical care (Table 3). In contrast, the highest percentage of clinically-inadequate responses at baseline (excluding item 12 which focused on adult learning), was attributed to the following items: question 11 (45%): health professionals in the local network who include active self-management; question 3 (38%): translating evidence into clinical practice; and question 1(35%): the current evidence based guidelines for the management of nsLBP, reinforcing the need to address these aspects in our intervention. Following gPEP however, an overwhelming movement for all questions towards a more evidence-based approach consistent with a positive benefit from the program was evident, with a very modest percentage (ranging from 1–3% for items 1–9) remaining guideline inconsistent.

The gPEP questionnaire was designed to reflect the knowledge and skills considered essential to effectively implement LBP guidelines into practice and translate this information into clinical behaviours. In this regard, a flexible “whole person” approach similar to that undertaken for our cases studies, may help to moderate some of the clinical tensions associated with matching patient expectation and guideline advice [19]. This approach is possibly reflected here, in the more guideline consistent responses evident post intervention for the gPEP questionnaire. In the case studies, we encouraged PCPs to recognise and directly address concurrent patient factors which may prove to be obstacles to recovery from LBP, including patient perceptions of personal control, the acute/chronic timeline, illness identification and pain self-efficacy [68] (see Case Example S1).

Also the use of screening tools that identify an individual's risk status, and are typically based on predictive psychosocial factors such as catastophizing and depression [69], was undertaken as part of the case studies. These tools included the Orebro Musculoskeletal Pain Screening Questionnaire [45] measuring pain and disability (although the 9 item STarT Back Screening Tool [70] may be more appropriate in primary care), the Depression, Anxiety and Stress Scale (DASS21) measuring emotional functioning [46] and the painDETECT [47] screening for neuropathic pain. Screening tools were perceived as potentially time-efficient for PCPs, and allowing a more comprehensive view of pain and disability, thereby assisting PCPs in planning their management, including the need for engaging other health professionals. Along with a thorough physical examination, and respecting patient preferences, expectations, and previous management, using these screening tools can enable a more comprehensive approach to triaging patients with nsLBP in primary care. A final step in each case study was to outline an evidence-informed, multimodal management plan; one which utilised patient-focused pain education including the use of self-management strategies, pharmacological and behavioural approaches (to movement re-education, time-contingent paced activity and short term goal setting) [30], [71]. Furthermore, we propose that a key factor relevant to the shifts demonstrated on the gPEP questionnaire may relate to the clinical background of the interprofessional team, although we acknowledge that the current research design does not allow us to draw firm conclusions in this regard. Here, the gPEP team reflected a mix of clinicians (physiotherapists, clinical psychologists, rheumatologist) working in primary care (private and hospital settings) alongside specialised clinicians (pain medicine; clinical psychologists and physiotherapists; rheumatologist) working in tertiary facilities (hospitals) and clinical researchers from universities. This primary-tertiary mix, we believe, provided a well balanced and real-world perspective on clinical issues confronted in primary and tertiary care settings and these perspectives were deliberately reflected in the design of the case studies and presented using a ‘shared’ stage, in terms of time and focus.

Collectively, our findings suggest there is a matching of PCP beliefs and clinical practice behaviours. We acknowledge that the self-report measures in this study may not faithfully reflect clinical practice and that patient vignettes have limitations [72], but vignettes are also useful surrogates for understanding PCP approaches to LBP [19], [60], [61]. Clinical case presentations and interactive discussions may potentially help to bridge the gap between evidence and practice, a vital outcome given the alternative to evidence based practice is ‘anything goes’. In this context, gPEP was designed with specific practice enablers to assist PCPs in their main role of delivering an evidence-based patient-centred ‘whole person’ approach to people with nsLBP. Furthermore, the significant shift in the gPEP questions 1, 3, 4, 6, 7 and 8 are consistent with this interpretation, as these questions incorporate specific knowledge and use of practical skills in assessing and managing the multidimensional aspects of nsLBP. An important point to highlight was the bidirectional nature of the gPEP intervention, as the PCPs articulated the practical barriers to implementation such as consultation time constraints for complex pain problems and a lack of funding for integrated interprofessional care.

Furthermore, while at baseline, a small number of PCPs, never recommended a specific exercise program for acute LBP (Table 4), and a small proportion also never recommended self management, regardless of the duration of nsLBP, this was not the case at post-intervention. Following gPEP, the general move in these questions away from the ‘never’ category and towards giving advice or assistance at least some of the time, was in accordance with the patient vignette work and activity recommendations and the favourable shift in HC-PAIRS score. PCPs recommendations for the use of various strategies for managing nsLBP, were made more frequently at post-intervention for the use of a specific exercise program and the use of self management strategies. These data align with the current evidence advocating for the use of active self management as an integral part of the co-care of LBP, and as a result patient outcomes are associated with reduced disability and health care utilisation [30], [37].

The themes that emerged as most important to PCPs following gPEP, were significant in the context of implementing guideline consistent approaches to managing people with nsLBP. In this regard, the reporting of management strategies and patient education are consistent with recommendations from current LBP guidelines [48], [50], [51], [55], [56]. PCPs' improved understanding of the complexity of pain, the importance of belief systems to patient outcomes and their improved knowledge of evidence-based assessment and management all lead to their increased confidence in the skills associated with educating a patient and designing an appropriate management plan. Furthermore, the importance of networking with and using an interdisciplinary team approach as appropriate was highlighted. Given the vast majority of PCPs reported having ‘access to health professionals from other disciplines to assist [with] a team approach to [nsLBP] management’, this clinical network combined with greater confidence in their approach to nsLBP, should facilitate improved health service delivery. However, we acknowledge that the current research design does not allow us to draw firm conclusions regarding the influence of the interprofessional nature of the education intervention as a factor in the positive study outcomes. The appropriate use of psychologists and physiotherapists was frequently mentioned, possibly reflecting the multimodal management approach taken in the case studies, which included a discussion of when to refer to these health professionals and what constituted current best practice by them. The limitations of imaging (39.5%) as a theme was a potentially significant outcome given that imaging is not recommended in over 95% of nsLBP cases, except when red flags are present [50]. Despite this, considerable overuse is still documented [12]. Finally, the pharmacological options available and the appropriate use of medicines for the management of people with nsLBP were noted as important themes. In gPEP, the pharmacological management of people with nsLBP was oriented towards using any therapeutic window as an opportunity for the patient to engage in active self management. We reinforced the importance of taking a multimodal approach to nsLBP management and combining pharmacological approaches with non-pharmacological management including the use of active self management.

Our findings would be strengthened if replicated using a stronger study design such as a randomised controlled trial and expanding the study to include cost and practice effectiveness comparisons made between face-to-face and online educational interventions such as gPEP. Critically, such studies need to include the monitoring of real-world practice behaviours (including PCP referral patterns for prescriptions and radiological imaging and referrals to tertiary facilities). However, it is also clear that the implementation and uptake of clinical guidelines in clinical practice is problematic. This issue highlights the similar factors that complicate such real-world clinical research where time and cost constrain what is possible, sustainable and optimal in high quality research with what is possible, sustainable and optimal for real-world clinicians. Furthermore, the applicability of our findings to other populations of PCPs managing people with persistent LBP may be limited because of the following factors: PCP data were based on self-report measures and these measures may serve to over-estimate the actual change in real practice; responder bias (unlikely here as a good response rate was achieved); and selection bias (PCPs self-referred to gPEP and their motivations for attending this educational intervention may differentiate them from other PCPs). While HC-PAIRS has undergone the most thorough testing to date of any tool for the measurement of health care professionals' attitudes and beliefs to LBP, gaps in the properties of this tool remain, particularly test-retest reliability and responsiveness [73]. The current research design does not allow us to draw firm conclusions regarding the role of the interprofessional nature of the education intervention as a factor in the positive study outcomes.

The adoption of a health policy framework can help when implementing an evidence-based model of care for the management of low back pain in primary care, and has shown promising outcomes in this before/after study with volunteer PCPs. We propose that using a contemporary biopsychosocial perspective of pain combined with evidence-informed knowledge and practical skills delivered by an interprofessional team may be an effective strategy to increase the uptake of clinical guidelines. To better manage the complexities experienced by people with persistent low back pain, we argue for the use of a whole person engagement model. Additionally, both health professionals and people with persistent low back pain may have to modify their expectations from treatment and focus more on the role of care rather than cure.

Supporting Information

Acknowledgments

The following authors received an honorarium* for developing and delivering the gPEP programs: SD, SS and JQ. The contribution of the following people, to the development and delivery of the educational materials, is gratefully acknowledged: Tracy Antill, Nicky Fortescue, Carl Graham, Felicity Kermode, Paul Knight, Luke Parkitny and Eric Visser. The support of the Royal Australian College of General Practitioners (WA chapter), in particular Alistair Vickery in the review of educational materials and Beth McEwan with accreditation, is kindly acknowledged. The Pain Medicine Unit, Fremantle Hospital and Health Service, is kindly acknowledged for use of their patient pain information booklets and Penny Hamilton for her assistance with preparation of these materials. The Fremantle and Osborne (now North Metropolitan Medicare Local) GP networks provided invaluable support and assistance with recruitment, collection of post course responses and compliance with Royal Australian College of General Practitioners requirements for awarding of Quality Assurance & Continuing Professional Development points. We are grateful to Dr Jim Codde (Department of Health WA) for advice on the experimental design and Chris Perrey (Curtin University) for data entry. The support of Pfizer Australia in funding the food and non-alcoholic beverages for the workshops is acknowledged. No input from Pfizer was provided regarding any aspect of the research including the study design, protocol, the data collection, analyses, or manuscript preparation or input regarding any of the educational materials.