Quality in Primary Care Open Access

Keywords

chronic pain management, Community Health Centers, evidence-based medicine, primary health care

Introduction

Chronic pain management is a challenging and important element of primary care. Approximately half of all patients suffering fromchronic pain are cared for by primary care providers (PCPs).[1] However, evidence suggests that primary care providers are not well equipped to manage chronic pain effectively. Most PCPs express low confidence in their ability to effectively manage pain,[2–5] and pain management education is scant to non-existent in most training programmes.[6,7] Studies suggest that in primary care there is wide variation in the use of opioids[8–11] and in the adherence of PCPs to guidelines for documentation and management of pain.[9,12] With increasing evidence of the potential harm caused by overuse and misuse of opioids, strategies are clearly needed to promote the safe, effective management of chronic pain in primary care.

Effective models for pain management have been developed and implemented. The Stepped Care Model for Pain Management (SCM-PM) is an evidencebased model advocated by the American Academy of Pain Medicine.[13] It emphasises an individualised approach to managing pain in three steps, beginning with primary care. Step 1 involves the primary care physician identifying and discussing the patient’s pain concerns and developing a treatment plan emphasising self-management and primary-care-based interventions. Step 2 involves more active, collaborative treatment, including psycho-educational assessment and intervention, medication, and consultations with specialists from appropriate disciplines. Depending on the setting, Step 2 interventions can often be delivered on-site within a primary care delivery system. Step 3 targets patients with chronic pain requiring significantly more care and involvement from other members of a pain management team.[14] Implementation of this model across five Veterans Health Administration (VHA) primary care clinics resulted in improved pain-specific outcomes.[15] While the SCM-PM has been shown to be successful within the VHA, it has not been applied to other healthcare settings, such as Community Health Centers. Project STEP-ing Out is a 3-year quality improvement project aimed at improving the care of medically underserved patients suffering from chronic painful conditions such as chronic low back pain, myofacial pain syndromes, and other musculoskeletal conditions, cared for at a large, multisite, statewide Community Health Center. The project focuses on adapting the SCM-PM to the local context of the health centre and maximising access and adherence to evidence-based care. As part of the project we conducted a comprehensive, formative assessment of pain management at the health centre and used the findings to develop a multifaceted intervention to improve operational and patient-centred outcomes. In this paper we present the results of our assessment and describe how they are being used to inform the implementation of a comprehensive pain quality improvement strategy.

Methods

Community Health Center Inc. (CHCI) is a multisite FederallyQualifiedHealthCenter located in Connecticut. CHCI provides comprehensive primary care services in 12 primary care health centres across the state. Additional sites of care include school-based clinics, homeless shelters, and mobile dental sites. CHCI cares for over 100 000 medically underserved patients in the state. Over 60% of CHCI patients are racial/ethnic minorities; over 90% are below 200% federal poverty level, 60% are on Medicaid or state insurance, and 22% are uninsured.

Each CHCI patient has a designated primary care provider. Primary care is delivered by a team comprised of nurses, medical assistants, and primary care providers, including internists, paediatricians, and family practitioners as well as family practice nurse practitioners. These frontline primary care teams are supported by on-site behavioural health providers. Specialist care, including pain-related specialties, is available through outside consultation, but many patients face substantial barriers to securing such specialist care. All patient care at CHCI is documented in an integrated electronic health record system.

Project STEP-ing Out: This project, so named to reflect the project’s goal of taking the VHA’s Stepped Care Model outside the walls of the VA health system, is a 3-year quality improvement initiative, usingSCMPM as a conceptual model, aimed at improving chronic pain management in primary care. The project includes an assessment phase to evaluate current pain care management by PCPs, appropriateness of the SCM-PM, the organisation’s readiness to change and its capacity to implement the new model. The results of the assessment will be used to help design a quality improvement intervention in the second phase and will also serve as a baseline fromwhich to evaluate the impact of the intervention in the final phase. Improvement will be measured by evaluating the impact of the intervention on provider pain management knowledge and adherence to standards of care as well as selected patient pain-specific outcomes. In the third phase, we will concentrate on integration and sustainability of the model. This study was reviewed and approved by the Institutional Review Board of the Community Health Center, Inc.

Promoting Action on Research Implementation in Health Services (PARIHS) Framework: We chose the PARIHS theoretical framework to design the formative assessment. This framework has been successfully utilised by practices to identify critical elements and to design an appropriate implementation strategy for evidence-based practice.23 PARIHS defines context, evidence and facilitation as the three key interacting elements determining success of an implementation. 16–22 Using these elements and their respective subelements,weselected a variety of assessment methods to evaluate the areas that were deemed important and relevant to the project (Table 1). Data on context and evidence were collected from the electronic health records system and staff surveys. For the third element, facilitation, a robust facilitation plan was developed to be employed in the implementation phase of the project.

Table 1: PARIHS Framework Application for Project STEP-ing Out: SI = function of E,C,F

Identification of patients with chronic pain

Identifying patients with chronic pain is challenging. Pain is measured subjectively and treated in myriad ways. We deliberately avoided using the few specific ICD-9 codes for chronic pain because of the wide variation in the use of coding to record the presence of various disease states and painful conditions. We identified a cohort of patients with chronic pain using two methods of identification. One involved patient self-reported pain scores and the other involved opioid prescriptions.

Allmedical assistants at CHCI are trained to include the pain score as part of the vital signs process. At every visit patients are asked to rate their level of pain on a scale from one to ten, with 10 indicating the highest level of pain; pain scores are recorded in the electronic health record. The patients in the ‘pain score cohort’ in this study were chosen according to the following set of criteria: (a) age > 18 years; and (b) two or more pain scores of 4 or greater (moderate to severe pain) separated by 90 days or more during the measurement timeframe.

The criteria for identifying the ‘chronic opioid cohort’ included: (a) age > 18 years; and (b) receipt of prescription opioid medications for a total of 90 days or more during the measurement timeframe.

It was judged that using pain scores would identify a large number of patients with chronic pain, but with less specificity, while using the opioid-prescribing method would identify chronic pain patients with more certainty but exclude large numbers of patients not managed with such medications. All analyses were conducted on both cohorts.

Data preparation and processing

CHCI uses the eClinicalWorks (ECW) Electronic Medical Record (EMR) in conjunction with GE Centricity Practice Solution for billing and scheduling management. All electronic data were retrieved from these two systems, de-identified, and analysed by the study team. For our analyses we used data that were documented in theEMRby CHCI clinical providers as part of routine clinical practices at CHCI during a 1- year time period (1 March 1 2010 through 28 February 2011). All database retrieval queries were validated by random chart reviews of at least 25 records. Data elements included the patient’s primary care provider name, their demographics, self-reported pain scores, medication-prescribing records, laboratory results, opioid agreement use, and behavioural health and medical referrals. The primary care provider panel to which each patient belongs is assigned in the EMR for each patient. While patients may see other providers occasionally for urgent visits if their own provider is unavailable, the identified primary care provider is responsible for all longitudinal care, and pain management issues are handled by the patient’s assigned primary care provider. When a provider leaves CHCI, their patients are assigned to a new provider who assumes their care, including their pain management.

Survey instruments and administration

The Organisational Change Manager (OCM) is a validated survey designed to measure subjective probability estimates for predicting success and failure of healthcare improvement projects.24 The survey assesses staff perception of the effectiveness of the following domains: the project start-up phase, the exploration of the stated problem, team solution development, and the process of implementation and testing. It is intended to be administered longitudinally over the course of the duration of a project. Many of the questions contained in the OCM map closely to elements of the PARIHS framework. The OCM survey was distributed at an agency-wide Grand Rounds conducted by video conference. All CHCI primary care providers, behavioural health providers and nurses who care for adult patients were eligible to complete this anonymous survey. A study information sheet was provided to each potential participant. Informed consent was indicated by the participant’s completion of the survey.

The KnowPain-50 Survey (KP50) is a validated tool for assessing physician knowledge regarding pain management25 in the following domains: initial pain assessment, defining treatment goals and expectations, development of a treatment plan, implementation of a treatment plan, reassessment and management, and management of environmental issues. The KP50 con tains multiple-choice questions as well as questions scored on a Likert scale. With five points given for each correct answer, the total possible score for the survey is 250. To gain contextual information on primary care providers’ attitudes and beliefs regarding pain care and the STEP-ing Out Project we added 11 survey questions taken from a VHA-developed survey2 to the KP50 (Box 1).

Box 1: Additional survey questions

All CHCI primary care providers who care for adult patients (internists, family physicians, family nurse practitioners) were offered the opportunity to complete the KnowPain-50 survey as well as the 11 questions from the VA-developed survey. The survey questions were distributed during weekly staff meetings. A study information sheet was provided to each potential participant. Informed consent was indicated by the participant’s completion of the survey.

Chart reviews: Manual chart reviews were conducted to determine the extent to which primary care providers were adhering to current guidelines for documenting care for patients with chronic pain. The review process used a detailed abstraction protocol developed by pain specialists at the VHA. Reviewed were medical visit records from the two chronic pain patient cohorts – the ‘pain score cohort’ and the ‘chronic opioid cohort’. Using a random number generator 150 records were chosen from each cohort. All records were reviewed by a trained research assistant; 5% of records were randomly spot-checked by the project director to ensure data integrity and accuracy

Multimodal care: To gauge the extent to which patients with chronic pain were receiving multimodal care that included additional pain-related specialists we evaluated referrals generated through the EMR. All CHC referrals are ordered electronically and managed by a central referral team. We queried the database for all referrals for patients with chronic pain during the measurement timeframe to pain-related disciplines including orthopaedics, rheumatology, pain management, physical therapy, psychiatry, chiropractic, acu acupuncture and all behavioural health disciplines, including addiction medicine.

Opioid prescribing: All data on opioid prescribing were collected from the electronic prescribing record within the EMR. All medications belonging to the opioid family were included in the analysis except suboxone, which is not used for pain management at CHCI. The medication data were cleansed extensively to adjust for variation in free text charting, particularly in data fields containing medication frequency and dosages. For example, medication frequency indicating that a patient was taking a medication three times daily might have been written as ‘TID’, ‘Q8hrs’, or ‘three times daily’. All possible frequency notations were analysed and combined to ensure an accurate assessment of actual medication use frequency. All decisions were made by the research team and reviewed by a senior clinician to ensure accuracy.

Chronic opioid use was defined as the use of any individual or combination of opioids for 90 days or more during the measurement year. For prescriptions without an identified duration, such as ‘14 days’ or ‘one month’, the duration was calculated by dividing the total number of pills dispensed by the daily frequency ordered. For prescriptions written as ‘PRN’ (as needed), the maximum frequency indicated was used in this calculation. For example, an opioid prescription with instructions to take two pills BID, dispense 60 pills would account for 15 days of opioid use. A prescription with instructions to take one pill BID PRN, dispense 40 would account for 20 days of opioid use.

Results

Patient demographic characteristics

In both cohorts patients were predominantly female, aged 30–60 years, and covered by Medicaid insurance (Table 2). Hispanic/Latino patients accounted for a larger percentage in the pain score cohort than in the chronic opioid cohort. Among patients with two or more pain scores
4, 40% were Hispanic/Latino, and 43% were Caucasian. By contrast, in the chronic opioid cohort, only 25% were Hispanic/Latino, and 67% were Caucasian.

Table 2: Demographics

Frequency and utilisation

Chronic pain was highly prevalent in the adult CHCI patient population. Patients with pain accounted for a substantial portion of primary care providers’ daily schedules. There were 33,254 adult patients treated in primary care during the measurement year, accounting for 138,614 primary care visits. During 36% of these visits, the patient provided a pain score of 4 or greater (moderate pain) and for 17% of the visits, the pain score offered by the patient was 8 or greater (severe pain). Patients in the ‘pain score cohort’, defined as having two or more pain scores > 4 separated by 90 days during the measurement year accounted for 59% of all adult visits. Patients in the ‘chronic opioid cohort’, defined as receiving opioid medications for 90 days or more (chronic opioids), accounted for 11% adult primary care visits. Patients in both cohorts had an average of 12 and 15 visits to primary care, respectively, compared with six visits on average for all adult primary care patients. These results suggest that a significant portion of visits in primary care involve a relatively small, select population of patients with chronic pain, and that patients with chronic pain utilise primary care services at a substantially higher rate than patients without pain.

Survey results

There were 127 OCM Surveys distributed and 54 (43%) collected for analysis. Results indicate a range of responses suggesting areas of strength as well as weakness (Table 3). Responses suggest that staff perceived strong advantages to the proposed project and felt that leadership had been effective at communicating a mandate for change and establishing a climate conducive to making change. Low scores suggest that primary care staff were not confident in their ability to develop new skills related to the project, and that they question whether sufficient money, time and personnel will be allocated to accomplish the project. Staff gave an intermediate score, 47%, on how well the intervention fitted with the agency’s culture and practice.

Table 3: Staff-perceived readiness for organisational change (n = 53)

The KnowPain-50 was distributed to 58 primary care providers at 11CHCI practice sites, and 47 (81%) were returned for analysis. Results demonstrated a wide distribution of scores, with an agency-wide average score of 150 out of a total possible score of 250. These results were similar to those seen in other primary care practices25 and suggest an opportunity to improve pain care knowledge (Figure 1).

Figure 1: Average CHCI KP50 baseline score comparison (2 standard deviations)

Responses to the 11 additional survey questions added to the KP50 to assess attitudes and beliefs about pain management issues provided important additional information regarding primary care providers’ perspectives about pain care (Figure 2). Providers placed a high priority on being able to provide effective pain care but expressed dissatisfaction with pain management resources and support. In contrast to other studies in primary care,6,10,26,27 CHCI providers generally did not cite fear of regulation, side effects, or fear of contributing to addiction as factors affecting their care for patients with pain.

Figure 2: CHCI Clinician Attitudes Survey score (n = 47)

Chart reviews

Chart reviews demonstrated substantial gaps in adherence to standards of care for documentation and management of chronic pain. Although presence, cause and source of pain were documented fairly consistently, a functional assessment was rarely documented. In addition, follow-up assessments for pain at subsequent visits were infrequently performed for both cohorts. Pain treatment plans were noted as being present in 81% and 96% of charts reviewed for patients in the ‘pain score’ and ‘chronic opioid’ cohorts, respectively (Figure 3). For patients in the pain score cohort, 36% of documented treatment plans contained a pain medication prescription only, and for patients in the chronic opioid cohort, 69% of treatment plans documented contained medication prescribing as the only element. Treatment plans rarely included patient education or referral to other specialists. Patients taking opioid medications chronically had a documented opioid agreement 68% of the time, and a documented toxicology screening test 66% of the time.

Figure 3: Chart review of adherence to pain care documentation standards (n = 150 per cohort)

Multimodal care

Referral to other pain-related specialties or to behavioural health was fairly uncommon. Referral rates are low for all specialties. Figure 4 shows the percentage of patients in each cohort that were referred to specialties including physical medicine and rehabilitation, orthopaedics, painmanagement, physical therapy and chiropractic services. Despite the existence of on-site behavioural health in individual and group formats at nearly allCHCI sites, fewer than 25% of patients in the two cohorts had been seen by an onsite behavioural health provider.

Figure 4: Specialty referrals for patients with pain

Opioid prescribing

Opioid prescribing varied widely across the agency. The number of patients in each provider’s panel being prescribed chronic opioids ranged from0 to 110 out of an average patient panel size of 803. From these numbers we calculated the ‘percentage panel on chronic opioids’ (PPCO) to further explore this variability. The PPCO is defined as the total number of patients in a provider’s panel receiving 90 days or more of an opioid medication, over the total number of adult patients in that provider’s panel. Results ranged from 0–10% (Figure 5). A wide range of opioids were prescribed, with oxycodone and hydrocodone the most common (Table 4).

Figure 5: Number of individual patients prescribed 90 days ormore of opioid medications in 1 year, by provider

Table 4: Frequency of individual opioid prescriptions

Discussion

The data from this detailed assessment of pain care at CHCI provided extremely useful information to inform the development and implementation of Project STEP-ing Out. Many of the results from this baseline assessment, such as providers’ knowledge, attitudes and adherence to standards of pain care will serve as outcome measures with which to evaluate the impact of the intervention. In the SCM-PM, the primary care provider provides pain care management for the majority of patients with pain (Step 1). One of the principal findings from our assessment is primary care providers’ need for additional knowledge and skill in pain management. The KnowPain 50 showed variability in pain knowledge and an average score of 60% of the total possible score. These findings are consistent with the growing literature suggesting the need for enhanced competency for pain management in primary care.6,7,9,28,29 The additional questions assessing attitudes and beliefs regarding pain management suggested that providers have low confidence in their ability tomanage pain, and in the Organisational Change Figure 3 Chart review of adherence to pain care documentation standards (n = 150 per cohort) Figure 4 Specialty referrals for patients with pain Manager, respondents expressed fear that they would not be able to develop the new skills needed for the intervention. Based on these findings, Project STEPing Out will focus substantial resources on skill building and knowledge acquisition. Agency-wide education on pain care will be provided in a variety of formats and made a standard requirement for all primary care providers.

An additional important finding was the relative paucity of referrals to pain-related specialists and behavioural health providers. The SCM-PM calls for multimodal care involving specialists in behavioural health, physical medicine and complementary/alternative medicine (CAM) for patients with pain that is not improving with primary care management. Patients with publicly funded insurance or without medical insurance face significant barriers to accessing specialty care, which may partially explain these findings. However, a survey in the VA system in which patients have ready access to such specialists found equally low rates of multimodal care.9 Primary care provider survey responses suggest dissatisfaction with resources available to support them in caring for patients with chronic pain. Based on these findings, project STEP-ing Out is focusing on creative ways to provide more multimodal support, including the use of video conferencing, on-site CAM offerings, and building internal capacity to provide behavioural health interventions focused on pain. In particular, the project is planning to provide primary care clinicians with access to specialist consultation via video conferencing, using the Project Extension for Community Healthcare Outcomes (ECHO) model.30,31

Opioid treatment agreements and urine toxicology monitoring for patients using opioids chronically are supported by evidence32 and part of pain care practice guidelines.13 We found moderate rates of adherence to these practices. Adherence to standards of care for documentation of pain care was also poor. These findings led to our placing greater emphasis in the implementation on the use of templates in the electronic health record, and the creation of a new standardised follow-up protocol. The use of clinical registries and performance report cards will be used to provide audit and feedback to individual providers on these elements of the intervention as well.

The PARIHS framework provided a useful theoretical construct for the formative assessment in preparation for a challenging project implementation. We observed limitations to operationalising PARIHS that are similar to those observed by others.23Westruggled with the lack of clarity and substantial overlap in the various subelements in the Evidence and Context domains. In particular we found it challenging to map PARIHS variables to specific tools, and struggled to choose appropriate tools that were both effective and practical given limited resources and the busy nature of the practice we were studying.

Our approach has several strengths. With a fully integrated electronic health record we were able to access a large amount of data and information on pain care including labs, medications, referrals and utilisation. This detail was critical for understanding the local context, the practice patterns at each individual site, and the patient population potentially being impacted on by the implementation. Using a model that has been implemented and tested in the VHA, a health system with some similarities to CHCI added credibility and promoted acceptance of the new model by the CHCI staff. While there were differences between VHA patients and CHCI patients, both systems emphasised comprehensive primary care, use of an electronic health record, and care for patients at the lower end of the socio-economic spectrum.

One of the principal weaknesses of our approach was the difficulty in accurately identifying patients with chronic pain. Identifying patients with chronic pain using pain scores is problematic. Pain scores are self-rated and have been shown to be poorly accepted by primary care providers and of limited accuracy.33 Our use of two or more pain scores >4 likely substantially overcounted the chronic pain cohort. Our second approach, using opioid use as a marker for chronic pain, is likely to be more specific, but less sensitive, missing patients using non-opioid medications or non-pharmacologic treatments. We chose not to use ICD9 codes, due to the poor reliability of coding and the lack of a specific code or set of codes for chronic pain.

An additional weakness was the limited use of assessment tools and other qualitative methods of evaluation. We administered three different survey instruments but relied heavily on the OCM for many elements of the PARIHS Context domain. The OCM had the lowest response rate (43%) of all the surveys employed. This low response rate limits our ability to interpret staff perceptions regarding the implementation of our project. Additional survey tools and a higher response rate would have strengthened our assessment. Work published subsequent to our evaluation has identified a toolkit of assessment toolsmapped to specific PARIHS subdomains.34 These tools would have been extremely helpful for this project.

Conclusion

Frontline health systems increasingly face the need to redesign practices and implement new models and system of care. Innovations such as electronic health records, Patient-Centred Medical Homes, Accountable Care Organisations, and a growing emphasis on outcomes-driven, evidence-based care will all require systems redesign and large-scale implementations. Designing an implementation based on evidence-based practice, a strong conceptual model and a thorough assessment of local context is critical to maximise the likelihood of implementation success.

Funding

This study was supported by a grant from the Mayday
Fund.

Ethical Approval

Not required as this was a quality improvement
project.

Peer Review

Not commissioned; externally Peer Reviewed.

Conflicts of Interest

None.

References

1. Breuer B, Cruciani R and Portenoy RK. Pain management by primary care physicians, pain physicians, chiropractors, and acupuncturists: a national survey. Southern Medical Journal 2010;103(8):738–47.
2. Dobscha SK, Corson K and Flores JA. Veterans Affairs primary care clinicians’ attitudes toward chronic pain and correlates of opioid prescribing rates. Pain Med 2008;9(5):564–71.
3. Matthias MS, Parpart AL and Nyland KA. The patient– provider relationship in chronic pain care: provider0s perspective. JAMA 2010;11:1688–97.
4. Upshur CC, Luckmann RS and Savageau JA. Primary care provider concerns about management of chronic pain in community clinic populations. J Gen Intern Med 2006;21(6):652–5.
5. Barry DT. Opioids, chronic pain, and addiction in primary care. J Pain 2010;11(12):1442–50.
6. Green CR, Wheeler JR, Marchant B, LaPorte F and Guerrero E. Analysis of the physician variable in pain management. Pain Med 2001;2(4):317–27.
7. Ponte C and Johnson-Tribino J. Attitudes and knowledge about pain: an assessment of West Virginia family physicians. Fam Med 2005;37(7):477–80.
8. Hutchinson K, Moreland AME, de C Williams AC, Weinman J and Horne R. Exploring beliefs and practice of opioid prescribing for persistent non-cancer pain by general practitioners. European Journal of Pain 2007; 11(1):93–8.
9. Phelan SM, van Ryn M, Wall M and Burgess D. Understanding primary care physicians0 treatment of chronic low back pain: the role of physician and practice factors. Pain Med 2009;10(7):1270–9.
10. Bhamb B, Brown D, Hariharan J, Anderson J, Balousek S and Fleming MF. Survey of select practice behaviours byprimary care physicians on the use of opioids for chronic pain. Curr Med Res Opin 2006;22(9):1859–65.
11. Potter M, Schafer S, Gonzalez-Mendez E et al. Opioids for chronic non-malignant pain. Attitudes and practices of primary care physicians in the UCSF/Stanford Collaborative Research Network. University of California, San Francisco. J FamPract 2001;50(2):145–51.
12. Zubkoff L, Lorenz KA, Lanto AB et al. Does screening for pain correspond to high quality care for veterans? J Gen Intern Med 2010;25(9):900–5.
13. Dubois MY, Gallagher RM and Lippe PM. Pain medicine position paper. Pain Med 2009;10(6):927–1000.
14. James LC and Folen RA (eds). The Management of Chronic Pain in the Primary Care Setting. Primary care clinical health psychology: A model for the next frontier. Washington DC: American Psychological Association Press; 2005.
15. Dobscha SK, Corson K and Perrin NA. Collaborative care for chronic pain in primary care: a cluster randomised trial. JAMA 2009;301(12):1242–52.
16. Kitson A, Harvey G and McCormack B. Enabling the implementation of evidence based practice: a conceptual framework. Qual Health Care 1998;7(3):149–58.
17. McCormack B, Kitson A, Harvey G, Rycroft-Malone J, Titchen A and Seers K. Getting evidence into practice: the meaning of 0context0. J AdvNurs 2002;38(1):94–104.
18. Harvey G, Loftus-Hills A, Rycroft-Malone J et al. Getting evidence into practice: the role and function of facilitation. J AdvNurs 2002;37(6):577–88.
19. Rycroft-Malone J, Seers K, Titchen A, Harvey G, Kitson A and McCormack B. What counts as evidence in evidence-based practice? J AdvNurs 2004;47(1):81–90.
20. Kitson AL, Rycroft-Malone J, Harvey G, McCormack B, Seers K and Titchen A. Evaluating the successful implementation of evidence into practice using the PARiHS framework: theoretical and practical challenges. Implement Sci 2008;3:1.
21. Rycroft-Malone J, Harvey G, Seers K, Kitson A, McCormack B and Titchen A. An exploration of the factors that influence the implementation of evidence into practice. J ClinNurs 2004;13(8):913–24.
22. Rycroft-Malone J. The PARIHS framework – a framework for guiding the implementation of evidence-based practice. J Nurs Care Qual 2004;19(4):297–304.
23. Helfrich CD, Damschroder LJ, Hagedorn HJ, Daggett GS, Sahay A, Ritchie M, et al. A critical synthesis of literature on the promoting action on research implementation in health services (PARIHS) framework. Implement Sci 2010;5:82.
24. Gustafson DH, Sainfort F, Eichler M, Adams L, Bisognano M and Steudel H. Developing and testing a model to predict the outcomes of organisational change. Health Serv Res 2003;38(2):751–76.
25. Harris JM, Jr, Fulginiti JV, Gordon PR et al. KnowPain- 50: a tool for assessing physician pain management education. Pain Med 2008;9(5):542–54.
26. Nwokeji ED, Rascati KL, Brown CM and Eisenberg A. Influences of attitudes on family physicians0 willingness to prescribe long-acting opioid analgaesics for patients with chronic non-malignant pain. ClinTher 2007;29 Suppl:2589–602.
27. Potter M, Schafer S and Gonzalez-Mendez E. Opioids for chronic non-malignant pain: attitudes and practices of primary care physicians in the UCSF/Stanford Collaborative Research Network. J FamPract 2001; 50(2):145–51.
28. Turk DC, Brody MC and Okifuji EA. Physicians0 attitudes and practices regarding the long-term prescribing of opioids for non-cancer pain. Pain 1994;59(2):201–8.
29. Morley-Forster PK, Clark AJ, Speechley M and Moulin DE. Attitudes toward opioid use for chronic pain: a Canadian physician survey. Pain Res Manag 2003;8(4): 189–94.
30. Arora S, Kalishman S, Dion D et al. Partnering urban academic medical centres and rural primary care clinicians to provide complex chronic disease care. Health Aff (Millwood) 2011;30(6):1176–84.
31. Arora S, Thornton K, Murata G, et al. Outcomes of treatment for hepatitis C virus infection by primary care providers. N Engl J Med 2011;364(23):2199–207.
32. Starrels JL, Becker WC, Alford DP, Kapoor A, Williams AR and Turner BJ. Systematic review: treatment agreements and urine drug testing to reduce opioid misuse in patients with chronic pain. Ann Intern Med 2010; 152(11):712–20.
33. Krebs EE, Carey TS and Weinberger M. Accuracy of the pain numeric rating scale as a screening test in primary care. J Gen Intern Med 2007;22(10):1453–8.
34. Stetler CB, Damschroder LJ, Helfrich CD and Hagedorn HJ. A guide for applying a revised version of the PARIHS framework for implementation. Implement Sci 2011; 6:99.
35. Nelson EC, Batalden PB and Godfrey MM. Quality By Design,AClinical Microsystems Approach. San Francisco: Jossey-Bass; 2007.