Self-reported practices in opioid management of chronic noncancer pain : A survey of Canadian family physicians

1Continuing Medical Education; 2Department of Community Health and Epidemiology; 3Department of Emergency Medicine; 4Department of Anesthesia, Dalhousie University, Halifax, Nova Scotia; 5Institute for Work and Health, Toronto Rehabilitation Institute; 6Department of Medicine, University of Toronto, Toronto; 7Rehabilitation Services, Woodstock General Hospital, Woodstock, Ontario Correspondence: Dr Michael JM Allen, Continuing Medical Education, Dalhousie University, 5849 University Avenue, PO Box 15000, Halifax, Nova Scotia B3H 4R2. Telephone 902-494-2173, fax 902-494-1479, e-mail michael.allen@dal.ca Chronic noncancer pain (CNCP) is a major health problem, the estimated prevalence of which varies according to methodology and settings. Recent data indicate that 25% of the general adult population (1,2) and 40% of seniors living in institutions (2) are affected by CNCP. Opioids are frequently prescribed to decrease pain and improve function in patients with CNCP (3). While evidence for the long-term efficacy of opioids in treating CNCP is weak, over the past several years, there has been a trend toward increased prescribing of opioids, particularly oxycodone and fentanyl. This trend has occurred in several countries (4-6), including the United States (7) and Canada (8), and has been accompanied by an increase in reported opioid abuse and deaths (8-11). In 2007, the medical regulatory authorities of all Canadian provinces formed the National Opioid Use Guideline Group (NOUGG). NOUGG developed an evidence-based national ‘Guideline for the Safe and Effective Use of Opioids for Chronic Noncancer Pain’ that was released in early 2010 (12). The Canadian Guideline provides a consistent, evidence-based approach to managing CNCP patients with opioids. It will be important to assess changes in family physicians’ (FPs) prescribing of opioids for CNCP patients following the release of the guideline. This requires some knowledge of FPs’ practices before the guideline was widely disseminated. However, there are little data on opioid prescribing practices of Canadian FPs, being limited to a total of 219 respondents in three studies (1,13,14). Several Canadian and American surveys have found that approximately 30% of FPs do not prescribe opioids for CNCP (13-15), and that FPs are more cautious with prescribing strong opioids than weak opioids (15,16). Factors originAl ArtiCle

Working in conjunction with the team that developed the Canadian guideline, the present study provides a baseline assessment of opioid prescribing practices before the release of the guideline by surveying FPs across the country.Drawing on the Canadian guideline as the gold standard, the present study examined two main questions: How consistent are FPs' practices and knowledge in prescribing opioids for CNCP relative to the Canadian guideline; and what factors hinder or enable FPs in their prescribing of opioids for CNCP?While we recognized that FPs did not have access to the guideline, we believed it was likely that some recommendations were already being followed because they were already considered best practices.

Questionnaire design
The survey questions were developed with reference to the recommendations of the Canadian Guideline.NOUGG provided access to the guideline recommendations before its release on May 3, 2010, for the sole purpose of designing the survey.The guideline itself was released by posting it on the website of the National Pain Centre at McMaster University (Hamilton, Ontario) (12) and through publication in the Canadian Medical Association Journal (20).Some questions differentiated between weak opioids (codeine, tramadol, pentazocine, propoxyphene and meperidine, with or without acetylsalicylic acid or acetaminophen) and strong opioids (morphine, oxycodone, hydromorphone, fentanyl and methadone).The online survey questions were tested for face and content validity by members of NOUGG (n=3), pain specialists (n=2), FPs (n=4) and information technology specialists (n=2).Their comments and suggestions were reviewed by a team consisting of the lead author (MA), an author who was involved in developing the guideline (AF), a pain specialist (PM), a methodologist (MA) and a medical resident (OT).Modifications to the survey were made based on the feedback received.The survey was available in French and English and was accessible from March 30, 2010, to July 10, 2010.The goal was to have the survey accessible before the release of the guideline and it was left open to obtain as many responses as possible.There was no incentive for completing the survey, which is presented in Appendix 1.The Dalhousie University Research Ethics Board (Halifax, Nova Scotia) approved the project.

Data collection
The present cross-sectional descriptive study used Opinio (21), an online survey program hosted at Dalhousie University.The study population included FPs who manage patients with CNCP, who were registered with the College of Family Physicians of Canada and who practiced in any Canadian province.To invite FPs to complete the survey, the College of Family Physicians of Canada, the provincial medical regulatory authorities and university continuing professional development offices sent e-mails and electronic newsletters with embedded links to the survey to their FP constituents.There were variations in the number and type of contacts made with FPs (Appendix 2).The invitation and introduction to the survey specified that FPs who do not manage patients with CNCP should not participate.There are approximately 32,000 FPs in Canada (22) but the number that received an invitation to complete the survey is unknown because not all FPs may have received and opened their e-mails or electronic bulletins.Given the lack of a discrete sampling frame and the varied methods of contacting FPs, a nonprobability convenience sample was obtained.

Data analysis
Questions regarding FPs' practices listed recommended practices and asked respondents how frequently they performed them (never, <25% of patients, <50% of patients, >50% of patients, >75% of patients, always).For these questions, the percentage of respondents performing these practices are reported in three categories: never and <25% of patients; 25% to 50% of patients; and >75% of patients and always.
Questions regarding FPs' knowledge asked respondents if they agreed, disagreed or had no opinion about various statements.Questions regarding barriers and enablers to prescribing opioids asked respondents to rate the importance of various factors on a 5-point Likert scale (1 = not very important, 5 = very important).For each factor, the per cent of response is reported in three categories: 1 and 2 (not important); 3 (neutral); and 4 and 5 (important) on the 5-point scale.Analysis was performed using PASW Statistics version 18.0.2(IBM Corporation, USA).

Responses
After excluding respondents who were not primary care physicians, 710 responses were received for analysis (701 English and nine French).Responses according to province were: Ontario, n=367; British Columbia, n=79; Nova Scotia, n=71; Saskatchewan, n=30; Alberta, n=26; and Newfoundland and Labrador, n=24.The remaining provinces had <10 responses each and n=85 respondents did not indicate their province of practice.
Three respondents were excluded because they were not primary care physicians (internist, internal medicine resident and oncologist).Family medicine residents (n=2) and FPs with special interests, such as emergency medicine, psychotherapy, palliative care and anesthesia, were included in analysis.It is not possible to determine a precise response rate because this was a convenience sample with no formal sampling frame to draw on.Demographic and practice variables for all FPs are shown in Table 1.Not all respondents answered all questions.

knowledge of opioids
Table 2 shows responses (disagree/agree/no opinion) to knowledge questions, with correct answers in bold.Generally, there were marked knowledge gaps in most responses, with two exceptions.Responses were largely correct concerning randomized controlled trial evidence for short-term effectiveness of opioids in CNCP, and with respect to the restoration of function being a more important indicator of opioid effectiveness than pain relief.
Table 3 reviews recommended physician practices before starting a patient on opioids.Twelve practices were listed, two of which were distracters and not included in the guidelines.The three recommended practices most frequently reported were explaining the potential harms and benefits of long-term opioid therapy and assessing patients' level of function.The two distracters were the practices least frequently performed by FPs.
Table 4 reviews recommended practices while monitoring patients on opioids.As above, the three most frequently reported practices while monitoring patients on opioids all concerned patient safety: observe for aberrant drug-related behaviour; assess for adverse effects; and advise caution while driving or operating machinery.The practices least frequently performed by FPs were urine drug screening and discontinuing opioids because of insufficient pain relief.
The Canadian Guideline introduced the term 'watchful dose' of opioids -the daily dose at which patients may need to be reassessed or more closely monitored.Only 5% (n=10) of respondents correctly identified the 'watchful dose'.Nearly one-half had no opinion (n=147) and 45% (n=143) underestimated the watchful dose of 200 mg morphine equivalent (MEQ) recommended by the guideline.Thirty-eight per cent of respondents (n=211) correctly identified the minimum daily dose of opioid a patient should be taking before receiving the fentanyl patch (60 mg MEQ).Twenty-nine per cent (n=158) indicated that there is no minimum dose and that the amount varies with the patient's condition, and an additional 14% underestimated the minimum dose.

Barriers to and enabling factors for prescribing opioids
Questions regarding barriers and enablers to prescribing opioids focused on respondents who either did not prescribe opioids for CNCP (n=32) or who prescribed only weak opioids (n=58).The most highly rated reasons for not prescribing opioids and for prescribing only weak opioids were concerns regarding potential long-term adverse events such as addiction and misuse.Concern of strong opioids being diverted and abused in the   *Per cent of respondents indicating they perform practices never or in <25% of their patients, in 25% to 50% of their patients, or in >75% of their patients or always; † Practices not recommended in guideline.Included in survey as distracters to reveal whether respondents tended to report they always performed the listed practices community was also highly rated.Importantly, concerns regarding regulatory body audits, inadequate knowledge of which opioid to use or the correct doses of opioids were not major barriers (Tables 5 and 6).FPs' ratings of various factors they identify as being important enablers to optimizing opioid therapy of CNCP were also examined (Table 7).The highest-rated factor was the ability to obtain a patient's opioid prescribing history from a provincial monitoring program, followed by knowledge of the risks and benefits of different opioids and improved access to pain or addiction specialists.Providing practical tips to help recognize patients at high risk of addiction was also deemed important.

DiSCuSSiOn
The intent of the present project was not to pass judgement on FPs' practices in relation to a guideline they had not had a chance to review and assimilate.The intent was to detect areas in which FPs were already following recommendations as part of best practices and areas in which they were not following recommendations as a baseline to detect future practice change as the guideline becomes widely implemented.The results of the present study provide a marker for FP knowledge of opioids before release of the new Canadian guideline; however, given the nonprobabilistic nature of the sample, we suggest caution in generalizing to the larger population of FPs.
We observed marked variability in how closely respondents' practices matched those recommended by the Canadian guideline.Concern for patient safety when prescribing opioids was reflected in FPs' emphasis on explaining the potential harms of long-term opioid therapy, observing for aberrant drug-related behaviour, assessing adverse effects and advising patient caution while driving.Respondents were also conscientious about assessing function, more so than assessing pain intensity.
The Canadian guideline recommends that long-term opioid treatment be viewed as a therapeutic trial in which physicians and patients define therapeutic goals when starting therapy.If the goals are not reached despite higher doses, it is reasonable to taper patients off the opioids.However, many FPs do not appear to be taking that approach,  which is a substantial practice gap.Perhaps physicians are reluctant to discontinue opioids because they mistakenly believe there is long-term randomized controlled trial evidence for the efficacy of opioids in treating CNCP, while in fact the longest trials lasted only 13 weeks (23).Another possibility is that opioids are started by another physician, such as a pain specialist, and respondents are reluctant to make changes.More concerning, a patient safety issue was identified regarding the minimum daily dose of strong opioids that patients should be taking before prescribing the fentanyl patch.To decrease the potential for overdose from fentanyl, the guideline states that patients should be taking 60 mg to 90 mg MEQ of strong opioid for two weeks (12).Forty-three per cent of respondents believed that there is no minimum dose or that the minimum dose was less than 60 mg MEQ.
Some knowledge gaps identified have cost implications.Most respondents believed some strong opioids provide better pain relief and were more likely to lead to addiction than others.Because there is no consistent evidence to support these differences in efficacy and harms (24), it makes economic sense to start treatment with the least expensive opioid.Similarly, controlled-release preparations are more expensive than immediate-release preparations.While they may be more convenient, there is no conclusive evidence that they offer increased pain relief or decreased potential for addiction (24).Therefore, FPs should feel confident prescribing inexpensive preparations if cost is a concern.
The present study also identified potential enablers and barriers to effective opioid prescribing for CNCP.A number of factors were important to FPs to improve opioid prescribing, particularly being able to obtain patients' opioid prescribing history from a provincial monitoring program.Also important were support services for FPs, such as access to pain or addiction specialists.Many respondents reported having to wait more than 12 months for a nonurgent referral to a pain specialist.Having access to an up-to-date guideline was also highly rated, which, when combined with low knowledge levels, speaks to the need for improved training and continuing education and support.
Information about barriers to care came from the 90 respondents who did not prescribe opioids or prescribed only weak opioids.For this group of FPs, the main barriers were concern regarding addiction and misuse, diversion for illicit use and being regarded as a target prescriber.These concerns echo those found in other studies on opioid prescribing (1,13,(17)(18)(19).
To our knowledge, the present study was the first national online survey on opioid prescribing for CNCP that attempted to elicit responses from FPs across Canada.The survey was developed with input from a wide variety of professionals involved in pain management and the development of the Canadian guideline.
A limitation of the present study was that the data were self-reported.However, respondents reported infrequently conducting practices included as distracters, giving credence to the findings.Another limitation was that the number of responses represents a small percentage of the approximately 32,000 FPs in Canada.Other limitations were that we received few French-language responses and the responses varied markedly according to province, likely due to different methods of publicizing the survey.Because the guideline was released online on May 3, 2010 (12), it is possible that some respondents had read it and altered their responses.However, we received only 93 responses after that date and analysis of responses to questions about practices in starting and monitoring opioids received after release of the guideline showed no statistically significant differences compared with responses received before its release.Similarly, there were no statistically significant differences in response to knowledge questions in those who responded before and after the release of the guideline.It is not surprising that publishing the guideline did not affect responses because guideline implementation and adherence is a complex knowledge translation process (25).
Volunteer bias is another potential limitation.Respondents may have had more interest and knowledge of this clinical area than nonresponders.Thus, our findings may represent a 'best-case' scenario.Demographic and practice responses on this survey were similar to those of the 2007 National Physician Survey, which received responses from approximately 10,000 FPs (26) (data not shown).While this finding does not guarantee the respondents were representative of the entire FP population, it is reassuring.
In contrast to other surveys, which found that 25% to 35% of FPs do not prescribe opioids (13)(14)(15)19), we found that only 5% were unwilling to do so.This may be because our survey was directed at FPs who treat CNCP while some other surveys were directed at FPs in general.However, our results are similar to those of unpublished data from the Nova Scotia Prescription Monitoring Program, which found that in 2010, only 8% of FPs did not prescribe opioids for CNCP.Responses to barriers and enablers to care were similar to other surveys.In a survey of Ontario FPs, the most highly rated concern when prescribing opioids was addiction and misuse, the same as we observed.As in other surveys from Canada (13,17) and the United States (18,27), we found that concern regarding audit from a regulatory body was not an important barrier to prescribing opioids.The time required to titrate and monitor opioids was also not reported as a substantial barrier.This may be because FPs recognize the significance of chronic pain and its effects on patients' lives and are willing to take the time to help their patients if they can.Previous Canadian surveys have found that chronic pain was a significant factor in their practices (16) and that pain management was not overly time consuming (13).
Our study provides a reasonable snapshot of FPs' current opioid-prescribing practices and knowledge with respect to the new Canadian guideline.It would be informative to repeat the survey with other health care professionals involved in managing CNCP with opioids -pharmacists, pain specialists and nurses -as well as repeating the survey with FPs in two to five years to detect changes since dissemination of the guideline.

COnCluSiOnS
Given that the responses represent only a small sample of Canadian FPs, the present survey identified a number of knowledge and practice gaps that have implications for patient care and the health care system.A reluctance to discontinue opioids if patients are not meeting treatment goals may lead to patients being left on the medications inappropriately and exposed to possible long-term adverse effects.Misunderstandings about increased efficacy and decreased adverse effects with long-acting opioids may lead to increased costs.Unawareness of the hazards of prescribing fentanyl to opioid-naive patients may lead to increased risk of overdose.Despite these gaps, FPs appear to take a precautionary approach to prescribing opioids, advising their patients of possible adverse effects and monitoring them for aberrant drug-related behaviour.The availability of their patients' opioid prescription history from a monitoring program was highly rated as an enabler to optimal prescribing and not regarded as a barrier to prescribing opioids.Thus, all provinces and territories should consider implementing such a system even though evidence for their effect on prescribing is lacking (28).Finally, FPs identified a current guideline as a valuable resource, an auspicious indicator for uptake of the new Canadian guideline.

______ patients per month
What is the size of the community in which you practice?19.
______ Under 5,000 people OF WORk: This work was performed as an unfunded MSc thesis in Community Health and Epidemiology at Dalhousie University.None of the authors have any conflicts of interest relevant to the present project to declare.Page 12 of 12.For approximately how many patients per month do you write prescriptions for WEAK opioids for Chronic Non-Cancer Pain?Weak opioids are Codeine, Tramadol, Propoxyphene, Meperidine, Pentazocine ______ 1 to 5 patients per month ______ 6 to 10 patients per month ______ 11 to 20 patients per month ______ more than 20 patients per month 13.For approximately how many patients per month do you write prescriptions for STRONG opioids for Chronic Non-Cancer Pain?Strong opioids are Morphine, Oxycodone, Hydromorphone, Fentanyl patch, Methadone ______ 1 to 5 patients per month ______ 6 to 10 patients per month ______ 11 to 20 patients per month ______ more than 20 patients per month 14.What type of health care professional are you?______ Family physician ______ Specialist physician -Please specify________________ ______ Other health care professional -Please specify___________________ 15.What is your gender ______ Female ______ Male 16.What year did you start practicing as a family physician?_____________ 17.Have you had any advanced training in pain management such as a diploma course or clinical traineeship?______ Yes ______ No We would like to know how busy your practice is.Approximately how many patients in TOTAL do you see in your office or outpatient clinic per month?18.

20 . 22 . 23 .
What is the waiting time for your patients to see a PAIN specialist for a NONis the waiting time for your patients to see an ADDICTION specialist for a NON-URGENT referral?In what province do you spend most of your time practicing?Respondents will be able to choose from list of provinces.The first three characters of your postal code at work indicate whether you practice in a rural or urban setting.Is the second character of your postal code a zero?______ Yes ______ No Your responses have been submitted.Thank you for taking our survey.If you have any questions or comments please contact Dr Michael Allen michael.allen@dal.ca

TAble 4 Frequency of following recommended practices performed while monitoring patients on opioids
*Per cent of respondents indicating they perform practices never or in <25% of their patients, in 25% to 50% of their patients, or in >75% of their patients or always

TAble 5 Rating of factors affecting decision not to prescribe opioids for chronic noncancer pain
*Per cent of respondents rating importance of factor as 1 or 2 (not important), 3 (neutral), or 4 or 5 (important) on 5-point Likert scale.Percentage may not total 100% because some respondents indicated 'no opinion'

TAble 6 Rating of factors affecting decision not to prescribe strong opioids for chronic noncancer pain
*Per cent of respondents rating importance of factor as 1 or 2 (not important), 3 (neutral), or 4 or 5 (important).Percentage may not total 100% because some respondents indicated 'no opinion'