Abstract
Objective To examine trends in chronic pain (CP) practice patterns among community-based family physicians (FPs).
Design Population-based descriptive study using health administrative data.
Setting British Columbia from fiscal years 2008-2009 to 2017-2018.
Participants Patients with an algorithm-defined CP condition and community-based FPs, both registered with the British Columbia Medical Services Plan.
Main outcome measures Using British Columbia health administrative data and a CP algorithm adapted from a previous study, the following were compared between fiscal years 2008-2009 and 2017-2018: CP patient volumes, pain-related medication prescriptions, referrals to pain specialists, musculoskeletal imaging requests, and interventional procedures.
Results In the fiscal year 2017-2018, among community-based family physicians (N=4796), an average of 32.5% of their patients had CP. Between 2008-2009 and 2017-2018, the proportion of CP patients per FP who were prescribed long-term opioids increased by an average absolute change of 0.56%; the proportion prescribed long-term neuropathic pain medications increased by 1.1%; and the proportion prescribed long-term nonsteroidal anti-inflammatory drugs decreased by 0.49%. The proportion of musculoskeletal imaging out of all imaging requests made by FPs increased by 2.0%; pain-related referrals increased by 1.73%; there was a 4.6% increase in the proportion of community-based FPs who performed 1 or more pain injections; and 10% more FPs performed 1 or more trigger point injections within a fiscal year.
Conclusion Findings show that the work of providing care to patients with CP increased while CP patient volumes per FP decreased. Workforce planning for community-based FPs should consider these increased demands and ensure FPs are adequately supported to provide CP care.
An estimated 1 in 5 people in Canada live with chronic pain (CP), defined as pain (cancer or noncancer) lasting 3 months or longer, representing substantial burden for both the patient and the health care system.1-3 Interdisciplinary pain programs have shown to be effective at treating patients who have CP4; however, these programs are difficult to access.5-7 Not all patients with CP have access to or require an interdisciplinary program, but nearly all seek primary care for CP from a family physician (FP).8,9 Simultaneously, there are reported barriers in providing CP primary care, including lack of time, lack of access to interdisciplinary treatment options, and challenges related to opioid prescribing.10
Studies report high primary care service usage among CP patients, such as having more visits compared with patients without pain.11-14 However, these studies were limited to surveys, focused on 1 type of CP condition, did not standardize at the physician level, and did not evaluate changes over time.11-14 A better description of CP management provided by FPs can inform education to support providers in the management of chronic pain, as well as support workforce planning (ie, measuring current capacity and predicting future supply and demand) and expansion of interdisciplinary team-based models to ensure provision of high-quality care.
This study investigates trends in CP management from 2008-2009 to 2017-2018 among community-based FPs. This study is part of a larger mixed-methods early-career primary care (ECPC) physician research program, the protocol for which is published elsewhere.15
METHODS
Study design
This is a population-based, repeated cross-sectional descriptive study using health administrative data. This project received ethics approval from the University of British Columbia–Simon Fraser University Harmonized Behavioural Research Ethics Board (Ethics No. H18-03291).
Study setting and population
We included individuals of all ages who were registered with the British Columbia Medical Services Plan (MSP) for at least 275 days of the year from 2008-2009 to 2017-2018. We included all FPs licensed to bill under the MSP who had a minimum of 100 patient contacts and billed more than 50 days each study year. We excluded physicians who graduated from medical school less than 2 years earlier or who had a missing year of graduation in the practitioner data set (only 5 FPs were excluded based on this criteria). To define community-based FPs in our sample, we used MSP billings to identify FPs and excluded physicians who conducted 80% or more of patient visits in hospital settings. Out of the community-based physicians, those providing 80% or more of their care in a subspeciality area (eg, obstetrics, palliative care, addictions medicine, mental health) made up only 1% to 2% of community-based physicians; we retained these physicians in the analysis as some might have focused pain practices.
Data sources
We used linked administrative health data accessed through Population Data BC.16 Data sets include the consolidation file, MSP billings, Discharge Abstract Database, PharmaNet, National Ambulatory Care Reporting System, and the College of Physicians and Surgeons of British Columbia practitioner data set. Access to data provided by any of these data stewards is subject to approval but can be requested for research projects through each data steward or its designated service providers. Further information regarding these data sets can be found by visiting the Popdata project webpage (https://my.popdata.bc.ca/project_listings/19-044/collection_approval_dates). All inferences, opinions, and conclusions drawn in this article are those of the authors, and do not reflect the opinions or policies of the data stewards.
Defining individuals with chronic pain
To identify patients with CP, we modified a previously validated algorithm for CP (Appendix A, available from CFPlus*).17 Patients were defined as a prevalent case of likely CP in a fiscal year if they had either 5 or more pain-related International Classification of Diseases (ICD) diagnostic codes in any of the MSP, Discharge Abstract Database, or National Ambulatory Care Reporting System data sets within the past 5 years, with at least 183 days separating at least any 2 ICD codes, and at least 1 visit with a pain-related ICD code in the respective fiscal year; or 1 CP-specific ICD code in that fiscal year. Clinically, CP is often defined as pain lasting 3 or more months, but we used 183 days to maintain consistency with the previously validated algorithm.17,18
We tabulated ICD code frequencies to assess algorithm validity by evaluating coding trends that might represent data changes versus real change in disease prevalence (Appendix A[b]). There is no CP population reference standard to use for validation of this algorithm in British Columbia, but this approach was validated for use on similar administrative data from Newfoundland and Labrador.17
Variables
Patient and provider characteristics. Patient characteristics included number of patients experiencing CP as defined by the algorithm, age, sex,† health authority, and neighbourhood income. Practitioner characteristics included number of practitioners, age, reported gender, location of practice, graduation year, years in practice, number of days with 1 or more billings per year (days billed), number of unique patients per year (all and CP), and number of visits per year (all and CP visits).
Health service utilization and practice patterns. Chronic pain–related practice patterns of interest calculated per FP included unique patient count; visit count; medications prescribed, including opioids that were not prescribed for the purposes of opioid agonist therapy; referrals to pain specialists; musculoskeletal radiologic investigations; and interventional procedures (Appendix B, available from CFPlus*).
Data analysis
As this was a descriptive study, data are presented as counts and proportions. Service delivery per FP was expressed as an average count of services per year or a proportion of annual services. Changes in this service delivery were expressed as either a relative percent change ([mean count (time 2) – mean count (time 1)]/mean count [time 1] × 100) or an absolute percent change (% of all of FP’s patients with CP with specific service [time 2] – % time of all of FP’s patients with CP with specific service [time 1]). Since these data represent population data, no statistical tests were done. Standard deviation was used as a measure of variability.
RESULTS
We identified 4060 and 4796 community-based FPs and 852,895 and 946,641 algorithm-defined CP patients for analysis in the 2008-2009 and 2017-2018 fiscal years, respectively (Figure 1). The prevalence of CP cases remained stable between 2008-2009 and 2017-2018 (20.2% to 19.7%). Overall, more CP patients were female (approximately 61%). The proportion of older CP patients increased, with 26.0% in the 60-79 age group in 2008-2009 versus 32.5% in 2017-2018 (Table 1).16
Family physicians and patients included in study
Comparison of 2008-2009 and 2017-2018 fiscal years of CP patient characteristics in British Columbia
Table 2 describes the community-based FPs included in this study.16 Across all metrics, there was a reduction in patient volumes per FP per year from 2008-2009 to 2017-2018, with a decrease in the average number of unique patients (any diagnosis) seen per year.
Comparison of 2008-2009 and 2017-2018 fiscal years of community-based FP characteristics in British Columbia
Service usage by community-based FPs for CP patients
The number of algorithm-defined CP patients seen by FPs decreased annually and gradually from 2008-2019 to 2017-2018, as is consistent with changes to service volume more broadly (Tables 2 and 3).16 However, when expressed as a percentage of total patient volume, the average percentage of all FP patients with CP increased slightly, from 31.9% in 2008-2009 to 32.5% in 2017-2018.
Average relative percent change in CP service volume or management provided per FP from fiscal years 2008-2009 to 2017-2018 in British Columbia
Over the same period, most metrics of service usage for pain-related management increased (Table 3). Prescription of long-term neuropathic pain medication per FP increased by an average absolute change (AAC) of 1.1%; musculoskeletal imaging requests over all imaging requests increased by an AAC of 2.0% (although absolute count of musculoskeletal imaging requests dropped, all imaging requests fell more over the same period); and pain-related referrals over all referrals made increased by an AAC of 1.7% (Figure 2). Long-term opioid prescribing per FP for their CP patients initially increased by an AAC of 0.69% from 2008-2009 to 2015-2016, then plateaued in 2016-2017, and slightly fell 0.13% in 2017-2018. In contrast, there was an 0.5% AAC decrease of all CP patients per FP being prescribed long-term nonsteroidal anti-inflammatory drug (NSAID) therapy.
Annual service usage by community-based family physicians among their algorithm-defined chronic pain patients from 2008-2009 to 2017-2018 in British Columbia
Median number of injection procedures commonly used for pain increased from 4 to 6 per year among all FPs between 2008-2009 and 2017-2018. Comparing 2008-2009 with 2017-2018 among FPs who performed at least 1 pain-related procedure per year, 4.6% more FPs in 2017-2018 performed at least 1 injection procedure (including soft tissue injections, hip joint injections, and trigger point injections) for pain (Figure 3).
Percentage of community-based family physicians performing each specific procedure of interest for algorithm-defined chronic pain patients among family physicians who performed at least 1 pain-related procedure each year
DISCUSSION
Our study shows that CP patient volumes per FP decreased between 2008-2009 and 2017-2018 and that this is consistent with the general reduction in all patient volumes observed in this study and elsewhere.19 Over the same period, the average percentage of patients with algorithm-defined CP and the percentage of visits with CP patients per FP remained relatively constant or slightly increased. Simultaneously, the amount of CP management work being done by FPs, including opioid and neuropathic prescriptions, referrals, and injection procedures for pain, increased.
Reasons for the increase in observed management practices for CP patients over time are likely multifactorial, possibly related to aging patient demographic characteristics that are associated with increasing patient complexity, or to greater awareness and evidence of CP and CP care owing to recent advancements.3,20,21 Moreover, chronic pain was recognized as a disease in its own right by the ICD-11 only in 2019.18 The pain specialist residency established in British Columbia in 2016 allowed for better recognition of CP and availability of pain specialty care, and was associated with an increase in pain-related referrals seen in this study. Newer evidence-based guidance on pharmacologic management for neuropathic pain might explain the increase in long-term prescribing of neuropathic pain medications,22,23 while a relative 25% decrease in long-term NSAID prescriptions might be due to improved provider education around NSAID side effects, reflecting more appropriate CP care, especially among elderly populations.24
The proportion of each FP’s patients with CP on long-term opioid therapy prescribed by the FP increased between 2008-2009 and 2014-2015 by 0.69%, held steady in 2015-2016 and 2016-2017, and then fell by 0.13% in 2017-2018. This slight drop might relate to the College of Physicians and Surgeons of British Columbia’s 2016 release of practice standards on opioid prescribing, as a previous study observed a 2% drop in monthly or chronic users of opioid therapy following the release of this standard.25,26 We cannot comment on the appropriateness of opioid prescribing from these data to explore reasons for the overall increase in long-term opioid prescribing over the data period. However, this observed increase might relate to patients being treated for cancer pain or reflect recognition of pain treatment as a human right in a landscape of poor access to alternatives.7,27,28
We found a 3-fold increase in the percentage of community FPs performing trigger point injections on CP patients. This trend also occurred elsewhere, with Ontario seeing a 3-fold increase in administration of pain injections by physicians (approximately 70% of whom were general practitioners) from 2010 to 2019.29 It is possible this change represents improved access to care but unclear if this large increase represents improved quality of care or follows financial incentivization. Future studies are needed to discern the cost-effectiveness of more pain injections for CP patients at the population level.3,21
In this study there are population-level indicators that FPs are using more tools to treat CP. The observed increase in management volume suggests that evidence-based practice solutions might be beneficial to support FPs providing patient care for CP. Such solutions might be of particular benefit regarding pain injections, which have surged in number since the early 2000s, as seen in our study and mentioned in others.20,29,30 Given the increasing proportion of work dedicated to care for patients with CP, and the ongoing long waits for community-based CP supports, more tools and structures to support FPs in this work are needed.31
Access to FPs has also been declining across Canada, possibly further limiting access for CP patients.32 Improving and strengthening access to primary care that is well-resourced might help improve patient outcomes. Previous studies indicate that FPs more often prefer team-based care,33,34 which has also been shown to improve outcomes for patients with CP.35 Taken together with FP preferences for collaborative care, our findings of increasing CP management demands align with efforts to fund team-based primary longitudinal care with embedded CP resources.36,37
Limitations
We used a CP algorithm adapted from the literature that has at best a sensitivity and specificity of 67%.17 Therefore, we have almost certainly misclassified cases, and there might be patients included without current CP that might inflate CP patient volumes. We cannot differentiate CP that is intermittent or of varying severity, and recognize that CP is a heterogenous condition where patients can have different needs that cannot be captured by administrative data.3 However, the purpose of this study is not to evaluate effectiveness of interventions for CP, which requires accurate identification, but to describe trends in population-level management of a CP conditions for the purposes of system-level workforce planning.
Conclusion
Compared with 2008-2009, community-based FPs in 2017-2018 do more CP management work for a smaller number of algorithm-defined patients with CP. It is uncertain if this represents a higher quality of care or an isolated increase in management practices. Future research should investigate whether quantity of service delivery translates into improved patient outcomes.
Acknowledgments
This study received funding from the Canadian Institutes of Health Research (CIHR-Project Grant 155965). The authors thank members of the Early-Career Primary Care research program who obtained ethics approval from the University of British Columbia–Simon Fraser University Harmonized Behavioural Research Ethics Board, as well as data from Population Data BC. Dr Curtis May thanks his public health colleagues for providing helpful feedback around population-level data checking and suggesting possible future directions for research.
Footnotes
↵* Appendices A and B are available from https://www.cfp.ca. Go to the full text of the article online and click on the CFPlus tab.
↵† The Medical Services Plan registration form contains a variable labelled “gender” with the options “M” and “F” provided (presumed to be abbreviations of “male” and “female”). Whether responses reflect gender, sex assigned at birth, or legal sex cannot be determined. We refer to this variable as “administrative sex.”
Contributors
Dr Curtis May conceived of the idea for the article, and contributed to primary data tabulation, interpretation, and writing of paper. Sandra Peterson analyzed the data, interpreted the technical data, and made major manuscript edits. Ellie Gooderham, Dr Lindsay Hedden, and Dr Rita K. McCracken contributed to data interpretation and major manuscript edits. Dr M. Ruth Lavergne was principal investigator of the study, providing support on idea inception, data interpretation, and major manuscript edits.
Competing interests
None declared
This article has been peer reviewed.
Cet article a fait l’objet d’une révision par des pairs.
- Copyright © 2024 the College of Family Physicians of Canada
