Opioid therapy for chronic pain is common,12,17 although recent data suggest plateauing30,31 or even declining11 opioid-prescribing rates. Reduced opioid prescribing for chronic pain may be due to a combination of factors including heightened awareness of increases in opioid-related adverse events,11 lack of information on the effectiveness of long-term opioid therapy (LTOT),7 and availability of state Prescription Drug Monitoring Programs (PDMPs) to ascertain concurrent prescribing of controlled substances.51 These factors may also contribute to opioid therapy discontinuation.
The 2009 American Pain Society5 and 2010 U.S. Department of Veterans Affairs/Department of Defense (VA/DoD)48 clinical practice guidelines recommend discontinuation of opioid therapy under several circumstances, including the presence of side effects, failure to meet therapeutic goals, and patient behaviors that heighten the risk of opioid overdose and death. The 2016 CDC Guidelines for Prescribing Opioids for Chronic Pain14 further emphasize a risk–benefit evaluation and recommend opioid therapy discontinuation when potential or actual harms exceed benefits.
A substantial body of empirical evidence indicates that patients with substance use disorder (SUD) histories prescribed opioid therapy for chronic pain are at increased risk of engaging in opioid-related hazardous behaviors such as opioid misuse and abuse, diversion, and concurrent use and abuse of alcohol, and other illicit substances.16,36,47 Patients with SUD are also at increased risk of experiencing opioid-related adverse events such as overdose and death.2,3 However, patients with SUD are more likely to be prescribed long-term13,50 and high-dose opioid therapy,28,35,50 a phenomenon scholars have termed “adverse selection.”15,45 In light of the increased risk patients with SUD histories face, opioid treatment guidelines recommend close monitoring of patients with SUD.14,48
Limited data about opioid therapy discontinuation are available. In a national cohort of 550,616 veterans prescribed LTOT between 2008 and 2012,49 discontinuation of LTOT was associated with younger and older ages (relative to middle age), African-American race, residing in an urban area, having mental health and SUD diagnoses, and greater number of medical comorbidities. A decreased likelihood of LTOT discontinuation was associated with longer duration on opioid therapy, higher daily doses of opioids, greater number of pain diagnoses, and tobacco use disorder diagnosis. A study of patients from a national private health network and state Medicaid found similar results, although patients who engaged in opioid misuse behaviors were less likely to discontinue LTOT.33 Despite these robust descriptions of demographic and clinical correlates of LTOT discontinuation, little is known about reasons for discontinuation of LTOT, particularly in patients at “high risk” for opioid-related adverse events.
The aim of this study was to compare reasons for discontinuation of LTOT between patients with and without SUD receiving care within a major U.S. health care system in the years after release of the 2009 and 2010 opioid therapy clinical practice guidelines. We hypothesized that patients with SUD would be more likely to discontinue LTOT because of aberrant behaviors, including illicit substance use, aberrant urine drug tests, opioid misuse, and diversion.
This study was approved by the Veterans Health Administration (VHA) Portland Health Care System Institutional Review Board.
2.1. Data source and sample selection
We used the VHA Corporate Data Warehouse (CDW) to identify a national cohort of VHA patients prescribed opioid therapy for all of 2011. The CDW provides comprehensive information contained in electronic medical records for all VHA patients.
We defined LTOT as having been prescribed opioid therapy for the entirety of 2011, allowing prescription opioid refill gaps of no more than 30 days between the completion of an opioid prescription and a refill of the next opioid prescription. Allowing 30-day gaps accounted for delayed scheduled refills due to travel, prescription mail order delays, or other circumstances. The allowance of 30-day gaps has been used in previous studies examining discontinuation of LTOT in the VHA.33,49 The mean number of days prescribed opioids in 2011 for this cohort was 351 of 365 days (SD = 17 days). From this cohort, we identified patients who discontinued LTOT—ie, had no VHA opioid prescriptions—for at least 12 consecutive months starting some time in 2012. We chose a 12-month discontinuation interval to ensure discontinuation was an intended treatment decision and not due to geographic relocation, switching care to another VHA facility, extended inpatient hospitalizations, or other reasons that may result in failure to renew an active opioid prescription. The date of the last opioid refill was used as an index date for each patient to ascertain patient sociodemographic and clinical characteristics before discontinuation. Because this study focused on discontinuation of LTOT in the context of chronic noncancer pain, we excluded patients with the following characteristics in the year before the index date: (1) the only opioid therapy prescribed was through a VHA opioid substitution program (ie, buprenorphine or methadone maintenance therapy), (2) a diagnosis of cancer, (3) enrollment in hospice or long-term care, or (4) having received surgery for which opioids may have been prescribed. We also excluded patients with no VHA contact (ie, no VHA clinical encounters or medications prescribed) or who died in the year after discontinuation, as well as nonveterans or veteran patients whose only medical care was obtained at a facility located in a U.S. territory. Figure 1 details sample selection and the number of patients meeting each exclusion criterion.
2.2. Propensity score matching
The aim of this study was to compare reasons for opioid discontinuation between patients with and without SUDs. From the cohort of 7247 patients who discontinued LTOT in 2012, 1868 (26%) had an SUD diagnosis based on ICD-9-CM codes in the year before the index date. These included diagnoses of abuse or dependence for alcohol, amphetamines, cannabis, cocaine, hallucinogens, opioids, sedatives/hypnotics, polysubstance use, or other/unspecified substances. For the purpose of this study, an SUD diagnosis was defined as having an ICD-9-CM code for an SUD diagnosis linked to a medical visit encounter. A medical visit encounter included encounters with any treating clinician—physician, physician assistant, nurse, psychologist, social worker, or other clinical provider—and could be for the purpose of medical, mental health, or other health care needs. We randomly sampled 300 patients with an SUD diagnosis for subsequent chart review. An additional 300 patients without an SUD diagnosis were matched as controls using propensity score matching procedures to ensure a similar distribution of sociodemographic and clinical characteristics to the sample of patients with an SUD diagnosis. We used this matching procedure to lessen the likelihood of confounding biases that may arise due to underlying differences between patients with and without SUD diagnoses. We first modeled the probability that a patient would receive an SUD diagnosis in the year before the index date using logistic regression that included sociodemographic, treatment utilization, and clinical factors present during the year before the window for SUD diagnoses (ie, 12-24 months before the index date). Propensity model variables were selected based on their associations in previous research with having an SUD diagnosis5,20,44 and from clinical judgment of clinicians who work in primary care and specialty SUD treatment settings. Variables included: age, sex, race/ethnicity, period of military service, geographic rurality based on patients' residential zip code, hospital complexity where patients receive their VA care, VA service connected disability status, medical comorbidity, history of SUD, mental health history, pain diagnoses, average daily dose of opioids in morphine equivalents, sustained-vs short-acting opioid, number of past-year opioid prescribers, as well as types of outpatient and inpatient clinical encounters in which patients engaged.
The resulting predicted probabilities, or propensity scores, were used to match non-SUD controls with a similar probability of receiving an SUD diagnosis in the year before the index date. Matching was conducted using a nearest neighbor matching algorithm.41 For each patient with an SUD diagnosis, this procedure selects the non-SUD diagnosis patient with the smallest absolute difference in propensity score. Standardized differences were then used to assess covariate balance between the matched groups,40 and kernel density plots of propensity scores were used to test for sufficient overlap.21 The online Supplemental Digital Content lists, for each variable in the propensity model, the standardized differences between patients with and without an SUD diagnosis in the full sample of N = 7247, and the matched sample of N = 600 (available online at http://links.lww.com/PAIN/A370).
2.3. Chart review tool development, pilot testing, and coding fidelity
After sample selection, 4 experienced chart reviewers (one internist, one psychiatrist, and 2 psychologists) with expertise treating SUD and non-SUD VHA patients receiving LTOT for chronic noncancer pain developed a chart review coding tool using group consensus procedures that identified reasons for LTOT discontinuation (see Supplemental Digital Content for the chart review tool, available online at http://links.lww.com/PAIN/A371). To refine the content of the chart review tool and review process, the developers pilot tested the tool on a randomly selected sample of 60 patients from the cohort of patients who discontinued LTOT who were not included in the analytic sample. A research associate (RA) experienced with reviewing and coding VHA medical charts for opioid-related studies was trained in coding procedures for this study. After training, the RA and study principal investigator double-coded 25 randomly selected medical charts of patients not included in the analytic sample. We used a benchmark of kappa ≥0.70 or simple agreement ≥95% for binary variables as measures of adequate intercoder agreement.29 The average kappa across all study variables was 0.88 (average percent simple agreement = 97%) and all study variables met our a priori standard of acceptable reliability.
The RA subsequently reviewed and coded the 600 charts from the analytic patient sample over a 4-month period. To ensure ongoing fidelity to the coding scheme, the study principal investigator double-coded 60 randomly selected charts (10%). The average kappa across all variables in the study phase was 0.85 (average percent simple agreement = 98%) and all study variables met our standard of acceptable reliability.
2.4.1. Administrative data abstraction
Data abstracted from the CDW included demographic characteristics (age, sex, race/ethnicity) and rurality of a patient's place of residence based on rural–urban commuting area codes.37 Medical comorbidities were assessed with the Elixhauser Comorbidity Measure,18 where higher scores indicate a greater number of comorbidities. Veterans Health Administration service–connected disability status, which is disability granted to veteran patients as a result of military service–related injuries or traumas, was obtained. Data did not permit ascertainment of specific medical conditions for which patients were service connected. Diagnoses of SUDs, mental health disorders, and chronic pain conditions were obtained for the 12 months before discontinuation of LTOT. Opioid-related variables assessed over the 12 months before discontinuation included type(s) of opioid(s) prescribed, average daily dose of opioids in morphine equivalents, and number of opioid prescribers in the 12 months before discontinuation. Finally, we identified patients prescribed benzodiazepines in the 12 months before discontinuation.
2.4.2. Chart review
Review of patients' electronic medical records identified reasons for discontinuation of LTOT. Reasons were grouped by patient- and clinician-initiated reason (see Table 1 for discontinuation reasons listed in the chart review coding tool). Clinician-initiated reasons were grouped into 3 categories to further assess discontinuation themes. The first, “Aberrant Behaviors,” included aberrant urine drug test results, suspected use of alcohol or other substances, opioid misuse, opioid diversion, and nonadherence to plan of care (eg, failing to present for a urine drug test when asked). The second, “Patient Safety Concerns,” included previous opioid overdose, high risk for an opioid-related adverse event, and contraindication with other prescribed medication. The final category, “Lack of Efficacy,” included opioids not indicated for type of chronic pain, opioids not decreasing pain, and opioids not improving functioning.
Patients could be coded as having multiple patient- or clinician-initiated discontinuation reasons (eg, a patient may have tested positive for an illicit substance and have been suspected of diverting opioid medication). However, all reasons were either patient initiated or clinician initiated in this sample. Opioid discontinuation reasons did not span both categories.
2.5. Statistical analysis
We utilized χ2 tests of association for categorical variables and independent sample t tests for continuous variables to compare demographic and clinical characteristics between patients with and without SUD diagnoses. We next used binary logistic regression to examine associations of SUD status with reasons for opioid discontinuation. Adjusted models controlled for variables associated with opioid discontinuation in previous studies.49 These included sociodemographic characteristics (age, sex, race/ethnicity, and rurality), mental health diagnoses (depressive disorders, bipolar disorders, post-traumatic stress disorder (PTSD) other anxiety disorders, and psychotic disorders), tobacco use disorder diagnosis, Elixhauser Comorbidity score, number of pain diagnoses, type of opioid prescribed, average daily dose of opioids in morphine equivalents in the year before the index date, benzodiazepine prescription in the year before the index date, and number of opioid prescribers in the year before the index date. The combination of both propensity score matching and covariate regression techniques is commonly used and provides estimates that are generally more robust to model misspecification and residual confounding when compared with either method in isolation.42,44
The study sample comprised 300 patients with an SUD diagnosis in the year before opioid therapy discontinuation and 300 propensity score–matched patients without an SUD diagnosis. Similar to the population of veterans prescribed opioid therapy through VHA,38,49 the study sample had a mean age of 55 years, was predominantly male (95%), non-Hispanic white (72%), and resided in urban locations (73%). High proportions of patients in this sample received diagnoses for mental health disorders in the year before discontinuation of LTOT, including PTSD (31%), anxiety disorders other than PTSD (25%), and depressive disorders (25%). Nearly half (45%) of patients had been diagnosed with tobacco use disorder.
Nearly all patients (86%) had been diagnosed with musculoskeletal pain, with smaller proportions having been diagnosed with headaches (including migraine; 11%) or neuropathic pain (6%). Hydrocodone, oxycodone, methadone, and morphine were the most commonly prescribed opioid medications in the year before discontinuation, with 57%, 38%, 29%, and 26% of patients being prescribed these medications, respectively. Forty-five percent of patients were prescribed 2 or more opioids concurrently in the year before discontinuation. The average morphine equivalent daily dose (MEDD) of prescribed opioids in the year before discontinuation was 76 mg. Fifty-six percent of patients were prescribed <50 mg MEDD, 20% 50 mg to <90 mg MEDD, 7% 90 mg to <120 mg MEDD, and 17% ≥120 mg MEDD. On average, patients had nearly 3 different VHA opioid prescribers in the year before discontinuation.
Among the sample of 300 patients with an SUD diagnosis, alcohol use disorder was the most common SUD (52%), followed by opioid use disorder (29%), cocaine use disorder (14%), cannabis use disorder (11%), sedative/hypnotic/anxiolytic use disorder (5%), amphetamine use disorder (4%), and other SUDs (12%; eg, hallucinogen, polysubstance, and unspecified SUD). Table 2 provides descriptive statistics for patient sociodemographic and clinical characteristics, as well as bivariate comparisons between patients with and without SUD.
3.1. Reasons for discontinuation of long-term opioid therapy
For most patients (85%), discontinuation of LTOT was initiated by a clinician, rather than patient, decision. Clinician-initiated discontinuation reasons included aberrant behaviors (64%), which comprised suspected substance abuse (44%), aberrant urine drug test results (37%), opioid misuse behaviors (15%), nonadherence to the pain plan of care (11%—eg, failing to present for urine drug tests or primary care appointments when asked), and concerns about opioid diversion (4%). Of the 223 patients whose clinician-initiated discontinuation reasons included aberrant urine drug tests, results of urine drug tests included negative for prescribed opioid(s) (26%), positive for cannabis (47%), positive for cocaine (22%), positive for a nonprescribed opioid (11%), positive for amphetamines (10%), positive for a nonprescribed sedative/hypnotic/anxiolytic (6%), and positive for alcohol (4%). Some patients had multiple aberrant urine drug test results that led to discontinuation. Few patients were discontinued by their clinicians because of patient safety concerns (6%) or because of a lack of efficacy of LTOT (5%). Reasons for LTOT discontinuation for the full sample and the subsamples of patients with and without SUD are presented in Table 3.
In unadjusted models, patients with SUD were more likely to have opioids discontinued by clinicians because of aberrant behaviors (odds ratio [OR] = 1.79, 95% confidence interval [CI] = 1.28-2.51), most notably known or suspected abuse of illicit substances (OR = 2.04, 95% CI = 1.47-2.83). However, they were no more likely than patients without SUD to be discontinued because of aberrant urine drug test results, opioid diversion, opioid misuse behaviors, or nonadherence to their pain plan of care. Patients with SUD were less likely to discontinue opioids for reasons that were not documented in the electronic medical record, and this was true for both patient- (OR = 0.24, 95% CI = 0.07-0.87) and clinician-initiated (OR = 0.44, 95% CI = 0.25-0.76) discontinuation. Patients with and without SUD did not significantly differ on any other patient- or clinician-initiated discontinuation reasons in unadjusted models (Table 3).
Results of covariate-adjusted models were consistent with those of unadjusted models. Namely, patients with SUD were more likely to discontinue LTOT because of aberrant behaviors, known or suspected substance abuse, and for reasons not documented in the medical record. Effect size magnitudes as measured by the OR were equivalent or slightly greater in covariate-adjusted models. Similar to unadjusted models, patients with SUD did not differ from patients without SUD on any other patient- or clinician-initiated opioid discontinuation reason. See Table 3 for results of multivariable models.
In a sample of 600 VHA patients with and without SUD who discontinued LTOT, an overwhelming majority (85%) of discontinuations were initiated by clinicians rather than patients. Of patients whose clinicians initiated the discontinuation, 75% were due to an aberrant behavior, including substance abuse, aberrant urine drug tests, other opioid misuse behaviors, opioid diversion, and nonadherence to the chronic pain plan of care. Clinician-initiated discontinuations in response to patients' aberrant and potentially high-risk behaviors are recommended by the 2009 American Pain Society5 and 2010 VA/DoD48 clinical practice guidelines that were in place at the time patients in this study discontinued LTOT. The 2016 CDC clinical practice guidelines14 recommend ongoing evaluation of clinical outcomes of opioid therapy for each patient and to taper or discontinue opioid therapy if harms outweigh benefits. Coupled with CDC recommendations for clinicians to review PDMP data quarterly and perform annual urine drug testing, rates of LTOT discontinuation due to aberrant behaviors may increase as CDC guidelines are adopted by clinicians and health care systems.
Clinicians face challenges when discontinuing LTOT for patients who engage in behaviors that heighten risk of opioid-related adverse events. Continuing to prescribe opioids to patients who are misusing, abusing, or diverting medication could result in significant harm to individual patients and society. Conversely, discontinuing LTOT when aberrant behaviors are identified, even when these actions represent good clinical practice, may erode the patient–provider relationship. Patient-centered models of care1 may be difficult to use in situations when patients perceive punitive action is being directed toward them. The ways in which clinicians communicate with patients before, during, and after discontinuation of LTOT are paramount to ensuring patients continue to receive pain care and other needed services rather than disengaging from a clinic or health care system entirely. If mishandled, patients may be at increased risk of engaging in illicit behaviors to obtain opioids, attempt to obtain opioids from other clinics or the emergency room, or resort to street drugs such as heroin.8,32
In this study, patients with an SUD diagnosis were more likely than those without to be discontinued for aberrant behaviors, specifically substance abuse. Previous data indicate patients with SUD are more likely to misuse opioids46 and experience adverse opioid-related events such as overdose and death.3 Data from this study indicate that these patients are also more likely to be discontinued from LTOT for substance-related reasons. For many patients with comorbid chronic pain and SUD, pain predates SUD and inadequately treated pain may lead to increased substance use24 or relapse to substance use after specialty SUD treatment.25 Treating chronic pain in SUD patients is further complicated by high rates of medical and psychiatric comorbidities.26 The complexity of this patient population points to the need for comprehensive, multidisciplinary care that, at a minimum, concurrently addresses substance use and pain, as well as other psychiatric and medical conditions that exacerbate pain and SUD symptoms. Indeed, patients with SUD are less likely than their non-SUD counterparts to experience improvements in pain-related functioning when receiving standard pain care,34 pointing to the fact that these patients often require a higher level of care. Cognitive behavioral therapy for co-occurring pain and SUD delivered in specialty SUD treatment settings has demonstrated efficacy in improving pain and SUD outcomes.10,22,23 For most patients with SUD who are unwilling or otherwise unable to engage in specialty SUD treatment, the primary care medical home may be an alternative treatment setting for meeting these patients' pain and SUD needs. Although many health systems currently lack capacity for such integrated models of care, others (eg, Veterans Health Administration, Kaiser Permanente) have integrated primary medical and mental health across their health care systems. Behavioral approaches to pain management administered through telehealth also show promise for expanding the reach of services to underserved rural areas.39 As capacity for behavioral health specialists in primary care continues to grow,9 incorporating specialists with backgrounds in treating both chronic pain and SUD or providing training to specialists already working in these clinical settings will be paramount.
Notably, 37% of patients in this study discontinued LTOT because of aberrant urine drug tests, and results did not differ between those with and without SUD diagnoses. A previous study found that 21% of patients on LTOT with no opioid-related behavioral issues tested positive for one or more nonprescribed controlled or illicit substances.27 These findings combined with those of this study suggest a guideline-concordant universal precautions approach to urine drug testing,19 in which all patients are randomly tested, may be necessary to accurately identify patients who misuse or abuse prescription medications, and illicit substances. Targeting urine drug testing to high-risk patients only, such as those with SUD diagnoses or other behaviors indicative of opioid misuse, will miss a substantial proportion of patients using substances.
Nearly half of aberrant urine drug tests that resulted in discontinuation of LTOT included a positive test for cannabis. Although we were unable to ascertain from the medical record the reasons for cannabis use, cannabis may be used by some patients to treat chronic pain.4 Currently, 29 states and the District of Columbia have passed legislation allowing the use of cannabis for medical purposes. Eight states—Alaska, California, Colorado, Maine, Massachusetts, Nevada, Oregon, and Washington—and the District of Columbia allow recreational use of cannabis. As the medicinal and recreational use of cannabis becomes legal across a growing number of jurisdictions, clinicians will increasingly be faced with questions from patients and their family members about the use of cannabis as a form of pain management. Clinicians and patients will also be confronted with clinical decisions concerning combined cannabis and opioid use for chronic pain and the challenge of assessing and safely balancing the potential risks and benefits of this strategy. Unfortunately, there is no empirical evidence to guide clinicians in the management of chronic noncancer pain using opioid therapy for patients concurrently prescribed cannabis for a medical or psychiatric condition.
This study has several limitations. First, the sample comprises VHA patients and results may not generalize to non-VHA patients who discontinue LTOT. Second, to ensure identification of a cohort with chronic noncancer pain, we excluded patients with cancer diagnoses and surgeries in the year before discontinuation of LTOT; however, it is possible that some of these excluded patients' pain may not have been due to cancer nor was it acute in nature. Third, we identified SUD patients as having a clinical encounter in the year before LTOT discontinuation in which SUD was linked to the clinical encounter. This may not fully capture all patients with SUD. Fourth, patients without SUD in this study were selected to match the cohort of patients with SUD, rather than randomly sampled from the population. Because of this, results for patients without SUD may not be fully generalizable to all non-SUD patients in VHA undergoing LTOT discontinuation. Fifth, discontinuation of LTOT was defined as 12 continuous months without filling an opioid prescription. Patients who were discontinued and then subsequently restarted opioid therapy within a year were not captured and may represent a selection bias. Sixth, some previous studies33,49 have defined LTOT as at least 90 days of opioid therapy. We sought to identify patients for whom opioids had become a very long-term approach to pain management and therefore implemented a more conservative definition of 12 consecutive months of opioid therapy. Finally, the sample comprises patients on LTOT for all of 2011 who discontinued opioids in 2012 and were followed through 2013. We chose this period because clinical practice guidelines for opioid therapy for chronic pain had been released one to 2 years prior,6,48 allowing us to examine discontinuation reasons after guidelines would likely have been implemented by clinicians. The data may not, however, reflect reasons for discontinuation of LTOT in more contemporary patient cohorts.
Discontinuation of LTOT is a decision overwhelmingly initiated by clinicians. Patients with SUD are more likely than patients without SUD to discontinue LTOT for aberrant behaviors. However, many patients without SUD screen positive for illicit and other substances and engage in aberrant behaviors that lead to opioid discontinuation. Increasing rates of opioid discontinuation are likely to occur due to policies and programs that encourage close monitoring of patients on LTOT for opioid misuse behaviors.14 Ensuring patients have access after opioid discontinuation to nonopioid analgesic pharmacotherapies, nonpharmacologic pain management approaches, and SUD treatment is critically important as inadequately treated pain can exacerbate other comorbid conditions—such as psychiatric disorders and SUDs—resulting in poorer quality of life. Integrating nonopioid pain therapies and SUD treatment into multiple settings such as primary care and specialty SUD care is one possible approach. Additional research is needed to determine if such models will prove efficacious for patients who discontinue LTOT.
Conflicts of interest statement
T. I. Lovejoy, J. W. Frank, and S. K. Dobscha report grants from the U.S. Department of Veterans Affairs during the conduct of the study. The remaining authors have no conflicts of interest to declare.
Presented at the 2016 American Pain Society Annual Conference. Abstract Citation: Lovejoy TI, Morasco BJ, Demidenko MI, Meath THA, Frank JW, Dobscha SK. Reasons for discontinuation of LTOT in patients with and without substance use disorders. J Pain. 2016;17:S87.
This work was supported by Locally Initiated Project Award # QLP 59-048 (PI: Lovejoy) from the United States (U.S.) Department of Veterans Affairs Substance Use Disorder Quality Enhancement Research Initiative. T. I. Lovejoy received additional support from Career Development Award IK2HX001516 from the U.S. Department of Veterans Affairs Health Services Research and Development during preparation of this manuscript. J. W. Frank received support from Career Development Award IK2HX001914 from the U.S. Department of Veterans Affairs Health Services Research and Development.
The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the U.S. Department of Veterans Affairs or U.S. Government.
We thank the VA Portland Health Care System and the U.S. Department of Veterans Affairs Health Services Research and Development Center to Improve Veteran Involvement in Care (CIVIC; CIN 13-404, PI: Dobscha) at the VA Portland Health Care System for the provision of support and resources for this project.
Supplemental Digital Content associated with this article can be found online at http://links.lww.com/PAIN/A370.
Supplemental Digital Content Supplemental media
Video content associated with this article can be found online at http://links.lww.com/PAIN/A371.
. Blubaugh S. The patient-centered medical home. Part I: a primer. J Med Pract Manage 2016;31:346–50.
. Bohnert AS, Ilgen MA, Galea S, McCarthy JF, Blow FC. Accidental poisoning mortality among patients in the department of veterans affairs health system. Med Care 2011;49:393–6.
. Bohnert AS, Ilgen MA, Ignacio RV, McCarthy JF, Valenstein M, Blow FC. Risk of death from accidental overdose associated with psychiatric and substance use disorders. Am J Psychiatry 2012;169:64–70.
. Carter GT, Javaher SP, Nguyen MH, Garret S, Carlini BH. Re-branding cannabis: the next generation of chronic pain medicine? Pain Manag 2015;5:13–21.
. Cerda M, Bordelois PM, Keyes KM, Galea S, Koenen KC, Pardini D. Cumulative and recent psychiatric symptoms as predictors of substance use onset: does timing matter? Addiction 2013;108:2119–28.
. Chou R, Fanciullo GP, Fine PG, Adler JA, Ballantyne JC, Davies P, et al. Clinical guidelines for the use of chronic opioid therapy in chronic noncancer pain. J Pain 2009;10:113–30.
. Chou R, Turner JA, Devine EB, Hansen RN, Sullivan SD, Blazina I, Dana T, Bougatsos C, Deyo RA. The effectiveness and risks of long-term opioid therapy for chronic pain: a systematic review for a National Institutes of Health Pathways to Prevention Workshop. Ann Intern Med 2015;162:276–86.
. Compton WM, Jones CM, Baldwin GT. Relationship between nonmedical prescription-opioid use and heroin use. N Engl J Med 2016;374:154–63.
. Crowley RA, Kirschner N; Health and Public Policy Committee of the American College of Physicians. The integration of care for mental health, substance abuse, and other behavioral health conditions into primary care: executive summary of an American College of Physicians position paper. Ann Intern Med 2015;163:298–9.
. Currie SR, Hodgins DC, Crabtree A, Jacobi J, Armstrong S. Outcome from integrated pain management treatment for recovering substance abusers. J Pain 2003;4:91–100.
. Dart RC, Surrat HL, Cicero TJ, Parrino MW, Severtson G, Bucher-Bartelson B, Green JL. Trends in opioid analgesic abuse and mortality in the United States. N Engl J Med 2015;372:241–8.
. Deyo RA, Von Korff M, Duhrkoop D. Opioids for low back pain. BMJ 2015;350:g6380.
. Dobscha SK, Morasco BJ, Duckart JP, Macey T, Deyo RA. Correlates of prescription opioid initiation and long-term opioid use in veterans with persistent pain. Clin J Pain 2013;29:102–8.
. Dowell D, Haegerich TM, Chou R. CDC guideline for prescribing opioids for chronic pain—United States, 2016. MMWR Recomm Rep 2016;65:1–49.
. Edlund MJ. Chronic opioid therapy for chronic noncancer pain in the United States: long day's journey into the night? Gen Hosp Psychiatry 2011;33:416–18.
. Edlund MJ, Steffick D, Hudson T, Harris KM, Sullivan M. Risk factors for clinically recognized opioid abuse and dependence among veterans using opioids for chronic non-cancer pain. PAIN 2007;129:355–62.
. Edlund MJ, Austen MA, Sullivan MD, Martin BC, Williams JS, Fortney JC, Hudson TJ. Patterns of opioid use for chronic noncancer pain in the Veterans Health Administration from 2009 to 2011. PAIN 2014;155;2337–43.
. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care 1998;36:8–27.
. Gourlay DL, Heit HA, Almahrezi A. Universal precautions in pain medicine: a rational approach to the treatment of chronic pain. Pain Med 2005;6:107–12.
. Hawkins EJ, Malte CA, Baer JS, Kivlahan DR. Prevalence, predictors, and service utilization of patients with recurrent use of veterans affairs substance use disorder specialty care. J Subst Abuse Treat 2012;43:221–30.
. Ho DE, Imai K, King G, Stuart EA. Matching as nonparametric preprocessing for reducing model dependence in parametric causal inference. Polit Anal 2007;15:199–236.
. Ilgen MA, Bohnert AS, Chermack S, Conran C, Jannausch M, Trafton J, Blow FC. A randomized trial of a pain management intervention for adults receiving substance use disorder treatment. Addiction 2016;111:1385–93.
. Ilgen MA, Haas E, Czyz E, Webster L. Treating chronic pain in veterans presenting to an addictions treatment program. Cog Behav Pract 2011;18:149–60.
. Ilgen MA, Peron B, Czyz EK, McCammon RJ, Trafton J. The timing of onset of pain and substance use disorders. Am J Addict 2010;19:409–15.
. Ilgen MA, Trafton JA, Humphreys K. Response to methadone maintenance treatment of opiate dependent patients with and without significant pain. Drug Alcohol Depend 2006;82:187–93.
. Jane-Llopis E, Matytsina I. Mental health and alcohol, drugs and tobacco: a review of the comorbidity between mental disorders and the use of alcohol, tobacco and illicit drugs. Drug Alcohol Rev 2006;25:515–36.
. Katz NP, Sherburne S, Beach M, Rose RJ, Vielguth J, Bradley J, Fanciullo GJ. Behavioral monitoring and urine toxicology testing in patients receiving long-term opioid therapy. Anesth Analg 2003;97:1097–102.
. Kobus AM, Smith DH, Morasco BJ, Johnson ES, Yang X, Petrik AF, Deyo RA. Correlates of higher-dose opioid medication use for low back pain in primary care. J Pain 2012;13:1131–9.
. Lacey S, Watson BR, Riffe D, Lovejoy J. Issues and best practices in content analysis. J Mass Commun Q 2015;92:791–811.
. Larochelle MR, Zhang F, Ross-Degnan D, Wharam JF. Rates of opioid dispensing and overdose after introduction of abuse-deterrent extended-release oxycodone and withdrawal of propoxyphene. JAMA Intern Med 2015;175:978–87.
. Levy B, Paulozzi L, Mack KA, Jones CM. Trends in opioid analgesic-prescribing rates by specialty, U.S., 2007–2012. Am J Prev Med 2015;49:409–13.
. Manchikanti L, Helm S, Fellows B, Janata JW, Pampati V, Grider JS, Boswell MV. Opioid epidemic in the United States. Pain Physician 2012;15:ES9–ES38.
. Martin BC, Fan MY, Edlund MJ, Devries A, Braden JB, Sullivan MD. Long-term chronic opioid therapy discontinuation rates from the TROUP study. J Gen Intern Med 2011;26:1450–7.
. Morasco BJ, Corson K, Turk DC, Dobscha SK. Association between substance use disorder status and pain-related function following 12 months of treatment in primary care patients with musculoskeletal pain. J Pain 2011;12:352–9.
. Morasco BJ, Duckart JP, Carr TP, Deyo RA, Dobscha SK. Clinical characteristics of veterans prescribed high doses of opioid medications for chronic non-cancer pain. PAIN 2010;151:625–32.
. Morasco BJ, Turk DC, Donovan DM, Dobscha SK. Risk for prescription opioid misuse among patients with a history of substance use disorder. Drug Alcohol Depend 2013;127:193–9.
. Morrill R, Cromartie J, Hart LG. Metropolitan, urban, and rural commuting areas: toward a better depiction of the U.S. settlement system. Urban Geogr 1999;20:727–48.
. Mosher HJ, Krebs EE, Carrel M, Kaboli PJ, Wander Weg MW, Lund BC. Trends in prevalent and incident opioid receipt: an observational study in Veterans Health Administration 2004–2012. J Gen Intern Med 2015;30:597–604.
. Palyo SA, Schopmeyer KA, McQuaid JR. Tele-pain management: use of videoconferencing technology in the delivery of an integrated cognitive-behavioral and physical therapy group intervention. Psychol Serv 2012;9:200–2.
. Rosenbaum PR, Rubin DB. Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. Am Stat 1985;39:33–8.
. Rubin DB. Matching to remove bias in observational studies. Biometrics 1973;29:159–83.
. Rubin DB. Using multivariate matched sampling and regression adjustment to control bias in observational studies. J Am Stat Assoc 1979;74:318–28.
. Seal KH, Cohen G, Waldrop A, Cohen BE, Maguen S, Ren L. Substance use disorders in Iraq and Afghanistan veterans in VA healthcare, 2001–2010: implications for screening diagnosis, and treatment. Drug Alcohol Depend 2011;116:93–101.
. Stuart EA. Matching methods for causal inference: a review and a look forward. Stat Sci 2010;25:1–21.
. Sullivan MD, Howe CQ. Opioid therapy for chronic pain in the United States: promises and perils. PAIN 2013;153(suppl 1):S94–S100.
. Turk DC, Wilson HD, Cahana A. Predicting opioid misuse by chronic pain patients: a systematic review and literature synthesis. Clin J Pain 2008;24:497–508.
. Turner JA, Saunders K, Shortreed SM, LeResche L, Riddell K, Rapp SE, Von Korff M. Chronic opioid therapy urine drug testing in primary care: prevalence and predictors of aberrant results. J Gen Intern Med 2014;29:1663–71.
. VA/DoD. VA/DoD clinical practice guideline for management of opioid therapy for chronic pain. Washington, DC: Veterans Administration, 2010. Available at: http://http://www.va.gov
/painmanagement/docs/cpg_opioidtherapy_fulltext.pdf. Accessed December 21, 2016.
. Vanderlip ER, Sullivan MD, Edlund MJ, Martin BC, Fortney J, Austen M, Williams JS, Hudson T. National study of discontinuation of long-term opioid therapy among veterans. PAIN 2014;155:2673–9.
. Weisner CM, Campbell CI, Ray GT, Saunders K, Merrill JO, Banta-Green C, Sullivan MD, Sliverberg MJ, Mertens JR, Boudreau D, Von Korff M. Trends in prescribed opioid therapy for non-cancer pain for individuals with prior substance use disorders. PAIN 2009;145:287–93.
. Worley J. Prescription drug monitoring programs, a response to doctor shopping: purpose, effectiveness, and directions for future research. Issues Ment Health Nurs 2012;33:319–28.