The dramatic rise in sales of prescription opioids in the United States over the past 2 decades has coincided with the rise in opioid overdose deaths and other opioid-related adverse outcomes.5,6,23 Prescribing of higher total daily dosages of opioids3,9,11 and co-prescribing of benzodiazepines7,24 have been associated with greater likelihood of opioid overdose. Unfortunately, these practices remain prevalent in U.S. health care systems.20 In response, there has been interest in developing guidelines and other interventions, including the recent Centers for Disease Control and Prevention (CDC) guidelines on opioid prescribing for chronic pain, to help providers better weigh risks and benefits when prescribing opioids.8
Guideline-based interventions for providers have been shown to reduce use of long-acting opioid medications10 and to reduce high-dosage opioid prescribing.29 However, most research has been limited to studies of local and state-specific interventions, and impacts of such interventions are unknown in larger health care systems, which likely encounter unique challenges and opportunities for implementation. The Veterans Health Administration (VHA) is the largest integrated health care system in the United States with 141 facilities, and it serves a national patient population at elevated risk for overdose.1 The Veterans Health Administration leadership developed and implemented the Opioid Safety Initiative (OSI) to promote safer opioid-related prescribing in the VHA. The OSI was built around a computerized “dashboard” tool that aggregates electronic medical record data so the leadership at each facility could audit the data and provide feedback using facility-, provider-, and patient-level opioid prescribing data. Key targets of the OSI included reducing the number of patients receiving higher daily opioid dosages and those receiving concurrent benzodiazepines. Piloted versions of this intervention in select VHA facilities were associated with decreased use of high-dosage regimens, but each of these local sites supplemented the OSI with additional elements that were not implemented nationwide.17,31
This study used an interrupted time series design to examine the impact of OSI implementation nationally on the number and percent of patients receiving high-dosage opioid and concurrent benzodiazepine regimens. The interrupted time series approach examines the trends in these outcomes prior to OSI and whether the intervention was associated with a significant change in these trends. Facility-level data were used to examine the impact of the OSI, which was implemented at the facility level, and variability of changes was also examined across facilities.
2.1. Opioid Safety Initiative in the Veterans Health Administration
Early in 2013, the Under Secretary for Health in the VHA chartered a task force to develop and deploy an opioid surveillance system that provides information about opioid prescribing to all VHA facilities. The goals were to develop tools to provide information and to support facilities and their providers in improving safety for patients receiving opioids. The VHA had previously developed guidelines on use of opioids for chronic pain, which were widely available to providers.27 In addition, the task force recommended a newly developed OSI dashboard tool that aggregates electronic medical record opioid-related prescribing data.17 The dashboard tool specifically provides information on high-dosage opioid prescribing and concurrent benzodiazepine prescribing. This analytic tool is designed to assist educational and quality improvement practices to enhance safety of opioid-related prescribing. At least 1 key leader was identified at each facility who was required to review the report, identify prescribing variation, and take appropriate action to enhance patient safety. These actions may include notifying providers and providing feedback to promote safe prescribing, while maintaining an individualized approach to treatments. The determination of appropriateness of prescribing was left to the specific facilities, given that the case mix may require higher use of opioids at some facilities and within particular clinical settings (eg, specialty pain clinics).
October 2013 was the end of the rollout of the OSI dashboard tool across all VHA facilities. In this study, outcomes were analyzed at the facility level on a monthly basis divided into 2 segments—preintervention (October 2012-September 2013) and post-intervention (October 2013-September 2014). In late 2014, VHA disseminated additional goals to facilitate safer prescribing of opioids, including increasing use of effective urine drug screening and facilitating use of state prescription drug monitoring databases; the outcomes of these later goals were not specifically evaluated in this study.
2.2. Design and data sources
This study is a retrospective system-wide interrupted time series analysis. Outpatient-prescription medication fill data included the fill date, days' supply, formulation, dose, quantity, and instructions for use of medications. Outpatient-prescription medication data came from VHA's Pharmacy Benefits Management service. VHA patients who filled an outpatient prescription for opioids from October 2012 to September 2014 in each of the 141 VHA facilities were included in these analyses. Study protocols were approved by the Ann Arbor VA's Institutional Review Board.
This analysis focused on total daily opioid dosages at thresholds >100 morphine equivalents (MEQ) and >200 MEQ because these thresholds have been associated with increased risk for unintentional overdose and are displayed on the dashboard tool.3,9,12 Patients who had total daily opioid dosages above threshold on any day of the month were considered to be above threshold for that month. Analyses also evaluated the number of patients receiving a benzodiazepine concurrently with an opioid.
Opioid medications included were codeine, morphine, oxycodone, hydrocodone, oxymorphone, hydromorphone, fentanyl, meperidine, pentazocine, propoxyphene, butorphanol, levorphanol, nalbuphine, tapentadol, and methadone. These medications were converted to morphine-equivalent milligrams using conversion ratios compiled by the CDC National Center for Injury Prevention and Control.21 Dosage calculations also accounted for the difference between routes of administration. Methods to calculate the prescribed daily opioid dosage each month during the study period followed the methods of prior research in this study population,3 which are very similar to the dashboard tool; although there may have been minor differences between methods for how to calculate MEQ for overlapping doses. However, this should not impact estimates of changes over time because the same dosage summation strategy was used across the study period. The calculation of opioid dosages used an “as-prescribed” approach, which assumed that patients take all prescribed opioids at the dosage and on the schedule described in their prescriptions; for prescriptions written for “as needed”/prn use, the pills are assumed to be consumed at the maximum frequency permitted by the prescription.
Benzodiazepines included were alprazolam, chlordiazepoxide, clonazepam, diazepam, lorazepam, temazepam, estazolam, flurazepam, oxazepam, quazepam, and triazolam. Patients were excluded if the only opioids prescribed were tramadol (because of the unique mechanism of action and evidence suggesting lower risk for overdose2,32), methadone prescribed for opioid substitution, or buprenorphine (because it is mainly on formulary for opioid use disorder treatment in the VHA13). Patients who received buprenorphine in addition to another opioid were excluded from the analysis from the time of their first buprenorphine fill forward.
Interrupted time series analyses with segmented regressions were used to estimate changes in the levels and trends of opioid prescribing following implementation of the OSI. This method controls for baseline levels and trends to estimate change in outcomes after intervention.18,30 The primary outcomes are the monthly change in high-dosage opioid and concurrent benzodiazepine prescribing post-OSI compared to pre-OSI. October 2013 was the deadline for implementation of the OSI dashboard tool. The data comprised measures of opioid prescribing in 141 VHA facilities each month. Segmented regression analysis was used to estimate the additional change in number of patients receiving daily opioids >100 MEQ and >200 MEQ and co-prescriptions of benzodiazepines in the post-intervention period compared to the pre-intervention period. In addition, because we found that there was a decreasing trend in number of outpatients prescribed any opioids, we examined a ratio with the value for each outcome divided by the total number of outpatients each month who obtained an opioid prescription.
Analyses accounted for clustering of observations over time using the autoregression procedure. If the Durbin–Watson statistic for first-order autocorrelation was significant, the model controlled for the autocorrelation parameter in the segmented regression model. Analyses also examined distributions of the average change in differences before and after OSI ([post-OSI level − pre-OSI level]/pre-OSI level) across facilities. All analyses were performed using SAS version 9.4 and SAS Enterprise Guide 6.1.
2.5. Sensitivity analyses
Sensitivity analyses assessed robustness of findings when varying assumptions. First, we excluded data from a 3-month period from October 2013 to end of December 2013 to examine the impact of accounting for a longer intervention phase-in period. Second, data from 4 geographic regions, known as Veterans Integrated Service Networks (VISNs), were removed because these VISNs were known to have piloted OSI-related interventions prior to the national rollout. Third, the preimplementation period was expanded to include data for 24 months prior to OSI implementation in October 2013. Finally, we also examined changes associated with the OSI in subgroups of patients on chronic opioid therapy and those not receiving opioids chronically. Chronic opioid use was defined as receiving opioids for at least 70 of the previous 90 days, and number of patients with chronic use and not with chronic use and meeting the high-dosage opioid or concurrent benzodiazepine outcome criteria were examined.29
3.1. Overall trends
Across the VHA, in October 2012, 55,722 patients received daily dosages of opioids >100 MEQ, which decreased to 46,780 patients in September 2014 (Fig. 1A) for an overall reduction of 16.05% (and 0.67% per month). In October 2012, 19,952 patients received total daily dosages of opioids >200 MEQ (Fig. 1B), which decreased to 15,121 patients by September 2014 (24.21% reduction overall and 1.01% per month). In October 2012, 112,907 patients received benzodiazepines concurrently with opioids (Fig. 1C), which decreased to 89,564 patients by September 2014 (20.67% reduction overall and 0.86% per month). Across the same time period, the total number of outpatients receiving opioids (ie, the denominator for all ratios) also declined from 571,476 to 514,883 patients (9.90% reduction overall and 0.41% per month), despite increases in the patient population of VHA nationally.28
3.2. Changes in total number of patients on high-dosage opioids and concurrent benzodiazepines
Segmented regression analyses were conducted to examine the extent to which the implementation of the OSI in October 2013 was associated with shifts in opioid-related prescribing. Prior to OSI implementation, there was already a decreasing trend of 218 (95% confidence interval [CI] −251 to −185) patients per month receiving daily opioid dosages >100 MEQ (Table 1). After OSI, there was a further decrease on top of the decreasing trend pre-OSI of 331 patients (95% CI −378 to −284) each month (see the last column of Table 1). Prior to OSI implementation, there was a decreasing trend of 131 (−147, −115) patients per month with prescribed dosages >200 MEQ. The OSI was associated with a significant immediate increase of 221 patients in the month of implementation (95% CI 66-376), but over time, was associated with a further decrease of 164 (95% CI −186 to −142) patients per month. With concurrent benzodiazepines, prior to OSI implementation, there was a decreasing trend of 574 (95% CI −707 to −441) patients per month with an additional decrease of 781 (95% CI −970 to −592) patients per month after OSI implementation.
3.3. Changes in proportion of patients receiving high-dosage opioids and concurrent benzodiazepines among all patients receiving opioids
Finally, we examined the number of patients receiving high-dosage opioids and concurrent benzodiazepines as a proportion of the total number of outpatients receiving any opioids. The OSI was associated with a further estimated decrease of 0.04% (95% CI −0.06 to −0.02) each month in percent of patients receiving >100 MEQ and a decrease of 0.02% (95% CI −0.03 to −0.01) per month in percent of patients receiving daily dosages >200 MEQ (Table 1). The OSI was also associated with a further decrease of 0.11% per month (95% CI −0.15 to −0.07) in percent of patients concurrently prescribed benzodiazepines.
3.4. Examining changes in opioid-related prescribing across facilities
The spread in the percent change in patients receiving daily opioid dosages >100 MEQ across facilities was −30.80% to +26.06%, with median −8.22% and quartiles (−13.09% to −3.53%) (Fig. 2). The spread of the percent change in patients receiving daily opioid dosages >200 MEQ across facilities ranged from −43.02% to +18.02%, with a median of −11.67% and quartiles (−18.53% to −5.35%). The spread of the percent change in patients receiving concurrent benzodiazepines across facilities ranged from −66.69% to +9.98, with a median of −10.42% and quartiles (−14.76% to −6.86%). The great majority of all facilities had reductions in total numbers of patients and proportion of patient receiving high-dosage opioid and concurrent benzodiazepines (see supplemental Table 1, available online at http://links.lww.com/PAIN/A380).
3.5. Sensitivity analyses
First, when analyses accounted for an implementation period by excluding data from October to December 2013, the immediate level change at time of intervention was no longer significant for the >200 MEQ threshold, but other results remained similar to primarily analyses (see supplemental Table 2, available online at http://links.lww.com/PAIN/A380). Second, when the 4 VISNs that had piloted the OSI were excluded, results were similar to those of primary analyses in that there was a significant further decrease in high-dosage opioid prescriptions and concurrent benzodiazepines filled after OSI. Third, results were also similar when the pre-OSI segment was expanded 1 year earlier to October 2011, with post-OSI changes even greater in magnitude. Finally, outcomes were examined in subgroups of patients receiving opioids chronically and those not receiving opioids chronically. Main results remained similar when restricting analysis to the subgroup of patients receiving chronic opioids. This group represented over 75% of patients on high-dosage opioids and over 65% of patients receiving concurrent benzodiazepines. Among patients on chronic opioids, the OSI was associated with significant decreases in patients receiving high-dosage opioids and concurrent benzodiazepines. However, there was no longer a significant downward trend in these outcomes prior to the OSI. Among patients not receiving opioids chronically, the OSI was not associated with further decreases in the trend compared to the already decreasing trend prior to OSI.
Across the national Veterans Health Administration, from October 2012 to September 2014, there was on average a decrease in number of patients receiving risky opioid regimens. In this 2-year time period, there was a 16% reduction in patients receiving opioid dosages >100 MEQ, a 24% reduction in patients receiving opioid dosages >200 MEQ, and a 21% reduction in patients receiving concurrent benzodiazepines. This is the first study, of which we are aware, that has examined an intervention promoting safer opioid use throughout a national health care system. The national implementation of the OSI dashboard tool was associated with a significant decrease in these outcomes, which highlights the possibility of system-wide approaches to address high-risk opioid prescribing. However, a large number of patients remained on these regimens at the end of the study period, which also underscores the challenges of making meaningful change in the context of health care systems that treat large numbers of complex patients.
Given that there is tremendous heterogeneity in prescribers' attitudes on use of opioids for chronic pain26 and wide variability in prevalence of opioid use across hospital systems and providers,14,16 provider-focused interventions could potentially lead to safer opioid prescribing. A distinct aspect of the OSI is that it supplements the dissemination of guidelines by also providing tools to facilitate audit and feedback to improve prescribing. Although the present design does not allow for the determination of the effectiveness of specific elements of the intervention, it is important to note that the intervention went above and beyond guidelines alone, suggesting that a combination of elements may be important. In particular, provider and facility-level audit and feedback interventions, which provide summaries of provider-level clinical performance, have been shown to have significant impacts on practice and patient-level outcomes across other clinical interventions.15
Although the OSI was associated with significant changes in high-dosage opioid prescribing, changes were more modest compared to interventions in localized health systems. Piloted versions of the OSI, which encompassed a more comprehensive intervention than the nationally implemented OSI, were associated with a 1.9% per month decrease in patients >200 MEQ in one VHA health system31 and 2.2% in another17 (calculated from primary results) compared to 1.0% in our results. In addition, a study of a Washington state-based insurance group practice initiative to improve opioid prescribing found that there were greater reductions in high-dosage opioid prescribing by the physicians who received additional initiatives promoting safer opioid prescribing compared to other physicians who were just affected by state-wide initiatives.29 There may be a “dose-response” effect, where additional components of an intervention may produce larger effects. It may also be that interventions, such as the OSI, targeting larger health systems across states known to have a high degree of variation in opioid-related outcomes, may, on average, have more modest effects overall.25
The OSI was associated with significant reductions in high-dosage opioid and concurrent benzodiazepines specifically in the large subgroup receiving opioids chronically. It is important to consider that appropriate tapering from opioids should largely occur gradually to minimize risk of withdrawal. There were significant reductions across all targeted outcomes in those receiving opioids chronically, but greater reductions may be seen over longer follow-up time as more time is allowed for patients to taper from very high-opioid dosages to a point that results in a dosage below the 200 MEQ or 100 MEQ thresholds. Overall, these results suggest that there may be opportunity to reduce dosages in patients receiving opioids chronically. Perhaps, the ongoing nature of their care allows for opportunity in tapering. The OSI was not associated with reductions in high-dosage opioid and concurrent benzodiazepine use in those receiving shorter-term opioid treatment. There may be a subsample of patients receiving high-dosage opioids or benzodiazepines concurrently with opioids because of an acute reason. Further work is needed to examine possible factors contributing to these findings.
This evaluation of the OSI also demonstrated wide variability in outcomes across facilities. Implementation of the OSI occurred at the facility level, and some facilities had large decreases in use of high-dosage opioid regimens, while others showed little change or increased use, though we cannot conclude that this was directly related to the OSI. Similar to prior studies that have shown variation in guideline implementation across facilities,4 different facilities may have prioritized addressing different opioid dosage thresholds; a facility with fewer resources may have elected to focus on patients only on dosages of opioids much higher than 100 MEQ. Although audit and feedback and education and training were discussed as the main tools to implement the OSI, because no specific guidance on requirements of implementation were given to facility leaders, actual implementation likely differed across facilities and deserves further study. For example, some may have offered more intensive education and other resources to providers; future work could seek to measure and quantify differences in implementation. From our analyses, the overall number of patients receiving opioids declined, so it is possible that some facilities prioritized decreasing overall use. This could make the proportion of patients on high-dosage regimens appear higher. Allowing facilities more autonomy in implementation decisions could have increased the local buy-in. However, this flexibility may have allowed some facilities to de-emphasize key national goals of the OSI. Determining the appropriate mix of flexibility and mandatory elements is an important step in increasing broader adoption and impact of the OSI and related initiatives.
This study had several limitations. Timing of OSI implementation across facilities was likely heterogeneous. However, our results were almost identical when we re-examined the data after excluding 4 VISNs that were known to have piloted the intervention earlier and when we factored in a 3-month phase-in period for the intervention. Second, the OSI did not have a concurrent control group of facilities. The strength of the segmented regression analysis is that it allowed us to control for patterns prior to the date of implementation. Nonetheless, it is possible that other influences, such as increasing media and public awareness about risks of opioids through events like deaths of notable figures, could have contributed to the changes in prescribing. There had been ongoing research findings suggesting risks of opioids,22 but the highly publicized CDC guidelines were released after the study period.8 Finally, this was a study of patients receiving care in the VHA. Our results may not generalize to a different health care system, though the integrated nature of the VHA system and electronic medical records create a unique opportunity to examine dissemination of large-scale intervention projects. In addition, system-level interventions like the OSI have been associated with safer prescribing in other health care systems, and other studies of system-level interventions are underway.19 In addition, a portion of VHA patients (ie, those with additional insurance) may also have access to care outside of the VHA, and it is possible that the OSI may have been associated with some of those patients pursuing opioid treatment outside of the VA, which warrants future study.
In conclusion, this national evaluation of the VHA's implementation of the OSI to promote safer opioid prescribing found that the intervention was associated with reductions in high-dosage opioid regimens and concurrent benzodiazepine prescribing. In addition, the degree of change in these outcomes varied notably across facilities. As policy-makers continue to consider how to reduce adverse outcomes associated with opioids, system-wide interventions employing guidelines along with tools to provide feedback on provider behavior can potentially be useful. However, enhancing the impact of such tools will likely require a greater understanding of potential barriers to implementation.
Conflict of interest statement
The authors have no conflicts of interest to declare.
This work was supported by VA QUERI Grant No. RRP 13-251.
The abstract of this manuscript was presented at the College on Problems of Drug Dependence in 2016.
Supplemental Digital Content
Supplemental Digital Content associated with this article can be found online at http://links.lww.com/PAIN/A380.
Video content associated with this article can be found online at http://links.lww.com/PAIN/A381.
. 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, Trafton JA, Kerns RD, Eisenberg A, Ganoczy D, Blow FC. Trends and regional variation in opioid
overdose mortality among Veterans Health Administration patients, fiscal year 2001 to 2009. Clin J Pain 2014;30:605–12.
. Bohnert AS, Valenstein M, Bair MJ, Ganoczy D, McCarthy JF, Ilgen MA, Blow FC. Association between opioid
prescribing patterns and opioid
overdose-related deaths. JAMA 2011;305:1315–21.
. Buscaglia AC, Paik MC, Lewis E, Trafton JA. Baseline variation in use of VA/DOD clinical practice guideline recommended opioid
prescribing practices across VA health care systems. Clin J Pain 2015;31:803–12.
. Compton WM, Boyle M, Wargo E. Prescription opioid
abuse: problems and responses. Prev Med 2015;80:5–9.
. Compton WM, Volkow ND. Major increases in opioid
analgesic abuse in the United States: concerns and strategies. Drug Alcohol Depend 2006;81:103–7.
. Dasgupta N, Funk MJ, Proescholdbell S, Hirsch A, Ribisl KM, Marshall S. Cohort study of the impact of high-dose opioid
analgesics on overdose mortality. Pain Med 2015;17:85–98.
. Dowell D, Haegerich TM, Chou R. CDC guideline for prescribing opioids for chronic Pain-United States, 2016. JAMA 2016;315:1624–45.
. Dunn KM, Saunders KW, Rutter CM, Banta-Green CJ, Merrill JO, Sullivan MD, Weisner CM, Silverberg MJ, Campbell CI, Psaty BM, Von Korff M. Opioid
prescriptions for chronic pain and overdose: a cohort study. Ann Intern Med 2010;152:85–92.
. Garcia MM, Angelini MC, Thomas T, Lenz K, Jeffrey P. Implementation of an opioid
management initiative by a state Medicaid program. J Manag Care Spec Pharm 2014;20:447–54.
. Gomes T, Mamdani MM, Dhalla IA, Paterson JM, Juurlink DN. Opioid
dose and drug-related mortality in patients with nonmalignant pain. Arch Intern Med 2011;171:686–91.
. Goodman F. Helping to achieve safe medication use: potential dose-related risk of opioid
deaths in veterans. Medication safety in seconds: a monthly publication from VA MedSAFE. 2011;1:1–2.
. Goodman A, Kivlahan D, Liberto J, McNicholas L, Oslin D, Saxon A, Trafton J. Buprenorphine/Naloxone and Buprenorphine for opioid
dependence—criteria for use in for office based opioid
treatment settings: VHA Pharmacy Benefits Management, 2014. http://http://www.pbm.va.gov
/clinicalguidance/criteriaforuse/Buprenorphine_for_Opioid_Dependence_or_Opioid_Use_Disorder_Criteria_for_Use.pdf. Accessed 9 September 2016.
. Herzig SJ, Rothberg MB, Cheung M, Ngo LH, Marcantonio ER. Opioid
utilization and opioid
-related adverse events in nonsurgical patients in US hospitals. J Hosp Med 2014;9:73–81.
. Ivers N, Jamtvedt G, Flottorp S, Young JM, Odgaard-Jensen J, French SD, O'Brien MA, Johansen M, Grimshaw J, Oxman AD. Audit and feedback: effects on professional practice and healthcare outcomes. Cochrane Database Syst Rev 2012;6:CD000259.
. Kim H, Hartung DM, Jacob RL, McCarty D, McConnell KJ. The concentration of opioid
prescriptions by providers and among patients in the Oregon medicaid program. Psychiatr Serv 2016;67:397–404.
. Kryskalla J, Kern S, Gray D, Hauser P. Using dashboard technology to monitor overdose risk: federal practitioner. Fed Pract 2014;31:8–14.
. 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.
. Lasser KE, Shanahan C, Parker V, Beers D, Xuan Z, Heymann O, Lange A, Liebschutz JM. A multicomponent intervention
to improve primary care provider adherence to chronic opioid
therapy guidelines and reduce opioid
misuse: a cluster randomized controlled trial protocol. J Subst Abuse Treat 2016;60:101–9.
. Liu Y, Logan JE, Paulozzi LJ, Zhang K, Jones CM. Potential misuse and inappropriate prescription practices involving opioid
analgesics. Am J Manag Care 2013;19:648–65.
. National Center for Injury Prevention and Control. CDC compilation of opioid
analgesic formulations with morphine milligram equivalent conversion factors, 2014 Available at: version. https://http://www.cms.gov
-Morphine-EQ-Conversion-Factors-March-2015.pdf. Accessed 7 January 2016.
. Nuckols TK, Anderson L, Popescu I, Diamant AL, Doyle B, Di Capua P, Chou R. Opioid
prescribing: a systematic review and critical appraisal of guidelines for chronic pain. Ann Intern Med 2014;160:38–47.
. Okie S. A flood of opioids, a rising tide of deaths. N Engl J Med 2010;363:1981–5.
. Park TW, Saitz R, Ganoczy D, Ilgen MA, Bohnert AS. Benzodiazepine
prescribing patterns and deaths from drug overdose among US veterans receiving opioid
analgesics: case-cohort study. BMJ 2015;350:h2698.
. Rudd RA, Aleshire N, Zibbell JE, Gladden RM; National Center for Injury Prevention and Control, CDC. Increases in drug and opioid
overdose deaths—United States, 2000–2014. MMWR Morb Mortal Wkly Rep 2016;64:1378–82.
. Tamayo-Sarver JH, Dawson NV, Cydulka RK, Wigton RS, Baker DW. Variability in emergency physician decision making about prescribing opioid
analgesics. Ann Emerg Med 2004;43:483–93.
. The Management of Opioid
Therapy for Chronic Pain Working Group. VA/DoD clinical practice guideline: management of opioid
therapy for chronic pain: Department of Veterans Affairs and Department of Defense, 2010. http://http://www.va.gov
/painmanagement/docs/cpg_opioidtherapy_summary.pdf. Accessed 7 January 2016.
. U.S. Department of Veteran
Affairs. National center for veterans analysis and statistics. number of veteran
patients by healthcare priority group: FY2000 to FY2014. Available at: http://http://www.va.gov
/vetdata/Utilization.asp. Accessed July 16, 2016.
. Von Korff M, Dublin S, Walker RL, Parchman M, Shortreed SM, Hansen RN, Saunders K. The impact of opioid
risk reduction initiatives on high-dose opioid
prescribing for patients on chronic opioid
therapy. J Pain 2016;17:101–10.
. Wagner AK, Soumerai SB, Zhang F, Ross-Degnan D. Segmented regression analysis of interrupted time series studies in medication use research. J Clin Pharm Ther 2002;27:299–309.
. Westanmo A, Marshall P, Jones E, Burns K, Krebs EE. Opioid
dose reduction in a VA health care system–implementation of a primary care population-level initiative. Pain Med 2015;16:1019–26.
. Zedler B, Xie L, Wang L, Joyce A, Vick C, Kariburyo F, Rajan P, Baser O, Murrelle L. Risk factors for serious prescription opioid
-related toxicity or overdose among veterans health administration patients. Pain Med 2014;15:1911–29.