BMJ Open Is Committed to Open Peer Review. As Part of This Commitment We Make the Peer Review History of Every Article We Publish Publicly Available

BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

BMJ Open is committed to open peer review. As part of this commitment we make the peer review history of every article we publish publicly available.

When an article is published we post the peer reviewers’ comments and the authors’ responses online. We also post the versions of the paper that were used during peer review. These are the versions that the peer review comments apply to.

The versions of the paper that follow are the versions that were submitted during the peer review process. They are not the versions of record or the final published versions. They should not be cited or distributed as the published version of this manuscript.

BMJ Open is an open access journal and the full, final, typeset and author-corrected version of record of the manuscript is available on our site with no access controls, subscription charges or pay-per-view fees (http://bmjopen.bmj.com).

If you have any questions on BMJ Open’s open peer review process please email [email protected] http://bmjopen.bmj.com/ on September 26, 2021 by guest. Protected copyright. BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

Characterization of Concurrent Use of Prescription Opioids and Benzodiazepine/Z Drugs In Alberta: A Population- Based Study ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2019-030858

Article Type: Research

Date Submitted by the 04-Apr-2019 Author:

Complete List of Authors: Sharma, Vishal; University of Alberta, School of Public Health Weir, Daniala; Mcgill Clinical and Health Informatics, Samanani, Salim; Okaki Health Intelligence, Simpson, Scot H.; University of Alberta, Pharmacy and Pharmaceutical Sciences Gilani, Fizza; The College of Physicians & Surgeons of Alberta Jess, Ed; The College of Physicians & Surgeons of Alberta, Eurich, Dean; University of Alberta, School of Public Health

opioids, benzodiazepines, benzodiazepine receptor agonists, concurrent Keywords: use, PUBLIC HEALTH, concurrent prescribing

http://bmjopen.bmj.com/ on September 26, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 27 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Characterization of Concurrent Use of Prescription Opioids and Benzodiazepine/Z Drugs In Alberta: A 4 Population-Based Study 5 6 7 8 Author list 9 10 Vishal Sharma (0000-0001-7907-1183), Daniala L Weir (0000-0001-7044-2443), Salim Samanani (0000-0001- 11 6751-4805), Scot Simpson (0000-0002-9880-2129), Fizza Gilani (0000-0002-1708-7451), Ed Jess, Dean Eurich 12 (0000-0003-2197-0463) 13 14 Address for each author 15 16 2-040 Li Ka Shing CenterFor for Health peer Research Innovation,review School of only Public Health, University of Alberta, 17 Edmonton, Alberta, Canada, T6G 2E1 Vishal Sharma BPharm MSc Candidate, Clinical and Health 18 Informatics Research Group, Department of Epidemiology, Biostatistics and Occupational Health, McGill 19 University, Montreal, Quebec, Canada Daniala L Weir PhD Candidate, OKAKI Health intelligence, Calgary, 20 21 Alberta, Canada Salim Samanani CEO and Medical Director OKAKI, Faculty of Pharmacy and 22 Pharmaceutical Sciences, University of Alberta, Edmonton, Alberta, Canada, T6G 2E1 Scot Simpson 23 professor, The College of Physicians & Surgeons of Alberta, Edmonton, Alberta, Canada, T5J 0N3 Fizza 24 Gilani Program Manager Prescribing & Analytics, The College of Physicians & Surgeons of Alberta, 25 Edmonton, Alberta, Canada, T5J 0N3 Ed Jess Director of Prescribing & Analytics, 2-040 Li Ka Shing Center 26 27 for Health Research Innovation, School of Public Health, University of Alberta, Edmonton, Alberta, 28 Canada, T6G 2E1 Dean Eurich professor 29 30 Corresponding Author: 31 32 Dean Eurich, 2-040 Li Ka Shing Center for Health Research Innovation, University of Alberta, Edmonton,

33 Alberta, Canada, T6G 2E1; Phone 780-492-6333; fax 780-492-7455; email: [email protected] http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 26, 2021 by guest. Protected copyright. 42 ACKNOWLEDGEMENT – This study is based in part on data provided by Alberta Health and was 43 commissioned by the College of Physicians & Surgeons of Alberta (CPSA). The interpretation and 44 conclusions contained herein are those of the researchers and do not necessarily represent the views of the 45 Government of Alberta or CPSA. Neither the Government of Alberta nor Alberta Health expresses any 46 47 opinion in relation to this study. 48 49 50 51 52 53 54 55 56 57 58 1 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 27 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Contributors: VS DE SS and DLW were involved in the conception and design of the study. VS and DE 4 analyzed the data. VS and DE drafted the article. EJ SS FG DLW and SS revised the article. All authors 5 6 gave final approval of the version to be published. The corresponding author attests that all listed 7 authors meet authorship criteria and that no others meeting the criteria have been omitted. DE is the 8 guarantor. 9 10 Funding: This study received no funding. 11 12 Copyright/license for publication: The Corresponding Author has the right to grant on behalf of all 13 authors and does grant on behalf of all authors, a worldwide licence to the Publishers and its licensees in 14 perpetuity, in all forms, formats and media (whether known now or created in the future), to i) publish, 15 reproduce, distribute, display and store the Contribution, ii) translate the Contribution into other 16 For peer review only 17 languages, create adaptations, reprints, include within collections and create summaries, extracts 18 and/or, abstracts of the Contribution, iii) create any other derivative work(s) based on the Contribution, 19 iv) to exploit all subsidiary rights in the Contribution, v) the inclusion of electronic links from the 20 Contribution to third party material where-ever it may be located; and, vi) licence any third party to do 21 22 any or all of the above. 23 24 Competing Interest: All authors have completed the ICMJE uniform disclosure form at 25 www.icmje.org/coi_disclosure.pdf and declare: no support from any organisation for the submitted 26 27 work; Salim Samanani has received research grants from the College of Physicians & Surgeons 28 29 of Alberta; no other relationships or activities that could appear to have influenced the submitted 30 31 work. 32

33 Ethical approval: This study was approved by the health ethics research board at the University of http://bmjopen.bmj.com/ 34 Alberta (#Pro00083807). 35 36 Data Sharing: The data used in this study is not available for external analysis. However, administrative 37 health data can be accessed from AH by following defined research protocols and confidentiality 38 agreements. 39 40 Transparency: The lead author (the manuscript's guarantor) affirms that the manuscript is an honest,

41 accurate, and transparent account of the study being reported; that no important aspects of the study on September 26, 2021 by guest. Protected copyright. 42 have been omitted; and that any discrepancies from the study as originally planned (and, if relevant, 43 44 registered) have been explained. 45 46 Summary Box: 47 Section 1: What is already known on this topic 48 49 Prescribed opioids are a major contributor to the opioid crisis and a substantial number of opioid related 50 deaths also involve benzodiazepines. 51 52 Numerous health authorities and medical associations warn against concurrent prescribing of opioids 53 54 and benzodiazepines. 55 56 57 58 2 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 27 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Although it is known that women, the elderly and chronic opioid users have a higher prevalence of 4 concurrent use, characterization of concurrent use of benzodiazepines within the opioid using 5 6 population has not been adequately carried out using potential markers of inappropriate use. 7 8 Section 2: What this study adds 9 The prevalence of concurrent use of opioids and benzodiazepines is alarmingly high especially among 10 11 chronic and high dose opioid users, high users of the health care system, opioid dependency treatment 12 patients, and the elderly. 13 14 The warnings against concurrent use of opioids and benzodiazepines from health authorities and 15 medical associations need to be further emphasized through education and policies. 16 For peer review only 17 More health care resources must be allocated towards treatment programs for substance abuse 18 disorders, chronic pain and mental health. 19 20 21 22 Word Count: 2909 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 3 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 27 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Abstract 4 5 6 Objective: The objective of this study is to characterize concurrent use of benzodiazepines among 7 8 prescription opioid users in Alberta in 2017. 9 10 11 Design: A population based retrospective study. 12 13 14 Setting: Alberta, Canada in the year 2017. 15 16 For peer review only 17 Participants: All individuals in Alberta, Canada with at least one dispensation record from a community 18 19 pharmacy for an opioid in the year 2017. 20 21 22 Exposure: Concurrent use with benzodiazepines, defined as overlap of opioids and benzodiazepines for 23 24 25 one or more days. 26 27 Main outcome measures: Prevalence of concurrency was estimated among subgroups of patient 28 29 30 characteristics that were considered clinically relevant or associated with inappropriate medication use. 31 32 Results: Among the 547,709 Albertans who were dispensed opioid prescriptions in 2017, 132,156 (24%) 33 http://bmjopen.bmj.com/ 34 35 also received prescriptions for BZRAs (Benzodiazepine Receptor Agonists). There were 96,581 (17.6%) 36 37 prescription opioid users who concurrently used BZRAs with an average of 98 days (sd=114, 95% CI 97- 38 39 99) of total cumulative concurrency and a median of 37 days (IQR 10-171). The average longest duration 40

41 on September 26, 2021 by guest. Protected copyright. 42 of consecutive days of concurrency was 45 (sd=60, 95% CI 44.6-45.4) with a median of 24 days (IQR 8- 43 44 59). Concurrency was more prevalent in females, patients using an average daily oral morphine 45 46 equivalent >90 mg, opioid dependence therapy patients, chronic opioid users, patients with a high 47 48 number of unique providers, lower median household incomes, and those older than 65 (p-value < 49 50 0.001 for all comparisons). 51 52 53 Conclusions: Concurrent prescribing of opioids and BZRAs is common in Alberta despite the ongoing 54 55 56 guidance of many clinical resources. Older patients, those taking higher doses of opioids, and for longer 57 58 4 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 27 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 durations may be at particular risk of adverse outcomes and may be worthy of closer follow-up for 4 5 6 assessment for dose tapering or discontinuations. Continued surveillance of concurrent use of these 7 8 medications is warranted to ensure that safe drug use recommendations are being followed by health 9 10 providers. 11 12 13 Strengths and Limitations 14 15 16 Strengths of this study:For peer review only 17 18 19  Large population-based sample using administrative database with near complete capture of all 20 21 opioid and BZRA dispensations occurring in the community 22 23  Analysis included prevalence of concurrency among subgroups of patient characteristics 24 25 26 including daily oral morphine equivalents, opioid dependence treatment, duration of opioid use 27 28 and number of unique providers. 29 30 31 Limitations of this study: 32

33 http://bmjopen.bmj.com/ 34  This study assumed that patients took their medications as prescribed and captured in the 35 36 database. This has never been validated. 37 38  This study assumed that days supply was entered correctly by pharmacies when calculating oral 39 40 morphine equivalents and daily defined doses. This has never been validated.

41 on September 26, 2021 by guest. Protected copyright. 42 43  Information on the indication for concurrent prescribing was not available from the 44 45 administrative database. 46 47 48 Introduction 49 50 51 Canada has among the highest rates of opioid prescribing in the world and since 1980, the 52 53 volume of opioids sold to hospitals and pharmacies has increased by 3000% despite increasing 54 55 recognition of the significant prescribing risks associated with such practices, including fatal overdoses, 56 57 58 5 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 27 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 dependency, motor vehicle collisions, and falls and fractures among the elderly.[1-3] Individuals older 4 5 6 than 65 are especially prone to the consequences of opioids, with this group accounting for 63% of 7 8 unintentional opioid poisonings and having the highest rate of opioid poisoning hospitalizations.[3, 4] A 9 10 similar picture exists for benzodiazepines and Z-drugs (zopiclone and zolpidem), collectively known as 11 12 BZRAs. BZRAs are one of the most widely prescribed psychotropic compounds for anxiety disorders and 13 14 insomnia.[5] Canadian clinical practice guidelines for the management of anxiety disorders and 15 16 For peer review only 17 insomnia suggest that BZRA treatment is appropriate for short term use in adults (aged 20-64) and in 18 19 some cases as second line treatment.[6, 7] Use of BZRAs outside of these recommendations is 20 21 considered “potentially inappropriate” given the potential for adverse effects, especially in those over 22 23 65.[5, 6, 8] For example, the risk of motor vehicle accidents, falls and hip fractures leading to 24 25 26 hospitalization and death can more than double in older adults taking benzodiazepines.[9] A 2006 study 27 28 in British Columbia found that 3.5% of the population were considered “long term” users of 29 30 benzodiazepines and 47% were over the age of 65.[10] Furthermore, a recent study reported that 10% 31 32 of Albertans in 2015 received a BZRA with the prevalence of use increasing with age.[11]

33 http://bmjopen.bmj.com/ 34 35 Given the similar concerns with prescribing and associated adverse outcomes, concurrent use of 36 37 38 opioids and BZRA’s is strongly discouraged for most patients.[1, 12, 13] Few studies, however, have 39 40 evaluated the characteristics of concurrent use. Only one American study, that we are aware of, found

41 on September 26, 2021 by guest. Protected copyright. 42 that concurrency was more common in chronic opioid users, women and the elderly.[14] However, this 43 44 study did not stratify concurrent use by daily oral morphine equivalents. No studies were found that 45 46 used Canadian data. There are no specific clinical guidelines on indications for concurrent use of these 47 48 49 medications and in fact, expert perspectives warn that opioids and BZRAs should very rarely be 50 51 prescribed together. [1, 12, 15] Furthermore, studies and safe medication use guidelines have identified 52 53 concurrent use of these medications as a risk factor for fatal opioid overdose.[3, 12] In Canada, national 54 55 56 57 58 6 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 27 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 and provincial initiatives have aimed at reducing inappropriate opioid and BZRA prescribing, as well as 4 5 6 decreasing the potential for harm. [1, 12, 13] 7 8 Alberta has implemented procedures around the individual prescribing of opioids and BZRAs. 9 10 11 Both of these medication classes are actively monitored in Alberta by the Triplicate Prescription Program 12 13 (TPP), a prescription drug monitoring program in which prescribers must register with in order to 14 15 prescribe a TPP medication. However, previous literature suggests that BZRAs and opioids cannot be 16 For peer review only 17 targeted by safe use policies in isolation.[16] There is very little published data on concurrent use, and 18 19 20 none in Alberta, Canada. Thus, the objective of this study is to expand our understanding of concurrent 21 22 use of opioids and BZRAs by characterizing the prevalence of concurrent use among opioid users using 23 24 administrative data from the province of Alberta in 2017. 25 26 27 Methods 28 29 30 Study Population 31 32

33 This study included all individuals in Alberta with at least one dispensation record from community http://bmjopen.bmj.com/ 34 35 pharmacies for an opioid between January 1, 2017 to December 31, 2017. 36 37 38 Data Sources 39 40

41 Data from Alberta Netcare’s Pharmaceutical Information Network (PIN) was used for this study. PIN on September 26, 2021 by guest. Protected copyright. 42 43 data includes >95% of all dispensations from community pharmacies in Alberta irrespective of insurance 44 45 coverage, thus providing comprehensive data on all medication dispensations (from all prescriber types) 46 47 48 occurring in the province outside of the hospital setting.[17] Information on dispensed medication (drug 49 50 identification number, dispense date, days supply, quantity, strength, name, directions), patient (age, 51 52 sex, unique patient identifier) and prescriber (type and license number) was available. The validity of the 53 54 days supply variable for each dispensation was evaluated to ensure it fell within a plausible clinical range 55 56 57 58 7 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 27 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 based on the defined daily dose for a single dispensation; less than 0.01% of the days supply values were 4 5 6 deemed to be outside of this range and a new days supply was imputed based on an individual’s 7 8 historical average for a particular ingredient. All unique identifiers (patient, prescriber) were 9 10 anonymized for the purposes of this analysis which was approved by the health ethics research board at 11 12 the University of Alberta (#Pro00083807). 13 14 15 Study Measures 16 For peer review only 17 18 All opioid and BZRA dispensations were retrieved from PIN for 2017. An opioid user was defined as 19 20 anyone who received at least 1 dispensation for an opioid. Similarly, a BZRA user was defined as anyone 21 22 23 who received at least 1 dispensation for a BZRA. Prevalence of concurrent use was estimated among 24 25 subgroups of patient characteristics that were considered clinically relevant or associated with 26 27 inappropriate medication use. Chronic opioid use was defined as total opioid days greater than 90, as 28 29 others have [1], or more than 10 opioid dispensations in one year. An opioid dependency treatment 30 31 (ODT) user was anyone that was dispensed a prescription for methadone or buprenorphine/naloxone. 32

33 http://bmjopen.bmj.com/ 34 Postal codes (forward sortation index) were used to categorize individuals as rural/urban and into 35 36 income categories (<50k, 50-75k, 75-100k, 100-125k, >125k). Average daily oral morphine equivalents 37 38 (OME’s) and number of daily defined doses (DDDs) were calculated for all opioids and BZRAs, 39 40 respectively, using the conversion factors specified by the TPP[18]. Methadone and buprenorphine 41 on September 26, 2021 by guest. Protected copyright. 42 43 were excluded from OME specific analyses. We used daily OME thresholds of <50, 50-90 and > 90 as 44 45 categories in our analyses since these are clinically accepted in the guidelines for determining the 46 47 risk/benefit profile when prescribing opioids for pain.[12] 48 49 50 The key variable of interest was whether an opioid user also used a BZRA concurrently in 2017. 51 52 Although we were not able to directly observe utilization of these medications by individuals, we 53 54 55 considered “use” as any day on which an individual had a supply of medication on hand based on the 56 57 58 8 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 27 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 date and days supply of each dispensation. Using the dispensation information from PIN, we generated 4 5 6 binary variables for each day of the year to indicate if it was “covered” by an opioid or BZRA. Beginning 7 8 on the dispensation day, each day was categorized as covered until the end of the days supplied. For 9 10 each patient, a day was categorized as concurrent if it was covered by both an opioid and BZRA. We 11 12 then calculated the number of days, both cumulative and consecutive, that were categorized as 13 14 concurrent. For example, if a patient received a 30-day opioid dispensation on Jan 1 and a 20-day BZRA 15 16 For peer review only 17 dispensation on Jan 20, this would be quantified as 11 days of concurrent use. In our main analyses, 18 19 concurrency was defined as having 1 or more days categorized as concurrent. 20 21 22 Statistical Analyses 23 24 25 We conducted a descriptive analysis to examine the characteristics of concurrent use of opioids and 26 27 BZRAs. All summary statistics were calculated using the denominator of total population of opioid users 28 29 in Alberta for 2017. 30 31 32 The measure of interest was prevalence of concurrent use by age, sex, average daily OME thresholds

33 http://bmjopen.bmj.com/ 34 (<50, 50-90, >90), duration of opioid use (chronic (as defined previously) vs intermittent), opioid 35 36 37 dependence therapy (ODT), rural vs urban residence , number of unique providers, median annual 38 39 household income thresholds, and number of DDD’s of BZRAs. Analyses were also stratified by the total 40

41 days of cumulative concurrency (1-7, 8-30, 31-90, >90) and consecutive days of concurrency (1-7, 8-30, on September 26, 2021 by guest. Protected copyright. 42 43 31-60, 61-90, >90). We used chi2 tests of independence to compare prevalence proportions between the 44 45 46 different groups in the above-named characteristics. All analyses were performed using STATA/MP 13.1 47 48 (StataCorp., College Station, TX). 49 50 51 Patient and Public Involvement 52 53 54 This research was done without patient involvement. Patients were not invited to comment on the 55 56 study design and were not consulted to develop patient relevant outcomes or interpret the results. 57 58 9 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 27 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Patients were not invited to contribute to the writing or editing of this document for readability or 4 5 6 accuracy. There are no plans to disseminate the results of the research to study participants. 7 8 9 10 11 Results 12 13 14 There were 547,709 Albertans who received at least one dispensation for an opioid and qualified as an 15 16 opioid user (Figure 1).For Females peerrepresented 53%review (n=292,396) of opioidonly users, 18% (n=98,083) were over 17 18 19 the age of 65, and the majority of patients were from urban areas (84%) (Table 1). Overall, 20% 20 21 (n=108,604) of opioid users were considered chronic users and ODT patients represented 1.7% (n=9139) 22 23 of opioid users. When methadone and buprenorphine were excluded, 88% (n=468,863), 9.5% 24 25 (n=51,033) and 2.8% (n=14,933) represented those in the <50, 50-90, and >90 OME categories, 26 27 28 respectively. A substantial number of patients received opioids from 3 or more pharmacies (7%) or from 29 30 3 or more prescribers (16%). 31 32

33 Among the 547,709 opioid users, 24% (n=132,156) received a BZRA and 17.6% (n=96,581) had at least http://bmjopen.bmj.com/ 34 35 one day of concurrent use of an opioid and a BZRA during 2017. The mean total days of concurrency 36 37 over the entire year was 98 (SD=114) days (median of 37 (IQR 10-171)). Among patients with 38 39 40 concurrent use, a substantial number had high durations of concurrent use during the year; 53% had

41 on September 26, 2021 by guest. Protected copyright. 42 over 30 days of concurrency and 36% had over 90 days of concurrency (Table 2). When we examined 43 44 the duration of consecutive days of concurrency, the mean longest duration was 45 (SD=60) with a 45 46 median of 24 days (IQR 8-59). Most concurrent patients (64%) had concurrent use for less than 30 47 48 consecutive days (Table 2). 49 50 51 Differences in concurrency were noted based on sex, urban/rural status, and median household 52 53 54 incomes, with the prevalence of concurrency being highest among the lowest incomes, as well as a 55 56 strong trend in age (Table 1, Figure 2). Indeed, <2% of all opioid users under the age of twenty used a 57 58 10 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 27 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 BZRA concurrently relative to nearly 30% of those over the age of 65. The highest concurrence was 4 5 6 observed in the highest age groups, who are also most at risk of severe adverse events (Table 3). 7 8 Concurrency was more common in chronic opioid users compared to intermittent users (Table 1). 9 10 Similarly, chronic opioid users had a higher number of concurrent days in the year compared to 11 12 intermittent users (Table 4). 13 14 15 Characteristics associated with potentially inappropriate use of opioids (e.g., older age, high OME’s, 16 For peer review only 17 multiple providers) had substantially higher concurrent BZRA use (Table 1, Figures 2-4). Although the 18 19 20 absolute number of patients using an average daily OME >90 was low (2.8% of opioid users), 46% had 21 22 concurrent use with a BZRA. Among concurrent users in the > 90 OME category, 58.8% had concurrent 23 24 use >90 days (Figure 3) and 12.8% of those with > 90 days of concurrent use were also taking > 90 OME 25 26 per day (Figure 4). 27 28 29 There were also clear trends with respect to providers. As the number of unique providers increased, so 30 31 too did the prevalence of concurrency. Although the absolute numbers were low (<5%), the opioid 32

33 http://bmjopen.bmj.com/ 34 users that visited more than 5 pharmacies or prescribers in 2017 both had a prevalence of concurrency 35 36 of 62% (Table 1). Opioid users who received a BZRA, either concurrently or not, visited more providers 37 38 compared to those who received only an opioid or BZRA (Figure 5). Interestingly, among concurrent 39 40 users, 78% (n=74882) received an opioid and BZRA from the same prescriber and 94% (n=90561) from 41 on September 26, 2021 by guest. Protected copyright. 42 43 the same pharmacy. Moreover, 58% of concurrent users (n=56098) received an opioid and BZRA on the 44 45 same day from the same prescriber while 64% (n=61715) received an opioid and BZRA from the same 46 47 pharmacy on the same day. 48 49 50 The trend between number of DDD’s of BZRAs and concurrency is similar to that with average daily 51 52 OME’s. Most of the opioid patients concurrently used a BZRA at the lowest number of DDD’s (66%). 53 54 55 However, around 88% of those using >2-3 times the DDD were concurrent users (Table 1). 56 57 58 11 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 27 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Interpretation 4 5 6 Many reputable clinical resources indicate that BZRAs should not be combined with opioids, yet this 7 8 study showed that nearly 20% of patients using an opioid did so in combination with a BZRA in Alberta 9 10 11 [1, 12, 13]. Those on >90 mg OME had the highest prevalence of concurrency when compared to lower 12 13 doses. Moreover, among concurrent users, total days of concurrency was high with about half of these 14 15 patients using opioids and BZRAs at the same time for more than 30 days. Perhaps not so surprising is 16 For peer review only 17 the high prevalence of concurrency in those with a greater number of distinct prescribers. In addition, 18 19 20 our observation of a higher prevalence of concurrency in chronic opioid users compared to intermittent 21 22 users was expected since prolonged opioid use provided more opportunities for concurrent use of 23 24 BZRAs. These results should be concerning to clinicians and policy makers because the potential for 25 26 adverse outcomes associated with opioid use is greatly increased since a significant proportion of opioid 27 28 29 fatalities involve BZRAs.[3, 19] 30 31 Our results are similar to an American study that showed a prevalence of concurrent use ranging from 32

33 http://bmjopen.bmj.com/ 34 6.8% to 9.6%.[14] The high prevalence of concurrent use among those over 65 and those taking >90 35 36 OME per day is especially concerning as these groups are at high-risk for adverse clinical outcomes. Our 37 38 finding that nearly 30% of opioid users > 65 using an opioid concurrently with a BZRA is higher than 39 40 others have observed. This same American study observed ~10-12% concurrent use among elderly 41 on September 26, 2021 by guest. Protected copyright. 42 43 patients using opioids from 2002-2014.[14] The reason for the discrepancy is unknown but could be 44 45 related to our inclusion of Z drugs in our analysis whereas the American study only included 46 47 benzodiazepines. Regardless, this finding for elderly adults is alarming since there are many clinical 48 49 guidelines that advise against prescribing BZRAs in most seniors, let alone in combination with an 50 51 52 opioid.[20, 21] Furthermore, patients aged 65 and older consistently have the highest rates of 53 54 hospitalization due to opioid poisoning.[4] Consistent with the previous American study[14], our study 55 56 also suggests that women and chronic opioid users are more likely to receive BZRAs concurrently with 57 58 12 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 27 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 opioids. Women are more likely to seek medical care compared to men and are therefore also more 4 5 6 likely to be diagnosed with a mental health disorder, which may party explain this result.[22, 23] 7 8 To date, we are unaware of other studies that have suggested those taking very high daily doses of 9 10 11 opioids (>90 OME per day) also have high concurrency rates. Irrespective of the reason for concurrent 12 13 use (i.e., opioid use disorder and doctor shopping or when used for more appropriate indications) the 14 15 evidence suggests that high dose opioid users have up to 5x the risk of overdose and those above 100 16 For peer review only 17 OME have a much higher risk of fatal over dose.[12, 24] Combining opioids and BZRAs in these groups 18 19 20 could certainly contribute to further adverse outcomes already at high rates. 21 22 23 Although concurrent use of opioids and BZRAs is often deemed clinically inappropriate, beyond 24 25 substance use disorder situations, one has to question why the observed prevalence is so high despite 26 27 the numerous efforts across the country, and in Alberta, to mitigate this high-risk prescribing. In the 28 29 groups with the highest concurrent use (females, ODT patients, chronic and high dose opioid users, 30 31 elderly, etc.), most, if not all, are known to have a higher prevalence of conditions related to pain and 32

33 http://bmjopen.bmj.com/ 34 mental health.[22, 23, 25]. Our results showed that 78% of concurrent users received both medications 35 36 from the same prescriber and 94% from the same pharmacy with over half receiving these drugs on the 37 38 same day. There is an opportunity here to educate providers about the risks of concurrent use and to 39 40 verify if concurrent use is truly appropriate. Furthermore, treatment emphasis in chronic pain and 41 on September 26, 2021 by guest. Protected copyright. 42 43 mental health patients is changing where opioids and BZRAs are no longer first line treatment options 44 45 and where integrated and multidisciplinary treatments are preferred.[26] Connecting patients with 46 47 these preferred treatment modalities is often difficult because of cost and time and often opioids and 48 49 /or BZRAs are used to address the unmet needs of patients. 50 51 52 The strengths of our study included the large population-based sample with near complete capture of all 53 54 55 opioid and BZRA dispensations occurring in community pharmacies within the province. Pharmacies in 56 57 58 13 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 27 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Alberta are mandated by the College of Pharmacy to ensure accurate prescription records such that use 4 5 6 of PIN data can accurately capture most, if not all, of the opioid and BZRA dispensations and the 7 8 information provided with each of these dispensations. Another strength is that our analyses included 9 10 average daily OME’s when characterizing concurrent use, something that we have not seen in other 11 12 studies. There are, however, some limitations in our study. First, we are assuming that patients took 13 14 their medications as dispensed. Medication adherence in opioid users is a challenging issue.[27] We 15 16 For peer review only 17 assumed that days supply was entered correctly by pharmacies when calculating our OME and DDD 18 19 values, however no validation of the PIN days supply field has been completed to date. Second, our 20 21 study was limited to descriptive analyses and does not provide outcomes data from concurrent use. 22 23 Clinically, there are instances where concurrent use may be considered appropriate, especially in 24 25 26 palliative care and cancer treatment settings. Information on the indications for concurrent prescribing 27 28 were not available in the PIN database used for this study. 29 30 31 Despite these limitations, Alberta still has an alarming prevalence of concurrent use. The opioid crisis in 32

33 Alberta and Canada is being driven in part, by prescription opioids.[28] However, due to wide spread http://bmjopen.bmj.com/ 34 35 attention to the opioid crisis, the number of opioid prescriptions and morphine milligram equivalents 36 37 38 prescribed sharply declined in all provinces in 2016 and 2017, including Alberta.[29, 30]. It is clear that 39 40 continued efforts are required to curb the concurrent utilization of opioids and BZRAs in the province,

41 on September 26, 2021 by guest. Protected copyright. 42 and elsewhere as it is unlikely Alberta is unique in this regard. Furthermore, as increasing clinical 43 44 emphasis is being placed on non-pharmacologic management of chronic pain and not prescribing 45 46 opioids to patients with mental-health disorders, as well as ongoing monitoring and educational 47 48 49 campaigns, we will hopefully see a decrease in concurrent use.[12, 13, 25, 31] 50 51 52 53 54 55 56 57 58 14 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 27 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Table 1. Summary statistics of prevalence of concurrency^ among opioid users and possible high -risk 4 markers. 5 6 Characteristic N (%) Prevalence of concurrency Prevalence of Percent of 7 (among opioid within characteristic. concurrency among all concurrent 8 users) Percent (n)# opioid users. Percent users (n) 9 (n=547709) 10 Opioid users 547709 (100) 17.6 (96581)## ------11 Gender: 12 Male 255293 (46.6) 14.9 (37955) 6.9 39.3 (37955) 13 Female 292396 (53.4) 20.0 (58620) 10.7 60.7 (58620) 14 Average daily 15 OME*: 16 <50 468863 (87.7) 15.7 (73411) 13.7 80.2 (73411) 50-90 For51033 (9.5)peer review22.1 (11287) only2.1 12.3 (11287) 17 >90 14933 (2.8) 46.2 (6899) 1.3 7.5 (6899) 18 Duration of opioid 19 use**: 20 Chronic 108604 (19.8) 47.1 (51214) 9.4 53.0 (51214) 21 Intermittent 439105 (80.2) 10.3 (45367) 8.3 47.0 (45367) 22 ODT 9139 (1.7) 37.2 (3401) 0.62 3.5 (3401) 23 Not on ODT 538570 (98.3) 17.3 (93180) 17.1 96.5 (93180) 24 Postal code zone: 25 Rural 85666 (15.6) 22.0 (18809) 3.4 19.5 (18809) 26 Urban 462043 (84.4) 16.8 (77772) 14.2 80.5 (77772) 27 Median Household 28 Income (x 1000) 29 <50 107240 (19.6) 23.1 (24781) 4.5 25.7 (24781) 50-75 261354 (47.2) 17.9 (46725) 8.5 48.4 (46725) 30 75-100 151352 27.6) 14.2 (21496) 3.9 22.3 (21496) 31 100-125 27314 (5.0) 12.9 (3514) 0.6 3.6 (3514) 32 >125 448 (0.08) 14.5 (65) 0.01 0.07 (65) 33 # of unique http://bmjopen.bmj.com/ 34 dispensing 35 pharmacies: 36 1 426557 (77.9) 12.0 (51413) 9.4 53.2 (51413) 37 2 82048 (15.0) 31.1 (25550) 4.7 26.4 (25550) 38 3 23155 (4.2) 45.3 (10482) 1.9 11.0 (10482) 39 4 8260 (1.5) 53.1 (4387) 0.80 4.5 (4387) 40 5+ 7689 (1.4) 61.8 (4749) 0.88 4.9 (4749) # of unique 41 on September 26, 2021 by guest. Protected copyright. prescribers: 42 1 352596 (64.4) 7.1 (25158) 4.6 26.0 (25158) 43 2 107347 (19.6) 25.9 (27805) 5.1 28.8 (27805) 44 3 42656 (7.8) 42.2 (17990) 3.3 18.6 (17990) 45 4 20126 (3.7) 50.5 (10163) 1.9 10.5 (10163) 46 5+ 24984 (4.6) 61.9 (15465) 2.8 16.0 (15465) 47 Age: 48 0-17 20366 (3.7) 1.5 (307) 0.06 0.3 (307) 49 18-65 429259 (78.4) 16.3 (70000) 12.8 72.5 (70000) 50 >65 98083 (17.9) 26.8 (26274) 4.8 27.2 (26274) 51 Number of BZRA 52 DDD’s 53 0-1 94192 (71.3) 67.4 (63531) 11.6 65.8 (63531) 1-2 30423 (23.0) 86.7 (26370) 4.8 27.3 (26370) 54 2-3 4761 (3.6) 89.1 (4243) 0.77 4.4 (4243) 55 >3 2780 (2.1) 87.7 (2437) 0.44 2.5 (2437) 56 57 58 15 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 27 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 ^ Concurrency is defined as 1 or more days of overlap between an opioid and BZRA. 4 5 *methadone and buprenorphine patients were excluded. 6 **chronic opioid users were defined by having at least 90 days of cumulative opioid use or at least 10 opioid prescriptions in 7 the year. This includes ODT patients. 8 9 # p-value for chi2 test of independence (difference between prevalence of concurrency between groups within characteristic) 10 <0.001 for all characteristics. 11 12 ## 95% confidence interval for prevalence of concurrency=17.5-17.7 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 16 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 27 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Table 2. Characteristics of concurrent use (n=96,581) 4 5 Characteristic % 6 7 Total days of cumulative concurrency 8 * * 9 Mean (SD) 98 (114) 10 1-7 21 11 8-30 26 12 31-90 17 13 >90 36 14 15 16 Longest Duration of consecutive concurrency For *peer review only * 17 Mean (SD) 45 (60) 18 1-7 24 19 8-30 40 20 31-60 13 21 61-90 8 22 23 >90 14 24 *Days 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 17 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 27 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 4 5 Table 3. Prevalence of Concurrency by Age group and total days of concurrency among opioid users (%) 6 7 8 9 Days of 10 Concurrency 11 Age Group 1-7 8-30 31-90 >90 Total (n=) 12 0-9 0.76 0.06 0.06 0.12 1669 13 14 10-19 1.4 0.4 0.1 0.1 30551 15 20-29 3 2 1 2 68710 16 30-39 For3 peer3 review2 4 only92549 17 40-49 4 4 3 7 93387 18 19 50-59 4 6 3 9 107917 20 60-69 4 6 4 10 83267 21 70-79 5 7 5 10 43973 22 80-89 6 9 6 9 21029 23 24 >90 8 8 5 9 4656 25 Total (n=) 20503 25614 15940 34524 547708 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 Table 4. Prevalence of Concurrency by category of Opioid Use and total days of concurrency. (p-value 36 37 < 0.001) 38 Days of cumulative % Intermittent users % Chronic users 39 40 concurrency (n=439,105) (n=108,604)

41 1-7 4.1 (18,163) 2.2 (2,340) on September 26, 2021 by guest. Protected copyright. 42 8-30 4.5 (19,712) 5.4 (5,902) 43 31-90 1.7 (7,492) 7.8 (8,448) 44 >90 0 (0) 31.8 (34,524) 45 46 47 48 49 50 51 52 53 54 55 56 57 58 18 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 27 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 References 4 5 1. Busse, J.W., et al., Guideline for opioid therapy and chronic noncancer pain. Canadian Medical 6 Association Journal, 2017. 189(18): p. E659-E666. 7 2. Guan, Q., et al., Evaluating the early impacts of delisting high-strength opioids on patterns of 8 prescribing in Ontario. Health Promotion and Chronic Disease Prevention in Canada, 2018. 38(6): 9 10 p. 256-262. 11 3. Belzak, L. and J. Halverson, Evidence synthesis - The opioid crisis in Canada: a national 12 perspective. Health Promotion and Chronic Disease Prevention in Canada, 2018. 38(6): p. 224- 13 233. 14 4. O’Connor, S., V. Grywacheski, and K. Louie, At-a-glance - Hospitalizations and emergency 15 department visits due to opioid poisoning in Canada. Health Promotion and Chronic Disease 16 Prevention inFor Canada, 2018.peer 38(6): p. 244-247.review only 17 5. CPSA, Clinical Toolkit Benzodiazepines: Use and Taper. CPSA, 2015. 18 19 6. Katzman, M.A., et al., Canadian clinical practice guidelines for the management of anxiety, 20 posttraumatic stress and obsessive-compulsive disorders. BMC Psychiatry, 2014. 14 Suppl 1: p. 21 S1. 22 7. Canadian Pharmacists Association. RxTx. 2019; Available from: https://www.e- 23 therapeutics.ca/search. 24 8. TOP, T.O.P. Guideline for Adult Primary Insomnia [Internet]. 2010; Available from: 25 http://www.topalbertadoctors.org/download/439/insomnia_management_guideline.pdf. 26 9. ChooseWiselyCanada. The Canadian Geriatrics Society has developed a list of 5 things physicians 27 28 and patients should question in geriatrics [Internet]. Available from: 29 https://choosingwiselycanada.org/geriatrics/. 30 10. Cunningham, C.M., G.E. Hanley, and S. Morgan, Patterns in the use of benzodiazepines in British 31 Columbia: examining the impact of increasing research and guideline cautions against long-term 32 use. Health Policy, 2010. 97(2): p. 122-129.

33 11. Weir, D.L., et al., Benzodiazepine receptor agonist dispensations in Alberta: a population-based http://bmjopen.bmj.com/ 34 descriptive study. CMAJ Open, 2018. 6(4): p. E678-E684. 35 12. ismp Canada. Essential Clinical Skills for Opioid Prescribers. 2017 Nov 2018]; Available from: 36 https://www.ismp-canada.org/download/OpioidStewardship/Opioid-Prescribing-Skills.pdf. 37 38 13. Centers for Disease Control and Prevention. Guideline for Prescribing Opioids for Chronic Pain. 39 2016; Available from: https://www.cdc.gov/drugoverdose/pdf/Guidelines_Factsheet-a.pdf. 40 14. Hwang, C.S., et al., Trends in the Concomitant Prescribing of Opioids and Benzodiazepines, 2002-

41 2014. Am J Prev Med, 2016. 51(2): p. 151-160. on September 26, 2021 by guest. Protected copyright. 42 15. Gudin, J.A., et al., Risks, Management, and Monitoring of Combination Opioid, Benzodiazepines, 43 and/or Alcohol Use. Postgraduate Medicine, 2013. 125(4): p. 115-130. 44 16. Jones, J.D., S. Mogali, and S.D. Comer, Polydrug abuse: a review of opioid and benzodiazepine 45 combination use. Drug and alcohol dependence, 2012. 125(1-2): p. 8-18. 46 47 17. Government of Alberta. Alberta Netcare EHR Usage Statistics. 2016 Jan 2019]; Available from: 48 http://www.albertanetcare.ca/Statistics.htm. 49 18. College of Physicians and Surgeons of Alberta. OME and DDD conversion factors. Available from: 50 http://www.cpsa.ca/wp-content/uploads/2017/06/OME-and-DDD-Conversion-Factors.pdf. 51 19. Sun, E.C., et al., Association between concurrent use of prescription opioids and benzodiazepines 52 and overdose: retrospective analysis. BMJ, 2017. 356: p. j760. 53 20. Farrell, K.P., et al., Deprescribing benzodiazepine receptor agonists. 2018. 54 55 56 57 58 19 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 27 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 21. Gallagher, P. and D. O'Mahony, STOPP (Screening Tool of Older Persons' potentially 4 inappropriate Prescriptions): application to acutely ill elderly patients and comparison with 5 6 Beers' criteria. Age Ageing, 2008. 37(6): p. 673-9. 7 22. Pinkhasov, R.M., et al., Are men shortchanged on health? Perspective on health care utilization 8 and health risk behavior in men and women in the United States. Int J Clin Pract, 2010. 64(4): p. 9 475-87. 10 23. Kessler, R.C., et al., Lifetime and 12-month prevalence of dsm-iii-r psychiatric disorders in the 11 united states: Results from the national comorbidity survey. Archives of General Psychiatry, 12 1994. 51(1): p. 8-19. 13 24. Ismp Canada, navigating-opioids-11x17-canada.pdf. 14 25. National Opioid Use Guideline Group (NOUGG) Canadian Guidelines for Safe and Effective Use of 15 16 Opioids for CNCP-PartFor B.peer 2010. review only 17 26. Centre for Effective Practice. Management of Chronic Non Cancer Pain. 2017; Available from: 18 thewellhealth.ca/cncp. 19 27. Graziottin, A., et al., Opioids: How to Improve Compliance and Adherence. Pain Practice, 2011. 20 11(6): p. 574-581. 21 28. Gomes, T., et al., Contributions of prescribed and non-prescribed opioids to opioid related 22 deaths: population based cohort study in Ontario, Canada. BMJ, 2018. 362: p. k3207. 23 29. Abdesselam, K., et al., At-a-glance - Opioid surveillance: monitoring and responding to the 24 25 evolving crisis. Health Promot Chronic Dis Prev Can, 2018. 38(9): p. 312-316. 26 30. College of Physicians and Surgeons of Alberta. Prescribing Trends. 2018 2018]; Available from: 27 http://www.cpsa.ca/physician-prescribing-practices/. 28 31. Centers for Disease Control and Prevention. Checklist for prescribing opioids for chronic pain. 29 2016; Available from: https://www.cdc.gov/drugoverdose/pdf/PDO_Checklist-a.pdf. 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 20 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 27 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Figure 1. Patient flow diagram of denominator used for analyses 4 5 6 785,240 BZRA & 7 Opioid Users 8 9 10 237,531 excluded 11 (never received an 12 opioid) 13 14 15 16 For peer547,709 review opioid users only 17 18 19 20 12,880 excluded (methadone & 21 buprenorphine users) 22 23 24 534,829 opioid users 25 (used for average 26 daily OME analyses) 27 28 29 Figure 2. Prevalence of concurrency by age group among all opioid users in 2017 (n=547,708). 30 31 PREVALENCE OF CONCURRENCY BY AGE GROUP 32

33 Overall Female Male http://bmjopen.bmj.com/ 34 35 35 36 37 30 38 39 25 40

41 on September 26, 2021 by guest. Protected copyright. 42 20 43 44 15 45 46 10 47 48 CONCURRENCY OF PREVALENCE 49 5 50 51 0 52 0 - 9 10- 19 20- 29 30- 39 40- 49 50- 59 60- 69 70- 79 80- 89 > 9 0 53 AGE GROUP 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 27 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Figure 3. Prevalence of concurrency by total days of concurrency and average daily OME category 4 in 2017 (n=91597)*. 5 6 7 Percent distribution of average daily OME categories across 8 9 days of concurrency 10 70 11 58.8 12 60 13 14 50 44.1 15 40 16 For peer29.4 review only 31 17 30 22.2 23.3 18 19.6 17.4 20 15.1 19 13.9 13 12.1 20

Prevalence Prevalence concurrencyof 10 21 22 0 23 1-7 8-30 31-90 90+ 24 Days of concurrency group 25 26 <50 OME 50-90 OME >90 OME 27

28 29 * Excludes methadone and buprenorphine. 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 27 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Figure 4. Percent distribution of average daily OME categories within categories of total cumulative 4 days of concurrent use (n=91597) * 5 6 7 BREAKDOWN OF TOTAL CUMULATIVE DAYS OF CONCURRENCY INTO 8 AVERAGE DAILY OME CATEGORIES 9 10 11 90+ 71.6 15.6 12.8 12 13 14 15 31-90 84.8 9.7 5.6 16 For peer review only 17 18 19 8-30 87.2 9 3.9 20 21 22 23 1-7 81.6 13.1 5.2 24 25 USE CONCURRENT OF DAYS TOTALCUMULATIVE 26 0 10 20 30 40 50 60 70 80 90 100 27 PERCENT 28 29 <50 OME 50-90 OME > 90 OME 30

31 32 * Excludes methadone and buprenorphine.

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 24 of 27 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 4 5 Figure 5. Distribution (percentage) of patient categories by number of unique prescribers. 6 7 8 Distribution of treatment groups across number of prescribers 9 90 10 11 80 76 77 12 13 14 70 15 16 60 For peer55 review only 17 18 50 19 20 Percent 40 21 22 29 30 23 26 24 18 18 19 20 17 16 25 14 26 11 27 10 6 5 4 4 28 1 2 1 2 29 0 30 1 2 3 4 5+ 31 Number of unique prescribers 32

33 http://bmjopen.bmj.com/ 34 35 BZRA dispensations only(n=237,531) 36 Opioid dispensations only(n=415,553) 37 Opioid & BZRA dispensations, no concurrency(n=35,575) Opioid & BZRA dispensations with concurrency (n=96,581) 38 39 40

41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 25 of 27 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

1 2 STROBE Statement—checklist of items that should be included in reports of observational studies

3 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from 4 5 Item Page 6 No Recommendation No 7 Title and abstract 1 (a) Indicate the study’s design with a commonly used term in the title or 1,4,5 8 the abstract 9 10 (b) Provide in the abstract an informative and balanced summary of what 4,5 11 was done and what was found 12 Introduction 13 14 Background/rationale 2 Explain the scientific background and rationale for the investigation being 5,6 15 reported 16 Objectives 3 State specific objectives, including any prespecified hypotheses 6 17 18 Methods For peer review only 19 Study design 4 Present key elements of study design early in the paper 6-9 20 Setting 5 Describe the setting, locations, and relevant dates, including periods of 6,7 21 recruitment, exposure, follow-up, and data collection 22 23 Participants 6 (a) Cohort study—Give the eligibility criteria, and the sources and 6,7 24 methods of selection of participants. Describe methods of follow-up 25 Case-control study—Give the eligibility criteria, and the sources and 26 27 methods of case ascertainment and control selection. Give the rationale 28 for the choice of cases and controls 29 Cross-sectional study—Give the eligibility criteria, and the sources and 30 methods of selection of participants 31 32 (b) Cohort study—For matched studies, give matching criteria and N/A 33 number of exposed and unexposed 34 Case-control study—For matched studies, give matching criteria and the 35 number of controls per case 36 Variables 7 Clearly define all outcomes, exposures, predictors, potential confounders, 7-9

37 http://bmjopen.bmj.com/ 38 and effect modifiers. Give diagnostic criteria, if applicable 39 Data sources/ 8* For each variable of interest, give sources of data and details of methods 7-9 40 measurement of assessment (measurement). Describe comparability of assessment 41 42 methods if there is more than one group 43 Bias 9 Describe any efforts to address potential sources of bias N/A 44 Study size 10 Explain how the study size was arrived at 6,7

45 on September 26, 2021 by guest. Protected copyright. 46 Quantitative variables 11 Explain how quantitative variables were handled in the analyses. If 7-9 47 applicable, describe which groupings were chosen and why 48 Statistical methods 12 (a) Describe all statistical methods, including those used to control for 8,9 49 confounding 50 51 (b) Describe any methods used to examine subgroups and interactions 8,9 52 (c) Explain how missing data were addressed 7 53 (d) Cohort study—If applicable, explain how loss to follow-up was N/A 54 addressed 55 56 Case-control study—If applicable, explain how matching of cases and 57 controls was addressed 58 Cross-sectional study—If applicable, describe analytical methods taking 59 account of sampling strategy 60 (e) Describe any sensitivity analyses N/A Continued on next page 1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 27 of 27 BMJ Open

1 2 Results 3 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from 4 Participants 13* (a) Report numbers of individuals at each stage of study—eg numbers 9,FIG.1 5 potentially eligible, examined for eligibility, confirmed eligible, included in the 6 study, completing follow-up, and analysed 7 (b) Give reasons for non-participation at each stage N/A 8 9 (c) Consider use of a flow diagram FIG.1 10 Descriptive 14* (a) Give characteristics of study participants (eg demographic, clinical, social) 9,11 11 data and information on exposures and potential confounders 12 (b) Indicate number of participants with missing data for each variable of interest N/A 13 14 (c) Cohort study—Summarise follow-up time (eg, average and total amount) N/A 15 Outcome data 15* Cohort study—Report numbers of outcome events or summary measures over N/A 16 time 17 18 Case-controlFor peer study—Report review numbers in each exposure only category, or summary N/A 19 measures of exposure 20 Cross-sectional study—Report numbers of outcome events or summary 9-11 21 measures 22 23 Main results 16 (a) Give unadjusted estimates and, if applicable, confounder-adjusted estimates 9-11 24 and their precision (eg, 95% confidence interval). Make clear which confounders 25 were adjusted for and why they were included 26 (b) Report category boundaries when continuous variables were categorized 8,9-11 27 28 (c) If relevant, consider translating estimates of relative risk into absolute risk for 29 a meaningful time period 30 Other analyses 17 Report other analyses done—eg analyses of subgroups and interactions, and 10 31 32 sensitivity analyses 33 Discussion 34 Key results 18 Summarise key results with reference to study objectives 11 35 36 Limitations 19 Discuss limitations of the study, taking into account sources of potential bias or 13

37 imprecision. Discuss both direction and magnitude of any potential bias http://bmjopen.bmj.com/ 38 Interpretation 20 Give a cautious overall interpretation of results considering objectives, 12,13,14 39 limitations, multiplicity of analyses, results from similar studies, and other 40 41 relevant evidence 42 Generalisability 21 Discuss the generalisability (external validity) of the study results 13,14 43 44 Other information Funding 22 Give the source of funding and the role of the funders for the present study and, N/A 45 on September 26, 2021 by guest. Protected copyright. 46 if applicable, for the original study on which the present article is based 47 48 49 *Give information separately for cases and controls in case-control studies and, if applicable, for exposed and 50 unexposed groups in cohort and cross-sectional studies. 51 52 Note: An Explanation and Elaboration article discusses each checklist item and gives methodological background and 53 54 published examples of transparent reporting. The STROBE checklist is best used in conjunction with this article (freely 55 available on the Web sites of PLoS Medicine at http://www.plosmedicine.org/, Annals of Internal Medicine at 56 http://www.annals.org/, and Epidemiology at http://www.epidem.com/). Information on the STROBE Initiative is 57 available at www.strobe-statement.org. 58 59 60

2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

Characterization of Concurrent Use of Prescription Opioids and Benzodiazepine/Z Drugs In Alberta, Canada: A Population-Based Study ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2019-030858.R1

Article Type: Research

Date Submitted by the 29-Jul-2019 Author:

Primary Subject Epidemiology Heading:

Secondary Subject Heading: Public health http://bmjopen.bmj.com/

opioids, benzodiazepines, concurrent use, PUBLIC HEALTH, concurrent Keywords: prescribing, Z-drugs

on September 26, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 28 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Characterization of Concurrent Use of Prescription Opioids and Benzodiazepine/Z Drugs In Alberta, 4 Canada: A Population-Based Study 5 6 7 8 Author list 9 10 Vishal Sharma (0000-0001-7907-1183), Daniala L Weir (0000-0001-7044-2443), Salim Samanani (0000-0001- 11 6751-4805), Scot Simpson (0000-0002-9880-2129), Fizza Gilani (0000-0002-1708-7451), Ed Jess, Dean Eurich 12 (0000-0003-2197-0463) 13 14 Address for each author 15 16 2-040 Li Ka Shing CenterFor for Health peer Research Innovation,review School of only Public Health, University of Alberta, 17 Edmonton, Alberta, Canada, T6G 2E1 Vishal Sharma BPharm MSc Candidate, Clinical and Health 18 Informatics Research Group, Department of Epidemiology, Biostatistics and Occupational Health, McGill 19 University, Montreal, Quebec, Canada Daniala L Weir PhD Candidate, OKAKI Health intelligence, Calgary, 20 21 Alberta, Canada Salim Samanani CEO and Medical Director OKAKI, Faculty of Pharmacy and 22 Pharmaceutical Sciences, University of Alberta, Edmonton, Alberta, Canada, T6G 2E1 Scot Simpson 23 professor, The College of Physicians & Surgeons of Alberta, Edmonton, Alberta, Canada, T5J 0N3 Fizza 24 Gilani Program Manager Prescribing & Analytics, The College of Physicians & Surgeons of Alberta, 25 Edmonton, Alberta, Canada, T5J 0N3 Ed Jess Director of Prescribing & Analytics, 2-040 Li Ka Shing Center 26 27 for Health Research Innovation, School of Public Health, University of Alberta, Edmonton, Alberta, 28 Canada, T6G 2E1 Dean Eurich professor 29 30 Corresponding Author: 31 32 Dean Eurich, 2-040 Li Ka Shing Center for Health Research Innovation, University of Alberta, Edmonton,

33 Alberta, Canada, T6G 2E1; Phone 780-492-6333; fax 780-492-7455; email: [email protected] http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 26, 2021 by guest. Protected copyright. 42 ACKNOWLEDGEMENT – This study is based in part on data provided by Alberta Health and was 43 commissioned by the College of Physicians & Surgeons of Alberta (CPSA). The interpretation and 44 conclusions contained herein are those of the researchers and do not necessarily represent the views of the 45 Government of Alberta or CPSA. Neither the Government of Alberta nor Alberta Health expresses any 46 47 opinion in relation to this study. 48 49 50 51 52 53 54 55 56 57 58 1 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 28 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

33 Ethical approval: This study was approved by the health ethics research board at the University of http://bmjopen.bmj.com/ 34 Alberta (#Pro00083807). 35 36 Data Sharing: The data sharing agreement specifies that the data cannot leave the province of Alberta 37 in any manner. Requests for data from Alberta Health related to the study can be sent to 38 [email protected]. The authors did not have any special access privileges to this data that 39 others would not have. 40

41 Transparency: The lead author (the manuscript's guarantor) affirms that the manuscript is an honest, on September 26, 2021 by guest. Protected copyright. 42 accurate, and transparent account of the study being reported; that no important aspects of the study 43 44 have been omitted; and that any discrepancies from the study as originally planned (and, if relevant, 45 registered) have been explained. 46 47 48 49 Word Count: 3074 50 51 52 53 54 55 56 57 58 2 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 28 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Abstract 4 5 6 Objective: The objective of this study is to characterize concurrent use of benzodiazepine receptor 7 8 modulators and opioids among prescription opioid users in Alberta in 2017. 9 10 11 Design: A population based retrospective study. 12 13 14 Setting: Alberta, Canada in the year 2017. 15 16 For peer review only 17 Participants: All individuals in Alberta, Canada with at least one dispensation record from a community 18 19 pharmacy for an opioid in the year 2017. 20 21 22 Exposure: Concurrent use of a benzodiazepine receptor modulator and opioid, defined as overlap of 23 24 25 supply for both drugs for at least one day. 26 27 Main outcome measures: Prevalence of concurrency was estimated among subgroups of patient 28 29 30 characteristics that were considered clinically relevant or associated with inappropriate medication use. 31 32 Results: Among the 547,709 Albertans who were dispensed opioid prescriptions in 2017, 132,156 (24%) 33 http://bmjopen.bmj.com/ 34 35 also received prescriptions for benzodiazepine receptor modulators. There were 96,581 (17.6%) 36 37 prescription opioid users who concurrently used benzodiazepine receptor modulators with an average 38 39 of 98 days (sd=114, 95% CI 97-99) of total cumulative concurrency and a median of 37 days (IQR 10- 40

41 on September 26, 2021 by guest. Protected copyright. 42 171). The average longest duration of consecutive days of concurrency was 45 (sd=60, 95% CI 44.6-45.4) 43 44 with a median of 24 days (IQR 8-59). Concurrency was more prevalent in females, patients using an 45 46 average daily oral morphine equivalent >90 mg, opioid dependence therapy patients, chronic opioid 47 48 users, patients utilizing a high number of unique providers, lower median household incomes, and those 49 50 older than 65 (p-value < 0.001 for all comparisons). 51 52 53 Conclusions: Concurrent prescribing of opioids and benzodiazepine receptor modulators is common in 54 55 56 Alberta despite the ongoing guidance of many clinical resources. Older patients, those taking higher 57 58 3 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 28 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 doses of opioids, and for longer durations may be at particular risk of adverse outcomes and may be 4 5 6 worthy of closer follow-up for assessment for dose tapering or discontinuations. As well, those with 7 8 higher health care utilization (seeking multiple providers) should also be closely monitored. Continued 9 10 surveillance of concurrent use of these medications is warranted to ensure that safe drug use 11 12 recommendations are being followed by health providers. 13 14 15 Strengths and Limitations 16 For peer review only 17 18 Strengths of this study: 19 20 21  Large population-based sample using administrative database with near complete capture of all 22 23 opioid and benzodiazepine receptor modulator dispensations occurring in the community 24 25  26 Analysis included prevalence of concurrency among subgroups of patient characteristics 27 28 including daily oral morphine equivalents, opioid dependence treatment, duration of opioid use 29 30 and number of unique providers. 31 32

33 Limitations of this study: http://bmjopen.bmj.com/ 34 35 36  This study assumed that patients took their medications as prescribed and captured in the 37 38 database. This has never been validated. 39 40  This study assumed that days supply was entered correctly by pharmacies when calculating oral

41 on September 26, 2021 by guest. Protected copyright. 42 43 morphine equivalents and daily defined doses. This has never been validated. 44 45  Information on the indication for concurrent prescribing was not available from the 46 47 administrative database. 48 49 50 Introduction 51 52 53 Canada has among the highest rates of opioid prescribing in the world and since 1980, the 54 55 volume of opioids sold to hospitals and pharmacies has increased by 3000% despite increasing 56 57 58 4 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 28 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 recognition of the significant prescribing risks associated with such practices, including fatal overdoses, 4 5 6 dependency, motor vehicle collisions, and falls and fractures among the elderly.[1-3] Individuals older 7 8 than 65 are especially prone to the consequences of opioids, with this group accounting for 63% of 9 10 unintentional opioid poisonings and having the highest rate of opioid poisoning hospitalizations.[3, 4] A 11 12 similar picture exists for benzodiazepines and Z-drugs (zopiclone and zolpidem), collectively known as 13 14  15 benzodiazepine receptor modulators because of their effects on -aminobutyric acid receptors.[5, 6] 16 For peer review only 17 Benzodiazepine receptor modulators are one of the most widely prescribed psychotropic compounds 18 19 for anxiety disorders and insomnia.[7] Canadian clinical practice guidelines for the management of 20 21 anxiety disorders and insomnia suggest that benzodiazepine receptor modulator treatment is 22 23 appropriate for short term use in adults (aged 20-64) and in some cases as second line treatment.[8, 9] 24 25 26 Use of benzodiazepine receptor modulators outside of these recommendations is considered 27 28 “potentially inappropriate” given the potential for adverse effects, especially in those over 65.[7, 8, 10] 29 30 For example, the risk of motor vehicle accidents, falls and hip fractures leading to hospitalization and 31 32 death can more than double in older adults taking benzodiazepines.[11] A 2006 study in British 33 http://bmjopen.bmj.com/ 34 35 Columbia found that 3.5% of the population were considered “long term” users of benzodiazepines and 36 37 47% were over the age of 65.[12] Furthermore, a recent study reported that 10% of Albertans in 2015 38 39 received a benzodiazepine with the prevalence of use increasing with age.[13] 40

41 on September 26, 2021 by guest. Protected copyright. 42 Given the similar concerns with prescribing and associated adverse outcomes, concurrent use of 43 44 opioids and benzodiazepine receptor modulators is strongly discouraged for most patients.[1, 14, 15] 45 46 47 Other studies have evaluated the characteristics of concurrent use. One American study found that 48 49 concurrency was more common in chronic opioid users, women and the elderly.[16] However, this 50 51 study did not stratify concurrent use by daily oral morphine equivalents. Two other studies using data 52 53 from the US further described a rising trend in concurrent use of opioids and benzodiazepine receptor 54 55 modulators.[17, 18] No studies were found that used Canadian data. There are no specific clinical 56 57 58 5 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 28 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 guidelines on indications for concurrent use of these medications and in fact, expert perspectives warn 4 5 6 that opioids and benzodiazepines should very rarely be prescribed together. [1, 14, 19] Furthermore, 7 8 studies and safe medication use guidelines have identified concurrent use of these medications as a risk 9 10 factor for fatal opioid overdose.[3, 14] In Canada, national and provincial initiatives have aimed at 11 12 reducing inappropriate opioid and benzodiazepine prescribing, as well as decreasing the potential for 13 14 harm. [1, 14, 15] 15 16 For peer review only 17 Alberta has implemented procedures around the individual prescribing of opioids and 18 19 20 benzodiazepine receptor modulators. Both of these medication classes have been actively monitored in 21 22 Alberta since 1986 through the Triplicate Prescription Program (TPP), a prescription drug monitoring 23 24 program in which prescribers must register with in order to prescribe a TPP medication. However, 25 26 previous literature suggests that benzodiazepine receptor modulators and opioids cannot be targeted by 27 28 29 safe use policies in isolation.[20] There is very little published data on concurrent use, and none in 30 31 Alberta, Canada. Thus, the objective of this study is to expand our understanding of concurrent use of 32

33 opioids and benzodiazepine receptor modulators by characterizing the prevalence of concurrent use http://bmjopen.bmj.com/ 34 35 among opioid users using administrative data from the province of Alberta in 2017. 36 37 38 Methods 39 40

41 Study Population on September 26, 2021 by guest. Protected copyright. 42 43 44 This study included all individuals in Alberta with at least one dispensation record from community 45 46 pharmacies for an opioid between January 1, 2017 to December 31, 2017. 47 48 49 Data Sources 50 51 52 Data from Alberta Netcare’s Pharmaceutical Information Network (PIN) was used for this study. PIN 53 54 data includes >95% of all dispensations from community pharmacies in Alberta irrespective of insurance 55 56 57 58 6 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 28 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 coverage, thus providing comprehensive data on all medication dispensations (from all prescriber types) 4 5 6 occurring in the province outside of the hospital setting.[21] Information on dispensed medication (drug 7 8 identification number, dispense date, days supply, quantity, strength, name, directions), patient (age, 9 10 sex, unique patient identifier) and prescriber (type and license number) was available. The validity of the 11 12 days supply variable for each dispensation was evaluated to ensure it fell within a plausible clinical range 13 14 based on the defined daily dose for a single dispensation; less than 0.01% of the days supply values were 15 16 For peer review only 17 deemed to be outside of this range and a new days supply was imputed based on an individual’s 18 19 historical average for a particular ingredient. All unique identifiers (patient, prescriber) were 20 21 anonymized for the purposes of this analysis which was approved by the health ethics research board at 22 23 the University of Alberta (#Pro00083807). 24 25 26 Study Measures 27 28 29 All opioid and benzodiazepine receptor modulator dispensations were retrieved from PIN for 2017. An 30 31 opioid user was defined as anyone who received at least 1 dispensation for an opioid. Patient 32

33 http://bmjopen.bmj.com/ 34 characteristics considered in other studies [16-18], as well as any additional clinically relevant 35 36 characteristics available in the administrative databases were examined to identify factors associated 37 38 with concurrent use. Chronic opioid use was defined as total opioid days greater than 90, as others have 39 40 [1], or more than 10 opioid dispensations in one year. An opioid dependency treatment (ODT) user was 41 on September 26, 2021 by guest. Protected copyright. 42 43 anyone that was dispensed a prescription for methadone or buprenorphine/naloxone. Postal codes 44 45 (forward sortation index) were used to categorize individuals as rural/urban and into income categories 46 47 (<50k, 50-75k, 75-100k, 100-125k, >125k). Average daily oral morphine equivalents (OME’s) and 48 49 number of daily defined doses (DDDs) were calculated for all opioids and benzodiazepine receptor 50 51 52 modulators, respectively, using the conversion factors specified by the TPP[22]. Methadone and 53 54 buprenorphine were excluded from OME specific analyses. We used daily OME thresholds of <50, 50-90 55 56 57 58 7 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 28 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 and > 90 as categories in our analyses since these are clinically accepted in the guidelines for 4 5 6 determining the risk/benefit profile when prescribing opioids for pain.[14] 7 8 The key variable of interest was whether an opioid user also used a benzodiazepine receptor modulator 9 10 11 concurrently in 2017. Although we were not able to directly observe utilization of these medications by 12 13 individuals, we considered “use” as any day on which an individual had a supply of medication on hand 14 15 based on the date and days supply of each dispensation. Using the dispensation information from PIN, 16 For peer review only 17 we generated binary variables for each day of the year to indicate if it was “covered” by an opioid or 18 19 20 benzodiazepine receptor modulator. Beginning on the dispensation day, each day was categorized as 21 22 covered until the end of the days supplied. For each patient, a day was categorized as concurrent if it 23 24 was covered by both an opioid and benzodiazepine receptor modulator. We then calculated the 25 26 number of days, both cumulative and consecutive, that were categorized as concurrent. For example, if 27 28 29 a patient received a 30-day opioid dispensation on Jan 1 and a 20-day benzodiazepine receptor 30 31 modulator dispensation on Jan 20, this would be quantified as 11 days of concurrent use. In our main 32

33 analyses, concurrency was defined as having 1 or more days categorized as concurrent. http://bmjopen.bmj.com/ 34 35 36 Statistical Analyses 37 38 39 We conducted a descriptive analysis to examine the characteristics of concurrent use of opioids and 40

41 benzodiazepine receptor modulators. All summary statistics were calculated using the denominator of on September 26, 2021 by guest. Protected copyright. 42 43 total population of opioid users in Alberta for 2017. 44 45 46 The measure of interest was prevalence of concurrent use by age, sex, average daily OME thresholds 47 48 (<50, 50-90, >90), duration of opioid use (chronic (as defined previously) vs intermittent), opioid 49 50 51 dependence therapy (ODT), rural vs urban residence , number of unique providers, median annual 52 53 household income thresholds, and number of DDD’s of benzodiazepine receptor modulators. Analyses 54 55 were also stratified by the total days of cumulative concurrency (1-7, 8-30, 31-90, >90) and consecutive 56 57 58 8 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 28 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 days of concurrency (1-7, 8-30, 31-60, 61-90, >90). We used chi2 tests of independence to compare 4 5 6 prevalence proportions between the different groups in the above-named characteristics. All analyses 7 8 were performed using STATA/MP 13.1 (StataCorp., College Station, TX). 9 10 11 Patient and Public Involvement 12 13 14 This research was done without patient involvement. Patients were not invited to comment on the 15 16 study design and wereFor not consulted peer to develop review patient relevant outcomesonly or interpret the results. 17 18 Patients were not invited to contribute to the writing or editing of this document for readability or 19 20 accuracy. There are no plans to disseminate the results of the research to study participants. 21 22 23 24 25 26 Results 27 28 29 There were 547,709 Albertans who received at least one dispensation for an opioid and qualified as an 30 31 opioid user (Figure 1). Females represented 53% (n=292,396) of opioid users, 18% (n=98,083) were over 32

33 http://bmjopen.bmj.com/ the age of 65, and the majority of patients were from urban areas (84%) (Table 1). Overall, 20% 34 35 36 (n=108,604) of opioid users were considered chronic users and ODT patients represented 1.7% (n=9139) 37 38 of opioid users. When methadone and buprenorphine were excluded, 88% (n=468,863), 9.5% 39 40 (n=51,033) and 2.8% (n=14,933) represented those in the <50, 50-90, and >90 OME categories,

41 on September 26, 2021 by guest. Protected copyright. 42 respectively. A substantial number of patients received opioids from 3 or more pharmacies (7%) or from 43 44 45 3 or more prescribers (16%). 46 47 48 Among the 547,709 opioid users, 24% (n=132,156) received a benzodiazepine receptor modulator and 49 50 17.6% (n=96,581) had at least one day of concurrent use of an opioid and a benzodiazepine receptor 51 52 modulator during 2017. The mean total days of concurrency over the entire year was 98 (SD=114) days 53 54 (median of 37 (IQR 10-171)). Among patients with concurrent use, a substantial number had high 55 56 57 58 9 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 28 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 durations of concurrent use during the year; 53% had over 30 days of concurrency and 36% had over 90 4 5 6 days of concurrency (Table 2). When we examined the duration of consecutive days of concurrency, the 7 8 mean longest duration was 45 (SD=60) with a median of 24 days (IQR 8-59). Most concurrent patients 9 10 (64%) had concurrent use for less than 30 consecutive days (Table 2). 11 12 13 Differences in concurrency were noted based on sex, urban/rural status, and median household 14 15 incomes, with the prevalence of concurrency being highest among the lowest incomes, as well as a 16 For peer review only 17 strong trend in age (Table 1, Figure 2). Indeed, <2% of all opioid users under the age of twenty used a 18 19 20 benzodiazepine receptor modulator concurrently relative to nearly 30% of those over the age of 65. The 21 22 highest concurrence was observed in the highest age groups, who are also most at risk of severe adverse 23 24 events (Table 3). Concurrency was more common in chronic opioid users compared to intermittent 25 26 users (Table 1). Similarly, chronic opioid users had a higher number of concurrent days in the year 27 28 29 compared to intermittent users (Table 4). 30 31 Characteristics associated with potentially inappropriate use of opioids (e.g., older age, high OME’s, 32

33 http://bmjopen.bmj.com/ 34 multiple providers) had substantially higher concurrent benzodiazepine receptor modulator use (Table 35 36 1, Figures 2-4). Although the absolute number of patients using an average daily OME >90 was low 37 38 (2.8% of opioid users), 46% had concurrent use with a benzodiazepine receptor modulator. Among 39 40 concurrent users in the > 90 OME category, 58.8% had concurrent use >90 days (Figure 3) and 12.8% of 41 on September 26, 2021 by guest. Protected copyright. 42 43 those with > 90 days of concurrent use were also taking > 90 OME per day (Figure 4). 44 45 46 There were also clear trends with respect to providers. As the number of unique providers increased, so 47 48 too did the prevalence of concurrency. Although the absolute numbers were low (<5%), the opioid 49 50 users that visited more than 5 pharmacies or prescribers in 2017 both had a prevalence of concurrency 51 52 of 62% (Table 1). Opioid users who received a benzodiazepine receptor modulator, either concurrently 53 54 55 or not, visited more providers compared to those who received only an opioid or benzodiazepine 56 57 58 10 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 28 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 receptor modulator (Figure 5). Interestingly, among concurrent users, 78% (n=74882) received an 4 5 6 opioid and benzodiazepine receptor modulator from the same prescriber and 94% (n=90561) from the 7 8 same pharmacy. Moreover, 58% of concurrent users (n=56098) received an opioid and benzodiazepine 9 10 receptor modulator on the same day from the same prescriber while 64% (n=61715) received an opioid 11 12 and benzodiazepine receptor modulator from the same pharmacy on the same day. 13 14 15 The trend between number of DDD’s of benzodiazepine receptor modulators and concurrency is similar 16 For peer review only 17 to that with average daily OME’s. Most of the opioid patients concurrently used a benzodiazepine 18 19 20 receptor modulator at the lowest number of DDD’s (66%). However, around 88% of those using >2-3 21 22 times the DDD were concurrent users (Table 1). 23 24 25 Interpretation 26 27 28 Many reputable clinical resources indicate that benzodiazepine receptor modulators should not be 29 30 combined with opioids, yet this study showed that nearly 20% of patients using an opioid did so in 31 32 combination with a benzodiazepine receptor modulator in Alberta [1, 14, 15]. Those on >90 mg OME

33 http://bmjopen.bmj.com/ 34 had the highest prevalence of concurrency when compared to lower doses. Moreover, among 35 36 37 concurrent users, total days of concurrency was high with about half of these patients using opioids and 38 39 benzodiazepine receptor modulators at the same time for more than 30 days. Perhaps not so surprising 40

41 is the high prevalence of concurrency in those with a greater number of distinct prescribers. In addition, on September 26, 2021 by guest. Protected copyright. 42 43 our observation of a higher prevalence of concurrency in chronic opioid users compared to intermittent 44 45 46 users was expected since prolonged opioid use provided more opportunities for concurrent use. These 47 48 results should be concerning to clinicians and policy makers because the potential for adverse 49 50 outcomes associated with opioid use is greatly increased since a significant proportion of opioid 51 52 fatalities involve benzodiazepine receptor modulators.[3, 23] 53 54 55 56 57 58 11 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 28 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Our observation that concurrent use of a benzodiazepine receptor modulator occurred in 20-25% of 4 5 6 opioid users was similar to the recent Vozoris study using data from the United States.[18] While both 7 8 studies also found a higher prevalence of concurrency in females than males, this difference was not 9 10 significant after adjusting for covariates in the Vozoris study. One reason for this discrepancy may be 11 12 the underlying patterns of benzodiazepine receptor modulator use; in Canada, these drugs are used 13 14 more frequently in females than males [24]. Our observations that concurrency increased with age and 15 16 For peer review only 17 was prevalent in nearly 30% of opioid users > 65 years of age contrasts with previous studies. For 18 19 example, Vozoris reported a trend towards decreased concurrency among patients 60 years and older 20 21 [18], and Hwang and colleagues reported concurrency in <20% of elderly patients.[16] Possible reasons 22 23 for the discrepancy in age related trends include differences in study methodology (survey data versus 24 25 26 administrative data), study population (increasing use of benzodiazepine receptor modulators amongst 27 28 the elderly in Canada and Alberta [24-26]), our inclusion of Z drugs to identify benzodiazepine receptor 29 30 modulators, and prescriber perception of safety of Z drugs over benzodiazepines [27]. Regardless, the 31 32 high prevalence of concurrent use among those over 65 is especially concerning because they are at

33 http://bmjopen.bmj.com/ 34 35 high-risk for adverse clinical outcomes. Indeed, many clinical guidelines advise against prescribing 36 37 benzodiazepines in most seniors, let alone in combination with an opioid.[28, 29] Furthermore, patients 38 39 aged 65 and older consistently have the highest rates of hospitalization due to opioid poisoning.[4] 40

41 on September 26, 2021 by guest. Protected copyright. 42 To date, we are unaware of other studies that have suggested those taking very high daily doses of 43 44 opioids (>90 OME per day) also have high concurrency rates. Irrespective of the reason for concurrent 45 46 use (i.e., opioid use disorder and doctor shopping or when used for more appropriate indications) the 47 48 49 evidence suggests that high dose opioid users have up to 5x the risk of overdose and those above 100 50 51 OME have a much higher risk of fatal over dose.[14, 30] Combining opioids and benzodiazepine 52 53 receptor modulators in these groups could certainly contribute to further adverse outcomes already at 54 55 high rates. 56 57 58 12 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 28 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Although concurrent use of opioids and benzodiazepine receptor modulators is often deemed clinically 4 5 6 inappropriate, beyond substance use disorder situations, one has to question why the observed 7 8 prevalence is so high despite the numerous efforts across the country, and in Alberta, to mitigate this 9 10 high-risk prescribing. In the groups with the highest concurrent use (females, ODT patients, chronic and 11 12 high dose opioid users, elderly, etc.), most, if not all, are known to have a higher prevalence of 13 14 conditions related to pain and mental health.[31-33]. Our results showed that 78% of concurrent users 15 16 For peer review only 17 received both medications from the same prescriber and 94% from the same pharmacy with over half 18 19 receiving these drugs on the same day. There is an opportunity here to educate providers about the 20 21 risks of concurrent use and to verify if concurrent use is truly appropriate. Furthermore, treatment 22 23 emphasis in chronic pain and mental health patients is changing where opioids and benzodiazepines are 24 25 26 no longer first line treatment options and where integrated and multidisciplinary treatments are 27 28 preferred.[34] Connecting patients with these preferred treatment modalities is often difficult because 29 30 of cost and time and often opioids and /or benzodiazepines are used to address the unmet needs of 31 32 patients.

33 http://bmjopen.bmj.com/ 34 35 The strengths of our study included the large population-based sample with near complete capture of all 36 37 38 opioid and benzodiazepine receptor modulator dispensations occurring in community pharmacies 39 40 within the province. Pharmacies in Alberta are mandated by the College of Pharmacy to ensure

41 on September 26, 2021 by guest. Protected copyright. 42 accurate prescription records such that use of PIN data can accurately capture most, if not all, of the 43 44 opioid and benzodiazepine receptor modulator dispensations and the information provided with each of 45 46 these dispensations. Another strength is that our analyses included average daily OME’s when 47 48 49 characterizing concurrent use, something that we have not seen in other studies. There are, however, 50 51 some limitations in our study. First, we are assuming that patients took their medications as dispensed. 52 53 Medication adherence in opioid users is a challenging issue.[35] We assumed that days supply was 54 55 entered correctly by pharmacies when calculating our OME and DDD values, however no validation of 56 57 58 13 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 28 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 the PIN days supply field has been completed to date. Second, our study was limited to descriptive 4 5 6 analyses and does not provide outcomes data from concurrent use. Clinically, there are instances where 7 8 concurrent use may be considered appropriate, especially in palliative care and cancer treatment 9 10 settings. Information on the indications for concurrent prescribing were not available in the PIN 11 12 database used for this study. 13 14 15 Despite these limitations, Alberta still has an alarming prevalence of concurrent use. The opioid crisis in 16 For peer review only 17 Alberta and Canada is being driven in part, by prescription opioids.[36] However, due to wide spread 18 19 20 attention to the opioid crisis, the number of opioid prescriptions and morphine milligram equivalents 21 22 prescribed sharply declined in all provinces in 2016 and 2017, including Alberta.[37, 38]. It is clear that 23 24 continued efforts are required to curb the concurrent utilization of opioids and benzodiazepine receptor 25 26 modulators in the province, and elsewhere as it is unlikely Alberta is unique in this regard. Furthermore, 27 28 29 as increasing clinical emphasis is being placed on non-pharmacologic management of chronic pain and 30 31 not prescribing opioids to patients with mental-health disorders, as well as ongoing monitoring and 32

33 educational campaigns, we will hopefully see a decrease in concurrent use.[14, 15, 33, 39] http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 14 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 28 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Table 1. Summary statistics of prevalence of concurrency^ among opioid users and possible high -risk 4 markers. 5 6 Characteristic N (%) Prevalence of concurrency Prevalence of Percent of 7 (among opioid within characteristic. concurrency among all concurrent 8 users) Percent (n)# opioid users. Percent users (n) 9 (n=547709) 10 Opioid users 547709 (100) 17.6 (96581)## ------11 Gender: 12 Male 255293 (46.6) 14.9 (37955) 6.9 39.3 (37955) 13 Female 292396 (53.4) 20.0 (58620) 10.7 60.7 (58620) 14 Average daily 15 OME*: 16 <50 468863 (87.7) 15.7 (73411) 13.7 80.2 (73411) 50-90 For51033 (9.5)peer review22.1 (11287) only2.1 12.3 (11287) 17 >90 14933 (2.8) 46.2 (6899) 1.3 7.5 (6899) 18 Duration of opioid 19 use**: 20 Chronic 108604 (19.8) 47.1 (51214) 9.4 53.0 (51214) 21 Intermittent 439105 (80.2) 10.3 (45367) 8.3 47.0 (45367) 22 ODT 9139 (1.7) 37.2 (3401) 0.62 3.5 (3401) 23 Not on ODT 538570 (98.3) 17.3 (93180) 17.1 96.5 (93180) 24 Postal code zone: 25 Rural 85666 (15.6) 22.0 (18809) 3.4 19.5 (18809) 26 Urban 462043 (84.4) 16.8 (77772) 14.2 80.5 (77772) 27 Median Household 28 Income (x 1000) 29 <50 107240 (19.6) 23.1 (24781) 4.5 25.7 (24781) 50-75 261354 (47.2) 17.9 (46725) 8.5 48.4 (46725) 30 75-100 151352 27.6) 14.2 (21496) 3.9 22.3 (21496) 31 100-125 27314 (5.0) 12.9 (3514) 0.6 3.6 (3514) 32 >125 448 (0.08) 14.5 (65) 0.01 0.07 (65) 33 # of unique http://bmjopen.bmj.com/ 34 dispensing 35 pharmacies: 36 1 426557 (77.9) 12.0 (51413) 9.4 53.2 (51413) 37 2 82048 (15.0) 31.1 (25550) 4.7 26.4 (25550) 38 3 23155 (4.2) 45.3 (10482) 1.9 11.0 (10482) 39 4 8260 (1.5) 53.1 (4387) 0.80 4.5 (4387) 40 5+ 7689 (1.4) 61.8 (4749) 0.88 4.9 (4749) # of unique 41 on September 26, 2021 by guest. Protected copyright. prescribers: 42 1 352596 (64.4) 7.1 (25158) 4.6 26.0 (25158) 43 2 107347 (19.6) 25.9 (27805) 5.1 28.8 (27805) 44 3 42656 (7.8) 42.2 (17990) 3.3 18.6 (17990) 45 4 20126 (3.7) 50.5 (10163) 1.9 10.5 (10163) 46 5+ 24984 (4.6) 61.9 (15465) 2.8 16.0 (15465) 47 Age: 48 0-17 20366 (3.7) 1.5 (307) 0.06 0.3 (307) 49 18-65 429259 (78.4) 16.3 (70000) 12.8 72.5 (70000) 50 >65 98083 (17.9) 26.8 (26274) 4.8 27.2 (26274) 51 Number of benzodiazepine 52 DDD’s 53 94192 (71.3) 67.4 (63531) 11.6 65.8 (63531) 0-1 30423 (23.0) 86.7 (26370) 4.8 27.3 (26370) 54 1-2 4761 (3.6) 89.1 (4243) 0.77 4.4 (4243) 55 2-3 2780 (2.1) 87.7 (2437) 0.44 2.5 (2437) 56 >3 57 58 15 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 28 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 ^ Concurrency is defined as 1 or more days of overlap between an opioid and benzodiazepine receptor modulator. 4 5 *methadone and buprenorphine patients were excluded. 6 **chronic opioid users were defined by having at least 90 days of cumulative opioid use or at least 10 opioid prescriptions in 7 the year. This includes ODT patients. 8 9 # p-value for chi2 test of independence (difference between prevalence of concurrency between groups within characteristic) 10 <0.001 for all characteristics. 11 12 ## 95% confidence interval for prevalence of concurrency=17.5-17.7 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 16 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 28 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 17 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 28 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

41 1-7 4.1 (18,163) 2.2 (2,340) on September 26, 2021 by guest. Protected copyright. 42 8-30 4.5 (19,712) 5.4 (5,902) 43 31-90 1.7 (7,492) 7.8 (8,448) 44 >90 0 (0) 31.8 (34,524) 45 46 47 48 49 50 51 52 53 54 55 56 57 58 18 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 28 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 References 4 5 1. Busse, J.W., et al., Guideline for opioid therapy and chronic noncancer pain. Canadian Medical 6 Association Journal, 2017. 189(18): p. E659-E666. 7 2. Guan, Q., et al., Evaluating the early impacts of delisting high-strength opioids on patterns of 8 prescribing in Ontario. Health Promotion and Chronic Disease Prevention in Canada, 2018. 38(6): 9 10 p. 256-262. 11 3. Belzak, L. and J. Halverson, Evidence synthesis - The opioid crisis in Canada: a national 12 perspective. Health Promotion and Chronic Disease Prevention in Canada, 2018. 38(6): p. 224- 13 233. 14 4. O’Connor, S., V. Grywacheski, and K. Louie, At-a-glance - Hospitalizations and emergency 15 department visits due to opioid poisoning in Canada. Health Promotion and Chronic Disease 16 Prevention inFor Canada, 2018.peer 38(6): p. 244-247.review only 17 5. Rudolph, U. and H. Möhler, GABA-based therapeutic approaches: GABAA receptor subtype 18 19 functions. Current opinion in pharmacology, 2006. 6(1): p. 18-23. 20 6. Roy-Byrne, P.P., The GABA-benzodiazepine receptor complex: structure, function, and role in 21 anxiety. The Journal of clinical psychiatry, 2005. 66: p. 14-20. 22 7. CPSA, Clinical Toolkit Benzodiazepines: Use and Taper. CPSA, 2015. 23 8. Katzman, M.A., et al., Canadian clinical practice guidelines for the management of anxiety, 24 posttraumatic stress and obsessive-compulsive disorders. BMC Psychiatry, 2014. 14 Suppl 1: p. 25 S1. 26 9. Canadian Pharmacists Association. RxTx. 2019; Available from: https://www.e- 27 28 therapeutics.ca/search. 29 10. TOP, T.O.P. Guideline for Adult Primary Insomnia [Internet]. 2010; Available from: 30 http://www.topalbertadoctors.org/download/439/insomnia_management_guideline.pdf. 31 11. ChooseWiselyCanada. The Canadian Geriatrics Society has developed a list of 5 things physicians 32 and patients should question in geriatrics [Internet]. Available from:

33 https://choosingwiselycanada.org/geriatrics/. http://bmjopen.bmj.com/ 34 12. Cunningham, C.M., G.E. Hanley, and S. Morgan, Patterns in the use of benzodiazepines in British 35 Columbia: examining the impact of increasing research and guideline cautions against long-term 36 use. Health Policy, 2010. 97(2): p. 122-129. 37 38 13. Weir, D.L., et al., Benzodiazepine receptor agonist dispensations in Alberta: a population-based 39 descriptive study. CMAJ Open, 2018. 6(4): p. E678-E684. 40 14. ismp Canada. Essential Clinical Skills for Opioid Prescribers. 2017 Nov 2018]; Available from:

41 https://www.ismp-canada.org/download/OpioidStewardship/Opioid-Prescribing-Skills.pdf. on September 26, 2021 by guest. Protected copyright. 42 15. Centers for Disease Control and Prevention. Guideline for Prescribing Opioids for Chronic Pain. 43 2016; Available from: https://www.cdc.gov/drugoverdose/pdf/Guidelines_Factsheet-a.pdf. 44 16. Hwang, C.S., et al., Trends in the Concomitant Prescribing of Opioids and Benzodiazepines, 2002- 45 2014. Am J Prev Med, 2016. 51(2): p. 151-160. 46 47 17. Hirschtritt, M.E., K.L. Delucchi, and M. Olfson, Outpatient, combined use of opioid and 48 benzodiazepine medications in the United States, 1993–2014. Preventive medicine reports, 49 2018. 9: p. 49-54. 50 18. Vozoris, N.T., Benzodiazepine and opioid co-usage in the US population, 1999–2014: an 51 exploratory analysis. Sleep, 2019. 42(4). 52 19. Gudin, J.A., et al., Risks, Management, and Monitoring of Combination Opioid, Benzodiazepines, 53 and/or Alcohol Use. Postgraduate Medicine, 2013. 125(4): p. 115-130. 54 20. Jones, J.D., S. Mogali, and S.D. Comer, Polydrug abuse: a review of opioid and benzodiazepine 55 combination use. Drug and alcohol dependence, 2012. 125(1-2): p. 8-18. 56 57 58 19 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 28 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 21. Government of Alberta. Alberta Netcare EHR Usage Statistics. 2016 Jan 2019]; Available from: 4 http://www.albertanetcare.ca/Statistics.htm. 5 6 22. College of Physicians and Surgeons of Alberta. OME and DDD conversion factors. Available from: 7 http://www.cpsa.ca/wp-content/uploads/2017/06/OME-and-DDD-Conversion-Factors.pdf. 8 23. Sun, E.C., et al., Association between concurrent use of prescription opioids and benzodiazepines 9 and overdose: retrospective analysis. BMJ, 2017. 356: p. j760. 10 24. Kassam, A. and S.B. Patten, Canadian trends in Benzodiazepine & Zopiclone use. Journal of 11 Population Therapeutics and Clinical Pharmacology, 2006. 13(1). 12 25. Kassam, A., B. Carter, and S.B. Patten, Sedative hypnotic use in Alberta. The Canadian Journal of 13 Psychiatry, 2006. 51(5): p. 287-294. 14 26. Brandt, J., et al., Novel Measures of Benzodiazepine and Z-Drug Utilisation Trends in a Canadian 15 16 Provincial AdultFor Population peer (2001-2016). review Journal of Population only Therapeutics and Clinical 17 Pharmacology, 2019. 26(1): p. e22-e38. 18 27. Hoffmann, F., Perceptions of German GPs on benefits and risks of benzodiazepines and Z-drugs. 19 Swiss medical weekly, 2013. 143(0304). 20 28. Farrell, K.P., et al., Deprescribing benzodiazepine receptor agonists. 2018. 21 29. Gallagher, P. and D. O'Mahony, STOPP (Screening Tool of Older Persons' potentially 22 inappropriate Prescriptions): application to acutely ill elderly patients and comparison with 23 Beers' criteria. Age Ageing, 2008. 37(6): p. 673-9. 24 25 30. Ismp Canada, navigating-opioids-11x17-canada.pdf. 26 31. Pinkhasov, R.M., et al., Are men shortchanged on health? Perspective on health care utilization 27 and health risk behavior in men and women in the United States. Int J Clin Pract, 2010. 64(4): p. 28 475-87. 29 32. Kessler, R.C., et al., Lifetime and 12-month prevalence of dsm-iii-r psychiatric disorders in the 30 united states: Results from the national comorbidity survey. Archives of General Psychiatry, 31 1994. 51(1): p. 8-19. 32 33. National Opioid Use Guideline Group (NOUGG) Canadian Guidelines for Safe and Effective Use of

33 http://bmjopen.bmj.com/ Opioids for CNCP-Part B. 2010. 34 35 34. Centre for Effective Practice. Management of Chronic Non Cancer Pain. 2017; Available from: 36 thewellhealth.ca/cncp. 37 35. Graziottin, A., et al., Opioids: How to Improve Compliance and Adherence. Pain Practice, 2011. 38 11(6): p. 574-581. 39 36. Gomes, T., et al., Contributions of prescribed and non-prescribed opioids to opioid related 40 deaths: population based cohort study in Ontario, Canada. BMJ, 2018. 362: p. k3207. 41 37. Abdesselam, K., et al., At-a-glance - Opioid surveillance: monitoring and responding to the on September 26, 2021 by guest. Protected copyright. 42 evolving crisis. Health Promot Chronic Dis Prev Can, 2018. 38(9): p. 312-316. 43 44 38. College of Physicians and Surgeons of Alberta. Prescribing Trends. 2018 2018]; Available from: 45 http://www.cpsa.ca/physician-prescribing-practices/. 46 39. Centers for Disease Control and Prevention. Checklist for prescribing opioids for chronic pain. 47 2016; Available from: https://www.cdc.gov/drugoverdose/pdf/PDO_Checklist-a.pdf. 48 49 50 51 52 53 54 55 56 57 58 20 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 28 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Figure Legend: 4 5 Figure 1. Patient flow diagram of denominator used for analyses 6 7 8 9 Figure 2. Prevalence of concurrency by age group among all opioid users in 2017 (n=547,708) 10 11 12 13 Figure 3. Prevalence of concurrency by total days of concurrency and average daily OME category in 14 2017 (n=91597) 15 16 For peer review only 17 18 Figure 4. Percent distribution of average daily OME categories within categories of total cumulative days 19 of concurrent use (n=91597) 20 21 22 23 Figure 5. Distribution (percentage) of patient categories by number of unique providers. 24 25 26 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 21 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 28 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Figure 1. Patient flow diagram of denominator used for analyses 4 5 785,240 Benzodiazepine 6 receptor modulator & 7 Opioid Users 8 9 10 237,531 excluded 11 (never received an 12 opioid) 13 14 15 16 For peer547,709 review opioid users only 17 18 19 20 12,880 excluded (methadone & 21 buprenorphine users) 22 23 24 534,829 opioid users 25 (used for average daily 26 OME analyses) 27 28 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 28 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 Figure 2. Prevalence of concurrency by age group among all opioid users in 2017 (n=547,708). 4 5 6 PREVALENCE OF CONCURRENCY BY AGE GROUP 7 8 Overall Female Male 9 35 10 11 12 30 13 14 25 15 16 20 For peer review only 17 18 19 15 20 21 10 22 PREVALENCE OF CONCURRENCY OF PREVALENCE 23 5 24 25 26 0 27 0 - 9 10- 19 20- 29 30- 39 40- 49 50- 59 60- 69 70- 79 80- 89 > 9 0 28 AGE GROUP 29 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 24 of 28 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

Prevalence Prevalence concurrencyof 10 21 22 0 23 1-7 8-30 31-90 90+ 24 Days of concurrency group 25 26 <50 OME 50-90 OME >90 OME 27

28 29 * Excludes methadone and buprenorphine. 30 31 32

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 25 of 28 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

31 32 * Excludes methadone and buprenorphine.

33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40

41 on September 26, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 26 of 28 BMJ Open: first published as 10.1136/bmjopen-2019-030858 on 6 September 2019. Downloaded from

1 2 3 4 5 Figure 5. Distribution (percentage) of patient categories by number of unique prescribers. 6 7 8 Distribution of treatment groups across number of prescribers 9 90 10 11 12 80 76 77 13 14 70 15 16 For peer review only 17 60 55 18 19 50 20 21

22 Percent 40 23 24 29 25 30 26 26 27 18 18 19 20 17 16 28 14 29 11 30 10 6 5 4 4 31 1 2 1 2 32 0 33 1 2 3 4 5+ http://bmjopen.bmj.com/ 34 35 Number of unique prescribers 36 37 Benzodiazepine/Z-drug dispensations only(n=237,531) 38 39 Opioid dispensations only(n=415,553) 40

41 Opioid & benzodiazepine/ Z-drugs dispensations, no on September 26, 2021 by guest. Protected copyright. 42 concurrency(n=35,575) 43 Opioid & benzodiazepine/Z-drugs dispensations with 44 concurrency(n=96,581) 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 27 of 28 BMJ Open

1 2 STROBE Statement—checklist of items that should be included in reports of observational studies

2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml