BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. 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 October 1, 2021 by guest. Protected copyright. BMJ Open BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

Opioid-Sparing effects of medical for chronic pain: A systematic review and meta-analysis of randomized and observational studies ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2020-047717

Article Type: Original research

Date Submitted by the 07-Dec-2020 Author:

Complete List of Authors: Noori, Atefeh; McMaster University, Department of Health Research Methods, Evidence, and Impact Miroshynchenko, Anna; McMaster University, Department of Health Research Methods, Evidence, and Impact Shergill, Yaad; McMaster University Ashoorion, Vahid; McMaster University, Department of Health Research Methods, Evidence, and Impact Rehman, Yasir ; McMaster University Faculty of Health Sciences, Department of Health Research Methods, Evidence, and Impact Couban, Rachel; McMaster University, Michael G. DeGroote Institute for Pain Research and Care Buckley, D; McMaster University Faculty of Health Sciences, Department

of Health Research Methods, Evidence, and Impact http://bmjopen.bmj.com/ Thabane, Lehana; McMaster University Bhandari, Mohit; McMaster University, Surgery Guyatt, Gordon; McMaster University Agoritsas, Thomas; University of Geneva Faculty of Medicine; McMaster University Busse, Jason; McMaster University, Anesthesia

PAIN MANAGEMENT, < ONCOLOGY, GENERAL MEDICINE Keywords: (see Internal Medicine) on October 1, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 85 BMJ Open

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 I, the Submitting Author has the right to grant and does grant on behalf of all authors of the Work (as defined 10 in the below author licence), an exclusive licence and/or a non-exclusive licence for contributions from authors 11 who are: i) UK Crown employees; ii) where BMJ has agreed a CC-BY licence shall apply, and/or iii) in accordance 12 with the terms applicable for US Federal Government officers or employees acting as part of their official 13 duties; on a worldwide, perpetual, irrevocable, royalty-free basis to BMJ Publishing Group Ltd (“BMJ”) its 14 licensees and where the relevant Journal is co-owned by BMJ to the co-owners of the Journal, to publish the 15 Work in this journal and any other BMJ products and to exploit all rights, as set out in our licence. 16 17 The Submitting Author accepts and understands that any supply made under these terms is made by BMJ to 18 the Submitting Author Forunless you peer are acting as review an employee on behalf only of your employer or a postgraduate 19 student of an affiliated institution which is paying any applicable article publishing charge (“APC”) for Open 20 Access articles. Where the Submitting Author wishes to make the Work available on an Open Access basis (and 21 intends to pay the relevant APC), the terms of reuse of such Open Access shall be governed by a Creative 22 Commons licence – details of these licences and which Creative Commons licence will apply to this Work are set 23 out in our licence referred to above. 24 25 Other than as permitted in any relevant BMJ Author’s Self Archiving Policies, I confirm this Work has not been 26 accepted for publication elsewhere, is not being considered for publication elsewhere and does not duplicate 27 material already published. I confirm all authors consent to publication of this Work and authorise the granting 28 of this licence. 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 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 2 of 85

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 -Sparing effects of for chronic pain: A systematic review and 4 5 meta-analysis of randomized and observational studies 6 7 8 9 Atefeh Noori PhD candidate1,2, Anna Miroshnychenko M.Sc. (Cand.) 1, Yaadwinder Shergill 10 11 M.Sc. (Cand.)1, Vahid Ashoorion PhD1, Yasir Rehman PhD candidate 1, Rachel Couban medical 12 2 1 1 13 librarian , Norman Buckley professor , Lehana Thabane professor , Mohit Bhandari 14 professor1,3, Gordon H. Guyatt distinguished professor1, Thomas Agoritsas assistant professor 15 16 1,4, Jason W. Busse Forassociate peerprofessor1,5,6,7 review* only 17 18 19 Affiliations 20 1. Department of Health Research Methods, Evidence, and Impact, McMaster University, 21 22 Hamilton, ON, Canada 23 2. The Michael G. DeGroote National Pain Center, McMaster University, Hamilton, ON, 24 Canada 25 3. Division of Orthopaedic Surgery, Department of Surgery, McMaster University, Hamilton, 26 ON, Canada 27 4. Division of of General Internal Medicine, Department of Medicine, University Hospitals 28 29 of Geneva, Geneva, Switzerland 30 5. Department of Anesthesia, McMaster University, Hamilton, ON, Canada 31 6. The Chronic Pain Centre of Excellence for Canadian Veterans, Hamilton, ON, Canada 32 7. The Michael G. DeGroote Centre for Medicinal Cannabis Research, McMaster University, 33 Hamilton, ON, Canada http://bmjopen.bmj.com/ 34 35 36 Corresponding Author: Jason W. Busse, Michael G. DeGroote Centre for Medicinal Cannabis 37 Research, McMaster University, HSC–2V9, 1280 Main St. West, Hamilton, Canada, L8S 4K1; 38 39 email: [email protected] 40 41 on October 1, 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 3 of 85 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 Strengths and limitations of this study 4 5 6  This is the first systematic review of randomized controlled trials and observational 7 8 studies exploring the impact of medical cannabis on prescription opioid use among 9 10 11 people living with chronic pain. 12 13  We conducted a comprehensive search for eligible studies, without language restrictions, 14 15 and evaluated the certainty of evidence using the GRADE approach. 16 For peer review only 17 18  All eligible randomized trials enrolled patients with chronic cancer-related pain, and the 19 20 generalizability of their results to non-cancer chronic pain is uncertain. 21 22  Most observational studies incorporated inadequate adjustment for confounding, and all 23 24 25 randomized trials, despite reporting this outcome, were not designed to address the effect 26 27 of medical cannabis on opioid use. 28 29 30 31 32 33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on October 1, 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 4 of 85

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 ABSTRACT 4 5 6 Objective: To assess the efficacy and harms of adding medical cannabis to prescription 7 8 among people living with chronic pain. 9 10 Design: Systematic review and meta-analysis. 11 12 Data sources: CENTRAL, EMBASE, and MEDLINE, from inception to March 2020, with no 13 14 15 language restrictions. 16 For peer review only 17 Main outcomes and measures: Opioid dose reduction, pain relief, sleep disturbance, physical 18 19 and emotional functioning, and three adverse events. 20 21 22 Study selection criteria and methods: We included randomized trials and observational studies 23 24 that enrolled patients with chronic pain receiving prescription opioids and explored the impact of 25 26 adding medical cannabis. Pairs of trained reviewers independently screened studies for 27 28 29 eligibility, extracted data, and assessed risk of bias. We performed random-effects meta-analyses 30 31 and used GRADE to assess the certainty of evidence for each outcome. 32 33 Results: Eligible studies included five randomized trials (all enrolling chronic cancer-pain http://bmjopen.bmj.com/ 34 35 patients) and 12 observational studies. All randomized trials instructed participants to maintain 36 37 38 their opioid dose, which resulted in a very low certainty evidence that adding cannabis has little 39 40 or no impact on opioid use (weighted mean difference [WMD] -3.4 milligram 41 on October 1, 2021 by guest. Protected copyright. 42 equivalent [MME]; 95% confidence interval [CI] -12.7 to 5.8). Randomized trials provided high 43 44 45 certainty evidence that cannabis addition had little or no effect on pain relief (WMD -0.18cm; 46 47 95%CI -0.38 to 0.02; on a 10 cm visual analogue scale [VAS] for pain) or sleep disturbance 48 49 (WMD -0.22 cm; 95%CI -0.4 to -0.06; on a 10 cm VAS for sleep disturbance; minimally 50 51 52 important difference [MID] is 1 cm) among chronic cancer-pain patients. Addition of cannabis 53 54 likely increases nausea (relative risk [RR] 1.43; 95%CI 1.04 to 1.96; risk difference [RD] 4%, 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 85 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 95%CI 0% to 7%) and vomiting (RR 1.5; 95%CI 1.01 to 2.24; RD 3%; 95%CI 0% to 6%) (both 4 5 6 moderate certainty), and may have little or no effect on constipation (RR 0.85; 95%CI 0.54 to 7 8 1.35; RD -1%; 95%CI -4% to 2%) (low certainty). Eight observational studies provided very-low 9 10 certainty evidence that adding cannabis reduced opioid use (WMD -22.5 MME; 95%CI -43.06 to 11 12 -1.97; 8 studies). 13 14 15 Conclusion: Opioid-sparing effects of medical cannabis for chronic pain remain uncertain due to 16 For peer review only 17 very-low certainty evidence. Based on moderate-to-high certainty evidence, adding medical 18 19 cannabis to opioid therapy, among chronic cancer-pain patients, influences neither pain relief nor 20 21 22 sleep disturbance and increases the risk of nausea and vomiting. 23 24 25 26 27 28 29 Keywords: chronic pain; opioids; cannabis; ; drug substitution; sparing effect; 30 31 tapering 32 33 http://bmjopen.bmj.com/ 34 35 Systematic review registration PROSPERO CRD42018091098 36 37 38 39 40 Funding Source: This review received no external funding 41 on October 1, 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 6 of 85

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 Introduction 4 5 6 Chronic pain affects approximately one in five adults and is a common reason for seeking 7 8 medical care.1, 2 Opioids are commonly prescribed for this condition, particularly in North 9 10 America;3 however, they are associated with harms such as overdose and death,4, 5 which are 11 12 dose-dependent.6-9 As a result, there is considerable interest in therapies that may allow patients 13 14 15 with chronic pain using opioid therapy to reduce their opioid intake. 16 For peer review only 17 One promising approach is adding cannabis therapy. Experimental studies have shown 18 19 that opioids and cannabis have similar signal transduction systems,10 and observational studies in 20 21 22 the US demonstrated that the rates of opioid-related mortality reduced after cannabis was 23 24 legalized.11-13 Between 64% and 77% of patients with chronic pain responding to cross-sectional 25 26 surveys reported a reduction in long-term opioid use after adding medical cannabis to their 27 28 14, 15 29 treatment. A 2017 systematic review concluded that pre-clinical studies provided robust 30 31 evidence for the opioid-sparing effects of cannabis.16 To clarify the issue, we undertook a 32 33 systematic review of randomized controlled trials and observational studies to explore the impact http://bmjopen.bmj.com/ 34 35 of adding medical cannabis on opioid dose, other patient-important outcomes, and related harms 36 37 38 in patients with chronic pain using prescribed opioid therapy. 39 40 This systematic review is part of the BMJ Rapid Recommendations project, a 41 on October 1, 2021 by guest. Protected copyright. 42 collaborative effort from the MAGIC Evidence Ecosystem Foundation(www.magicevidnece.org) 43 44 17 45 and BMJ. This systematic review informed a parallel guideline published on BMJ.com and 46 47 MAGICapp (https://app.magicapp.org/#/guideline/jMMYPj). 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 7 of 85 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 METHOD 4 5 We followed standards for meta-analysis of observational studies in epidemiology (MOOSE)18 6 7 19 8 and preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines 9 10 and registered our review (PROSPERO Identifier: CRD42018091098). 11 12 13 14 15 Eligibility criteria 16 For peer review only 17 We included randomized controlled trials (RCTs), observational studies, including cohort studies 18 19 and case-control studies, in any language, that explored the impact of adding medicinal cannabis 20 21 22 (i.e. phytocannabinoids, endocannabinoids, or synthetic cannabinoids) on the use of prescription 23 24 opioids among people living with chronic pain. We defined pain as chronic if patients reported 25 26 that symptoms had persisted for ≥3 months.20 We excluded editorials, letters to the editor, pre- 27 28 clinical studies, conference abstracts, case reports, case series, cross-sectional studies, studies 29 30 31 with less than 2-weeks follow-up, and studies of recreational cannabis use. We classified 32 33 observational study designs according to recommendations by the Cochrane Observational http://bmjopen.bmj.com/ 34 35 Studies Methods Group.21 36 37 38 39 40 Literature search and study selection 41 on October 1, 2021 by guest. Protected copyright. 42 We searched the Cochrane Central Register of Controlled Trials (CENTRAL), EMBASE, and 43 44 45 MEDLINE from inception to March 2020 with no restriction on language of publication. An 46 47 experienced medical librarian (RC) developed our database-specific search strategies (Appendix 48 49 A). We also searched the ClinicalTrials.gov registry to identify ongoing trials, and reference lists 50 51 of all eligible studies and related systematic reviews for additional eligible studies. Two teams of 52 53 54 paired reviewers independently screened titles, abstracts and full-text studies for eligibility using 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 85

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 online systematic review software (Rayyan QCRI, Qatar Computing Research Institute). 4 5 6 Reviewers resolved disagreements through discussion. 7 8 9 10 Data collection 11 12 Using standardized forms and a detailed instruction manual, pairs of reviewers independently 13 14 15 abstracted data from each eligible study, including study and patient characteristics, and details 16 For peer review only 17 of treatment (e.g. dose, formulation, and duration of cannabis add-on therapy). Our primary 18 19 outcome was opioid dose. We also captured patient-important outcomes, as guided by the 20 21 22 22 Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials, including pain 23 24 relief, sleep disturbance, physical and emotional functioning. Regarding adverse events, we 25 26 focused on vomiting, nausea, and constipation as a systematic review of values and preferences23 27 28 29 demonstrated that patients living with chronic pain experience gastrointestinal complaints as the 30 31 most important opioid-induced adverse events. We contacted authors to obtain unpublished data. 32 33 http://bmjopen.bmj.com/ 34 35 Risk of bias assessment 36 37 38 Following training and calibration exercises two independent reviewers used a modified 39 40 Cochrane risk of bias tool24, 25 to assess the risk of bias among eligible RCTs according to the 41 on October 1, 2021 by guest. Protected copyright. 42 following domains: allocation concealment, blinding of participants, study personnel, outcome 43 44 45 assessors and data analyst, and loss to follow-up (≥20% missing data was assigned high risk of 46 47 bias). Response options for each item were 'definitely or probably yes' (assigned a low risk of 48 49 bias) and 'definitely or probably no' (assigned a high risk of bias). (Supplement Table 1) We used 50 51 26 52 criteria suggested by the CLARITY group to assess the risk of bias of observational studies 53 54 including selection bias, confidence that all patients had the condition of interest, control for 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 85 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 confounding variables, validity of outcome assessment(s), and infrequent missing data (<20%) 4 5 6 (details available at www.evidencepartners.com/resources/methodological-resources/). 7 8 (Supplement Tables 2-3). 9 10 11 12 Data analysis 13 14 15 We calculated inter-rater agreement regarding the eligibility of full-text studies using an adjusted 16 For peer review only 17 κ statistic.27 We conducted separate analyses for randomized controlled trials and observational 18 19 studies. All continuous measures for pain intensity and sleep disturbance were converted to a 10 20 21 28, 22 cm visual analogue scale (VAS); the minimally important difference (MID) for both was 1 cm. 23 24 29 All continuous outcomes that were reported by more than one study were pooled to derive the 25 26 weighted mean difference (WMD) and associated 95% confidence interval (95% CI). We pooled 27 28 29 binary outcomes (adverse events) as relative risks (RRs) and risk differences (RDs) and their 30 31 associated 95% CIs. We conducted all meta-analyses with random-effects models and the 32 33 DerSimonian-Laird method.30 http://bmjopen.bmj.com/ 34 35 When studies reported effects on continuous outcomes as the median and interquartile 36 37 38 range, we derived the mean and SD using the method presented by Wan et al. 31 We also 39 40 converted medians to means using the approach recommended by the Cochrane Handbook as a 41 on October 1, 2021 by guest. Protected copyright. 42 sensitivity analysis. When authors failed to report a measure of precision associated with mean 43 44 45 differences, we imputed the SD from eligible studies that reported these measures (Technical 46 47 appendix).32 We included each comparison reported by multi-arm studies and calculated a 48 49 correction factor to account for the unit of analysis error (i.e. when information from a treatment 50 51 33 52 arm is used more than once in the same meta-analysis). We explored the consistency of the 53 54 association between our pooled results and studies reporting the same outcome domains that 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 85

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 were not possible to pool. We used Stata (StataCorp, Release 15.1, College Station, Texas) for 4 5 6 all analyses. Comparisons were 2-tailed using a threshold of p ≤ 0.05. 7 8 9 10 Subgroup analyses and meta-regression 11 12 We examined heterogeneity among pooled RCTs using the I2 statistic, and through visual 13 14 15 inspection of forest plots for pooled observational data, because statistical tests of heterogeneity 16 For peer review only 17 can be misleading when sample sizes are large and associated confidence intervals, are therefore 18 19 narrow.34 When we had at least two studies in each subgroup, we explored sources of 20 21 22 heterogeneity with five pre-specified subgroup hypotheses, assuming greater benefits with (1) 23 24 shorter vs. longer duration of follow-up; (2) higher vs. lower risk of bias; (3) enriched vs non- 25 26 enriched study design; (4) chronic non-cancer vs. chronic cancer-related pain; and (5) higher vs 27 28 29 lower [THC] content. We assumed similar directions of subgroup effects 30 31 for harms, except for study design and THC content in which we expected greater harms with 32 33 non-enriched trials and higher THC content. However, apart from item two (risk of bias), studies http://bmjopen.bmj.com/ 34 35 did not report sufficient data to undertake subgroup analyses. 36 37 38 39 40 The certainty of the evidence 41 on October 1, 2021 by guest. Protected copyright. 42 We used the Grading of Recommendations Assessment, Development and Evaluation (GRADE) 43 44 45 approach to assess the certainty of evidence on an outcome-by-outcome basis as high, moderate, 46 47 low or very low.35 With GRADE, RCTs begin as high-certainty evidence, but can be rated down 48 49 because of risk of bias, imprecision, inconsistency, indirectness, or publication bias. We rated 50 51 52 down for imprecision if the 95% CI associated with a pooled continuous outcome included ½ the 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 85 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 MID, or if the estimate of precision associated with the RR for binary outcomes included no 4 5 6 effect. 7 8 Using GRADE, observational studies begin as low certainty evidence, and while they can 9 10 be rated down further for the same reasons as RCTs, they can also be rated up in the presence of 11 12 a large magnitude of the effect, a dose-response gradient, or presence of plausible confounders or 13 14 15 other biases that increase confidence in the estimated effect.36 We only reported the pooling 16 For peer review only 17 results of observational studies when they resulted in the same or higher certainty of evidence 18 19 than evidence from RCTs. When there were at least 10 studies for meta-analysis, we explored for 20 21 37 22 small-study effects by visual assessment of funnel plot asymmetry and Egger’s statistical test. 23 24 25 26 27 28 29 30 31 32 33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on October 1, 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 12 of 85

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 RESULTS 4 5 6 Of 5133 records identified, we reviewed 133 articles in full text, and 18 studies reported in 17 7 8 publications proved eligible (Figure 1); five RCTs in four publications38-41 and 13 observational 9 10 studies.42-54 One study enrolled a mixed group of opioid and non-opioid users;45 however, our 11 12 attempts to contact the authors to acquire pain intensity data for the sub-group of patients 13 14 15 prescribed opioids proved unsuccessful. All five RCTs38-41 and three observational studies46, 49, 50 16 For peer review only 17 enrolled patients with chronic cancer-related pain; the remaining 9 observational studies 18 19 explored adding cannabis to opioids for patients with chronic non-cancer pain,43, 47, 48, 51-53 or a 20 21 42, 44, 45, 54 22 mix of cancer and non-cancer pain (Table 1). 23 24 Among the 18 included studies, the percentage of female participants was 48% (median 25 26 of individual trials 48.3%, interquartile range [IQR] 42.7% to 58.4%), and the median of the 27 28 29 mean age was 56.3 (IQR 51.2 to 59.9). Follow-up ranged from 2 to 5 weeks among RCTs, and 30 31 from 4 weeks to 6.4 years for observational studies. Only 1 RCT38 used an enrichment design 32 33 (following the open-label phase, patients with at least 15% improvement in pain were http://bmjopen.bmj.com/ 34 35 randomized to the intervention and control groups) and all RCTs advised patients to maintain 36 37 38 stable doses of all other prescribed pain medications, including opioids, during the study period 39 40 (Table 1). All included RCTs, and three of the observational studies42, 46, 47 administered 41 on October 1, 2021 by guest. Protected copyright. 42 synthetic cannabis products (i.e. , , and nabiximole), five observational 43 44 44, 45, 52-54 45 studies reported different combinations of THC: CBD products, and 5 other 46 47 observational studies43, 48-51 did not provide details of cannabis type (Table1, Supplement Table 48 49 4). 10 studies reported receiving industry funding,38-41, 44, 46, 47, 51, 52 five studies45, 48-50, 54 reported 50 51 42, 43, 53 52 no-industry funding, and three studies did not report funding information (Table 1). 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 85 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 Risk of bias of included studies 4 5 6 All included RCTs reported adequate allocation concealment and blinding of patients and health- 7 8 care providers; however, three trials38, 40, 41 were at risk of bias due to high loss to follow up 9 10 (Supplement Table 5). All observational studies were at high risk of bias, typically due to lack of 11 12 confidence in the assessment of exposure, non-representative samples, and insufficient control 13 14 15 for confounding (Supplement Tables 6-7). 16 For peer review only 17 18 19 20 Outcomes for medical cannabis add-on therapy 21 22 Opioid dose reduction 23 24 The primary limitation of RCTs was that all investigators instructed patients to not alter their 25 26 dose of opioids. This represents a very serious indirectness of the findings regarding the research 27 28 29 question, warranting rating down two levels, and was the primary reason for very low certainty 30 31 evidence from the 1176 patients.38-40 Their results raised the possibility that adding medical 32 33 cannabis may not be associated with a reduction in opioid use (WMD -3.4 MME; 95%CI -12.7 http://bmjopen.bmj.com/ 34 35 36 to 5.9; table 2; Supplement Figure 1). There were no differences in effect based on the loss to 37 38 follow-up (Supplement Figure 2; test of interaction P=0.79). 39 40 Very-low certainty evidence from 8 observational studies42, 43, 45, 46, 48-50, 52 raised the 41 on October 1, 2021 by guest. Protected copyright. 42 43 possibility that adding medical cannabis may reduce the use of opioids among patients with 44 45 chronic pain (WMD -22.5 MME; 95%CI -43.06 to -1.97; Table 2; Supplement Figure 3). Three 46 47 observational studies that could not be pooled reported consistent results. The first study assessed 48 49 the impact of providing medical cannabis to 61 patients with chronic low back pain who were 50 51 52 prescribed opioid therapy (median opioid dose was 21 mg MME/day) and reported that 52% of 53 54 patients (32 of 61) stopped all use of opioids at a median follow-up of 6.4 years.51 The second 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 85

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 study 44 reported that of 94 patients with chronic pain (both cancer and non-cancer pain) who 4 5 6 began using CBD hemp extract, 53.2% were able to decrease their use of prescription opioids at 7 8 8 weeks. An additional study 54 included 600 patients with chronic pain who all were indicated 9 10 willingness to taper their opioid dose and were administered 0.5g daily of medicinal cannabis for 11 12 each 10% reduction in opioid dose. After 6 months’ follow-up, 55% of patients reported a 30% 13 14 15 reduction in opioid dose on average and 26% of them discontinued opioid use. 16 For peer review only 17 18 19 Pain relief 20 21 38-41 22 High-certainty evidence from 5 RCTs demonstrated that adding medical cannabis to opioid 23 24 therapy resulted in trivial or no difference in pain (WMD -0.18 cm; 95%CI -0.38 to 0.02 on the 25 26 10 cm VAS for pain; MID 1cm; Table 2; Supplement Figure 4). Results did not differ depending 27 28 29 on loss to follow-up (Supplement Figure 5, a test of interaction P=0.44). 30 31 32 33 Sleep disturbance http://bmjopen.bmj.com/ 34 35 Five RCTs38-41 provided high certainty evidence that adding medical cannabis to prescription 36 37 38 opioids results in a trivial improvement in sleep disturbance (WMD -0.22 cm; 95%CI -0.4 to - 39 40 0.06 on the 10 cm VAS for sleep disturbance; MID 1cm; Table 2; Supplement Figure 7). Results 41 on October 1, 2021 by guest. Protected copyright. 42 did not differ between trials reporting the low and high loss to follow-up (Supplement Figure 8, a 43 44 45 test of interaction P =0.82). 46 47 48 49 Other reported outcomes 50 51 39 52 A single RCT reported moderate certainty evidence that adding cannabis likely has little or no 53 54 effect on emotional and physical functioning (Supplement Tables 8-9). 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 85 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 Adverse events 4 5 6 Nausea, vomiting, or constipation 7 8 4 RCTs38-41 provided moderate certainty evidence that adding medical cannabis to opioid therapy 9 10 likely increases the incidence of nausea (RR 1.43, 95%CI 1.04 to 1.96; RD 4%, 95%CI 0% to 11 12 7%; Supplement Figure 9-10) and vomiting (RR 1.50; 95%CI 1.01 to 2.24; RD 3%; 95%CI 0% 13 14 15 to 6%; Supplement Figure 11-12) in patients with cancer-related chronic pain prescribed opioid 16 For peer review only 17 therapy. 3 RCTs38, 40, 41 provided low certainty evidence that adding medical cannabis to opioid 18 19 therapy may not increase constipation (RR 0.85, 95%CI 0.54 to 1.35; RD -1%; 95%CI -4% to 20 21 22 2%; Supplement Figure 13-14). 23 24 25 26 27 28 29 30 31 32 33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on October 1, 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 16 of 85

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 DISCUSSION 4 5 Very-low certainty evidence from randomized trials and observational studies was conflicting 6 7 8 and leaves uncertain whether the addition of medical cannabis affects the use of prescribed 9 10 opioids among patients living with chronic pain. Compared with long-term opioid therapy for 11 12 chronic pain without medical cannabis, high certainty evidence showed that adding medical 13 14 15 cannabis did not reduce pain or sleep disturbance. Results provided moderate certainty evidence 16 For peer review only 17 that adding cannabis therapy to opioids likely increases both nausea (RR 1.43, 95%CI 1.04 to 18 19 1.96) and vomiting (RR 1.50; 95%CI 1.01 to 2.24), and low certainty evidence suggested that it 20 21 22 may have no effect on constipation (RR 0.85, 95%CI 0.54 to 1.35). 23 24 25 Strengths of our review include a comprehensive search for eligible randomized and 26 27 observational studies, appraisal of the risk of bias among individual studies, and use of the 28 29 GRADE approach to rate the certainty of evidence. Our review has limitations, primarily due to 30 31 32 features of primary studies eligible for review. All eligible RCTs enrolled patients with chronic 33 http://bmjopen.bmj.com/ 34 cancer-related pain, and the generalizability to non-cancer chronic pain is uncertain. Most 35 36 observational studies incorporated inadequate adjustment for confounding, and all randomized 37 38 trials, despite reporting this outcome, were not designed to address the effect of medical cannabis 39 40 41 on opioid use. on October 1, 2021 by guest. Protected copyright. 42 43 44 A meta-analysis of pre-clinical studies,16 a narrative systematic review,55 and several 45 46 cross-sectional and case studies have reported an apparent reduction in opioid use with addition 47 48 8, 9, 56-60 61 49 of cannabis therapy. In a national US population-based survey of 2,774 cannabis users 50 51 (both medical and non-medical use) 36% of respondents reported substituting cannabis for 52 53 prescription opioids (discontinued opioid use). In this survey, 60% of participants who identified 54 55 as medical cannabis users were much more likely to substitute cannabis for prescription drugs 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 85 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 than recreational users (OR 4.59; 95%CI 3.87 to 5.43). Another US survey62 that included 841 4 5 6 patients prescribed long-term opioid therapy for chronic pain reported that 61% used medical 7 8 cannabis, and 97% of this subgroup reported coincident reduction of their opioid use. 9 10 11 Consistent with these findings, very low certainty evidence from observational studies in 12 13 our review also suggests that adding medical cannabis allows patients with chronic pain to 14 15 16 reduce their use of opioids.For Althoughpeer RCT review results do not support only reduction in opioid dose by 17 18 adding medical cannabis for opioids, the evidence is also very low certainty, primarily because 19 20 investigators instructed patients to maintain their current opioid dose. One could argue that this 21 22 23 limitation makes the evidence irrelevant to the issue of opioid reduction with cannabis use. 24 25 Results showed that, among patients with chronic cancer pain prescribed opioid therapy, the 26 27 addition of medical cannabis does not result in important reductions in pain, and likely does not 28 29 improve sleep quality. 30 31 32 33 http://bmjopen.bmj.com/ 34 35 Conclusion 36 37 The opioid-sparing effects of medical cannabis for chronic pain remain uncertain. Based on 38 39 moderate-to-high certainty evidence, adding medical cannabis to opioid therapy among chronic 40 41 on October 1, 2021 by guest. Protected copyright. 42 cancer pain patients influences neither pain relief nor sleep disturbance and increases the risk of 43 44 nausea and vomiting. The accompanying BMJ Rapid Recommendation17 provides contextualized 45 46 guidance based on this evidence, as well as three other systematic reviews on benefits,63 harms64 47 48 65 49 and patients' values and preferences. 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 18 of 85

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 Disclosures/Conflicts of Interest: All authors have no financial relationships with any 4 5 6 organizations that might have an interest in the submitted work. 7 8 Role of the Funding Source: This review received no external funding or other support. 9 10 Ethical approval: ethical approval is not required because this study retrieved and synthesised 11 12 data from already published studies. 13 14 15 Data: Details of the characteristics of the included studies were shared in the supplementary 16 For peer review only 17 materials. Data will be made available upon publication and can be obtained from the 18 19 corresponding author at [email protected]. 20 21 22 Disclaimers: None. 23 24 Transparency: All authors affirm that the manuscript is an honest, accurate, and transparent 25 26 account of the study being reported; that no important aspects of the study have been omitted; 27 28 29 and that any discrepancies from the study as planned (and, if relevant, registered) have been 30 31 explained. 32 33 Patients and public involvement: Patients and public were not involved in this research. http://bmjopen.bmj.com/ 34 35 Contribution: JWB, AN, GG, conceived and designed the study. RC performed the literature 36 37 38 search. AN, AM, YSh, VA, YR selected the studies, extracted the relevant information, and 39 40 assessed the risk of bias of selected studies. AN synthesised the data. AN wrote the first draft of 41 on October 1, 2021 by guest. Protected copyright. 42 the paper. AN, JWB, GG, and TA critically revised the manuscript for important intellectual 43 44 45 content. JWB, LT, GG, MB, and NB interpreted the findings. JWB, LT, and GG provided 46 47 methodological support. All authors reviewed the paper and approved the final version. The 48 49 corresponding author attests that all listed authors meet authorship criteria and that no others 50 51 52 meeting the criteria have been omitted. 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 85 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 Figure 1: Study selection process in review of opioid-sparing effects of cannabis in chronic pain 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 41 on October 1, 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: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

BMJ Open Page 20 of 85

1 2 3 Table 1: Characteristics of included studies (n=18) 4 5 Author- Study Participan Pain Age Fema Opioid FU Daily dose Co- Funding 6 year design ts # classification mean le regimen duratio of medical interven source 7 (country) (opioid (specific (SD) (%) (baseline dose n cannabis tion condition) in MME 8 user %) 9 mean ± SD) 10 11 Bellnier et One-arm n= 29 90% CNCP; 61 65% Different 13 10mg NR NR 12 al-2018 observationa (100%)For10% peer cancer (10) reviewopioids weeksonlycapsules of 13 (US)42 l study pain (79.94; ranged THC/ 14 0 to 450) CBD in a 15 1:1 ratio 3- 16 times daily 17 Barlow et al- Retrospectiv Enrolled in 100% CNCP 49.9 45% Different Ranged NR NR NR 18 2019 (US)43 e chart- MCP=34; (chronic (10.5) opioids :34 to http://bmjopen.bmj.com/ 19 review not painful (not enrolled 297 20 enrolled in pancreatitis) in MCP weeks 21 MCP=19 183±284; 22 (100%) enrolled in 23 MCP 24 190±273) 25 26 on October 1, 2021 by guest. Protected copyright. 27 Capano et One-arm n= 131 Chronic pain 56.1 68% On a stable 8 30mg NR Funded 28 al-2020 observationa (100%) (mixed of (rang opioid for at weeks CBD/1mg by (US)44 l study cancer and e: 39 least THC Ananda 29 non-cancer) to 70) 1 year Professio 30 (defined as nal. 31 less than 10% 32 change in its 33 severity) 34 Fallon et al- Parallel arm n=399; 100% chronic 59.8 43% On a stable 5 THC 27 Patients Otsuka 35 2017-study I RCT Nabiximol cancer pain (10.9) maintenance weeks mg/mL; allowed Pharmace 36 (multicenter s =20 opioid therapy CBD 25 to take utical 37 trial£)38 =1 with <500mg mg/mL not more Developm 38 99 MME/day (maximum than one ent & 39 (100%) (Nabiximols: allowed type of Commerc 40 199±131; daily breakthro ialization, 41 dosage ugh Inc., 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

Page 21 of 85 BMJ Open

1 2 3 placebo: of 10 medicati Rockville, 4 207±135) sprays) ons/other MD, USA 5 interventi 6 on that 7 could 8 affect 9 pain 10 were 11 restricted 12 Fallon et al- Parallel arm n=206;For 100% peer chronic 61.5 review49% Same as above only5 Same as Same as Otsuka 13 2017- study RCT Nabiximol cancer pain (11.3) (Nabiximols: weeks above. above Pharmace 14 II s=103, 212±136; Patients utical 15 (multicenter placebo=1 placebo: with 15% Developm £ 38 16 trial ) 03 209±121) improveme ent & 17 (100%) nt in pain Commerc 18 entered http://bmjopen.bmj.com/ ialization, 19 into the Inc., double- Rockville, 20 blinded MD, USA 21 phase 22 Haroutounia One-arm n=73 93.2% CNCP; 51.2 38%¥ - 26 Cigarettes: On a No- 23 n et al-2016 observationa (35%) 6.8% chronic (15.4) , weeks 6% to 14% stable external 24 (Israel)45 l study cancer pain ¥ Morphine, THC, medicati funding 25 , 0.2% to on; 26 3.8% however, on October 1, 2021 by guest. Protected copyright. 27 , CBD; patients 28 were 29 (median 60; Oral: 11% encourag 30 ranged 45, 90) to 19% ed to 31 THC, discontin 32 0.5% to ue or 33 5.5% CBD decrease 34 using of 35 other 36 pain 37 killers 38 Johnson et Parallel arm n=177; 100% chronic 60.2 46% On a stable 2 One spray: Intervent GW 39 al-2010 RCT THC: CBD cancer pain (12.3) maintenance weeks 2.7mg ion that Pharma 40 (multicenter extract= strong opioid THC/2.5m could Ltd £ 39 41 trial ) 60, THC therapy g CBD. affect 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

BMJ Open Page 22 of 85

1 2 3 extract=58, for at least The pain 4 placebo=5 one-week maximum within 2 5 9 before permitted weeks of 6 (100%) included into dose: screening 7 the study 8 were 8 (THC:CBD: actuations restricted 9 258±789; in any 3- 10 THC: hour and 11 188±234; 48 12 For peer reviewplacebo: onlyactuations 13 367±886) in any 24- 14 hour 15 Lichtman et Parallel arm n=398; 100% chronic 60 46% On a stable 5 THC 27 Patients Otsuka 16 al-2018 RCT Nabiximol cancer pain (11.5) maintenance weeks mg/mL; allowed Pharmace £ 17 (multicenter =199 opioid therapy CBD 25 to take utical 40 18 ) Placebo=1 with <500 mg/mL not morehttp://bmjopen.bmj.com/ Developm 19 98 MME/day (maximum than one ent & (100%) (nabiximols: allowed type of Commerc 20 193±130; daily breakthro ialization, 21 placebo: dosage ugh Inc., 22 186±131) of 10 medicati Rockville, 23 sprays per ons/other MD, USA 24 day) interventi 25 on that 26 could on October 1, 2021 by guest. Protected copyright. 27 affect 28 pain 29 were 30 restricted 31 Maida et al- Prospective Enrolled in 100% Chronic 69.7 42% Different 4 On Patients Valeant 32 2008 cohort MCP=47 cancer pain (10.1) opioids weeks average were Pharma- 33 (Canada)46 not (nabilone 1.79 mg permitted ceuticals 34 enrolled in treated:60±64; twice daily to use Canada 35 MCP=65 untreated: nabilone concomit Ltd 36 (100%) 67±101) ant 37 medicati 38 ons 39 Narang et al- One-arm n=30 100% CNCP Medi 53% Methadone, Phase Flexible NR Solvay 40 2008 (US)47 observationa (100%) an=4 Morphine, 2: open dose Pharmace 41 l study 3.5 Oxycodone, schedule, uticals, 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

Page 23 of 85 BMJ Open

1 2 3 (rang , label dronabinol Inc. 4 e=21- Hydromorpho for 4 5mg to 5 67) ne weeks. 20mg 3 6 (68±57) times daily 7 O’Connell et One-arm n=77 100% CNCP 54.1 58% Different 26 NR Patients No- 8 al-2019 observationa (100%) (mixed (rang opioids weeks were industry 9 (US)48 l study conditions) e=26- (140±184) permitted funding 10 76) to use 11 concomit 12 For peer review only ant 13 medicati 14 ons 15 Portenoy et Parallel arm n=360; 100% chronic 58 48% On a stable 5 THC 27 Concomi Supported 16 al-2012 RCT nabiximols cancer pain (12.2) maintenance weeks mg/mL; tant use by GW 17 (multicenter low- opioid therapy CBD 25 of Pharmace 41 18 £) dose=91, with <500 mg/mL medicatihttp://bmjopen.bmj.com/ uticals 19 medium- MME/day (maximum ons was and 20 dose=88, (median MME allowed restricted Otsuka high- of 120 mg daily 21 dose=90, ranged 3 to dosage of 22 placebo=9 16660) 10 sprays 23 1 per day) 24 (100%) 25 Pritchard- Retrospectiv cannabis 100% chronic 53.1 23% Different 26 NR NR No- on October 1, 2021 by guest. Protected copyright. 26 2020 e cohort and cancer pain (11.7) opioids (MCP weeks industry 27 (US)49 opioids co- enrolled=144± funding 28 use=22 129; MCP not 29 Opioids enrolled=119± 30 only=61 100) 31 (100%) 32 Pawasarat- Retrospectiv Enrolled in 100% chronic 58 56% Different Betwee NR NR No- 33 2020 e chart MCP=137, cancer pain (IQR: opioids (MCP n 39 industry 34 (US)50 review not 14.7) enrolled=medi and 52 funding 35 enrolled in an 45 weeks 36 MCP=95 IQR=135; for 37 (100%) MCP not MCP 38 enrolled=97,1 enrolle 39 50) d; <26 40 weeks 41 for not 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

BMJ Open Page 24 of 85

1 2 3 enrolle 4 d 5 Rod-2019 One-arm n=600 Chronic pain NR NR Different 6 CBD and All No- 6 (Canada)54 observationa (not specified opioids months THC participa external 7 l study type of pain) ranging from ranged nts funding 8 90 to 240mg between indicated 9 MME 4% to 6%. ready to 10 (average Doses reduce 11 120mg MME) related opioid 12 For peer review onlydirectly to dose and 13 the opioid also 14 taper. received 15 psycholo 16 gical 17 supports 18 (e.g.http://bmjopen.bmj.com/ 19 CBT, mindfuln 20 ess, 21 relaxatio 22 n) 23 24 Takakuwa et One-arm n=61 100% CNCP 50 38% Different Median NR NR The 25 al-2020 observationa (100%) (back pain) (11.4) opioids of 6.4 Society of 26 (US)51 l study divided into years on October 1, 2021 by guest. Protected copyright. Cannabis 27 intermittent among Clinicians 28 users and short patients paid for 29 intermittent who the IRB 30 users ceased review 31 (median 21 opioids and 32 ranged 1.1, comple statistical 33 500) tely analysis 34 Vigil et al- Retrospectiv Enrolled in 100% CNCP 56.3 36% Different 52 Varied in NR the 35 2017 e chart MCP *=37 (11.8) opioids with weeks individuals Universit 36 (US)52 review not maximum based on y of New 37 enrolled=2 daily dosages their Mexico 38 9 (100%) of less than selection Medical 39 200 Cannabis 40 (enrolled in Research 41 MCP: 24±23; Fund 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

Page 25 of 85 BMJ Open

1 2 3 not enrolled: 4 16±14) 5 Yassin et al- One-arm n=31 100% CNCP 33.4 90% Oxycodone 5 26 THC Patients NR 6 2019 observationa (100%) (fibromyalgia) (12.3) mg three weeks to CBD were 7 (Israel)53 l study times/daily ratio: ¼, allowed 8 (not reported) 20 to use 9 g/month concomit 10 for 3 ant pain 11 months, therapy 12 For peer review onlyincreased in a 13 up to 30 stable 14 g/month at dose 15 the end of 16 6 months 17 *CNCP: Chronic non-cancer pain; MCP: Medical Cannabis Program; MME: milligram morphine equivalent; FU: follow-up; NR: not 18 reported http://bmjopen.bmj.com/ 19 ¥ Based on the whole population including opioid users and non-users 20 £In Belgium, Bulgaria, Czech Republic, Estonia, Germany, Hungary, Latvia, Lithuania, Poland, Romania, the United Kingdom and 21 the United States 22 23 24 25 26 on October 1, 2021 by guest. Protected copyright. 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

BMJ Open Page 26 of 85

1 2 3 Table 2: GRADE Evidence Profile of cannabis for patients with chronic pain prescribed long-term opioid therapy 4 5 # of studies # of FU Risk of Inconsistency Indirectnessc Imprecisiond Publication Treatment Overall 6 Patients Duration biasa (I2, P-value)b bias association certainty of 7 (Weeks) (95% CI) evidence 8 Opioid dose reduction: morphine milligram equivalents (MME) per day 9 10 4 RCTs38-40 1,176 2 to 5 No serious No serious Very serious Serious Not WMD 11 risk of bias e inconsistency indirectness f imprecision g detected -3.4MME Very Low 12 For peer[40.4%, review only (-12.7 to 5.9) 13 P=0.15] 14 8 453 4 to 297 Serious risk Serious No serious No serious Not WMD Very low 15 Observational of bias h inconsistency indirectness imprecision detected -22.5MME 16 studies42, 43, 45, [visual (-43.06 to - 17 46, 48-50, 52 inspection] 1.97) 18 http://bmjopen.bmj.com/ 19 Pain: 10 cm VAS for pain; lower is better; the MID = 1 cm 20 5 RCTs38-41 1,536 2 to 5 No serious No serious No serious No serious Not WMD -0.18 21 risk of biase inconsistency indirectness imprecision detected (-0.38 to High 22 [28%, P=0.20] 0.02) 23 24 Sleep disturbance: 10 cm VAS for sleep disturbance; lower is better; the MID= 1 cm 25 38-41 26 5 RCTs 1,536 2 to 5 No serious No serious No serious No serious Not on October 1, 2021 by guest. Protected copyright. WMD -0.22 High e 27 risk of bias inconsistency indirectness imprecision detected (-0.39 to - 28 [0%, P=0.45] 0.06) 29 30 Nausea 31 32 4 RCTs*38-41 1330 2 to 5 Serious risk No serious No serious No serious Not RR 1.43 Moderate 33 of biasi inconsistency indirectness imprecision detected (1.04 to 1.96) 34 [0%, P=0.88] 35 Vomiting 36 37 4 RCTs*38-41 1330 2 to 5 Serious risk No serious No serious No serious Not RR 1.5 Moderate 38 of biasi inconsistency indirectness imprecision detected (1.01 to 2.24) 39 [0%, P=0.50] 40 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

Page 27 of 85 BMJ Open

1 2 3 Constipation 4 5 3 RCTs*38, 40, 1153 5 Serious risk No serious No serious Serious Not RR 0.85 Low 6 41 of bias i inconsistency indirectness imprecision g detected (0.54 to 1.35) 7 [0%, P=0.92] 8 9 10 WMD: weighted mean difference; RR: relative risk; 95% CI: 95% confidence interval; VAS: visual analogue scale; MID: minimally 11 important difference; FU: follow-up 12 a We assessed risk of bias usingFor a modified Cochranepeer risk of biasreview instrument; only 13 b Inconsistency refers to unexplained heterogeneity of results. For RCTs an I2 of 75-100% indicates that heterogeneity may be 14 considerable. We assessed heterogeneity of pooled observational studies through visual inspection of forest plots. 15 c Indirectness results if the intervention, control, patients or outcomes are different from the research question under investigation. 16 d 17 Serious imprecision refers to situations in which the confidence interval includes both benefit and harm (the 95%CI includes 1 MID). e 18 Some of the included RCTs were at high risk of bias, due to loss to follow-up (>20%); however, we did not http://bmjopen.bmj.com/ rate down for risk of bias 19 as subgroup analysis showed no difference in treatment effect between trials at high and low risk of bias for missing outcome data 20 (test of interaction p= 0.791). 21 f downgraded twice due to indirectness since all trials instructed participants to maintain their opioid dose during the study period. 22 g The 95%CI around the WMD includes no effect. 23 h 24 Studies are based on non-representative samples. i 25 Most of the included RCTs were at high risk of bias due to loss to follow-up (>20%). 26 on October 1, 2021 by guest. Protected copyright. 27 *Fallon et al-2017 (the results of two separate RCTs reported in this publication): only study number 1 reported these outcomes and 28 subsequently included in the meta-analysis. 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 28 of 85

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 References: 4 5 6 1. Schopflocher D, Taenzer P, Jovey R. The prevalence of chronic pain in Canada. Pain Res 7 8 Manag 2011;16(6):445-50. doi:10.1155/2011/876306. 9 10 2. Schappert SM, Burt CW. Ambulatory care visits to physician offices, hospital outpatient 11 12 departments, and emergency departments: United States, 2001-02. Vital and Health Statistics 13 14 15 Series 13, Data from the National Health Survey 2006(159):1-66. 16 For peer review only 17 3. International Narcotics Control Board. Narcotic drugs: estimated world requirements for 18 19 2019: https://www.incb.org/documents/Narcotic-Drugs/Technical-Publications/2018/INCB- 20 21 22 Narcotics_Drugs_Technical_Publication_2018.pdf (Accessed: 02 Nov 2020). Vienna: United 23 24 Nations. 25 26 4. Gomes T, Greaves S, Martins D, et al. Latest Trends in Opioid-Related Deaths in 27 28 29 Ontario: 1991 to 2015. Toronto: Ontario Drug Policy Research Network; 2017. 30 31 5. Busse JW, Craigie S, Juurlink DN, et al. Guideline for opioid therapy and chronic 32 33 noncancer pain. CMAJ 2017;189(18):E659-e66. doi:10.1503/cmaj.170363. http://bmjopen.bmj.com/ 34 35 6. Gomes T, Mamdani MM, Paterson JM, Dhalla IA, Juurlink DN. Trends in high-dose 36 37 38 opioid prescribing in Canada. Can Fam Physician 2014;60(9):826-32. 39 40 7. Bedson J, Chen Y, Ashworth J, et al. Risk of adverse events in patients prescribed long- 41 on October 1, 2021 by guest. Protected copyright. 42 term opioids: A cohort study in the UK Clinical Practice Research Datalink. Eur J Pain 43 44 45 2019;23(5):908-22. doi:10.1002/ejp.1357. 46 47 8. Gwira Baumblatt JA, Wiedeman C, Dunn JR, et al. High-risk use by patients prescribed 48 49 opioids for pain and its role in overdose deaths. JAMA Intern Med 2014;174(5):796-801. 50 51 52 doi:10.1001/jamainternmed.2013.12711. 53 54 55 56 57 58 27 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 29 of 85 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 9. Bohnert AS, Logan JE, Ganoczy D, Dowell D. A Detailed Exploration Into the 4 5 6 Association of Prescribed Opioid Dosage and Overdose Deaths Among Patients With Chronic 7 8 Pain. Med Care 2016;54(5):435-41. doi:10.1097/mlr.0000000000000505. 9 10 10. Attal N, Mazaltarine G, Perrouin-Verbe B, Albert T. Chronic 11 12 management in spinal cord injury patients. What is the efficacy of pharmacological treatments 13 14 15 with a general mode of administration? (oral, transdermal, intravenous). Ann Phys Rehabil Med 16 For peer review only 17 2009;52(2):124-41. doi:10.1016/j.rehab.2008.12.011. 18 19 11. Livingston MD, Barnett TE, Delcher C, Wagenaar AC. Recreational Cannabis 20 21 22 Legalization and Opioid-Related Deaths in Colorado, 2000-2015. Am J Public Health 23 24 2017;107(11):1827-9. doi:10.2105/ajph.2017.304059. 25 26 12. Bradford AC, Bradford WD. Medical Marijuana Laws Reduce Prescription Medication 27 28 29 Use In Medicare Part D. Health Aff (Millwood) 2016;35(7):1230-6. 30 31 doi:10.1377/hlthaff.2015.1661. 32 33 13. Bachhuber MA, Saloner B, Cunningham CO, Barry CL. Medical cannabis laws and http://bmjopen.bmj.com/ 34 35 opioid overdose mortality in the United States, 1999-2010. JAMA Intern Med 36 37 38 2014;174(10):1668-73. doi:10.1001/jamainternmed.2014.4005. 39 40 14. Boehnke KF, Litinas E, Clauw DJ. Medical Cannabis Use Is Associated With Decreased 41 on October 1, 2021 by guest. Protected copyright. 42 Medication Use in a Retrospective Cross-Sectional Survey of Patients With Chronic Pain. 43 44 45 J Pain 2016;17(6):739-44. doi:10.1016/j.jpain.2016.03.002. 46 47 15. Piper BJ, DeKeuster RM, Beals ML, et al. Substitution of medical cannabis for 48 49 pharmaceutical agents for pain, anxiety, and sleep. J Psychopharmacol 2017;31(5):569-75. 50 51 52 doi:10.1177/0269881117699616. 53 54 55 56 57 58 28 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 30 of 85

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 16. Nielsen S, Sabioni P, Trigo JM, et al. Opioid-Sparing Effect of Cannabinoids: A 4 5 6 Systematic Review and Meta-Analysis. Neuropsychopharmacology 2017;42(9):1752-65. 7 8 doi:10.1038/npp.2017.51. 9 10 17. Busse JW, Vankrunkelsven P, Zeng L, et al. Medical cannabis for chronic pain: a clinical 11 12 practice guideline. BMJ 2020(submitted). 13 14 15 18. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in 16 For peer review only 17 epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology 18 19 (MOOSE) group. JAMA 2000;283(15):2008-12. doi:10.1001/jama.283.15.2008. 20 21 22 19. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic 23 24 reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and 25 26 elaboration. BMJ 2009;339:b2700. doi:10.1136/bmj.b2700. 27 28 29 20. Classification of chronic pain. Descriptions of chronic pain syndromes and definitions of 30 31 pain terms. Prepared by the International Association for the Study of Pain, Subcommittee on 32 33 Taxonomy. Pain Suppl 1986;3:S1-226. http://bmjopen.bmj.com/ 34 35 21. Reeves BC, Deeks JJ, Higgins JPT, et al. Chapter 24: Including non-randomized studies 36 37 38 on intervention effects. . In: Higgins JPT TJ, Chandler J, Cumpston M, Li T, Page MJ, Welch 39 40 VA (editors). editor: Cochrane Handbook for Systematic Reviews of Interventions version 6.1. 41 on October 1, 2021 by guest. Protected copyright. 42 Available from www.training.cochrane.org/handbook; 2020. 43 44 45 22. Dworkin RH, Turk DC, Farrar JT, et al. Core outcome measures for chronic pain clinical 46 47 trials: IMMPACT recommendations. Pain 2005;113(1-2):9-19. doi:10.1016/j.pain.2004.09.012. 48 49 23. Goshua A, Craigie S, Guyatt GH, et al. Patient Values and Preferences Regarding 50 51 52 Opioids for Chronic Noncancer Pain: A Systematic Review. Pain Med 2018;19(12):2469-80. 53 54 doi:10.1093/pm/pnx274. 55 56 57 58 29 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 31 of 85 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 24. Akl EA, Sun X, Busse JW, et al. Specific instructions for estimating unclearly reported 4 5 6 blinding status in randomized trials were reliable and valid. J Clin Epidemiol 2012;65(3):262-7. 7 8 doi:10.1016/j.jclinepi.2011.04.015. 9 10 25. Higgins JP, Altman DG, Gøtzsche PC, et al. The Cochrane Collaboration's tool for 11 12 assessing risk of bias in randomised trials. BMJ 2011;343:d5928. doi:10.1136/bmj.d5928. 13 14 15 26. Tool to Assess Risk of Bias. CLARITY Group at McMaster University. Access 16 For peer review only 17 www.evidencepartners.com/resources/methodological-resources/: Evidence partner; 2020 [ 18 19 27. Landis JR, Koch GG. The measurement of observer agreement for categorical data. 20 21 22 Biometrics 1977;33(1):159-74. 23 24 28. Farrar JT, Young JP, Jr., LaMoreaux L, Werth JL, Poole RM. Clinical importance of 25 26 changes in chronic pain intensity measured on an 11-point numerical pain rating scale. Pain 27 28 29 2001;94(2):149-58. doi:10.1016/s0304-3959(01)00349-9. 30 31 29. Zisapel N, Nir T. Determination of the minimal clinically significant difference on a 32 33 patient visual analog sleep quality scale. J Sleep Res 2003;12(4):291-8. doi:10.1046/j.0962- http://bmjopen.bmj.com/ 34 35 1105.2003.00365.x. 36 37 38 30. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 39 40 1986;7(3):177-88. doi:10.1016/0197-2456(86)90046-2. 41 on October 1, 2021 by guest. Protected copyright. 42 31. Wan X, Wang W, Liu J, Tong T. Estimating the sample mean and standard deviation 43 44 45 from the sample size, median, range and/or interquartile range. BMC Med Res Methodol 46 47 2014;14:135. doi:10.1186/1471-2288-14-135. 48 49 32. Ma J, Liu W, Hunter A, Zhang W. Performing meta-analysis with incomplete statistical 50 51 52 information in clinical trials. BMC Med Res Methodol 2008;8:56. doi:10.1186/1471-2288-8-56. 53 54 55 56 57 58 30 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 32 of 85

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 33. Rücker G, Cates CJ, Schwarzer G. Methods for including information from multi-arm 4 5 6 trials in pairwise meta-analysis. Res Synth Methods 2017;8(4):392-403. doi:10.1002/jrsm.1259. 7 8 34. Rücker G, Schwarzer G, Carpenter JR, Schumacher M. Undue reliance on I(2) in 9 10 assessing heterogeneity may mislead. BMC Med Res Methodol 2008;8:79. doi:10.1186/1471- 11 12 2288-8-79. 13 14 15 35. Guyatt GH, Oxman AD, Vist GE, et al. GRADE: an emerging consensus on rating 16 For peer review only 17 quality of evidence and strength of recommendations. BMJ 2008;336(7650):924-6. 18 19 doi:10.1136/bmj.39489.470347.AD. 20 21 22 36. Guyatt GH, Oxman AD, Sultan S, et al. GRADE guidelines: 9. Rating up the quality of 23 24 evidence. J Clin Epidemiol 2011;64(12):1311-6. doi:10.1016/j.jclinepi.2011.06.004. 25 26 37. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a 27 28 29 simple, graphical test. BMJ 1997;315(7109):629-34. doi:10.1136/bmj.315.7109.629. 30 31 38. Fallon MT, Albert Lux E, McQuade R, et al. Sativex oromucosal spray as adjunctive 32 33 therapy in advanced cancer patients with chronic pain unalleviated by optimized opioid therapy: http://bmjopen.bmj.com/ 34 35 two double-blind, randomized, placebo-controlled phase 3 studies. Br J Pain 2017;11(3):119-33. 36 37 38 doi:10.1177/2049463717710042. 39 40 39. Johnson JR, Burnell-Nugent M, Lossignol D, et al. Multicenter, double-blind, 41 on October 1, 2021 by guest. Protected copyright. 42 randomized, placebo-controlled, parallel-group study of the efficacy, safety, and tolerability of 43 44 45 THC:CBD extract and THC extract in patients with intractable cancer-related pain. J Pain 46 47 Symptom Manage 2010;39(2):167-79. doi:10.1016/j.jpainsymman.2009.06.008. 48 49 40. Lichtman AH, Lux EA, McQuade R, et al. Results of a Double-Blind, Randomized, 50 51 52 Placebo-Controlled Study of Nabiximols Oromucosal Spray as an Adjunctive Therapy in 53 54 55 56 57 58 31 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 33 of 85 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 Advanced Cancer Patients with Chronic Uncontrolled Pain. J Pain Symptom Manage 4 5 6 2018;55(2):179-88.e1. doi:10.1016/j.jpainsymman.2017.09.001. 7 8 41. Portenoy RK, Ganae-Motan ED, Allende S, et al. Nabiximols for opioid-treated cancer 9 10 patients with poorly-controlled chronic pain: a randomized, placebo-controlled, graded-dose 11 12 trial. J Pain 2012;13(5):438-49. doi:10.1016/j.jpain.2012.01.003. 13 14 15 42. Bellnier T, Brown GW. Preliminary evaluation of the efficacy, safety, and costs 16 For peer review only 17 associated with the treatment of chronic pain with medical cannabis. Ment Health Clin 18 19 2018;8(3):110-5. doi:10.9740/mhc.2018.05.110. 20 21 22 43. Barlowe TS, Koliani-Pace JL, Smith KD, Gordon SR, Gardner TB. Effects of Medical 23 24 Cannabis on Use of Opioids and Hospital Visits by Patients With Painful Chronic Pancreatitis. 25 26 Clin Gastroenterol Hepatol 2019;17(12):2608-9.e1. doi:10.1016/j.cgh.2019.01.018. 27 28 29 44. Capano A, Weaver R, Burkman E. Evaluation of the effects of CBD hemp extract on 30 31 opioid use and quality of life indicators in chronic pain patients: a prospective cohort study. 32 33 Postgrad Med 2020;132(1):56-61. doi:10.1080/00325481.2019.1685298. http://bmjopen.bmj.com/ 34 35 45. Haroutounian S, Ratz Y, Ginosar Y, et al. The Effect of Medicinal Cannabis on Pain and 36 37 38 Quality-of-Life Outcomes in Chronic Pain: A Prospective Open-label Study. Clin J Pain 39 40 2016;32(12):1036-43. doi:10.1097/ajp.0000000000000364. 41 on October 1, 2021 by guest. Protected copyright. 42 46. Maida V, Ennis M, Irani S, Corbo M, Dolzhykov M. Adjunctive nabilone in cancer pain 43 44 45 and symptom management: a prospective observational study using propensity scoring. J 46 47 Support Oncol 2008;6(3):119-24. 48 49 47. Narang S, Gibson D, Wasan AD, et al. Efficacy of dronabinol as an adjuvant treatment 50 51 52 for chronic pain patients on opioid therapy. J Pain 2008;9(3):254-64. 53 54 doi:10.1016/j.jpain.2007.10.018. 55 56 57 58 32 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 34 of 85

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 48. O'Connell M, Sandgren M, Frantzen L, Bower E, Erickson B. Medical Cannabis: Effects 4 5 6 on Opioid and Benzodiazepine Requirements for Pain Control. Ann Pharmacother 7 8 2019;53(11):1081-6. doi:10.1177/1060028019854221. 9 10 49. Pritchard ER, Dayer L, Belz J, et al. Effect of cannabis on opioid use in patients with 11 12 cancer receiving palliative care. J Am Pharm Assoc (2003) 2020;60(1):244-7. 13 14 15 doi:10.1016/j.japh.2019.10.013. 16 For peer review only 17 50. Pawasarat IM, Schultz EM, Frisby JC, et al. The Efficacy of Medical Marijuana in the 18 19 Treatment of Cancer-Related Pain. J Palliat Med 2020;23(6):809-16. 20 21 22 doi:10.1089/jpm.2019.0374. 23 24 51. Takakuwa KM, Hergenrather JY, Shofer FS, Schears RM. The Impact of Medical 25 26 Cannabis on Intermittent and Chronic Opioid Users with Back Pain: How Cannabis Diminished 27 28 29 Prescription Opioid Usage. Cannabis Res 2020;5(3):263-70. 30 31 doi:10.1089/can.2019.0039. 32 33 52. Vigil JM, Stith SS, Adams IM, Reeve AP. Associations between medical cannabis and http://bmjopen.bmj.com/ 34 35 prescription opioid use in chronic pain patients: A preliminary cohort study. PLoS One 36 37 38 2017;12(11):e0187795. doi:10.1371/journal.pone.0187795. 39 40 53. Yassin M, Oron A, Robinson D. Effect of adding medical cannabis to analgesic treatment 41 on October 1, 2021 by guest. Protected copyright. 42 in patients with low back pain related to fibromyalgia: an observational cross-over single centre 43 44 45 study. Clin Exp Rheumatol 2019;37 Suppl 116(1):13-20. 46 47 54. Rod K. A Pilot Study of a Medical Cannabis - Opioid Reduction Program. American 48 49 Journal of Psychiatry and Neuroscience 2019;7(3):74-7. doi:10.11648/j.ajpn.20190703.14 50 51 52 53 54 55 56 57 58 33 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 35 of 85 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 55. Okusanya BO, Asaolu IO, Ehiri JE, et al. Medical cannabis for the reduction of opioid 4 5 6 dosage in the treatment of non-cancer chronic pain: a systematic review. Syst Rev 2020;9(1):167. 7 8 doi:10.1186/s13643-020-01425-3. 9 10 56. Zaller N, Topletz A, Frater S, Yates G, Lally M. Profiles of medicinal cannabis patients 11 12 attending compassion centers in rhode island. J Psychoactive Drugs 2015;47(1):18-23. 13 14 15 doi:10.1080/02791072.2014.999901. 16 For peer review only 17 57. Hazekamp A, Ware MA, Muller-Vahl KR, Abrams D, Grotenhermen F. The medicinal 18 19 use of cannabis and cannabinoids--an international cross-sectional survey on administration 20 21 22 forms. J Psychoactive Drugs 2013;45(3):199-210. doi:10.1080/02791072.2013.805976. 23 24 58. Lucas P, Walsh Z, Crosby K, et al. Substituting cannabis for prescription drugs, 25 26 and other substances among medical cannabis patients: The impact of contextual factors. Drug 27 28 29 Alcohol Rev 2016;35(3):326-33. doi:10.1111/dar.12323. 30 31 59. Cooper ZD, Bedi G, Ramesh D, et al. Impact of co-administration of oxycodone and 32 33 smoked cannabis on analgesia and abuse liability. Neuropsychopharmacology http://bmjopen.bmj.com/ 34 35 2018;43(10):2046-55. doi:10.1038/s41386-018-0011-2. 36 37 38 60. Ishida JH, Wong PO, Cohen BE, et al. Substitution of marijuana for opioids in a national 39 40 survey of US adults. PloS one 2019;14(10):e0222577. 41 on October 1, 2021 by guest. Protected copyright. 42 61. Corroon JM, Jr., Mischley LK, Sexton M. Cannabis as a substitute for prescription drugs 43 44 45 - a cross-sectional study. J Pain Res 2017;10:989-98. doi:10.2147/jpr.s134330. 46 47 62. Reiman A, Welty M, Solomon P. Cannabis as a Substitute for Opioid-Based Pain 48 49 Medication: Patient Self-Report. Cannabis Cannabinoid Res 2017;2(1):160-6. 50 51 52 doi:10.1089/can.2017.0012. 53 54 55 56 57 58 34 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 36 of 85

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 63. Wang L, Hong PJ, May C, et al. Medical cannabis for chronic pain: a systematic review 4 5 6 and meta-analysis of randomized clinical trials. BMJ 2020(submitted). 7 8 64. Zeraatkar D, Cooper MA, Agarwal A, et al. Long-term and serious harms of medical 9 10 cannabis for chronic pain: A systematic review of non-randomized studies. BMJ 11 12 2020(submitted). 13 14 15 65. Zeng L, Lytvyn L, Wang X, et al. Values and preferences towards medical cannabis 16 For peer review only 17 among patients with chronic pain: A mixed methods systematic review. BMJ 2020(submitted). 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 October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 35 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 37 of 85 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 4

5 Records identified through Additional records identified 6 database searching through other sources 7 (n = 5132) (n =1)

8 tification 9

10 Iden 11 12 13 Records after duplicates removed (n =4289) 14 15

16 For peer review only 17 18 Titles & abstracts screened Records excluded 19 (n=4289) (n=4165)

20 Screening 21 22 23 Full-text articles excluded 24

25 Full-text articles assessed with reasons (n =116): 26 for eligibility (n=133) 27 Not opioid users (n=64); Not 28 chronic pain (n=6); 29 Recreational cannabis use 30 (n=2); Medical cannabis not

31 Eligibility 32 Studies included in added to opioid (n=4); less 33 qualitative synthesis (n than two weeks’ follow-up http://bmjopen.bmj.com/ 34 =18 studies in 17 (n=1); Case-series≤20 (n=4); 35 publications) Pre-clinical studies (n=2); 36 Wrong study design (e.g. 37 38 cross-sectional; narrative review, commentary) (n=32); 39 40 Duplicate (n=1) 41 Studies included in on October 1, 2021 by guest. Protected copyright.

42 Included quantitative synthesis 43 (meta-analysis) 44 (n=15) 45 46 47 48 49 50 Figure 1: Study selection process in review of opioid-sparing effects of cannabis in chronic pain 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 38 of 85

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Supplementary Material 4 5 Opioid-sparing effects of cannabis for chronic pain: A systematic review and meta-analysis 6 7 of randomized and observational studies 8 9 10 11 12 13 14 Contents 15 16 Appendix A: LiteratureFor Search Strategiespeer ...... review...... only ...... 3 17 Supplement Table 1: Detailed guidance for risk of bias assessment RCTs ...... 9 18 19 Supplement Table2: Detailed guidance for risk of bias assessment retrospective or prospective chart- 20 reviews with control group ...... 12 21 22 Supplement Table 3: Detailed guidance for risk of bias assessment retrospective or prospective chart- 23 reviews with no control group ...... 16 24 25 Supplement Table 4: Characteristics of Eligible studies ...... 18 26 27 Supplement Table 5: Risk of bias assessment for RCTs ...... 27 28 Supplement Table 6: Risk of bias assessments for chart reviews with control group ...... 28 29 30 Supplement Table 7: Risk of bias assessments for one-arm studies with no control group ...... 29 31 32 Supplement Table 8: GRADE evidence profile of cannabis adjuvant to opioids vs. opioid alone for 33 physical function among patients with chronic pain from 1 RCT ...... 30 http://bmjopen.bmj.com/ 34 35 Supplement Table 9: GRADE evidence profile of cannabis adjuvant to opioids vs. opioid alone for 36 emotional function among patients with chronic pain from 1 RCT...... 30 37 38 Supplement Table 10: Summary of adverse events among included observational studies ...... 31 39 Supplement Figure 1: forest plot for oral morphine equivalence dose reduction among patients with 40 41 Chronic Pain who received cannabis adjuvant to opioids vs. opioid alone in RCTs ...... 33 on October 1, 2021 by guest. Protected copyright. 42 Supplement Figure 2: Subgroup analysis for opioid dose reduction and risk of bias (high risk vs. low risk) 43 44 from 4 RCTs of Cannabis+opioids vs. placebo ...... 34 45 Supplement Figure 3: forest plot for oral morphine equivalence dose reduction among patients with 46 47 Chronic Pain who received cannabis adjuvant to opioids vs. opioid alone in observational studies ...... 35 48 Supplement Figure 4: forest plot for pain relief on a 10-cm Visual Analog Scale (VAS) among patients 49 50 with Chronic Pain who received cannabis adjuvant to opioids vs. opioid alone in RCTs ...... 36 51 Supplement Figure 5: Subgroup analysis for pain relief on a 10-cm VAS and risk of bias (high risk vs. low 52 53 risk) from 5 RCTs of Cannabis+opioids vs. placebo ...... 37 54 Supplement Figure 6: forest plot for pain relief on a 10-cm VAS among patients with Chronic Pain who 55 56 received cannabis adjuvant to opioids vs. opioid alone in observational studies with no control group . 38 57 58 1 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 39 of 85 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Supplement Figure 7: forest plot for sleep disturbance on a 10 cm VAS for sleep disturbance among 4 5 patients with Chronic Pain who received cannabis adjuvant to opioids vs. opioid alone in RCTs ...... 39 6 Supplement Figure 8: Subgroup analysis for sleep disturbance a 10-cm VAS for sleep disturbance and risk 7 8 of bias (high risk vs. low risk) from 5 RCTs of Cannabis+opioids vs. placebo ...... 40 9 Supplement Figure 9: Risk difference of nausea among patients with Chronic Pain who received cannabis 10 11 adjuvant to opioids vs. opioid alone in RCTs ...... 41 12 Supplement Figure 10: Relative Risk of nausea among patients with Chronic Pain who received cannabis 13 adjuvant to opioids vs. opioid alone in RCTs ...... 41 14 15 Supplement Figure 11: Relative Risk of vomiting among patients with Chronic Pain who received 16 cannabis adjuvant to Foropioids vs. peer opioid alone inreview RCTs ...... only...... 42 17 18 Supplement Figure 12: Risk Difference of vomiting among patients with Chronic Pain who received 19 cannabis adjuvant to opioids vs. opioid alone in RCTs ...... 42 20 21 Supplement Figure 13: Relative Risk of constipation among patients with Chronic Pain who received 22 cannabis adjuvant to opioids vs. opioid alone in RCTs ...... 43 23 24 Supplement Figure 14: Risk difference of constipation among patients with Chronic Pain who received 25 cannabis adjuvant to opioids vs. opioid alone in RCTs ...... 43 26 27 Appendix B: Reference List of Eligible studies ...... 44 28 29 Technical Appendix ...... 46 30 31 32 33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 2 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 40 of 85

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Appendix A: Literature Search Strategies 4 5 Database: OVID Medline Epub Ahead of Print, In-Process & Other Non-Indexed 6 Citations, Ovid MEDLINE(R) Daily and Ovid MEDLINE(R) 1946 to Present. 7 8 The search terminology included all types of chronic pain AND any kinds of cannabinoids: 9 ………………………………………………………………………………………………. 10 11 12 Database: OVID Medline Epub Ahead of Print, In-Process & Other Non-Indexed Citations, 13 14 Ovid MEDLINE(R) Daily and Ovid MEDLINE(R) 1946 to Present 15 Search Strategy: 16 ------For peer review only 17 1 exp , Opioid/ (111496) 18 19 2 opioid*.mp. (112576) 20 3 ( or alphaprodine or beta-casomorphin$ or buprenorphine or 21 or or deltorphin or dextromethorphan or or 22 or or enkephalin$ or ethylketocyclazocine or 23 or etorphine or fentanyl or or hydrocodone or or 24 or or or meperidine or or methadone or 25 methadyl acetate or morphine or or or oxycodone or or 26 or or or pirinitramide or promedol or 27 28 propoxyphene or or or or ).mp. [mp=title, 29 abstract, original title, name of substance word, subject heading word, floating 30 sub-heading word, keyword heading word, organism supplementary concept word, protocol 31 supplementary concept word, rare disease supplementary concept word, unique 32 identifier, synonyms] (150565) 33 4 or/1-3 (207118) http://bmjopen.bmj.com/ 34 5 exp Narcotics/ (119511) 35 6 (adolonta or Anpec or Ardinex or Asimadoline or Alvimopam or amadol or 36 37 biodalgic or biokanol or Codinovo or contramal or Demerol or Dicodid or 38 Dihydrocodeinone or dihydromorphinone or dihydrohydroxycodeinone or dihydrone or 39 dilaudid or dinarkon or dolsin or dolosal or dolin or dolantin or dolargan or 40 dolcontral or duramorph or duromorph or duragesic or durogesic or eucodal or 41 Fedotzine or Fentanest or Fentora or Fortral or Hycodan or Hycon or Hydrocodone or on October 1, 2021 by guest. Protected copyright. 42 Hydrocodeinonebitartrate or hydromorphon or hydroxycodeinon or isocodeine or 43 isonipecain or jutadol or laudacon or l dromoran or levodroman or levorphan or levo- 44 dromoran or levodromoran or lexir or lidol or lydol or morfin or morfine or morphia 45 or morphin or morphinium or morphinene or morphium or ms contin or n-methylmorphine 46 47 or n methylmorphine or nobligan or numorphan or oramorph or oxycodeinon or oxiconum 48 or oxycone or oxycontin or palladone or pancodine or or phentanyl or 49 prontofort or robidone or skenan or sublimaze or sulfentanyl or sulfentanil or 50 sufenta or takadol or talwin or theocodin or or tramadolhameln or tramadolor 51 or tramadura or tramagetic or tramagit or tramake or tramal or tramex or tramundin or 52 trasedal or theradol or tiral or topalgic or tradol or tradolpuren or tradonal or 53 tralgiol or tramadorsch or tramadin or tramadoc or ultram or zamudol or zumalgic or 54 zydol or zytram).mp. [mp=title, abstract, original title, name of substance word, 55 56 subject heading word, floating sub-heading word, keyword heading word, organism 57 58 3 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 41 of 85 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 supplementary concept word, protocol supplementary concept word, rare disease 4 supplementary concept word, unique identifier, synonyms] (10373) 5 7 or/1-6 (213683) 6 7 Annotation: opioid block 8 8 (Cannabis or or cannabinoid* or or bhang or cannador or 9 charas or ganja or ganjah or hashish or hemp or marihuana or marijuana or nabilone or 10 cesamet or cesametic or ajulemic acid or cannabichromene or cannabielsoin or 11 or tetrahydrocannabinol or dronabinol or or nabiximols or 12 palmidrol or tetrahydrocannabinolic acid or tetrahydro cannabinol or marinol or 13 tetranabinex or sativex or endocannabinoid*).mp. [mp=title, abstract, original title, 14 name of substance word, subject heading word, floating sub-heading word, keyword 15 16 heading word, organismFor supplementary peer concept review word, protocol supplementaryonly concept 17 word, rare disease supplementary concept word, unique identifier, synonyms] (52087) 18 9 Cannabis/ (8573) 19 10 exp CANNABINOIDS/ (13258) 20 11 8 or 9 or 10 (52087) 21 Annotation: cannabis block 22 12 7 and 11 (6089) 23 Annotation: opioid and cannabis 24 25 13 (chronic adj4 pain*).mp. [mp=title, abstract, original title, name of 26 substance word, subject heading word, floating sub-heading word, keyword heading 27 word, organism supplementary concept word, protocol supplementary concept word, rare 28 disease supplementary concept word, unique identifier, synonyms] (65717) 29 14 Chronic Pain/ (12620) 30 15 exp Osteoarthritis/ (59676) 31 16 osteoarthrit*.mp. (84419) 32 17 osteo-arthritis.mp. (375) 33 http://bmjopen.bmj.com/ 18 exp Arthritis, Rheumatoid/ (109607) 34 35 19 exp Neuralgia/ (19415) 36 20 Diabetic Neuropathies/ (14247) 37 21 (neuropath* adj5 pain*).mp. [mp=title, abstract, original title, name of 38 substance word, subject heading word, floating sub-heading word, keyword heading 39 word, organism supplementary concept word, protocol supplementary concept word, rare 40 disease supplementary concept word, unique identifier, synonyms] (23043) 41 22 neuralg*.mp. (26154) on October 1, 2021 by guest. Protected copyright. 42 23 zoster.mp. (20386) 43 44 24 Irritable Bowel Syndrome/ (6748) 45 25 IBS.mp. (8435) 46 26 Migraine Disorders/ (24388) 47 27 migraine.mp. (37040) 48 28 Fibromyalgia/ (8088) 49 29 fibromyalg*.mp. (11178) 50 30 complex regional pain syndromes/ or exp causalgia/ or exp reflex sympathetic 51 dystrophy/ (5426) 52 31 Pain, Intractable/ (6126) 53 54 32 Phantom Limb/ (1816) 55 33 Hyperalgesia/ (11136) 56 57 58 4 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 42 of 85

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 34 exp back pain/ or exp failed back surgery syndrome/ or exp low back pain/ 4 (37369) 5 35 radiculopathy.mp. (8722) 6 7 36 musculoskeletal pain/ or headache/ (29687) 8 37 exp Headache Disorders/ (33178) 9 38 headache*.mp. (89612) 10 39 exp Temporomandibular Joint Disorders/ (16711) 11 40 whiplash.mp. or exp whiplash injury/ (3896) 12 41 exp Cumulative Trauma Disorders/ (13326) 13 42 exp Peripheral Nervous System Diseases/dt [Drug Therapy] (14079) 14 43 Pain Measurement/de [Drug Effects] (6594) 15 16 44 (backache* or backpain*For orpeer dorsalgi* or arthralgi*review or polyarthralgi* only or 17 arthrodyni* or myalgi* or fibromyalgi* or myodyni* or neuralgi* or ischialgi* or crps 18 or rachialgi*).ab,ti. (43072) 19 45 ((noncancer* or non-cancer* or back or discogen* or chronic* or recurrent or 20 persist* or bone or musculoskelet* or muscle* or skelet* or spinal or spine or 21 vertebra* or joint* or arthritis or Intestin* or neuropath* or neck or cervical* or 22 head or facial* or complex or radicular or cervicobrachi* or orofacial or somatic or 23 non-malign* or shoulder* or knee* or hip or hips) adj3 pain).mp. (206944) 24 25 46 exp Pain/ (379991) 26 47 pain*.mp. (745044) 27 48 or/13-47 (1122771) 28 49 12 and 48 (1034) 29 30 31 Database: Embase <1974 to 2019 September 04> 32 Search Strategy: 33 http://bmjopen.bmj.com/ ------34 35 1 exp narcotic analgesic agent/ (317763) 36 2 (opioid* or opiate*).mp. [mp=title, abstract, heading word, drug trade name, 37 original title, device manufacturer, drug manufacturer, device trade name, keyword, 38 floating subheading word, candidate term word] (188237) 39 3 (alfentanil or alphaprodine or beta-casomorphin$ or buprenorphine or 40 carfentanil or codeine or deltorphin or dextromethorphan or dezocine or 41 dihydrocodeine or dihydromorphine or enkephalin$ or ethylketocyclazocine or on October 1, 2021 by guest. Protected copyright. 42 ethylmorphine or etorphine or fentanyl or heroin or hydrocodone or hydromorphone or 43 44 ketobemidone or levorphanol or lofentanil or meperidine or meptazinol or methadone or 45 methadyl acetate or morphine or nalbuphine or opium or oxycodone or oxymorphone or 46 pentazocine or phenazocine or phenoperidine or pirinitramide or promedol or 47 propoxyphene or remifentanil or sufentanil or tilidine or tapentadol).mp. (278150) 48 4 (adolonta or Anpec or Ardinex or Asimadoline or Alvimopam or amadol or 49 biodalgic or biokanol or Codinovo or contramal or Demerol or Dicodid or 50 Dihydrocodeinone or dihydromorphinone or dihydrohydroxycodeinone or dihydrone or 51 dilaudid or dinarkon or dolsin or dolosal or dolin or dolantin or dolargan or 52 dolcontral or duramorph or duromorph or duragesic or durogesic or eucodal or 53 54 Fedotzine or Fentanest or Fentora or Fortral or Hycodan or Hycon or Hydrocodone or 55 Hydrocodeinonebitartrate or hydromorphon or hydroxycodeinon or isocodeine or 56 isonipecain or jutadol or laudacon or l dromoran or levodroman or levorphan or levo- 57 58 5 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 43 of 85 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 dromoran or levodromoran or lexir or lidol or lydol or morfin or morfine or morphia 4 or morphin or morphinium or morphinene or morphium or ms contin or n-methylmorphine 5 or n methylmorphine or nobligan or numorphan or oramorph or oxycodeinon or oxiconum 6 7 or oxycone or oxycontin or palladone or pancodine or pethidine or phentanyl or 8 prontofort or robidone or skenan or sublimaze or sulfentanyl or sulfentanil or 9 sufenta or takadol or talwin or theocodin or tramadol or tramadolhameln or tramadolor 10 or tramadura or tramagetic or tramagit or tramake or tramal or tramex or tramundin or 11 trasedal or theradol or tiral or topalgic or tradol or tradolpuren or tradonal or 12 tralgiol or tramadorsch or tramadin or tramadoc or ultram or zamudol or zumalgic or 13 zydol or zytram).mp. [mp=title, abstract, heading word, drug trade name, original 14 title, device manufacturer, drug manufacturer, device trade name, keyword, floating 15 16 subheading word, candidateFor term peer word] (50642) review only 17 5 or/1-4 (403926) 18 6 exp cannabis/ (32390) 19 7 cannabinoid/ or cannabidiol/ or cannabinoid derivative/ or cannabinol/ or 20 cannabinol derivative/ or cannabis derivative/ or delta8 tetrahydrocannabinol/ or 21 delta8 tetrahydrocannabinol derivative/ or "delta9(11) tetrahydrocannabinol"/ or 22 dronabinol/ or medical cannabis/ or nabiximols/ or tetrahydrocannabinol/ or 23 tetrahydrocannabinol derivative/ or tetrahydrocannabinolic acid/ (26180) 24 25 8 (Cannabis or cannabinol or cannabidiol or bhang or cannador or charas or ganja 26 or ganjah or hashish or hemp or marihuana or marijuana or nabilone or cesamet or 27 cesametic or ajulemic acid or cannabichromene or cannabielsoin or cannabigerol or 28 tetrahydrocannabinol or dronabinol or levonantradol or nabiximols or palmidrol or 29 tetrahydrocannabinolic acid or tetrahydro cannabinol or marinol or tetranabinex or 30 sativex or endocannabinoid*).mp. [mp=title, abstract, heading word, drug trade name, 31 original title, device manufacturer, drug manufacturer, device trade name, keyword, 32 floating subheading word, candidate term word] (69860) 33 http://bmjopen.bmj.com/ 9 6 or 7 or 8 (75281) 34 35 10 5 and 9 (16412) 36 11 (chronic adj4 pain*).mp. [mp=title, abstract, heading word, drug trade name, 37 original title, device manufacturer, drug manufacturer, device trade name, keyword, 38 floating subheading word, candidate term word] (109897) 39 12 chronic pain/ (57642) 40 13 exp osteoarthritis/ (122475) 41 14 osteoarthrit*.mp. (136019) on October 1, 2021 by guest. Protected copyright. 42 15 osteo-arthritis.mp. (424) 43 44 16 degenerative arthrit*.mp. (1563) 45 17 exp rheumatoid arthritis/ (194747) 46 18 exp neuralgia/ (99958) 47 19 / (22699) 48 20 (neuropath* adj5 (pain* or diabet*)).mp. (71799) 49 21 neuralg*.mp. (29200) 50 22 zoster.mp. (36684) 51 23 irritable colon/ (24792) 52 24 (Irritable Bowel Syndrome or IBS).mp. [mp=title, abstract, heading word, drug 53 54 trade name, original title, device manufacturer, drug manufacturer, device trade 55 name, keyword, floating subheading word, candidate term word] (24025) 56 25 exp migraine/ (60235) 57 58 6 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 44 of 85

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 26 migraine.mp. (66593) 4 27 fibromyalgia/ (19402) 5 28 fibromyalg*.mp. (20958) 6 7 29 reflex sympathetic dystrophy.mp. (2356) 8 30 (complex regional pain syndromes or causalgia).mp. (1275) 9 31 intractable pain/ (4701) 10 32 phantom limb.mp. or agnosia/ or phantom pain/ or amputation stump/ (7388) 11 33 hyperalgesia/ (18711) 12 34 ((noncancer* or non-cancer*or chronic* or recurrent or persist* or non- 13 malign*) adj3 pain).mp. (27031) 14 35 exp backache/ (104042) 15 16 36 radiculopathy.mp.For or exp radiculopathy/peer review(37176) only 17 37 musculoskeletal pain/ (10292) 18 38 exp arthralgia/ (58208) 19 39 headache/ (204055) 20 40 headache*.mp. (264831) 21 41 temporomandibular joint disorder/ (13308) 22 42 ((TMJ or TMJD) and pain*).mp. (3648) 23 43 whiplash.mp. or whiplash injury/ (4815) 24 25 44 exp cumulative trauma disorder/ (20089) 26 45 exp pain/ (1249315) 27 46 pain*.mp. (1280762) 28 47 or/11-46 (1963522) 29 48 10 and 47 (3115) 30 31 Search Name: cannabis pain 32 Date Run: 05/09/2019 16:12:03 33 http://bmjopen.bmj.com/ Comment: 34 35 36 ID Search Hits 37 #1 MeSH descriptor: [Cannabis] explode all trees 293 38 #2 MeSH descriptor: [Cannabinoids] explode all trees 743 39 #3 MeSH descriptor: [Endocannabinoids] explode all trees 46 40 #4 MeSH descriptor: [Endocannabinoids] explode all trees 46 41 #5 (Cannabis or cannabinol or cannabinoid* or cannabidiol or bhang or cannador or charas or ganja on October 1, 2021 by guest. Protected copyright. 42 or ganjah or hashish or hemp or marihuana or marijuana or nabilone or cesamet or cesametic or 43 44 ajulemic acid or cannabichromene or cannabielsoin or cannabigerol or tetrahydrocannabinol or 45 dronabinol or levonantradol or nabiximols or palmidrol or tetrahydrocannabinolic acid or tetrahydro 46 cannabinol or marinol or tetranabinex or sativex or endocannabinoid*):ti,ab,kw (Word variations have 47 been searched) 4215 48 #6 #1 or #2 or #3 or #4 or #5 4215 49 #7 MeSH descriptor: [Pain] explode all trees 45094 50 #8 (pain*):ti,ab,kw (Word variations have been searched) 164064 51 #9 #7 or #8 169846 52 #10 #6 and #9 578 53 54 #11 [mh Osteoarthritis] or [mh ^"Arthritis, Rheumatoid"] or [mh Neuralgia] or [mh ^"Diabetic 55 Neuropathies"] or [mh ^"Irritable Bowel Syndrome"] or [mh ^"Migraine Disorders"] or [mh 56 Fibromyalgia] or [mh ^"complex regional pain syndromes"] or [mh causalgia] or [mh ^"reflex 57 58 7 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 45 of 85 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 sympathetic dystrophy"] or [mh ^"pain Intractable"] or [mh ^"Phantom Limb"] or [mh Hyperalgesia] or 4 [mh ^"back pain"] or [mh ^"failed back surgery syndrome"] or [mh ^"low back pain"] or [mh 5 Radiculopathy] or [mh ^"musculoskeletal pain"] or [mh headache] or [mh Arthralgia] or [mh ^"Headache 6 7 Disorders"] or [mh ^"Temporomandibular Joint Dysfunction Syndrome"] or [mh ^"whiplash injury"] or 8 [mh ^"Cumulative Trauma Disorders"] or [mh "Peripheral Nervous System Diseases"/DT] or [mh ^"Pain 9 Measurement"/DE] 28499 10 #12 (osteoarthrit* or osteo-arthritis or arthrit* or neuropath* or neuralgi* or zoster* or migraine* 11 or headache* or fibromyalgi* or causalgia or radiculopathy* or whiplash or backache* or backpain* or 12 dorsalgi* or arthralgi* or polyarthralgi* or arthrodyni* or myalgi* or myodyni* or ischialgi* or crps or 13 rachialgi*or TMJ or TMJD or IBS or crohn* or colitis* or enteritis* or ileitis*) 104465 14 #13 (irrita* or inflam*) near/4 (bowel or colon) 7249 15 16 #14 #11 or #12 orFor #13 peer113256 review only 17 #15 #6 and #14 in Trials 353 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 October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 8 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 46 of 85

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Characteristics of eligible studies and Risk of Bias Assessment 4 5 6 7 Supplement Table 1: Detailed guidance for risk of bias assessment RCTs 8 Domain Judgment 9 10 Random allocation concealment Definitely yes (low risk): used 11 12 central allocations (e.g. computer, 13 telephone) 14 15 Probably yes (low risk): 16 For peer reviewsequentially only numbered, opaque, 17 18 sealed envelopes; studies did not 19 provide enough information about 20 concealment approach; however, it 21 22 was placebo-control trial with 23 double blinded design. 24 25 Probably no (high risk): not 26 enough information was provided 27 28 and study was not blinded. 29 30 Definitely no (high risk): used any 31 unconcealed approach of allocation 32 33 (e.g. case record number, day of http://bmjopen.bmj.com/ 34 week, health-care decision). 35 36 Blinding of patients Definitely yes (low risk): explicitly 37 mentioned that patients were blinded 38 39 40 Probably yes (low risk): a placebo- 41 controlled double-blinded trial. on October 1, 2021 by guest. Protected copyright. 42 43 Probably no (high risk): no explicit 44 statement about blinding status and 45 46 not double-blinded placebo- 47 controlled trial. 48 49 Definitely no (high risk): explicitly 50 51 mentioned that patients were not 52 blinded. 53 54 55 56 57 58 9 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 47 of 85 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Blinding of health care providers Definitely yes (low risk): explicitly 4 5 mentioned that this group was 6 blinded. 7 8 Probably yes (low risk): mentioned 9 that it was a double-blinded study; 10 11 mentioned investigator were blinded. 12 13 Probably no (high risk) 14 15 Definitely no (high risk): explicitly 16 For peer reviewmentioned only that this group was not 17 blinded. 18 19 Blinding of data collector Definitely yes (low risk): explicitly 20 mentioned that this group was 21 blinded. 22 23 Probably yes (low risk): mentioned 24 25 that it was a double-blinded study; 26 mentioned investigator were blinded. 27 28 Probably no (high risk) 29 30 Definitely no (high risk): explicitly 31 mentioned that this group was not 32 33 blinded. http://bmjopen.bmj.com/ 34 Blinding of outcome assessor Definitely yes (low risk): explicitly 35 mentioned that this group was 36 blinded. 37 38 39 Probably yes (low risk): mentioned 40 that it was a double-blinded study. 41 on October 1, 2021 by guest. Protected copyright. 42 Probably no (high risk) 43 44 Definitely no (high risk): explicitly 45 46 mentioned that this group was not 47 blinded; open-blinded or unblended 48 trial. 49 50 Blinding data analyst Definitely yes (low risk): explicitly 51 52 mentioned that this group were 53 blinded 54 55 56 57 58 10 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 48 of 85

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Probably yes (low risk): 4 5 Probably no (high risk): no explicit 6 7 statement about blinding and only 8 mentioned double-blinded. 9 10 Definitely no (high risk): explicitly 11 12 mentioned that this group was not 13 blinded; open-blinded or unblended 14 trial. 15 16 Loss to follow-up For peer reviewDefinitely only yes: the retention rate was 17 18 at least 90% through the study. 19 20 Probably yes (low risk): the 21 retention rate approximately 80-89% 22 23 and loss to follow-up unlikely to be 24 related to the outcome, or missing 25 outcome data were balanced across 26 groups. 27 28 29 Probably no (high risk): the 30 retention rate approximately 80- 31 89%, however its rate likely to be 32 33 related to the loss to follow-up. http://bmjopen.bmj.com/ 34 35 Definitely no (high risk): the 36 retention rate was less than 80%. 37 Sample size We also considered the sample size 38 lower than 250 per arm as high risk 39 of bias and rated down on the basis 40 41 of imprecision in GRADE on October 1, 2021 by guest. Protected copyright. 42 assessment. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 11 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 49 of 85 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Supplement Table2: Detailed guidance for risk of bias assessment retrospective or 4 5 prospective chart-reviews with control group 6 Domain Judgment 7 8 1) Did the study match participants for all variables Definitely yes (low risk): 9 that are associated with the outcome of interest 10 studies that adjusted based on 11 or did the statistical analysis adjust for these all important covariates prognostic variables? (This item queries how 12 including age, sex, baseline 13 confident we are that the reported association or pain, baseline opioid dose, and 14 lack thereof is not due to confounding). 15 other disabilities. 16 For peer review only 17 Probably yes (low risk): 18 studies that adjusted at a 19 20 minimum for baseline pain and 21 baseline opioid dose. 22 23 Probably no (high risk): 24 studies that did not provide any 25 26 details about analysis method. 27 28 Definitely no (high risk): 29 30 Studies that did not adjust based 31 on baseline opioid dose or 32 33 baseline pain. http://bmjopen.bmj.com/ 34 35 2) Was selection of exposed and non-exposed Definitely yes (low risk): 36 cohorts drawn from the same population? (this Studies in which selection for 37 item queries whether participants who co-used participation is not dependent 38 cannabis and opioids or used opioids alone were 39 on exposure status (cannabis drawn from the same population) 40 and opioid co-use). 41 on October 1, 2021 by guest. Protected copyright. 42 Probably yes (low risk): 43 44 studies that did not provide 45 enough information about 46 recruitment to judge whether 47 48 recruitment into the study was 49 dependent on exposure status or 50 not. 51 52 Probably no (high risk): NA 53 54 55 56 57 58 12 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 50 of 85

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Definitely no (high risk): 4 5 studies that compared cannabis 6 and opioid co-users and non- 7 users from different cohort. 8 9 3) Can we be confident in the assessment of Definitely yes (low risk): if 10 11 exposure? (this item queries how confident we are study reported some 12 about the quantification of cannabis and opioids co- ascertainment methods for 13 use). cannabis use (e.g. urine 14 15 analysis), or study prescribed 16 For peer review onlythe specific dose of medical 17 cannabis to the participants. 18 19 20 Probably yes (low risk): self- 21 report of cannabis use. 22 23 Probably no (high risk): when 24 study did not provide any details 25 26 about assessing exposure status. 27 28 Definitely no (high risk): 29 participants self-reported 30 cannabis usage only at baseline, 31 32 or exposure status not assessed 33 during the 4-weeks follow-up at http://bmjopen.bmj.com/ 34 least one time, or level of 35 36 cannabis usage was not similar 37 among participants. For 38 example, some studies allowed 39 40 patients to select the type or 41 dose of cannabis themselves. on October 1, 2021 by guest. Protected copyright. 42 43 4) Can we be confident in the assessment of the Definitely yes (low risk): when 44 presence or absence of prognostic factors? patients self-reported the 45 46 prognostic factors. 47 48 Probably yes (low risk): when 49 the method of assessment was 50 not reported, it was considered 51 as probably yes. 52 53 54 *Note that for this item, we are 55 only concerned with the 56 measurement of the prognostic 57 58 13 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 51 of 85 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 factors that mentioned in item 4 number 1 as minimum adjusted 5 6 variables (baseline pain 7 intensity and opioid dose). 8 5) Were co-interventions similar between groups? Definitely yes (low risk): study 9 (this item queries how similar are the use of other reported that co-intervention 10 pain killers (e.g. NSAIDs) between cannabis users other than study intervention 11 and non-users. 12 were limited during the study 13 period. 14 15 Probably yes (low risk): when 16 For peer review only 17 co-intervention usage was 18 approximately balanced 19 between both intervention and 20 21 control groups. 22 23 Probably no (high risk): when 24 study did not provide enough 25 information about other drugs 26 27 that participants may use. 28 29 Definitely no (high risk): when 30 participants were allowed to use 31 32 all other co-interventions that 33 could affect the outcome of the http://bmjopen.bmj.com/ 34 study. 35 36 6) Was the follow up of cohorts adequate? (This Definitely yes (low risk): the 37 38 item queries the risk of bias associated with loss to retention rate was at least 90% 39 follow-up and missing outcome data). through the study. 40 41 Probably yes (low risk): the on October 1, 2021 by guest. Protected copyright. 42 43 retention rate approximately 80- 44 89% and loss to follow-up 45 unlikely to be related to the 46 47 outcome. 48 49 Probably no (high risk): the 50 retention rate approximately 80- 51 89%, however its rate likely to 52 53 be related to the loss to follow- 54 up. For instance, if patients were 55 required to come to clinic for 56 57 58 14 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 52 of 85

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 outcome measurement, patients 4 5 who had poorer outcomes, or on 6 the other hand, patients who 7 were feeling better, may be less 8 9 likely to attend the clinic. 10 11 Loss to follow-up did not report 12 or could not estimate. 13 14 Definitely no (high risk): loss 15 to follow-up more than 20%. 16 For peer review only 17 18 7) Can we be confident in the assessment of Definitely yes (low risk): study 19 outcome? (This item queries our confidence in the used a validated/reliable 20 accuracy of the measurement of the outcome). measurement for pain 21 22 assessment (e.g. VAS, NRS); 23 reported opioid dose in a 24 morphine equivalence dose by 25 26 assessing patients’ medical or 27 prescription records. 28 29 Probably yes (low risk): NA 30 31 Probably no (high risk): when 32 33 study did not provide enough http://bmjopen.bmj.com/ 34 information about the outcome 35 measurement. 36 37 Definitely no (high risk): study 38 39 used non-validated/reliable 40 instrument. 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 15 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 53 of 85 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Supplement Table 3: Detailed guidance for risk of bias assessment retrospective or 4 5 prospective chart-reviews with no control group 6 Domain Judgment 7 8 Is the source population (sampling frame) Definitely yes (low risk): 9 10 representative of the general population? participants were selected from a 11 representative sample (e.g. national 12 population registry) 13 14 Probably yes (low risk): single 15 16 For peer reviewcommunity only center, however the 17 center was the only referral center 18 that provided cannabis legally to 19 20 participants. 21 22 Probably no (high risk): based on 23 the provided information source 24 population could not be defined. 25 26 27 Definitely no (high risk): sampling 28 from one center or clinic or hospital 29 or patients selected through using 30 31 convenience sampling. 32 33 Is the assessment of the outcome accurate both at Definitely yes (low risk): study used http://bmjopen.bmj.com/ 34 baseline and at follow-up? a validated/reliable measurement for 35 pain assessment (e.g. VAS, NRS); 36 37 reported opioid dose in a morphine 38 equivalence dose by assessing 39 patients’ medical or prescription 40 41 records. on October 1, 2021 by guest. Protected copyright. 42 43 Probably yes (low risk): NA 44 45 Probably no (high risk): when 46 study did not provide enough 47 48 information about the outcome 49 measurement. 50 51 Definitely no (high risk): used of 52 different instruments at different 53 54 follow-up intervals with concern of 55 56 57 58 16 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 54 of 85

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 accuracy of responses, or used 4 5 invalidated/reliable instruments. 6 7 Is there little missing data? Definitely yes (low risk): the 8 retention rate was at least 90% 9 through the study. 10 11 12 Probably yes (low risk): the 13 retention rate approximately 80-89% 14 and loss to follow-up unlikely to be 15 16 For peer reviewrelated only to the outcome. 17 18 Probably no (high risk): the 19 retention rate approximately 80- 20 89%, however its rate likely to be 21 22 related to the loss to follow-up. For 23 instance, if patients were required to 24 come to clinic for outcome 25 26 measurement, patients who had 27 poorer outcomes, or on the other 28 hand, patients who were feeling 29 30 better, may be less likely to attend 31 the clinic. 32 33 Loss to follow-up did not report or http://bmjopen.bmj.com/ 34 could not estimate. 35 36 37 Definitely no (high risk): loss to 38 follow-up more than 20%. 39 40 41 on October 1, 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 Page 55 of 85 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Supplement Table 4: Characteristics of Eligible studies 4 5 6 1 7 Barlowe et al-2019 8 9 Study design Retrospective chart review. 10 11 Participants 34 chronic painful pancreatitis patients with chronic use of 12 opioids enrolled in a state therapeutic cannabis program 13 14 were compared to 19 non-enrolled patients. 15 16 Intervention (comparison)For peermedical review cannabis added toonly opioids (no cannabis). 17 18 Follow-up Cohort of patients who enrolled into the program had 19 20 received cannabis therapy with a range from 34 to 297 21 weeks. 22 23 Funding source No industry funding reported. 24 Outcome -Reduction of opioid (calculated in average daily 25 26 intravenous [IV] morphine equivalence dosages) 27 28 29 30 Bellnier et al-20182 31 32 Study design One-arm observational study (before/after). 33 Participants 29 patients with chronic pain who used opioids enrolled in a http://bmjopen.bmj.com/ 34 state therapeutic cannabis program. 35 36 Intervention (comparison) medical cannabis added to opioids (no cannabis). 37 Follow-up 3 months 38 Funding source Not reported. 39 Outcome -Reduction of opioid (calculated in average daily 40 intravenous [IV] morphine equivalence dosages) on October 1, 2021 by guest. Protected copyright. 41 -Pain Quality Assessment Scale (PQAS) paroxysmal 42 43 domain 44 45 46 Capano et al-20203 47 48 Study design One-arm observational study (before/after). 49 Participants 131 patients with chronic pain who used opioids enrolled in 50 51 a pain clinic cannabis therapy. 52 Intervention (comparison) medical cannabis added to opioids (no cannabis). 53 Follow-up 8 weeks 54 Funding source Industry fund reported. 55 Outcome - Pain disability index 56 57 58 18 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 56 of 85

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 - Pittsburgh Sleep Quality Index 4 - Pain intensity and interference index 5 6 7 8 Haroutounian et al-2016 4 9 10 Study design One-arm observational study (before/after). 11 12 Participants Chronic non-cancer pain (14 individuals had pain due to 13 14 cancer) with a duration of 3 months or longer, and a lack of 15 satisfactory analgesic response or intolerable adverse effects 16 For peerwith at reviewleast 2 analgesics fromonly 2 different drug classes at full 17 dose (Opioid user: N=73; 35%). 18 19 20 Intervention (comparison) The initial recommended medical cannabis dose was 20 21 g/mo added to opioids, which could be obtained as smoked 22 cannabis, baked cookies or oil taking from cannabis 23 dispensary centers. Cannabis could be titrated up to 3 times 24 a day until satisfactory pain relief was gained (before using 25 cannabis). 26 27 Follow-up 6 months. 28 29 Funding source No industry funding reported. 30 Outcome - Reduction of opioid (calculated in median daily 31 intravenous [IV] morphine equivalence dosages among 32 33 opioid users). http://bmjopen.bmj.com/ 34 35 36 37 Maida et al-20085 38 39 Study design Prospective cohort study. 40 on October 1, 2021 by guest. Protected copyright. 41 Participants 47 patients with chronic cancer pain who were opioid user 42 43 and treated with nabilone were compared to 65 non-treated 44 patients. 45 46 Intervention (comparison) nabilone added to opioids (no nabilone). 47 48 Follow-up 30 days. 49 50 51 Funding source Industry funding reported. 52 Outcome -Reduction of opioid (calculated in average daily morphine 53 equivalence dosages); 54 55 56 57 58 19 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 57 of 85 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 -Pain reduction (Edmonton Symptom Assessment System 0: 4 no pain-10: most severe pain); 5 6 -anxiety, nausea, depression. 7 8 9 10 Narange et al-20086 11 12 Study design Phase II: One-arm observational study (before/after). 13 14 Participants 30 patients with chronic non-cancer pain who were taking 15 16 For peeropioids review for a long time. only 17 18 Intervention (comparison) The starting dose was 5mg of dronabinol twice daily and 19 titrated up to 20 mg 3 times a day added to opioids (before 20 21 using dronabinol). 22 23 Follow-up 4 weeks 24 25 Funding source Industry funding reported. 26 Outcome -Pain reduction (VAS 0: no pain-10: most severe pain); 27 -patients’ satisfaction 28 -pain interfere with sleep (Brief pain inventory) 29 30 -social functioning 31 32 -sleep disturbance 33 http://bmjopen.bmj.com/ 34 -adverse events including anxiety, dizziness, and inability to 35 36 concentrate. 37 38 39 7 40 O’Connell et al-2019 41 on October 1, 2021 by guest. Protected copyright. 42 Study design One-arm observational study (before/after). 43 44 Participants 77 mixed type of chronic non-cancer pain patients who used 45 opioids (96%) or benzodiazepines. 46 47 48 Intervention (comparison) Medical cannabis including THC, CBD products added to 49 opioid (before using cannabis) 50 51 Follow-up 6 months 52 53 Funding source No industry funding reported. 54 55 Outcome - Reduction of opioid (calculated in mean daily morphine 56 equivalence dosages among opioid users). 57 58 20 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 58 of 85

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 -pain reduction (VAS 0: no pain-10: most severe pain). 4 5 6 8 7 Pritchard-2019 8 9 Study design Retrospective chart review. 10 11 Participants 22 patients who had chronic cancer-related pain and used 12 opioids with the presence of THC in their urine drug 13 14 screening were compared to 61 patients with opioid use 15 only. 16 For peer review only 17 Intervention (comparison) medical cannabis added to opioids (no cannabis). 18 19 Follow-up 26 weeks. 20 21 22 Funding source No industry funding reported. 23 Outcome -Reduction of opioid (calculated in average daily 24 intravenous [IV] morphine equivalence dosages) 25 26 27 28 9 29 Pawasarat-2020 30 31 Study design Retrospective chart review. 32 33 Participants 137 chronic cancer-related pain patients with chronic use of http://bmjopen.bmj.com/ 34 opioids enrolled in a State of New Jersey Medicinal 35 36 Marijuana Program Registry were compared to 95 non- 37 enrolled patients. 38 39 Intervention (comparison) medical cannabis added to opioids (no cannabis). 40 41 Follow-up Between 36 and 52 weeks for enrolled patients and 24 on October 1, 2021 by guest. Protected copyright. 42 43 weeks for non-enrolled patients. 44 45 Funding source No industry funding reported. 46 Outcome -Reduction of opioid (calculated in average daily 47 intravenous [IV] morphine equivalence dosages) 48 49 -Pain reduction. 50 51 52 53 54 55 56 57 58 21 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 59 of 85 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Rod-201910 4 5 Study design One-arm observational study (before/after). 6 7 8 Participants 600 of chronic pain patients who used opioids and indicated 9 they were prepared to reduce their opioid dose. 10 11 Intervention (comparison) Medical cannabis added to opioid (before using cannabis) 12 13 Follow-up 6 months 14 15 No industry funding reported. 16 Funding source For peer review only 17 Outcome - Reduction or cease of opioid use (reported as percentage of 18 patients who ceased or reduced their opioid use after 6 19 months). 20 21 22 Takakuwa et al-202011 23 24 25 Study design One-arm observational study (before/after). 26 27 Participants 61 of chronic non-cancer pain patients (low-back pain) who 28 used opioids. 29 30 Intervention (comparison) Medical cannabis added to opioid (before using cannabis) 31 32 33 Follow-up Median of 6.4 years among patients who ceased opioids http://bmjopen.bmj.com/ 34 completely 35 36 Funding source Industry funding reported. 37 Outcome - Reduction of opioid (calculated in median daily morphine 38 equivalence dosages among chronic and intermittent opioid 39 40 users). 41 on October 1, 2021 by guest. Protected copyright. 42 43 Vigil et al-2017 12 44 45 Study design Retrospective chart review. 46 47 Participants 37 habitual opioid using, severe CNCP patients enrolled in 48 49 the Medical Cannabis Program were compared to 29 non- 50 enrolled patients. 51 52 Intervention (comparison) Medical cannabis added to opioids (no cannabis). 53 54 Follow-up 1 year. 55 56 57 58 22 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 60 of 85

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Funding source No industry funding reported. 4 5 Outcome -Cessation of opioid (defined as the absence of opioid 6 prescriptions activity during the last three months of 7 observation) 8 9 -Reduction of opioid (calculated in average daily 10 11 intravenous [IV] morphine equivalence dosages); 12 13 -Pain reduction only among cannabis users; 14 15 -Quality of life (no effect; good benefit; great benefit; 16 For peernegative review effect; and extremely only negative effect of co- 17 18 prescription of cannabis on quality of life). 19 20 21 22 Yassin et al-201913 23 24 Study design One-arm observational study (before/after). 25 26 Participants 31 patients with fibromyalgia were treated for at least 12 27 28 months with 5 mg of oxycodone hydrochloride equivalent to 29 4.5 mg oxycodone and 2.5 mg naloxone hydrochloride twice 30 a day and duloxetine 30 mg once a day. 31 32

20 grams of smoked medical cannabis added to opioids http://bmjopen.bmj.com/ 33 Intervention (comparison) 34 (before cannabis inhalation). 35 36 Follow-up 6 months 37 38 Funding source No industry funding reported. 39 Outcome -Pain reduction 40 41 -Change in pain medication use in 5 categories: 1- increased on October 1, 2021 by guest. Protected copyright. 42 doses, 2- stable dose through medical cannabis therapy 43 duration, 3- less than half reduction in medication 44 consumption, 4- more than half reduction in analgesic 45 consumption, 5-deceased analgesic consumption. 46 - Owestry Disability Index reduction (scale 0: no disability, 47 100: total disability) 48 49 50 51 52 53 54 55 56 57 58 23 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 61 of 85 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Johnson et al-201014 4 5 Study design Parallel, multi-center randomized double-blinded, placebo- 6 7 controlled trial. 8 9 Participants 177 patients with chronic cancer pain who were under 10 treatment by opioid regimen. 11 12 13 Intervention (comparison) tetrahydrocannabinol: cannabidiol (THC:CBD) extract 14 added to opioids (placebo) 15 16 Follow-up For peer2 weeks review only 17 18 Funding source Industry funding reported. 19 Outcome - Reduction of opioid (calculated in mean daily morphine 20 21 equivalence dosages) 22 -Pain reduction (VAS 0: no pain-10: most severe pain) 23 -Sleep disturbance (NRS 0: no disturbance-10: most severe 24 disturbance) 25 -Physical, emotional, role, and social functioning (QLQ- 26 C30) 27 -Nausea, vomiting, constipation. 28 29 30 31 Portenoy et al-201215 32 33 Study design Parallel, randomized double-blinded, placebo-controlled http://bmjopen.bmj.com/ 34 trial. 35 36 37 Participants 360 patients with chronic cancer pain who were under 38 treatment by opioid regimen. 39 40 Intervention (comparison) Nabiximols at a low dose (1–4 sprays/day), medium dose 41 (6–10 sprays/day), or high dose (11–16 sprays/day) added to on October 1, 2021 by guest. Protected copyright. 42 43 opioids-(placebo) 44 45 Follow-up 5 weeks 46 47 Funding source Industry funding reported. 48 Outcome - Reduction of opioid (calculated in mean daily morphine 49 equivalence dosages) 50 51 -Pain reduction (VAS 0: no pain-10: most severe pain) 52 -Sleep disturbance (NRS 0: no disturbance-10: most severe 53 disturbance) 54 -Pain interference-BPI-SF 55 56 57 58 24 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 62 of 85

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 - Patient Assessment of Constipation Quality of Life (PAC- 4 QoL) 5 6 - Montgomery-Asberg Depression Rating Scale 7 -Opioid composite score 8 - Nausea, vomiting, constipation. 9 10 11 Fallon et al-2017-Study 116 12 13 Study design Parallel, multi-center randomized double-blinded, placebo- 14 15 controlled trial. 16 For peer review only 17 Participants 399 patients with chronic cancer pain who were under 18 treatment by opioid regimen. 19 20 21 Intervention (comparison) Sativex (Δ9-tetrahydrocannabinol (27 mg/mL): cannabidiol 22 23 (25 mg/mL) added to opioids (placebo) 24 25 26 Follow-up 5 weeks 27 28 Funding source Industry funding reported. 29 Outcome -Reduction of opioid (calculated in mean daily morphine 30 equivalence dosages) 31 -Pain reduction (VAS 0: no pain-10: most severe pain) 32 -Sleep disturbance (NRS 0: no disturbance-10: most severe http://bmjopen.bmj.com/ 33 disturbance) 34 35 -Mean constipation (NRS 0: no constipation-10: most severe 36 constipation) 37 -Global Impression of Change (SGIC), Patient Satisfaction 38 - Nausea, vomiting, constipation. 39 40 41 on October 1, 2021 by guest. Protected copyright. 42 Fallon et al-2017-Study 216 43 44 45 Study design Parallel, multi-center randomized double-blinded, placebo- 46 controlled trial. 47 48 Participants 206 patients with chronic cancer pain who were under 49 50 treatment by opioid regimen. 51 52 Intervention (comparison) Sativex (Δ9-tetrahydrocannabinol (27 mg/mL): cannabidiol 53 54 55 (25 mg/mL)) added to opioids (placebo)-patients who 56 tolerated titrated dose of cannabis and showed an 57 58 25 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 63 of 85 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 improvement of at least 15% on pain NRS score randomized 4 5 into this study (randomized withdrawal design). 6 7 Follow-up 5 weeks 8 9 Funding source Industry funding reported. 10 Outcome -Reduction of opioid (calculated in mean daily morphine 11 equivalence dosages) 12 13 -Pain reduction (VAS 0: no pain-10: most severe pain) 14 -Sleep disturbance (NRS 0: no disturbance-10: most severe 15 disturbance) 16 For peer- Global review Impression of Change only (SGIC), Patient Satisfaction 17 - mean constipation (NRS 0: no constipation-10: most 18 severe constipation) 19 20 21 22 23 Lichtman et al-201717 24 25 Study design Parallel, multi-center randomized double-blinded, placebo- 26 controlled trial. 27 28 29 Participants 397 patients with chronic cancer pain who were under 30 treatment by opioid regimen. 31 32 Intervention (comparison) Nabiximols was added to opioids and was titrated the 33 maximum allowed daily dosage of 10 sprays per day http://bmjopen.bmj.com/ 34 35 (placebo). 36 37 Follow-up 5 weeks 38 39 Funding source Industry funding reported. 40 Outcome -Reduction of opioid (calculated in mean daily morphine on October 1, 2021 by guest. Protected copyright. 41 equivalence dosages) 42 43 -Pain reduction (NRS 0: no pain-10: most severe pain) 44 -Sleep disturbance (NRS 0: no disturbance-10: most severe 45 disturbance) 46 -Mean constipation (NRS 0: no constipation-10: most severe 47 constipation) 48 -Global Impression of Change (SGIC), Patient Satisfaction 49 50 51 52

53 54 55 56 57 58 26 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 64 of 85

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Supplement Table 5: Risk of bias assessment for RCTs 4 5 6 outcome Blinding of Blinding Loss to Loss 7 Study of Blinding d concealment ata 8 of Blinding of Blinding of Blinding (author- Allocation p patients (≤20%) analyst 9 year) roviders collectors care

10 follow 11 assessors

health health

Data

12

-

up 13

14

15 € 16 Johnson et al- ForPY peerPY reviewPY PY onlyPY PN PY 17 2010 18 £ 19 Portenoy et DY DY PY PY PY PN DN 20 al-2012 21 Fallon et al- PY PY PY PY PY PN DN¥ 22 2017 23 Study 1 24 Fallon et al- PY PY PY PY PY PN PY€ 25 26 2017 27 Study 2 28 Lichtman et PY PY PY PY PY PN DN¥ 29 al-2017 30 31 32

*definitely/probably yes= low risk of bias; definitely//probably no=high risk of bias. http://bmjopen.bmj.com/ 33 34 £ 35 The rate of loss to follow-up was more than 27%. ¥ 36 The rate of loss to follow-up was approximately 26%. 37 €The rate of loss to follow-up was approximately less than 20% 38 39 40 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 27 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 65 of 85 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 4 5 Supplement Table 6: Risk of bias assessments for chart reviews with control group 6 population? unexposed drawn from same Were the exposed and assessment of exposure? Are we confident in the factors? absence of prognostic assessment of the presence or Can we be confident thein o Can we be confident thein up? Was there adequate follow similar? Were the co different confounders? Did the authors adjust for

7 Study utcome assessment? Overall 8 risk of

9

bias 10

-

11 interventions 12 13

14 15

16 For peer review - only 17 18 19 Vigil 2017 DY DN PY PN PY PN PY High 20 21 Maida 2008 DY DY PY DY PN PN PY High 22 23 Barlowe 2019 DY DN PY DY PN PN PN High 24 25 Pritchard-2020 DY DY PY DY DN PN PN High 26 27 Pawasarat-2020 DY DN PY DY DY PN PN High 28 29 30 31 *DY: definitely yes; DN: definitely no; PY: probably yes; PN: probably no; DY/PY= low 32 risk of bias; DN/PN=high risk of bias. 33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 28 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 66 of 85

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 4 Supplement Table 7: Risk of bias assessments for one-arm studies with no control 5 6 group 7 Study Is the source Is the assessment Is there little Overall risk 8 population (sampling of the outcome missing data? of bias 9 frame) representative accurate both at 10 11 of the general baseline and at 12 population? follow-up? 13 Haroutounian et DN DY PN High 14 al-2016 15 16 Narang et al-2008 For peerDN reviewDY only PY High 17 18 19 Yassin et al-2019 DN DY PY High 20 21 22 O’Connell et al- DN DY PY High 23 2019 24 Takakuwa et al- High 25 DN DY PY 26 2020 27 Vigil et al-2017 DN PN PY High 28 29 30 Bellnier-2018 DN DY DY High 31 32 Capano et al-2020 DN DY PN High 33 http://bmjopen.bmj.com/ 34 Rod-2019 DN PN PN High 35 36 37 38 39 *definitely/probably yes= low risk of bias; definitely//probably no=high risk of bias. 40 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49

50 51 52 53 54 55 56 57 58 29 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 67 of 85 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Supplement Table 8: GRADE evidence profile of cannabis adjuvant to opioids vs. 4 14 5 opioid alone for physical function among patients with chronic pain from 1 RCT 6 Outcome n of Follow- Mean difference Certainty Plain- 7 participants up of evidence language 8 (studies) (GRADE) summary 9 Physical Cannabis=118, Two THC: CBD vs. Moderate b Adding 10 11 functioning placebo=59 weeks placebo: -4.23 cannabis to 14 12 (1 RCT ) (P=0.108) opioids 13 THC vs. placebo: probably does 14 -1.25 (P=0.631) not improve 15 physical 16 For peer review only functioning. 17 a b 18 In favor of placebo; Due to imprecision. 19 20 21 Supplement Table 9: GRADE evidence profile of cannabis adjuvant to opioids vs. 22 opioid alone for emotional function among patients with chronic pain from 1 RCT14 23 24 Outcome n of Follow- Mean Certainty of Plain-language 25 participants up difference evidence summary 26 (studies) (GRADE) 27 Emotional Cannabis=118, Two THC: CBD vs. Moderateb Adding 28 functioning placebo=59 weeks placebo: 6.73 cannabis to 29 (1 RCT14) (P=0.084) opioids 30 31 THC vs. probably does 32 placebo: not improve 33 5.22 (P=0.174) emotional http://bmjopen.bmj.com/ 34 functioning. 35 a In favor of cannabis; b Due to imprecision. 36 37 38 39 40 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 30 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 68 of 85

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Supplement Table 10: Summary of adverse events among included observational 4 5 studies* 6 Study Method of Adverse events reported 7 8 assessment 9 10 11 Haroutounian et al4 Self-reported. Two participants discontinued treatment 12 13 due to serious side effects. 14 15 16 Maida et al5 ForSelf peer-reported reviewAnxiety only(P=0.028), nausea (P<0.001), and 17 18 distress (P=0.021) were decreased 19 20 21 significantly among patients who used 22 23 nabilone in comparison to patients who did 24 25 not use it. 26 27 6 28 Narang et al Self-reported (29-item Phase II: Dry mouth, tiredness (both 29 30 symptom Side Effect P<0.0001), abnormal thinking, anxiety, 31 32 Checklist). facial flushing, eye irritation, headache, 33 http://bmjopen.bmj.com/ 34 and ringing in the ears, and drowsiness (P< 35 36 37 0.05) showed a significantly higher 38 39 occurrence at the 20 mg dronabinol dose 40 41 compared with placebo. on October 1, 2021 by guest. Protected copyright. 42 43 44 -Dry mouth, difficulty speaking, 45 46 forgetfulness, confusion, dizziness, and 47 48 euphoria were more occurred in both 49 50 51 treatment group versus placebo (P= 0.01) 52 53 54 55 56 57 58 31 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 69 of 85 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Vigil et al12 Self-reported. No respondents reported any serious side 4 5 6 effects from cannabis use (only 9% of 7 8 patients reported cannabis affected 9 10 negatively their concentration). 11 12 13 13 Yassin et al Self-reported Mostly mild adverse events were reported 14 15 (e.g. red eye, sore throat, increase 16 For peer review only 17 appetite); only 6 patients out of withdrew 18 19 due to the side effects in non-cannabis 20 21 22 group. 23 24 *O’Connell et al7, Barlowe et al1, Rod 2019, and Takakuwa et al11 did not report adverse events. 25 26 27 28 29 30 31 32 33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 32 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 70 of 85

1 2 3 Additional results tables and figures 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

37 Supplement Figure 1: forest plot for oral morphine equivalence dose reduction http://bmjopen.bmj.com/ 38 among patients with Chronic Pain who received cannabis adjuvant to opioids vs. 39 40 opioid alone in RCTs 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

33

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 71 of 85 BMJ Open

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Supplement Figure 2: Subgroup analysis for opioid dose reduction and risk of 36

37 bias (high risk vs. low risk) from 4 RCTs of Cannabis+opioids vs. placebo http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

34

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 72 of 85

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 Supplement Figure 3: forest plot for oral morphine equivalence dose reduction 34 among patients with Chronic Pain who received cannabis adjuvant to opioids vs. 35 opioid alone in observational studies 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

35

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 73 of 85 BMJ Open

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 Supplement Figure 4: forest plot for pain relief on a 10-cm Visual Analog Scale 35 (VAS) among patients with Chronic Pain who received cannabis adjuvant to 36

37 opioids vs. opioid alone in RCTs http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

36

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 74 of 85

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Supplement Figure 5: Subgroup analysis for pain relief on a 10-cm VAS and risk of 36

37 bias (high risk vs. low risk) from 5 RCTs of Cannabis+opioids vs. placebo http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

37

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 75 of 85 BMJ Open

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 Supplement Figure 6: forest plot for pain relief on a 10-cm VAS among patients 32 33 with Chronic Pain who received cannabis adjuvant to opioids vs. opioid alone in 34 observational studies with no control group 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

38

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 76 of 85

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 Supplement Figure 7: forest plot for sleep disturbance on a 10 cm VAS for sleep 34 disturbance among patients with Chronic Pain who received cannabis adjuvant to 35 opioids vs. opioid alone in RCTs 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

39

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 77 of 85 BMJ Open

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

37 Supplement Figure 8: Subgroup analysis for sleep disturbance a 10-cm VAS for http://bmjopen.bmj.com/ 38 sleep disturbance and risk of bias (high risk vs. low risk) from 5 RCTs of 39 40 Cannabis+opioids vs. placebo 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

40

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 78 of 85

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 Supplement Figure 9: Risk difference of nausea among patients with Chronic Pain 28 29 who received cannabis adjuvant to opioids vs. opioid alone in RCTs 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 Supplement Figure 10: Relative Risk of nausea among patients with Chronic Pain 56 57 who received cannabis adjuvant to opioids vs. opioid alone in RCTs 58 59 60

41

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 79 of 85 BMJ Open

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Supplement Figure 11: Relative Risk of vomiting among patients with Chronic 33 Pain who received cannabis adjuvant to opioids vs. opioid alone in RCTs 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 Supplement Figure 12: Risk Difference of vomiting among patients with Chronic 58 Pain who received cannabis adjuvant to opioids vs. opioid alone in RCTs 59 60

42

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 80 of 85

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 Supplement Figure 13: Relative Risk of constipation among patients with Chronic 27 28 Pain who received cannabis adjuvant to opioids vs. opioid alone in RCTs 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 Supplement Figure 14: Risk difference of constipation among patients with 54 Chronic Pain who received cannabis adjuvant to opioids vs. opioid alone in RCTs 55 56 57 58 59 60

43

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 81 of 85 BMJ Open

1 2 3 Appendix B: Reference List of Eligible studies 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 1. Barlowe TS, Koliani-Pace JL, Smith KD, Gordon SR, Gardner TB. Effects of medical cannabis on 8 use of opioids and hospital visits by patients with painful chronic pancreatitis. Clinical 9 Gastroenterology and Hepatology 2019. 10 2. Bellnier T, Brown GW, Ortega TR. Preliminary evaluation of the efficacy, safety, and costs 11 associated with the treatment of chronic pain with medical cannabis. Mental Health Clinician 12 2018;8(3):110-5. 13 3. Capano A, Weaver R, Burkman E. Evaluation of the effects of CBD hemp extract on opioid 14 15 use and quality of life indicators in chronic pain patients: a prospective cohort study. Postgrad Med 16 2020;132(1):56-61. doi:10.1080/00325481.2019.1685298. 17 4. Haroutounian S, Ratz Y, Ginosar Y, et al. The Effect of Medicinal Cannabis on Pain and 18 Quality-of-Life OutcomesFor in Chronic peer Pain: A Prospectivereview Open- labelonly Study. Clin J Pain 19 2016;32(12):1036-43. doi:10.1097/ajp.0000000000000364. 20 5. Maida V, Ennis M, Irani S, Corbo M, Dolzhykov M. Adjunctive nabilone in cancer pain and 21 symptom management: a prospective observational study using propensity scoring. The journal of 22 supportive oncology 2008;6(3):119-24. 23 24 6. Narang S, Gibson D, Wasan AD, et al. Efficacy of dronabinol as an adjuvant treatment for 25 chronic pain patients on opioid therapy. J Pain 2008;9(3):254-64. doi:10.1016/j.jpain.2007.10.018. 26 7. O’Connell M, Sandgren M, Frantzen L, Bower E, Erickson B. Medical Cannabis: Effects on 27 Opioid and Benzodiazepine Requirements for Pain Control. Annals of Pharmacotherapy 28 2019:1060028019854221. 29 8. Pritchard ER, Dayer L, Belz J, et al. Effect of cannabis on opioid use in patients with cancer 30 receiving palliative care. J Am Pharm Assoc (2003) 2020;60(1):244-7. doi:10.1016/j.japh.2019.10.013. 31 9. Pawasarat IM, Schultz EM, Frisby JC, et al. The Efficacy of Medical Marijuana in the 32 Treatment of Cancer-Related Pain. J Palliat Med 2020;23(6):809-16. doi:10.1089/jpm.2019.0374. 33 34 10. Rod K. A Pilot Study of a Medical Cannabis - Opioid Reduction Program. American Journal of 35 Psychiatry and Neuroscience 2019;7(3):74-7. doi:10.11648/j.ajpn.20190703.14 36 11. Takakuwa KM, Hergenrather JY, Shofer FS, Schears RM. The Impact of Medical Cannabis on

37 Intermittent and Chronic Opioid Users with Back Pain: How Cannabis Diminished Prescription Opioid http://bmjopen.bmj.com/ 38 Usage. Cannabis and Cannabinoid Research 2020. 39 12. Vigil JM, Stith SS, Adams IM, Reeve AP. Associations between medical cannabis and 40 prescription opioid use in chronic pain patients: A preliminary cohort study. PLoS One 41 2017;12(11):e0187795. doi:10.1371/journal.pone.0187795. 42 43 13. Yassin M, Oron A, Robinson D. Effect of adding medical cannabis to analgesic treatment in 44 patients with low back pain related to fibromyalgia: an observational cross-over single centre study.

45 Clin Exp Rheumatol 2019;37(Suppl 116):S13-S20. on October 1, 2021 by guest. Protected copyright. 46 14. Johnson JR, Burnell-Nugent M, Lossignol D, et al. Multicenter, double-blind, randomized, 47 placebo-controlled, parallel-group study of the efficacy, safety, and tolerability of THC: CBD extract 48 and THC extract in patients with intractable cancer-related pain. Journal of pain and symptom 49 management 2010;39(2):167-79. 50 15. Portenoy RK, Ganae-Motan ED, Allende S, et al. Nabiximols for opioid-treated cancer 51 52 patients with poorly-controlled chronic pain: a randomized, placebo-controlled, graded-dose trial. 53 The Journal of Pain 2012;13(5):438-49. 54 16. Fallon MT, Albert Lux E, McQuade R, et al. Sativex oromucosal spray as adjunctive therapy in 55 advanced cancer patients with chronic pain unalleviated by optimized opioid therapy: two double- 56 blind, randomized, placebo-controlled phase 3 studies. British journal of pain 2017;11(3):119-33. 57 17. Lichtman AH, Lux EA, McQuade R, et al. Results of a double-blind, randomized, placebo- 58 controlled study of nabiximols oromucosal spray as an adjunctive therapy in advanced cancer 59 60

44

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 82 of 85

1 2 3 patients with chronic uncontrolled pain. Journal of pain and symptom management 2018;55(2):179- 4

88. e1. BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 5 6 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

45

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 83 of 85 BMJ Open

1 2 3 Technical Appendix 4 5 This appendix provides additional details on two different methods of estimation, including BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 1) estimating the mean and standard deviation (SD) from sample size, median, and 7 8 interquartile range (IQR); 2) estimating missing SD (for two non-randomized studies 5,7) 9 10 using the available SD from other included studies. 11 12 1) Estimating the mean and standard deviation (SD) from sample size, median, and 13 14 IQR: 15 16 1) Pawasarat et al 2020 original reported data: median total morphine equivalent=45, 17 n=137, and IQR=135. 18 For peer review only 19 -Using Wan et al method1 produced: mean=60, SD=101 20 21 -Method recommended by Cochrane as sensitivity analysis: 22 23 24 25 26 27 28 q3-q1=IQR. This method produced SD=100. 29 30 2) Bellnier et al 2018 original reported data: median total morphine equivalent (before 31 32 adding cannabis) =79.94, range=0 to 450, median (after adding cannabis) =19.65; range 33 34 =0 to 150, n=29. 35 36 -Using Wan et al method produced: mean (before)=152.4, SD=111; mean (after)=47.3, 37 http://bmjopen.bmj.com/ 38 SD=37.0 39 40 3 41 -Using Cochrane approach (Hozo et al ): Mean (before)= 152.4, SD= 112.5; mean 42 (after)= 47.3, SD= 37.5 43 44

45 We finally included estimation by Wan et al method. The excel sheet including all on October 1, 2021 by guest. Protected copyright. 46 1 47 formula was provided by Wan et al in supplementary file of their article . 48 49 2) Estimating missing SD using the available SD from other included studies: 50 51 1) Maida et al 2008 did not report SD around the mean at the end of follow-up for 52 53 pain intensity. Original reported data: mean (SD) before adding cannabis= 54 7.1(2.4); after adding cannabis mean=3 (missing) 55 56 2) Connell et al 2019 original reported data: mean (SD) before adding cannabis=6.25 57 58 (missing); mean after adding cannabis=6.57 (missing) 59 60

46

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 84 of 85

1 2 3 We imputed missing SDs for these two studies from the given SDs related to other 4 5 five included studies using prognostic method that presented by Ma et al2: BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 10 11 12 13 Assume there are k + l trials altogether where k trials are with full given information 14 15 SEM: value for trial j (missing) with sample size: 16 n : sample size for study with missing information. 17 j 18 SD (imputed) for firstFor study= peer 1.51 review only 19 20 SDs (imputed) for second study=1.76, 1.20 21 22 23 24 25 1 Wan X, Wang W, Liu J, et al. Estimating the sample mean and standard deviation from the sample size, 26 median, range and/or interquartile range. BMC medical research methodology 2014;14(1):135. 27 2 Ma J, Liu W, Hunter A, et al. Performing meta-analysis with incomplete statistical information in clinical 28 trials. BMC medical research methodology 2008;8(1):56. 29 3 Hozo, S.P., Djulbegovic, B. & Hozo, I. Estimating the mean and variance from the median, range, and the 30 size of a sample. BMC Med Res Methodol 5, 13 (2005). https://doi.org/10.1186/1471-2288-5-13 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

47

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

Page 85 of 85 BMJ Open Reported Section/topic # Checklist item 1 on page # 2 TITLE 3 4 Title 1 Identify the report as a systematic review, meta-analysis, or both. 1 5 ABSTRACT 6 7 Structured summary 2 Provide a structured summary including, as applicable: background; objectives; data sources; study eligibility criteria, 3-4 8 participants, and interventions; study appraisal and synthesis methods; results; limitations; conclusions and 9 implications of key findings; systematic review registration number. 10 INTRODUCTION 11 12 Rationale 3 DescribeFor the rationale peer for the review in thereview context of what is already only known. 5 13 Objectives 4 Provide an explicit statement of questions being addressed with reference to participants, interventions, comparisons, 5 14 outcomes, and study design (PICOS). 15 16 METHODS 17 Protocol and registration 5 Indicate if a review protocol exists, if and where it can be accessed (e.g., Web address), and, if available, provide 6 18 registration information including registration number. http://bmjopen.bmj.com/ 19 20 Eligibility criteria 6 Specify study characteristics (e.g., PICOS, length of follow-up) and report characteristics (e.g., years considered, 6 21 language, publication status) used as criteria for eligibility, giving rationale. 22 Information sources 7 Describe all information sources (e.g., databases with dates of coverage, contact with study authors to identify 6 23 additional studies) in the search and date last searched. 24 25 Search 8 Present full electronic search strategy for at least one database, including any limits used, such that it could be 6 26 repeated. on October 1, 2021 by guest. Protected copyright. 27 Study selection 9 State the process for selecting studies (i.e., screening, eligibility, included in systematic review, and, if applicable, 7 28 included in the meta-analysis). 29 Data collection process 10 Describe method of data extraction from reports (e.g., piloted forms, independently, in duplicate) and any processes 30 7 for obtaining and confirming data from investigators. 31 32 Data items 11 List and define all variables for which data were sought (e.g., PICOS, funding sources) and any assumptions and 7 33 simplifications made. 34 Risk of bias in individual 12 Describe methods used for assessing risk of bias of individual studies (including specification of whether this was 35 7-8 studies done at the study or outcome level), and how this information is to be used in any data synthesis. 36 37 Summary measures 13 State the principal summary measures (e.g., risk ratio, difference in means). 8 38 Synthesis of results 14 Describe the methods of handling data and combining results of studies, if done, including measures of consistency 39 8 (e.g., I2) for each meta-analysis. 40 41 Page 1 of 2 42 Reported 43 Section/topic # Checklist item on page # 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

BMJ Open Page 86 of 85 Risk of bias across studies 15 Specify any assessment of risk of bias that may affect the cumulative evidence (e.g., publication bias, selective 9 reporting within studies). 1 2 Additional analyses 16 Describe methods of additional analyses (e.g., sensitivity or subgroup analyses, meta-regression), if done, indicating 9-10 3 which were pre-specified. 4 RESULTS 5 6 Study selection 17 Give numbers of studies screened, assessed for eligibility, and included in the review, with reasons for exclusions at 11 7 each stage, ideally with a flow diagram. 8 Study characteristics 18 For each study, present characteristics for which data were extracted (e.g., study size, PICOS, follow-up period) and 11 9 provide the citations. 10 11 Risk of bias within studies 19 Present data on risk of bias of each study and, if available, any outcome level assessment (see item 12). 12 12 Results of individual studies 20 For all outcomesFor considered peer (benefits orreview harms), present, for each only study: (a) simple summary data for each 12-14 13 intervention group (b) effect estimates and confidence intervals, ideally with a forest plot. 14 15 Synthesis of results 21 Present results of each meta-analysis done, including confidence intervals and measures of consistency. 12-14 16 Risk of bias across studies 22 Present results of any assessment of risk of bias across studies (see Item 15). 9 17 18 Additional analysis 23 Give results of additional analyses, if done (e.g., sensitivity or subgroup analyses, meta-regressionhttp://bmjopen.bmj.com/ [see Item 16]). 12-14 19 20 DISCUSSION 21 Summary of evidence 24 Summarize the main findings including the strength of evidence for each main outcome; consider their relevance to 15 22 key groups (e.g., healthcare providers, users, and policy makers). 23 Limitations 25 Discuss limitations at study and outcome level (e.g., risk of bias), and at review-level (e.g., incomplete retrieval of 15 24 identified research, reporting bias). 25 26 Conclusions 26 Provide a general interpretation of the results in the context of other evidence, and implications on October 1, 2021 by guest. Protected copyright. for future research. 16 27 28 FUNDING 29 Funding 27 Describe sources of funding for the systematic review and other support (e.g., supply of data); role of funders for the 17 30 systematic review. 31 32 33 Page 2 of 2 34 35 36 37 38 39 40 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

Opioid-Sparing effects of medical cannabis or cannabinoids for chronic pain: A systematic review and meta-analysis of randomized and observational studies ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2020-047717.R1

Article Type: Original research

Date Submitted by the 11-Jun-2021 Author:

Complete List of Authors: Noori, Atefeh; McMaster University, Department of Health Research Methods, Evidence, and Impact; McMaster University, The Michael G. DeGroote National Pain Center Miroshnychenko, Anna; McMaster University, Department of Health Research Methods, Evidence, and Impact Shergill, Yaadwinder; McMaster University, Department of Health Research Methods, Evidence, and Impact Ashoorion, Vahid; McMaster University, Department of Health Research Methods, Evidence, and Impact Rehman, Yasir ; McMaster University, Department of Health Research Methods, Evidence, and Impact Couban, Rachel; McMaster University, The Michael G. DeGroote National

Pain Center http://bmjopen.bmj.com/ Buckley, D; McMaster University, Department of Anesthesia Thabane, Lehana; McMaster University, 1. Department of Health Research Methods, Evidence, and Impact Bhandari, Mohit; McMaster University, Division of Orthopaedic Surgery, Department of Surgery; McMaster University, Department of Health Research Methods, Evidence, and Impact Guyatt, Gordon; McMaster University, Department of Health Research Methods, Evidence, and Impact,

Agoritsas, Thomas; University of Geneva Faculty of Medicine, Division of on October 1, 2021 by guest. Protected copyright. of General Internal Medicine; McMaster University, Department of Health Research Methods, Evidence, and Impact Busse, Jason; McMaster University, Department of Anesthesia; McMaster University, Department of Health Research Methods, Evidence, and Impact,

Primary Subject General practice / Family practice Heading:

Secondary Subject Heading: Anaesthesia

PAIN MANAGEMENT, Cancer pain < ONCOLOGY, GENERAL MEDICINE Keywords: (see Internal Medicine)

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 88 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 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 41 on October 1, 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 2 of 88

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 I, the Submitting Author has the right to grant and does grant on behalf of all authors of the Work (as defined 10 in the below author licence), an exclusive licence and/or a non-exclusive licence for contributions from authors 11 who are: i) UK Crown employees; ii) where BMJ has agreed a CC-BY licence shall apply, and/or iii) in accordance 12 with the terms applicable for US Federal Government officers or employees acting as part of their official 13 duties; on a worldwide, perpetual, irrevocable, royalty-free basis to BMJ Publishing Group Ltd (“BMJ”) its 14 licensees and where the relevant Journal is co-owned by BMJ to the co-owners of the Journal, to publish the 15 Work in this journal and any other BMJ products and to exploit all rights, as set out in our licence. 16 17 The Submitting Author accepts and understands that any supply made under these terms is made by BMJ to 18 the Submitting Author Forunless you peer are acting as review an employee on behalf only of your employer or a postgraduate 19 student of an affiliated institution which is paying any applicable article publishing charge (“APC”) for Open 20 Access articles. Where the Submitting Author wishes to make the Work available on an Open Access basis (and 21 intends to pay the relevant APC), the terms of reuse of such Open Access shall be governed by a Creative 22 Commons licence – details of these licences and which Creative Commons licence will apply to this Work are set 23 out in our licence referred to above. 24 25 Other than as permitted in any relevant BMJ Author’s Self Archiving Policies, I confirm this Work has not been 26 accepted for publication elsewhere, is not being considered for publication elsewhere and does not duplicate 27 material already published. I confirm all authors consent to publication of this Work and authorise the granting 28 of this licence. 29 30 31 32 33 34 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 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 3 of 88 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 Opioid-Sparing effects of medical cannabis or cannabinoids for chronic pain: A systematic 4 5 review and meta-analysis of randomized and observational studies 6 7 8 9 Atefeh Noori PhD candidate1,2, Anna Miroshnychenko M.Sc. (Cand.) 1, Yaadwinder Shergill 10 11 M.Sc. (Cand.)1, Vahid Ashoorion PhD1, Yasir Rehman PhD candidate 1, Rachel Couban medical 12 2 3 1 13 librarian , Norman Buckley professor , Lehana Thabane professor , Mohit Bhandari 14 professor1,4, Gordon H. Guyatt distinguished professor1, Thomas Agoritsas assistant professor 15 16 1,5, Jason W. Busse Forassociate peerprofessor6,7* review only 17 18 19 Affiliations 20 1. Department of Health Research Methods, Evidence, and Impact, McMaster University, 21 22 Hamilton, ON, Canada 23 2. The Michael G. DeGroote National Pain Center, McMaster University, Hamilton, ON, 24 Canada 25 3. Department of Anesthesia, McMaster University, Hamilton, ON, Canada 26 4. Division of Orthopaedic Surgery, Department of Surgery, McMaster University, Hamilton, 27 ON, Canada 28 29 5. Division of of General Internal Medicine, Department of Medicine, University Hospitals 30 of Geneva, Geneva, Switzerland 31 6. The Chronic Pain Centre of Excellence for Canadian Veterans, Hamilton, ON, Canada 32 7. The Michael G. DeGroote Centre for Medicinal Cannabis Research, McMaster University, 33 Hamilton, ON, Canada http://bmjopen.bmj.com/ 34 35 36 Corresponding Author: Jason W. Busse, Michael G. DeGroote Centre for Medicinal Cannabis 37 Research, McMaster University, HSC–2V9, 1280 Main St. West, Hamilton, Canada, L8S 4K1; 38 39 email: [email protected] 40 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 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 4 of 88

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 ABSTRACT 4 5 6 Objective: To assess the efficacy and harms of adding medical cannabis to prescription opioids 7 8 among people living with chronic pain. 9 10 Design: Systematic review. 11 12 Data sources: CENTRAL, EMBASE, and MEDLINE. 13 14 15 Main outcomes and measures: Opioid dose reduction, pain relief, sleep disturbance, physical 16 For peer review only 17 and emotional functioning, and adverse events. 18 19 Study selection criteria and methods: We included studies that enrolled patients with chronic 20 21 22 pain receiving prescription opioids and explored the impact of adding medical cannabis. We 23 24 used GRADE to assess the certainty of evidence for each outcome. 25 26 Results: Eligible studies included five randomized trials (all enrolling chronic cancer-pain 27 28 29 patients) and 12 observational studies. All randomized trials instructed participants to maintain 30 31 their opioid dose, which resulted in a very low certainty evidence that adding cannabis has little 32 33 or no impact on opioid use (weighted mean difference [WMD] -3.4 milligram morphine http://bmjopen.bmj.com/ 34 35 equivalent [MME]; 95% confidence interval [CI] -12.7 to 5.8). Randomized trials provided high 36 37 38 certainty evidence that cannabis addition had little or no effect on pain relief (WMD -0.18cm; 39 40 95%CI -0.38 to 0.02; on a 10 cm VAS for pain) or sleep disturbance (WMD -0.22 cm; 95%CI - 41 on October 1, 2021 by guest. Protected copyright. 42 0.4 to -0.06; on a 10 cm VAS for sleep disturbance; minimally important difference [MID] is 1 43 44 45 cm) among chronic cancer-pain patients. Addition of cannabis likely increases nausea (relative 46 47 risk [RR] 1.43; 95%CI 1.04 to 1.96; risk difference [RD] 4%, 95%CI 0% to 7%) and vomiting 48 49 (RR 1.5; 95%CI 1.01 to 2.24; RD 3%; 95%CI 0% to 6%) (both moderate certainty) and may 50 51 52 have no effect on constipation (RR 0.85; 95%CI 0.54 to 1.35; RD -1%; 95%CI -4% to 2%) (low 53 54 55 56 57 58 2 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 88 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 certainty). Eight observational studies provided very-low certainty evidence that adding cannabis 4 5 6 reduced opioid use (WMD -22.5 MME; 95%CI -43.06 to -1.97). 7 8 Conclusion: Opioid-sparing effects of medical cannabis for chronic pain remain uncertain due to 9 10 very-low certainty evidence. 11 12 13 14 15 Systematic review registration PROSPERO CRD42018091098 16 For peer review only 17 18 19 Funding Source: This review received no external funding or other support 20 21 22 23 24 Keywords: chronic pain; opioids; cannabis; cannabinoids; drug substitution; sparing effect; 25 26 tapering 27 28 29 30 31 32 33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on October 1, 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 6 of 88

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 Strengths and limitations of this study 4 5 6  This is the first meta-analysis to pool the results of randomized controlled trials (RCTs) 7 8 and observational studies exploring the opioid-sparing effects of medical cannabis among 9 10 11 people living with chronic pain. 12 13  We conducted a comprehensive search for eligible studies, appraised the risk of bias of 14 15 included studies, and evaluated the certainty of evidence using the GRADE approach. 16 For peer review only 17 18  Most observational studies incorporated inadequate adjustment for confounding, and all 19 20 randomized trials, despite reporting this outcome, were not designed to address the effect 21 22 of medical cannabis on opioid use. 23 24 25 26 27 28 29 30 31 32 33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 4 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 88 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 Introduction 4 5 6 Chronic pain affects approximately one in five adults and is a common reason for seeking 7 8 medical care.1, 2 Opioids are commonly prescribed for this condition, particularly in North 9 10 America;3 however, they only provide benefit to a minority of patients. A 2018 systematic 11 12 review of 96 trials found high certainty evidence that, versus placebo, opioids provide important 13 14 15 pain relief (≥1cm improvement on a 10-cm visual analog scale for pain) to 12% of patients for 16 For peer review only 17 whom they are prescribed.4 Moreover, opioids are associated with harms such as overdose and 18 19 death,5, 6 which are dose-dependent.7-10 As a result, there is considerable interest in therapies that 20 21 22 may allow patients with chronic pain using opioid therapy to reduce their opioid intake. 23 24 One promising approach is adding cannabis therapy, which low certainty evidence 25 26 suggests may be similarly effective to opioids for reducing pain and improving physical 27 28 4 29 functioning among people living with chronic pain. Experimental studies have shown that 30 31 opioids and cannabis have similar signal transduction systems,11 and observational studies in the 32 33 US demonstrated that the rates of opioid-related mortality reduced after cannabis was http://bmjopen.bmj.com/ 34 35 legalized.12-14 Between 64% and 77% of patients with chronic pain responding to cross-sectional 36 37 38 surveys reported a reduction in long-term opioid use after adding medical cannabis to their 39 40 treatment.15, 16 A 2017 systematic review concluded that pre-clinical studies provided robust 41 on October 1, 2021 by guest. Protected copyright. 42 evidence for the opioid-sparing effects of cannabis.17 To clarify the issue, we undertook a 43 44 45 systematic review of randomized controlled trials and observational studies to explore the impact 46 47 of adding medical cannabis on opioid dose, other patient-important outcomes, and related harms 48 49 in patients with chronic pain using prescribed opioid therapy. 50 51 52 This systematic review is part of the BMJ Rapid Recommendations project, a 53 54 collaborative effort from the MAGIC Evidence Ecosystem Foundation 55 56 57 58 5 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 88

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 (www.magicevidnece.org) and BMJ. This systematic review informed a parallel guideline 4 5 18 6 published on BMJ.com and MAGICapp (https://app.magicapp.org/#/guideline/jMMYPj). 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 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 6 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 88 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 METHODS 4 5 We followed standards for meta-analysis of observational studies in epidemiology (MOOSE)19 6 7 20 8 and preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines 9 10 and registered our review (PROSPERO Identifier: CRD42018091098). 11 12 13 14 15 Eligibility criteria 16 For peer review only 17 We included randomized controlled trials (RCTs) and observational studies, including cohort 18 19 studies and case-control studies, in any language, that explored the impact of adding medical 20 21 22 cannabis (i.e. phytocannabinoids, endocannabinoids, or synthetic cannabinoids) on the use of 23 24 prescription opioids among people living with chronic pain. We defined pain as chronic if 25 26 patients reported that symptoms had persisted for ≥3 months.21 We excluded editorials, letters to 27 28 the editor, pre-clinical studies, conference abstracts, case reports, case series, cross-sectional 29 30 31 studies, and studies with less than 2-weeks follow-up. We also excluded studies of recreational 32 33 cannabis use as these products typically contain much higher amounts of the psychotropic http://bmjopen.bmj.com/ 34 35 cannabinoid tetrahydrocannabinol (THC) than would be administered for therapeutic purposes.22, 36 37 23 38 We classified observational study designs according to recommendations by the Cochrane 39 40 Observational Studies Methods Group.24 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 Literature search and study selection 46 47 We searched the Cochrane Central Register of Controlled Trials (CENTRAL), EMBASE, and 48 49 MEDLINE from inception to March 2020 with no restriction on language of publication. An 50 51 experienced medical librarian (RC) developed our database-specific search strategies (Appendix 52 53 54 A). We also searched the ClinicalTrials.gov registry to identify ongoing trials, and reference lists 55 56 of all eligible studies and related systematic reviews for additional eligible studies. Two teams of 57 58 7 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 88

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 paired reviewers independently screened titles, abstracts and full-text studies for eligibility using 4 5 6 online systematic review software (Rayyan QCRI, Qatar Computing Research Institute). 7 8 Reviewers resolved disagreements through discussion. 9 10 11 12 Data collection 13 14 15 Using standardized forms and a detailed instruction manual, pairs of reviewers independently 16 For peer review only 17 abstracted data from each eligible study, including study and patient characteristics, and details 18 19 of treatment (e.g. dose, formulation, and duration of cannabis add-on therapy). Our primary 20 21 22 outcome was opioid dose. We also captured all patient-important outcomes, as guided by the 23 24 Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials,25 including pain 25 26 relief, sleep disturbance, physical and emotional functioning. Regarding adverse events, we 27 28 26 29 focused on vomiting, nausea, and constipation as a systematic review of values and preferences 30 31 demonstrated that patients living with chronic pain experience gastrointestinal complaints as the 32 33 most important opioid-induced adverse events. We contacted authors to obtain unpublished data. http://bmjopen.bmj.com/ 34 35 36 37 38 Risk of bias assessment 39 40 Following training and calibration exercises two independent reviewers used a modified 41 on October 1, 2021 by guest. Protected copyright. 42 Cochrane risk of bias tool27, 28 to assess the risk of bias among eligible RCTs according to the 43 44 45 following domains: allocation concealment, blinding of participants, study personnel, outcome 46 47 assessors and data analyst, and loss to follow-up (≥20% missing data was assigned high risk of 48 49 bias). Response options for each item were 'definitely or probably yes' (assigned a low risk of 50 51 52 bias) and 'definitely or probably no' (assigned a high risk of bias). (Supplement Table 1) We used 53 54 criteria suggested by the CLARITY group 29 to assess the risk of bias of observational studies 55 56 57 58 8 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 88 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 including selection bias, confidence that all patients had the condition of interest, control for 4 5 6 confounding variables, validity of outcome assessment(s), and infrequent missing data (<20%) 7 8 (details available at www.evidencepartners.com/resources/methodological-resources/). 9 10 (Supplement Tables 2-3). 11 12 13 14 15 Data analysis 16 For peer review only 17 We calculated inter-rater agreement regarding the eligibility of full-text studies using an adjusted 18 19 kappa (κ) statistic.30 We conducted separate analyses for randomized controlled trials and 20 21 22 observational studies. All continuous measures for pain intensity and sleep disturbance were 23 24 converted to a 10 cm visual analogue scale (VAS); the minimally important difference (MID) for 25 26 both was 1 cm.31, 32 All continuous outcomes that were reported by more than one study were 27 28 29 pooled to derive the weighted mean difference (WMD) and associated 95% confidence interval 30 31 (95% CI). We pooled binary outcomes (adverse events) as relative risks (RRs) and risk 32 33 differences (RDs) and their associated 95% CIs. We conducted all meta-analyses with random- http://bmjopen.bmj.com/ 34 35 effects models and the DerSimonian-Laird method.33 36 37 38 When studies reported effects on continuous outcomes as the median and interquartile 39 40 range, we derived the mean and SD using the method presented by Wan et al. 34 We also 41 on October 1, 2021 by guest. Protected copyright. 42 converted medians to means using the approach recommended by the Cochrane Handbook as a 43 44 45 sensitivity analysis. When authors failed to report a measure of precision associated with mean 46 47 differences, we imputed the SD from eligible studies that reported these measures (Technical 48 49 appendix).35 We included each comparison reported by multi-arm studies and calculated a 50 51 52 correction factor to account for the unit of analysis error (i.e. when information from a treatment 53 54 arm is used more than once in the same meta-analysis).36 We explored the consistency of 55 56 57 58 9 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 88

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 association between our pooled results and studies reporting the same outcome domains that 4 5 6 were not possible to pool. We used Stata (StataCorp, Release 15.1, College Station, Texas) for 7 8 all analyses. Comparisons were 2-tailed using a threshold of p ≤ 0.05. 9 10 11 12 Subgroup analyses and meta-regression 13 14 15 We examined heterogeneity among pooled RCTs using the I2 statistic, and through visual 16 For peer review only 17 inspection of forest plots for pooled observational data, because statistical tests of heterogeneity 18 19 can be misleading when sample sizes are large and associated confidence intervals are therefore 20 21 37 22 narrow. When we had at least two studies in each subgroup, we explored sources of 23 24 heterogeneity with five pre-specified subgroup hypotheses, assuming greater benefits with: (1) 25 26 shorter vs. longer duration of follow-up; (2) higher vs. lower risk of bias; (3) enriched vs non- 27 28 29 enriched study design; (4) chronic non-cancer vs. chronic cancer-related pain; and (5) higher vs 30 31 lower tetrahydrocannabinol [THC] content. We assumed similar directions of subgroup effects 32 33 for harms, except for study design and THC content in which we expected greater harms with http://bmjopen.bmj.com/ 34 35 non-enriched trials and higher THC content. However, apart from item two (risk of bias), studies 36 37 38 did not report sufficient data to undertake subgroup analyses. 39 40 41 on October 1, 2021 by guest. Protected copyright. 42 The certainty of the evidence 43 44 45 We used the Grading of Recommendations Assessment, Development and Evaluation (GRADE) 46 47 approach to assess the certainty of evidence on an outcome-by-outcome basis as high, moderate, 48 49 low or very low.38 With GRADE, RCTs begin as high-certainty evidence, but can be rated down 50 51 52 because of risk of bias, imprecision, inconsistency, indirectness, or publication bias. We rated 53 54 down for imprecision if the 95% CI associated with a pooled continuous outcome included ½ the 55 56 57 58 10 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 88 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 MID, or if the estimate of precision associated with the RR for binary outcomes included no 4 5 2 6 effect. We considered an I value between 75% and 100% to represent considerable 7 8 inconsistency.39 We rated down the certainty of evidence for indirectness if there were important 9 10 differences between our research question and the patients enrolled, intervention tested, or 11 12 outcomes reported among studies contributing to our meta-analyses.40 13 14 15 Using GRADE, observational studies begin as low certainty evidence, and while they can 16 For peer review only 17 be rated down further for the same reasons as RCTs, they can also be rated up in the presence of 18 19 a large magnitude of the effect, a dose-response gradient, or consideration of plausible 20 21 41 22 confounders or other biases that increase confidence in the estimated effect. We only reported 23 24 the pooling results of observational studies when they resulted in the same or higher certainty of 25 26 evidence than evidence from RCTs. When there were at least 10 studies for meta-analysis, we 27 28 29 explored for small-study effects by visual assessment of funnel plot asymmetry and Egger’s 30 31 statistical test.42 32 33 http://bmjopen.bmj.com/ 34 35 Patients and public involvement 36 37 38 Patients and public were not involved in this research. 39 40 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 11 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 88

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 RESULTS 4 5 6 Of 5133 records identified, we reviewed 133 articles in full text, and 18 studies reported in 17 7 8 publications proved eligible (Figure 1, Appendix B); five RCTs in four publications43-46 and 13 9 10 observational studies.47-59 One study enrolled a mixed group of opioid and non-opioid users;50 11 12 however, our attempts to contact the authors to acquire pain intensity data for the sub-group of 13 14 15 patients prescribed opioids proved unsuccessful. All five RCTs43-46 and three observational 16 For peer review only 17 studies51, 54, 55 enrolled patients with chronic cancer-related pain; the remaining 10 observational 18 19 studies explored adding cannabis to opioids for patients with chronic non-cancer pain (e.g. 20 21 47, 52, 53, 57-59 22 chronic low back pain, fibromyalgia, painful chronic pancreatitis), or a mix of cancer 23 24 and non-cancer pain (Table 1).48-50, 56 25 26 Among the 18 included studies, the percentage of female participants was 48% (median 27 28 29 of individual trials 48%, interquartile range [IQR] 43% to 58%), and the median of the mean age 30 31 was 56.3 (IQR 51.2 to 59.9). Follow-up ranged from 2 to 5 weeks among RCTs, and from 4 32 33 weeks to 6.4 years for observational studies. Only 1 RCT43 used an enrichment design (following http://bmjopen.bmj.com/ 34 35 the open-label phase, patients with at least 15% improvement in pain were randomized to the 36 37 38 intervention and control groups) and all RCTs advised patients to maintain stable doses of all 39 40 other prescribed pain medications, including opioids, during the study period (Table 1). All 41 on October 1, 2021 by guest. Protected copyright. 42 included RCTs, and three of the observational studies48, 51, 52 administered synthetic cannabis 43 44 49, 50, 56, 58, 59 45 products (i.e. nabilone, dronabinol, and nabiximole), five observational studies 46 47 reported different combinations of THC: CBD products, and 6 other observational studies47, 53-55, 48 49 57 did not provide details on the type of cannabis or cannabinoids provided (Table 1, Supplement 50 51 43-46, 49, 51, 52, 57, 58 50, 53-56 52 Table 4). Ten studies reported receiving industry funding, five studies 53 54 55 56 57 58 12 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 88 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 reported no-industry funding, and three studies47, 48, 59 did not report funding information (Table 4 5 6 1). 7 8 9 10 Risk of bias of included studies 11 12 All included RCTs reported adequate allocation concealment and blinding of patients and health- 13 14 15 care providers; however, three trials43, 45, 46 were at risk of bias due to high loss to follow up 16 For peer review only 17 (Supplement Table 5). Each RCT specified that they employed an intention-to-treat analysis. All 18 19 observational studies were at high risk of bias, typically due to lack of confidence in the 20 21 22 assessment of exposure, non-representative samples, and insufficient control for confounding 23 24 (Supplement Tables 6-7). 25 26 27 28 29 Outcomes for medical cannabis add-on therapy 30 31 Opioid dose reduction 32 33 The primary limitation of RCTs was that all investigators instructed patients to not alter their http://bmjopen.bmj.com/ 34 35 36 dose of opioids. This represents a very serious indirectness of the findings regarding the research 37 38 question, warranting rating down two levels, and was the primary reason for very low certainty 39 40 evidence from the 1176 patients.43-45 Their results raised the possibility that adding medical 41 on October 1, 2021 by guest. Protected copyright. 42 43 cannabis may not be associated with a reduction in opioid use among patients living with chronic 44 45 cancer pain (WMD -3.4 MME; 95%CI -12.7 to 5.9; table 2; Supplement Figure 1). There were 46 47 no differences in effect based on the loss to follow-up (Supplement Figure 2; test of interaction 48 49 P=0.758). 50 51 52 Very-low certainty evidence from 8 observational studies (7 of which enrolled people 53 54 with chronic non-cancer pain)47, 48, 50, 51, 53-55, 58 raised the possibility that adding medical cannabis 55 56 57 58 13 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 88

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 may reduce the use of opioids among patients with predominantly chronic non-cancer pain 4 5 6 (WMD -22.5 MME; 95%CI -43.06 to -1.97; Table 2; Supplement Figure 3). Three observational 7 8 studies that could not be pooled, as they only reported opioid reduction as a percentage, also 9 10 found that providing medical cannabis allowed patients to decrease their opioid dose. The first 11 12 study assessed the impact of providing medical cannabis to 61 patients with chronic low back 13 14 15 pain who were prescribed opioid therapy (median opioid dose was 21 mg MME/day) and 16 For peer review only 17 reported that 52% of patients (32 of 61) stopped all use of opioids at a median follow-up of 6.4 18 19 years.57 The second study 49 reported that of 94 patients with chronic pain (both cancer and non- 20 21 22 cancer pain) who began using CBD hemp extract, 53% were able to decrease their use of 23 24 prescription opioids at 8 weeks. A third study56 included 600 patients with chronic pain who 25 26 indicated willingness to taper their opioid dose and were administered 0.5g daily of medicinal 27 28 29 cannabis for each 10% reduction in opioid dose. After 6 months’ follow-up, 55% of patients 30 31 reported a 30% reduction in opioid dose on average and 26% of them discontinued opioid use. 32 33 http://bmjopen.bmj.com/ 34 35 Pain relief 36 37 38 High-certainty evidence from 5 RCTs43-46 demonstrated that adding medical cannabis to opioid 39 40 therapy resulted in trivial or no difference in cancer related pain (WMD -0.18 cm; 95%CI -0.38 41 on October 1, 2021 by guest. Protected copyright. 42 to 0.02 on the 10 cm VAS for pain; MID 1cm; Table 2; Supplement Figure 4). Results did not 43 44 45 differ depending on loss to follow-up (Supplement Figure 5, a test of interaction P=0.623). Very 46 47 low certainty evidence from observational studies suggested a large decrease in pain when 48 49 medical cannabis was added to opioids (Supplement Figure 6). 50 51 52 53 54 Sleep disturbance 55 56 57 58 14 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 88 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 Five RCTs43-46 provided high certainty evidence that adding medical cannabis to prescription 4 5 6 opioids results in a trivial improvement in sleep disturbance in people living with cancer-related 7 8 chronic pain (WMD -0.22 cm; 95%CI -0.4 to -0.06 on the 10 cm VAS for sleep disturbance; 9 10 MID 1cm; Table 2; Supplement Figure 7). Results did not differ between trials reporting the low 11 12 and high loss to follow-up (Supplement Figure 8, a test of interaction P =0.93). Very low 13 14 15 certainty evidence from observational studies suggested an improvement in sleep disturbance 16 For peer review only 17 when medical cannabis was added to opioids (Supplement Table 8). 18 19 20 21 22 Other reported outcomes 23 24 A single RCT44 reported moderate certainty evidence that adding cannabis likely has little or no 25 26 effect on emotional and physical functioning (Supplement Tables 9-10). 27 28 29 30 31 32 Adverse events 33 http://bmjopen.bmj.com/ 34 Nausea, vomiting, or constipation 35 36 4 RCTs43-46 provided moderate certainty evidence that adding medical cannabis to opioid therapy 37 38 39 likely increases the incidence of nausea (RR 1.43, 95%CI 1.04 to 1.96; RD 4%, 95%CI 0% to 40 41 7%; Supplement Figure 9-10) and vomiting (RR 1.50; 95%CI 1.01 to 2.24; RD 3%; 95%CI 0% on October 1, 2021 by guest. Protected copyright. 42 43 to 6%; Supplement Figure 11-12) in patients with cancer-related chronic pain prescribed opioid 44 45 43, 45, 46 46 therapy. Three RCTs provided low certainty evidence that adding medical cannabis to 47 48 opioid therapy may not increase constipation (RR 0.85, 95%CI 0.54 to 1.35; RD -1%; 95%CI - 49 50 4% to 2%; Supplement Figure 13-14). Supplement Table 11 summarizes adverse events reported 51 52 in observational studies. 53 54 55 56 57 58 15 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 88

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 DISCUSSION 4 5 Very-low certainty evidence from randomized trials and observational studies was conflicting 6 7 8 and leaves uncertain whether the addition of medical cannabis affects the use of prescribed 9 10 opioids among people living with chronic pain. Compared with long-term opioid therapy for 11 12 chronic cancer pain without medical cannabis, high certainty evidence showed that adding 13 14 15 medical cannabis had little to no effect on pain or sleep disturbance. Results provided moderate 16 For peer review only 17 certainty evidence that adding cannabis therapy to opioids likely increases both nausea (RR 1.43, 18 19 95%CI 1.04 to 1.96) and vomiting (RR 1.50; 95%CI 1.01 to 2.24), and low certainty evidence 20 21 22 suggested no effect on constipation (RR 0.85, 95%CI 0.54 to 1.35). 23 24 25 Strengths of our review include a comprehensive search for eligible randomized and 26 27 observational studies, appraisal of the risk of bias among individual studies, and use of the 28 29 GRADE approach to rate the certainty of evidence. Our review has limitations, primarily due to 30 31 32 features of primary studies eligible for review, which failed to report all recommended outcomes 33 http://bmjopen.bmj.com/ 34 that have been established as important for people living with chronic pain. Most observational 35 36 studies incorporated inadequate adjustment for confounding. All randomized trials, despite 37 38 reporting this outcome, were not designed to address the effect of medical cannabis on opioid 39 40 41 use. All eligible RCTs enrolled patients with chronic cancer-related pain, and the generalizability on October 1, 2021 by guest. Protected copyright. 42 43 to non-cancer chronic pain is uncertain. Specifically, substitution effects of medical cannabis for 44 45 prescription opioids may also differ between chronic cancer and non-cancer pain; however, lack 46 47 48 of variability among studies eligible for our review precluded exploration of this subgroup effect. 49 50 Studies included in our review administered different formulations of cannabis and cannabinoid 51 52 products; however, pooled effects of outcomes reported in RCTs showed no important 53 54 55 heterogeneity. 56 57 58 16 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 88 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 A meta-analysis of pre-clinical studies,17 a narrative systematic review,60 and several 4 5 6 cross-sectional and case studies have reported an apparent reduction in opioid use with addition 7 8 of cannabis therapy.9, 10, 61-65 In a national US population-based survey66 of 2,774 cannabis users 9 10 (both medical and non-medical use) 36% of respondents reported substituting cannabis for 11 12 prescription opioids (discontinued opioid use). In this survey, the 60% of participants who 13 14 15 identified as medical cannabis users were much more likely to substitute cannabis for 16 For peer review only 17 prescription drugs than recreational users (OR 4.59; 95%CI 3.87 to 5.43). Another US survey67 18 19 that included 841 patients prescribed long-term opioid therapy for chronic pain reported that 20 21 22 61% used medical cannabis, and 97% of this subgroup reported coincident reduction of their 23 24 opioid use. Consistent with these findings, very low certainty evidence from observational 25 26 studies in our review also suggests that adding medical cannabis allows patients predominantly 27 28 29 with chronic non-cancer pain to reduce their use of opioids. Although RCT results do not 30 31 support reduction in opioid dose by adding medical cannabis for opioids, the evidence is also 32 33 very low certainty, primarily because investigators instructed patients to maintain their current http://bmjopen.bmj.com/ 34 35 opioid dose. This is a critical limitation, despite the 2019 NICE guideline having concluded that 36 37 38 providing medical cannabis for chronic pain does not reduce opioid use on the basis of these 39 40 trials.68 Future trials should randomize chronic pain patients who voluntarily agree to engage in a 41 on October 1, 2021 by guest. Protected copyright. 42 trial of opioid tapering to receive medical cannabis or placebo and report all patient-important 43 44 69 70 71 45 outcomes. Forced opioid tapering is ineffective and may cause harm. 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 20 of 88

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 Conclusion 4 5 6 The opioid-sparing effects of medical cannabis for chronic pain remain uncertain. Based on 7 8 moderate-to-high certainty evidence, adding medical cannabis to opioid therapy among chronic 9 10 cancer pain patients had little or no effect on neither pain relief nor sleep disturbance and likely 11 12 increases the risk of nausea and vomiting. The accompanying BMJ Rapid Recommendation18 13 14 15 provides contextualized guidance based on this evidence, as well as three other systematic 16 For peer review only 17 reviews on benefits,72 harms73 and patients' values and preferences74. 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 October 1, 2021 by guest. Protected copyright. 42 43 44 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 21 of 88 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 Disclosures/Conflicts of Interest: All authors have no financial relationships with any 4 5 6 organizations that might have an interest in the submitted work. 7 8 Role of the Funding Source: This review received no external funding or other support. 9 10 Ethical approval: ethical approval was not required because this study retrieved and synthesized 11 12 data from already published studies. 13 14 15 Data: Details of the characteristics of the included studies were shared in the supplementary 16 For peer review only 17 materials. Data will be made available upon publication and can be obtained from the 18 19 corresponding author at [email protected]. 20 21 22 Disclaimers: None. 23 24 Transparency: All authors affirm that the manuscript is an honest, accurate, and transparent 25 26 account of the study being reported; that no important aspects of the study have been omitted; 27 28 29 and that any discrepancies from the study as planned (and, if relevant, registered) have been 30 31 explained. 32 33 Contribution: JWB, AN and GHG conceived and designed the study. RC performed the http://bmjopen.bmj.com/ 34 35 literature search. AN, AM, YS, VA and YR selected the studies, extracted the relevant 36 37 38 information, and assessed the risk of bias of selected studies. AN synthesised the data. AN wrote 39 40 the first draft of the paper. AN, JWB, GHG and TA critically revised the manuscript for 41 on October 1, 2021 by guest. Protected copyright. 42 important intellectual content. AN, JWB, LT, GHG, MB and NB interpreted the findings. JWB, 43 44 45 LT and GHG provided methodological support. All authors reviewed the paper and approved the 46 47 final version. The corresponding author attests that all listed authors meet authorship criteria and 48 49 that no others meeting the criteria have been omitted. 50 51 52 53 54 55 56 57 58 19 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 88

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 Figure Legends 4 5 6 7 8 Figure 1: Study selection process in review of opioid-sparing effects of cannabis in chronic pain 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 41 on October 1, 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 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

Page 23 of 88 BMJ Open

1 2 3 Table 1: Characteristics of included studies (n=18) 4 5 Author- Study design No. of Type of Age % Baseline opioid Follow- Medical Analgesic Co- Funding 6 year participants chronic pain mean Femal dose up cannabis dose intervention source 7 (country) (% prescribed (specific (SD) e duration condition) 8 opioids) 9 10 11 12 For peer review only 13 Fallon et Parallel arm n=399; 100% 59.8 43% Receiving 5 weeks THC 27 Patients were Otsuka 14 al., 2017 RCT nabiximols [n chronic (10.9) opioid therapy mg/mL; CBD excluded if Pharmac 15 study I =20], cancer pain of <500 25 mg/mL they planned eutical 16 (multicente placebo MME/day (maximum to undergo Co., Ltd. 17 r trial£)43 [n=199] (Nabiximols allowed daily clinical 18 (100%) group: dosage http://bmjopen.bmj.com/ interventions 19 199MME/day± of 10 sprays) that would 20 131; placebo affect pain 21 group: 22 207MME/day± 23 135) 24 Fallon et Parallel arm n=206; 100% 61.5 49% Receiving 5 weeks THC 27 Patients were Otsuka 25 al., 2017 RCT nabiximols chronic (11.3) opioid therapy mg/mL; CBD excluded if Pharmac 26 study II [n=103], cancer pain of <500 25 mg/mL on October 1, 2021 by guest. Protected copyright. they planned eutical 27 (multicente placebo=103 MME/day (maximum to undergo Co., Ltd. £ 43 28 r trial ) (100%) (Nabiximols: allowed daily clinical 29 212MME/day± dosage interventions 30 136; placebo: of 10 sprays) that would 31 209MME/day± affect pain 121) 32 Johnson et Parallel arm n=177; THC: 100% 60.2 46% Receiving 2 weeks One spray: Patients were GW 33 al., 2010 RCT CBD extract chronic (12.3) opioid therapy 2.7mg excluded if Pharmac 34 (multicente [n= 60], THC cancer pain for at least one- THC/2.5mg they planned euticals 35 r trial£)44 extract [n=58], week before CBD. to undergo 36 placebo enrollment The maximum clinical 37 [n=59] (THC:CBD: permitted dose: interventions 38 (100%) 258MME/day± 8 actuations that would 39 789; THC: over 3-hours affect pain 40 188MME±234; and 41 42 43 21 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

BMJ Open Page 24 of 88

1 2 3 placebo: 48 actuations 4 367±886) over 24-hours 5 Lichtman Parallel arm n=398; 100% 60 46% Receiving 5 weeks THC 27 Patients were Otsuka 6 et al., 2018 RCT nabiximol chronic (11.5) opioid therapy mg/mL; CBD excluded if Pharmac 7 (multicente [n=199], cancer pain of <500 25 mg/mL they planned eutical 8 r£)45 placebo MME/day (maximum to undergo Co., Ltd. 9 [n=198] (nabiximols: allowed daily clinical 10 (100%) 193MME/day± dosage interventions 11 130; placebo: of 10 sprays per that would 12 For peer review186MME/day± only day) affect pain 13 131) 14 Portenoy Parallel arm n=360; 100% 58 48% Receiving 5 weeks THC 27 Patients were GW 15 et al., 2012 RCT nabiximols chronic (12.2) opioid therapy mg/mL; CBD allowed to use Pharmac 16 (multicente low dose (1-4 cancer pain of <500 25 mg/mL breakthrough euticals; 46 17 r£) sprays/day) MME/day (maximum opioid Otsuka 18 [n=91], (median was allowed daily http://bmjopen.bmj.com/ analgesic as Pharmac 19 medium dose 120MME/day; dosage of 10 required eutical 20 (6-10 range 3 to sprays per day) Co., Ltd. sprays/day) 16,660) 21 [n=88], high 22 dose (11-16 23 sprays/day) 24 [n=90], 25 placebo 26 [n=91] on October 1, 2021 by guest. Protected copyright. 27 (100%) 28 Barlow et Retrospectiv Enrolled in 100% CNCP 49.9 45% Not enrolled in Range 4 NR NR NR 29 al., 2019 e chart- MCP [n=34], (chronic (10.5) MCP to 297 30 (US)47 review not enrolled in painful 183MME/day± weeks 31 MCP [n=19] pancreatitis) 284; enrolled in 32 (100%) MCP 33 190MME/day± 34 273 35 36 Bellnier et One-arm n= 29 90% CNCP; 61 65% Patients were 13 weeks 10mg capsules NR NR 37 al., 2018 observationa (100%) 10% cancer (10) receiving a of 38 (US)48 l study pain median opioid THC/ CBD in a 39 dose of 1:1 ratio 3-times 40 79.94MME/day daily 41 42 43 22 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

Page 25 of 88 BMJ Open

1 2 3 Capano et One-arm n= 131 100% 56.1 68% Receiving at 8 weeks 30mg CBD/1mg NR Ananda 4 al., 2020 observationa (100%) chronic pain (rang least THC Professio 5 (US)49 l study (cancer and e: 39 50MME/day nal. 6 non-cancer) to 70) 7 Haroutouni One-arm n=73 93.2% 51.2 38%¥ Receiving a 26 weeks Cigarettes: 6% All No- 8 an et al., observationa (35%) CNCP; 6.8% (15.4) median opioid to 14% THC, participants external 9 2016 l study chronic ¥ dose of 0.2% to 3.8% were funding 10 (Israel)50 cancer pain 60MME/day CBD; encouraged to 11 (range 45 - 90) attempt 12 For peer review onlyOral: 11% to gradual dose 13 19% THC, reduction and 14 0.5% to 5.5% possible 15 CBD discontinuatio 16 n of other 17 analgesics 18 Maida et Prospective Enrolled in 100% 69.7 42% nabilone 4 weeks On average 1.79 http://bmjopen.bmj.com/ Patients were Valeant 19 al., 2008 cohort MCP [n=47], chronic (10.1) treated:60MME mg twice daily permitted to Pharma- 51 20 (Canada) not enrolled in cancer pain /day±64; nabilone use ceuticals MCP [n=65] nabilone concomitant Canada 21 (100%) untreated: analgesics Ltd 22 67MME/day±1 23 01 24 Narang et One-arm n=30 100% CNCP Medi 53% Receiving an 4 weeks Flexible dose NR Solvay 25 al., 2008 observationa (100%) an=4 average opioid schedule, Pharmac on October 1, 2021 by guest. Protected copyright. 26 (US)52 l study 3.5 dose of dronabinol 5mg euticals, 27 (rang 68MME/day±5 to 20mg 3 times Inc. 28 e=21- 7 daily 29 67) 30 O’Connell One-arm n=77 (100%) 100% CNCP 54.1 58% Receiving a 26 weeks NR NR No 31 et al., 2019 observationa (rang mean opioid industry 32 (US)53 l study e=26- dose of funding 33 76) 140MME/day± 34 184 35 Pritchard Retrospectiv cannabis and 100% 53.1 23% MCP enrolled: 26 weeks NR NR No 36 et al.,2020 e cohort opioids co-use chronic (11.7) 144MME/day± industry 37 (US)54 [n=22], cancer pain 129; MCP not funding 38 opioids only enrolled: 39 [n=61] 119MME/day 40 (100%) ±100 41 42 43 23 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

BMJ Open Page 26 of 88

1 2 3 Pawasarat Retrospectiv Enrolled in 100% 58 56% MCP enrolled: Between NR NR No 4 et al., 2020 e chart MCP [n=137], chronic (IQR: median 39 and industry 5 (US)55 review not enrolled in cancer pain 14.7) 45MME/day, 52 weeks funding 6 MCP [n=95] IQR=135; MCP for MCP 7 (100%) not enrolled: enrolled; 8 median <26 9 97.5MME/day, weeks 10 IQR=150 for not 11 enrolled 12 Rod et al., One-arm n=600For 100%peer NR reviewNR Receiving a 26only weeks CBD and THC All No 13 2019 observationa chronic pain mean opioid ranged between participants external 14 (Canada)56 l study (cancer and dose of 120 4% to 6%. indicated funding 15 non-cancer) MME/day Doses related ready to 16 (range 90 to directly to the reduce opioid 17 240MME/day) opioid taper. dose and also 18 http://bmjopen.bmj.com/ received 19 psychological supports (e.g. 20 CBT, 21 mindfulness, 22 relaxation) 23 24 Takakuwa One-arm n=61 100% CNCP 50 38% Receiving a Median NR NR The 25 et al., 2020 observationa (100%) (back pain) (11.4) median opioid of 6.4 Society 26 (US)57 l study dose of years on October 1, 2021 by guest. Protected copyright. of 27 21MME/day among Cannabis 28 patients Clinician 29 who s 30 ceased 31 opioids 32 complete 33 ly 34 Vigil et al., Retrospectiv Enrolled in 100% CNCP 56.3 36% Maximum daily 52 weeks NR NR Universit 35 2017 e chart MCP [n=37], (90% back (11.8) dosage y of New 36 (US)58 review not enrolled pain) of < Mexico 37 [n=29] 200MME/day Medical 38 (100%) (enrolled in Cannabis 39 MCP: mean Research 40 24MME/day±2 Fund 41 3; not enrolled 42 43 24 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

Page 27 of 88 BMJ Open

1 2 3 in MCP: mean 4 16MME/day±1 5 4) 6 Yassin et One-arm n=31 (100%) 100% CNCP 33.4 90% Receiving 26 weeks THC Patients were NR 7 al., 2019 observationa (fibromyalgi (12.3) Oxycodone 5 to CBD ratio permitted to 8 (Israel)59 l study a) mg three was 1:4; use 9 times/day 20 g/month for standardized 10 3 months, analgesic 11 increased up to therapy 12 For peer review only30 g/month at (duloxetine 30 13 the end of 6 mg once daily 14 months and Targin 15 5/2.5 mg 16 twice a day) 17 *CNCP: Chronic non-cancer pain; MCP: Medical Cannabis Program; MME: milligram morphine equivalent; FU: follow-up; NR: not 18 reported http://bmjopen.bmj.com/ 19 ¥ Based on the whole population including opioid users and non-users 20 £In Belgium, Bulgaria, Czech Republic, Estonia, Germany, Hungary, Latvia, Lithuania, Poland, Romania, the United Kingdom and 21 the United States 22 23 24 25 26 on October 1, 2021 by guest. Protected copyright. 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 25 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

BMJ Open Page 28 of 88

1 2 3 Table 2: GRADE Evidence Profile of medical cannabis or cannabinoids for patients with chronic pain prescribed long-term 4 5 opioid therapy 6 c d 7 # of studies # of FU Risk of Inconsistency Indirectness Imprecision Publication Treatment Overall a 2 b 8 Patients Duration bias (I , P-value) bias association certainty of 9 (Weeks) (95% CI) evidence 10 Opioid dose: morphine milligram equivalents (MME) per day 11 12 4 RCTs43-45 1,176 2 to 5 No serious No serious Very serious Serious Not WMD For peere reviewf onlyg 13 risk of bias inconsistency indirectness imprecision detected -3.4MME Very Low 14 [40%, P=0.15] (-12.7 to 5.8) 15 16 8 453 4 to 297 Serious risk Serious No serious No serious Not WMD Very low 17 Observational of bias h inconsistency indirectness imprecision detected -22.5MME 18 studies47, 48, 50, [visual http://bmjopen.bmj.com/ (-43.06 to - 19 51, 53-55, 58 inspection] 1.97) 20 Pain: 10 cm VAS for pain; lower is better; the MID = 1 cm 21 22 5 RCTs43-46 1,536 2 to 5 No serious No serious No serious No serious Not WMD -0.18 23 risk of bias inconsistency indirectness imprecision detected (-0.38 to High 24 [28%, P=0.20] 0.02) 25 Sleep disturbance: 10 cm VAS for sleep disturbance; lower is better; the MID= 1 cm 26 on October 1, 2021 by guest. Protected copyright. 27 5 RCTs43-46 1,536 2 to 5 No serious No serious No serious No serious Not WMD -0.22 High 28 risk of bias inconsistency indirectness imprecision detected (-0.39 to - 29 [0%, P=0.45] 0.06) 30 31 32 Nausea 33 43-46 34 4 RCTs 1330 2 to 5 Serious risk No serious No serious No serious Not RR 1.43 Moderate of bias inconsistency indirectness imprecision detected (1.04 to 1.96) 35 [0%, P=0.88] 36 37 Vomiting 38 39 40 41 42 43 26 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

Page 29 of 88 BMJ Open

1 2 3 4 RCTs43-46 1330 2 to 5 Serious risk No serious No serious No serious Not RR 1.5 Moderate 4 of bias inconsistency indirectness imprecision detected (1.01 to 2.24) 5 [0%, P=0.50] 6 Constipation 7 8 43, 45, 9 3 RCTs 1153 5 Serious risk No serious No serious Serious Not RR 0.85 Low 46 of bias i inconsistency indirectness imprecision g detected (0.54 to 1.35) 10 [0%, P=0.92] 11 12 For peer review only 13 WMD: weighted mean difference; RR: relative risk; 95% CI: 95% confidence interval; VAS: visual analogue scale; MID: minimally 14 important difference; FU: follow-up. 15 a We assessed risk of bias using a modified Cochrane risk of bias instrument. 16 b 2 17 Inconsistency refers to unexplained heterogeneity of results. For RCTs an I of 75-100% indicates that heterogeneity may be 18 considerable. We assessed heterogeneity of pooled observational studies through visual inspection of forest plots.http://bmjopen.bmj.com/ 19 c Indirectness results if the intervention, control, patients or outcomes are different from the research question under investigation. 20 d Serious imprecision refers to situations in which the confidence interval includes both benefit and harm (the 95%CI includes 1 MID). 21 e Some of the included RCTs were at high risk of bias, due to loss to follow-up (>20%); however, we did not rate down for risk of bias 22 as subgroup analysis showed no difference in treatment effect between trials at high and low risk of bias for missing outcome data 23 (test of interaction p= 0.758 and p=0.623 for opioid dose reduction and pain respectively). 24 f 25 Downgraded twice due to indirectness since all trials instructed participants to maintain their opioid dose during the study period. g 26 The 95%CI around the WMD includes no effect. on October 1, 2021 by guest. Protected copyright. 27 h Studies are based on non-representative samples. 28 i Most RCTs were at high risk of bias due to loss to follow-up (>20%). 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 27 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 30 of 88

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 References: 4 5 6 1. Schopflocher D, Taenzer P, Jovey R. The prevalence of chronic pain in Canada. Pain Res 7 8 Manag 2011;16(6):445-50. doi: 10.1155/2011/876306. 9 10 2. Schappert SM, Burt CW. Ambulatory care visits to physician offices, hospital outpatient 11 12 departments, and emergency departments: United States, 2001-02. Vital and Health 13 14 15 Statistics Series 13, Data from the National Health Survey 2006(159):1-66. 16 For peer review only 17 3. International Narcotics Control Board. Narcotic drugs: estimated world requirements for 18 19 2019: https://www.incb.org/documents/Narcotic-Drugs/Technical-Publications/2018/INCB- 20 21 22 Narcotics_Drugs_Technical_Publication_2018.pdf (Accessed: 02 Nov 2020). Vienna: 23 24 United Nations. 25 26 4. Busse JW, Wang L, Kamaleldin M, et al. Opioids for Chronic Noncancer Pain: A 27 28 29 Systematic Review and Meta-analysis. JAMA 2018;320(23):2448-60. doi: 30 31 10.1001/jama.2018.18472. 32 33 5. Gomes T, Greaves S, Martins D, et al. Latest Trends in Opioid-Related Deaths in Ontario: http://bmjopen.bmj.com/ 34 35 1991 to 2015. Toronto: Ontario Drug Policy Research Network, 2017. 36 37 38 6. Busse JW, Craigie S, Juurlink DN, et al. Guideline for opioid therapy and chronic noncancer 39 40 pain. CMAJ 2017;189(18):E659-e66. doi: 10.1503/cmaj.170363. 41 on October 1, 2021 by guest. Protected copyright. 42 7. Gomes T, Mamdani MM, Paterson JM, Dhalla IA, Juurlink DN. Trends in high-dose opioid 43 44 45 prescribing in Canada. Can Fam Physician 2014;60(9):826-32. 46 47 8. Bedson J, Chen Y, Ashworth J, et al. Risk of adverse events in patients prescribed long-term 48 49 opioids: A cohort study in the UK Clinical Practice Research Datalink. Eur J Pain 50 51 52 2019;23(5):908-22. doi: 10.1002/ejp.1357. 53 54 55 56 57 58 28 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 31 of 88 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 9. Gwira Baumblatt JA, Wiedeman C, Dunn JR, et al. High-risk use by patients prescribed 4 5 6 opioids for pain and its role in overdose deaths. JAMA Intern Med 2014;174(5):796-801. 7 8 doi: 10.1001/jamainternmed.2013.12711. 9 10 10. Bohnert AS, Logan JE, Ganoczy D, Dowell D. A Detailed Exploration Into the Association 11 12 of Prescribed Opioid Dosage and Overdose Deaths Among Patients With Chronic Pain. Med 13 14 15 Care 2016;54(5):435-41. doi: 10.1097/mlr.0000000000000505. 16 For peer review only 17 11. Attal N, Mazaltarine G, Perrouin-Verbe B, Albert T. Chronic neuropathic pain management 18 19 in spinal cord injury patients. What is the efficacy of pharmacological treatments with a 20 21 22 general mode of administration? (oral, transdermal, intravenous). Ann Phys Rehabil Med 23 24 2009;52(2):124-41. doi: 10.1016/j.rehab.2008.12.011. 25 26 12. Livingston MD, Barnett TE, Delcher C, Wagenaar AC. Recreational Cannabis Legalization 27 28 29 and Opioid-Related Deaths in Colorado, 2000-2015. Am J Public Health 30 31 2017;107(11):1827-9. doi: 10.2105/ajph.2017.304059. 32 33 13. Bradford AC, Bradford WD. Medical Marijuana Laws Reduce Prescription Medication Use http://bmjopen.bmj.com/ 34 35 In Medicare Part D. Health Aff (Millwood) 2016;35(7):1230-6. doi: 36 37 38 10.1377/hlthaff.2015.1661. 39 40 14. Bachhuber MA, Saloner B, Cunningham CO, Barry CL. Medical cannabis laws and opioid 41 on October 1, 2021 by guest. Protected copyright. 42 analgesic overdose mortality in the United States, 1999-2010. JAMA Intern Med 43 44 45 2014;174(10):1668-73. doi: 10.1001/jamainternmed.2014.4005. 46 47 15. Boehnke KF, Litinas E, Clauw DJ. Medical Cannabis Use Is Associated With Decreased 48 49 Opiate Medication Use in a Retrospective Cross-Sectional Survey of Patients With Chronic 50 51 52 Pain. J Pain 2016;17(6):739-44. doi: 10.1016/j.jpain.2016.03.002. 53 54 55 56 57 58 29 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 32 of 88

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 16. Piper BJ, DeKeuster RM, Beals ML, et al. Substitution of medical cannabis for 4 5 6 pharmaceutical agents for pain, anxiety, and sleep. J Psychopharmacol 2017;31(5):569-75. 7 8 doi: 10.1177/0269881117699616. 9 10 17. Nielsen S, Sabioni P, Trigo JM, et al. Opioid-Sparing Effect of Cannabinoids: A Systematic 11 12 Review and Meta-Analysis. Neuropsychopharmacology 2017;42(9):1752-65. doi: 13 14 15 10.1038/npp.2017.51. 16 For peer review only 17 18. Busse JW, Vankrunkelsven P, Zeng L, et al. Medical cannabis for chronic pain: a clinical 18 19 practice guideline. BMJ 2020(submitted). 20 21 22 19. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in 23 24 epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in 25 26 Epidemiology (MOOSE) group. JAMA 2000;283(15):2008-12. doi: 27 28 29 10.1001/jama.283.15.2008. 30 31 20. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic 32 33 reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and http://bmjopen.bmj.com/ 34 35 elaboration. BMJ 2009;339:b2700. doi: 10.1136/bmj.b2700. 36 37 38 21. Classification of chronic pain. Descriptions of chronic pain syndromes and definitions of 39 40 pain terms. Prepared by the International Association for the Study of Pain, Subcommittee 41 on October 1, 2021 by guest. Protected copyright. 42 on Taxonomy. Pain Suppl 1986;3:S1-226. 43 44 45 22. Lintzeris N, Driels J, Elias N, et al. Medicinal , 2016: the Cannabis as 46 47 Medicine Survey (CAMS-16). Med J Aust 2018;209(5):211-6. 48 49 23. ElSohly MA, Ross SA, Mehmedic Z, et al. Potency trends of delta9-THC and other 50 51 52 cannabinoids in confiscated marijuana from 1980-1997. J Forensic Sci 2000;45(1):24-30. 53 54 55 56 57 58 30 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 33 of 88 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 24. Reeves BC, Deeks JJ, Higgins JPT, et al. Chapter 24: Including non-randomized studies on 4 5 6 intervention effects. In: Higgins JPT TJ, Chandler J, Cumpston M, Li T, Page MJ, Welch 7 8 VA (editors). editor: Cochrane Handbook for Systematic Reviews of Interventions version 9 10 6.1. Available from www.training.cochrane.org/handbook; 2020. 11 12 25. Dworkin RH, Turk DC, Farrar JT, et al. Core outcome measures for chronic pain clinical 13 14 15 trials: IMMPACT recommendations. Pain 2005;113(1-2):9-19. doi: 16 For peer review only 17 10.1016/j.pain.2004.09.012. 18 19 26. Goshua A, Craigie S, Guyatt GH, et al. Patient Values and Preferences Regarding Opioids 20 21 22 for Chronic Noncancer Pain: A Systematic Review. Pain Med 2018;19(12):2469-80. doi: 23 24 10.1093/pm/pnx274. 25 26 27. Akl EA, Sun X, Busse JW, et al. Specific instructions for estimating unclearly reported 27 28 29 blinding status in randomized trials were reliable and valid. J Clin Epidemiol 30 31 2012;65(3):262-7. doi: 10.1016/j.jclinepi.2011.04.015. 32 33 28. Higgins JP, Altman DG, Gøtzsche PC, et al. The Cochrane Collaboration's tool for assessing http://bmjopen.bmj.com/ 34 35 risk of bias in randomised trials. BMJ 2011;343:d5928. doi: 10.1136/bmj.d5928. 36 37 38 29. Tool to Assess Risk of Bias. CLARITY Group at McMaster University. Access 39 40 www.evidencepartners.com/resources/methodological-resources/: Evidence partner; 2020 41 on October 1, 2021 by guest. Protected copyright. 42 [cited 2020 04 November, 2020]. 43 44 45 30. Landis JR, Koch GG. The measurement of observer agreement for categorical data. 46 47 Biometrics 1977;33(1):159-74. 48 49 31. Farrar JT, Young JP, Jr., LaMoreaux L, Werth JL, Poole RM. Clinical importance of 50 51 52 changes in chronic pain intensity measured on an 11-point numerical pain rating scale. Pain 53 54 2001;94(2):149-58. doi: 10.1016/s0304-3959(01)00349-9. 55 56 57 58 31 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 34 of 88

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 32. Zisapel N, Nir T. Determination of the minimal clinically significant difference on a patient 4 5 6 visual analog sleep quality scale. J Sleep Res 2003;12(4):291-8. doi: 10.1046/j.0962- 7 8 1105.2003.00365.x. 9 10 33. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7(3):177- 11 12 88. doi: 10.1016/0197-2456(86)90046-2. 13 14 15 34. Wan X, Wang W, Liu J, Tong T. Estimating the sample mean and standard deviation from 16 For peer review only 17 the sample size, median, range and/or interquartile range. BMC Med Res Methodol 18 19 2014;14:135. doi: 10.1186/1471-2288-14-135. 20 21 22 35. Ma J, Liu W, Hunter A, Zhang W. Performing meta-analysis with incomplete statistical 23 24 information in clinical trials. BMC Med Res Methodol 2008;8:56. doi: 10.1186/1471-2288- 25 26 8-56. 27 28 29 36. Rücker G, Cates CJ, Schwarzer G. Methods for including information from multi-arm trials 30 31 in pairwise meta-analysis. Res Synth Methods 2017;8(4):392-403. doi: 10.1002/jrsm.1259. 32 33 37. Rücker G, Schwarzer G, Carpenter JR, Schumacher M. Undue reliance on I(2) in assessing http://bmjopen.bmj.com/ 34 35 heterogeneity may mislead. BMC Med Res Methodol 2008;8:79. doi: 10.1186/1471-2288-8- 36 37 38 79. 39 40 38. Guyatt GH, Oxman AD, Vist GE, et al. GRADE: an emerging consensus on rating quality of 41 on October 1, 2021 by guest. Protected copyright. 42 evidence and strength of recommendations. BMJ 2008;336(7650):924-6. doi: 43 44 45 10.1136/bmj.39489.470347.AD. 46 47 39. Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, Welch VA (editors). 48 49 Cochrane Handbook for Systematic Reviews of Interventions version 6.0 (updated July 50 51 52 2019). Cochrane, 2019. Available from www.training.cochrane.org/handbook. 53 54 55 56 57 58 32 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 35 of 88 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 40. Guyatt GH, Oxman AD, Kunz R, et al. GRADE guidelines: 8. Rating the quality of 4 5 6 evidence--indirectness. J Clin Epidemiol 2011;64(12):1303-10. doi: 7 8 10.1016/j.jclinepi.2011.04.014. 9 10 41. Guyatt GH, Oxman AD, Sultan S, et al. GRADE guidelines: 9. Rating up the quality of 11 12 evidence. J Clin Epidemiol 2011;64(12):1311-6. doi: 10.1016/j.jclinepi.2011.06.004. 13 14 15 42. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a 16 For peer review only 17 simple, graphical test. BMJ 1997;315(7109):629-34. doi: 10.1136/bmj.315.7109.629. 18 19 43. Fallon MT, Albert Lux E, McQuade R, et al. Sativex oromucosal spray as adjunctive therapy 20 21 22 in advanced cancer patients with chronic pain unalleviated by optimized opioid therapy: two 23 24 double-blind, randomized, placebo-controlled phase 3 studies. Br J Pain 2017;11(3):119-33. 25 26 doi: 10.1177/2049463717710042. 27 28 29 44. Johnson JR, Burnell-Nugent M, Lossignol D, et al. Multicenter, double-blind, randomized, 30 31 placebo-controlled, parallel-group study of the efficacy, safety, and tolerability of 32 33 THC:CBD extract and THC extract in patients with intractable cancer-related pain. J Pain http://bmjopen.bmj.com/ 34 35 Symptom Manage 2010;39(2):167-79. doi: 10.1016/j.jpainsymman.2009.06.008. 36 37 38 45. Lichtman AH, Lux EA, McQuade R, et al. Results of a Double-Blind, Randomized, 39 40 Placebo-Controlled Study of Nabiximols Oromucosal Spray as an Adjunctive Therapy in 41 on October 1, 2021 by guest. Protected copyright. 42 Advanced Cancer Patients with Chronic Uncontrolled Pain. J Pain Symptom Manage 43 44 45 2018;55(2):179-88.e1. doi: 10.1016/j.jpainsymman.2017.09.001. 46 47 46. Portenoy RK, Ganae-Motan ED, Allende S, et al. Nabiximols for opioid-treated cancer 48 49 patients with poorly-controlled chronic pain: a randomized, placebo-controlled, graded-dose 50 51 52 trial. J Pain 2012;13(5):438-49. doi: 10.1016/j.jpain.2012.01.003. 53 54 55 56 57 58 33 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 36 of 88

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 47. Barlowe TS, Koliani-Pace JL, Smith KD, Gordon SR, Gardner TB. Effects of Medical 4 5 6 Cannabis on Use of Opioids and Hospital Visits by Patients With Painful Chronic 7 8 Pancreatitis. Clin Gastroenterol Hepatol 2019;17(12):2608-9.e1. doi: 9 10 10.1016/j.cgh.2019.01.018. 11 12 48. Bellnier T, Brown GW. Preliminary evaluation of the efficacy, safety, and costs associated 13 14 15 with the treatment of chronic pain with medical cannabis. Ment Health Clin 2018;8(3):110- 16 For peer review only 17 5. doi: 10.9740/mhc.2018.05.110. 18 19 49. Capano A, Weaver R, Burkman E. Evaluation of the effects of CBD hemp extract on opioid 20 21 22 use and quality of life indicators in chronic pain patients: a prospective cohort study. 23 24 Postgrad Med 2020;132(1):56-61. doi: 10.1080/00325481.2019.1685298. 25 26 50. Haroutounian S, Ratz Y, Ginosar Y, et al. The Effect of Medicinal Cannabis on Pain and 27 28 29 Quality-of-Life Outcomes in Chronic Pain: A Prospective Open-label Study. Clin J Pain 30 31 2016;32(12):1036-43. doi: 10.1097/ajp.0000000000000364. 32 33 51. Maida V, Ennis M, Irani S, Corbo M, Dolzhykov M. Adjunctive nabilone in cancer pain and http://bmjopen.bmj.com/ 34 35 symptom management: a prospective observational study using propensity scoring. J 36 37 38 Support Oncol 2008;6(3):119-24. 39 40 52. Narang S, Gibson D, Wasan AD, et al. Efficacy of dronabinol as an adjuvant treatment for 41 on October 1, 2021 by guest. Protected copyright. 42 chronic pain patients on opioid therapy. J Pain 2008;9(3):254-64. doi: 43 44 45 10.1016/j.jpain.2007.10.018. 46 47 53. O'Connell M, Sandgren M, Frantzen L, Bower E, Erickson B. Medical Cannabis: Effects on 48 49 Opioid and Benzodiazepine Requirements for Pain Control. Ann Pharmacother 50 51 52 2019;53(11):1081-6. doi: 10.1177/1060028019854221. 53 54 55 56 57 58 34 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 37 of 88 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 54. Pritchard ER, Dayer L, Belz J, et al. Effect of cannabis on opioid use in patients with cancer 4 5 6 receiving palliative care. J Am Pharm Assoc (2003) 2020;60(1):244-7. doi: 7 8 10.1016/j.japh.2019.10.013. 9 10 55. Pawasarat IM, Schultz EM, Frisby JC, et al. The Efficacy of Medical Marijuana in the 11 12 Treatment of Cancer-Related Pain. J Palliat Med 2020;23(6):809-16. doi: 13 14 15 10.1089/jpm.2019.0374. 16 For peer review only 17 56. Rod K. A pilot study of a medical cannabis-opioid reduction program. Am J Psychiatry 18 19 Neurosci 2019;7(3):74. doi: 10.11648/j.ajpn.20190703.14. 20 21 22 57. Takakuwa KM, Hergenrather JY, Shofer FS, Schears RM. The Impact of Medical Cannabis 23 24 on Intermittent and Chronic Opioid Users with Back Pain: How Cannabis Diminished 25 26 Prescription Opioid Usage. Cannabis Cannabinoid Res 2020;5(3):263-70. doi: 27 28 29 10.1089/can.2019.0039. 30 31 58. Vigil JM, Stith SS, Adams IM, Reeve AP. Associations between medical cannabis and 32 33 prescription opioid use in chronic pain patients: A preliminary cohort study. PLoS One http://bmjopen.bmj.com/ 34 35 2017;12(11):e0187795. doi: 10.1371/journal.pone.0187795. 36 37 38 59. Yassin M, Oron A, Robinson D. Effect of adding medical cannabis to analgesic treatment in 39 40 patients with low back pain related to fibromyalgia: an observational cross-over single 41 on October 1, 2021 by guest. Protected copyright. 42 centre study. Clin Exp Rheumatol 2019;37 Suppl 116(1):13-20. 43 44 45 60. Okusanya BO, Asaolu IO, Ehiri JE, et al. Medical cannabis for the reduction of opioid 46 47 dosage in the treatment of non-cancer chronic pain: a systematic review. Syst Rev 48 49 2020;9(1):167. doi: 10.1186/s13643-020-01425-3. 50 51 52 53 54 55 56 57 58 35 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 38 of 88

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 61. Zaller N, Topletz A, Frater S, Yates G, Lally M. Profiles of medicinal cannabis patients 4 5 6 attending compassion centers in rhode island. J Psychoactive Drugs 2015;47(1):18-23. doi: 7 8 10.1080/02791072.2014.999901. 9 10 62. Hazekamp A, Ware MA, Muller-Vahl KR, Abrams D, Grotenhermen F. The medicinal use 11 12 of cannabis and cannabinoids--an international cross-sectional survey on administration 13 14 15 forms. J Psychoactive Drugs 2013;45(3):199-210. doi: 10.1080/02791072.2013.805976. 16 For peer review only 17 63. Lucas P, Walsh Z, Crosby K, et al. Substituting cannabis for prescription drugs, alcohol and 18 19 other substances among medical cannabis patients: The impact of contextual factors. Drug 20 21 22 Alcohol Rev 2016;35(3):326-33. doi: 10.1111/dar.12323. 23 24 64. Cooper ZD, Bedi G, Ramesh D, et al. Impact of co-administration of oxycodone and smoked 25 26 cannabis on analgesia and abuse liability. Neuropsychopharmacology 2018;43(10):2046-55. 27 28 29 doi: 10.1038/s41386-018-0011-2. 30 31 65. Ishida JH, Wong PO, Cohen BE, et al. Substitution of marijuana for opioids in a national 32 33 survey of US adults. PloS One 2019;14(10). doi: 10.1371/journal.pone.0222577. eCollection http://bmjopen.bmj.com/ 34 35 2019. 36 37 38 66. Corroon JM, Jr., Mischley LK, Sexton M. Cannabis as a substitute for prescription drugs - a 39 40 cross-sectional study. J Pain Res 2017;10:989-98. doi: 10.2147/jpr.s134330. 41 on October 1, 2021 by guest. Protected copyright. 42 67. Reiman A, Welty M, Solomon P. Cannabis as a Substitute for Opioid-Based Pain 43 44 45 Medication: Patient Self-Report. Cannabis Cannabinoid Res 2017;2(1):160-6. doi: 46 47 10.1089/can.2017.0012. 48 49 68. The National Institute for Health and Care Excellence (NICE) Cannabis-based medicinal 50 51 52 products (NG144). 2019(Available at: 53 54 55 56 57 58 36 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 39 of 88 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 https://www.nice.org.uk/guidance/ng144/chapter/Recommendations#chronic-pain [accessed 4 5 6 21 May 2021]). 7 8 69. Mulla SM, Maqbool A, Sivananthan L, et al. Reporting of IMMPACT-recommended core 9 10 outcome domains among trials assessing opioids for chronic non-cancer pain. Pain 11 12 2015;156(9):1615-9. doi: 10.1097/j.pain.0000000000000241. 13 14 15 70. Frank JW, Carey E, Nolan C, et al. Association Between Opioid Dose Reduction Against 16 For peer review only 17 Patients' Wishes and Change in Pain Severity. J Gen Intern Med 2020;35(Suppl 3):910-7. 18 19 doi: 10.1007/s11606-020-06294-z. 20 21 22 71. Busse JW, Juurlink D, Guyatt GH. Addressing the limitations of the CDC guideline for 23 24 prescribing opioids for chronic noncancer pain. CMAJ 2016;188(17-18):1210-1. doi: 25 26 10.1503/cmaj.161023. 27 28 29 72. Wang L, Hong PJ, May C, et al. Medical cannabis or cannabinoids for chronic pain: a 30 31 systematic review and meta-analysis of randomized clinical trials. BMJ 2021 (Accepted for 32 33 publication). http://bmjopen.bmj.com/ 34 35 73. Zeraatkar D, Cooper MA, Agarwal A, et al. Long-term and serious harms of medical 36 37 38 cannabis or cannabinoids for chronic pain: A systematic review of non-randomized studies. 39 40 BMJ Open 2021 (submitted). 41 on October 1, 2021 by guest. Protected copyright. 42 74. Zeng L, Lytvyn L, Wang X, et al. Values and preferences towards medical cannabis or 43 44 45 cannabinoids among patients with chronic pain: A mixed methods systematic review. BMJ 46 47 Open 2021 (accepted for publication). 48 49 50 51 52 53 54 55 56 57 58 37 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 40 of 88

1 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 2 3 4

5 Records identified through Additional records identified 6 database searching through other sources 7 (n = 5132) (n =1)

8 tification 9

10 Iden 11 12 13 Records after duplicates removed (n =4289) 14 15

16 For peer review only 17 18 Titles & abstracts screened Records excluded 19 (n=4289) (n=4165)

20 Screening 21 22 23 Full-text articles excluded 24

25 Full-text articles assessed with reasons (n =116): 26 for eligibility (n=133) 27 Not opioid users (n=64); Not 28 chronic pain (n=6); 29 Recreational cannabis use 30 (n=2); Medical cannabis not

31 Eligibility 32 Studies included in added to opioid (n=4); less 33 qualitative synthesis (n than two weeks’ follow-up http://bmjopen.bmj.com/ 34 =18 studies in 17 (n=1); Case-series≤20 (n=4); 35 publications) Pre-clinical studies (n=2); 36 Wrong study design (e.g. 37 38 cross-sectional; narrative review, commentary) (n=32); 39 40 Duplicate (n=1) 41 Studies included in on October 1, 2021 by guest. Protected copyright.

42 Included quantitative synthesis 43 (meta-analysis) 44 (n=15) 45 46 47 48 49 50 Figure 1: Study selection process in review of opioid-sparing effects of cannabis in chronic pain 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 41 of 88 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Supplementary Material 4 5 6 Opioid-sparing effects of cannabis for chronic pain: A systematic review and meta-analysis 7 of randomized and observational studies 8 9 10 11 12 13 14 Contents 15 16 Appendix A: LiteratureFor Search Strategiespeer ...... review only ..... 3 17 Supplement Table 1: Detailed guidance for risk of bias assessment RCTs ...... 9 18 19 Supplement Table 2: Detailed guidance for risk of bias assessment retrospective or prospective chart- 20 reviews with control group ...... 12 21 22 Supplement Table 3: Detailed guidance for risk of bias assessment retrospective or prospective chart- 23 reviews with no control group ...... 16 24 25 Supplement Table 4: Characteristics of Eligible studies ...... 18 26 27 Supplement Table 5: Risk of bias assessment for RCTs ...... 27 28 Supplement Table 6: Risk of bias assessments for chart reviews with control group ...... 28 29 30 Supplement Table 7: Risk of bias assessments for one-arm studies with no control group ...... 29 31 32 Table 8: Other reported outcomes in observational studies ...... 30 33 http://bmjopen.bmj.com/ 34 Sleep disturbance results from two observational studies ...... 30 35 Other reported outcomes in one observational study ...... 30 36 37 Table 9: GRADE evidence profile of cannabis adjuvant to opioids vs. opioid alone for physical function 38 among patients with chronic pain from 1 RCT ...... 31 39 40 Table 10: GRADE evidence profile of cannabis adjuvant to opioids vs. opioid alone for emotional function 41 among patients with chronic pain from 1 RCT ...... 31 on October 1, 2021 by guest. Protected copyright. 42 43 Supplement Table 11: Summary of adverse events among included observational studies* ...... 32 44 45 Supplement Figure 1: forest plot for oral morphine equivalence dose reduction among patients with 46 Chronic Pain who received cannabis adjuvant to opioids vs. opioid alone in RCTs ...... 34 47 48 Supplement Figure 2: Subgroup analysis for opioid dose reduction and risk of bias (high risk vs. low risk) 49 from 4 RCTs of Cannabis+opioids vs. placebo ...... 35 50 51 Supplement Figure 3: forest plot for oral morphine equivalence dose reduction among patients with 52 Chronic Pain who received cannabis adjuvant to opioids vs. opioid alone in observational studies ...... 36 53 54 Supplement Figure 4: forest plot for pain relief on a 10-cm Visual Analog Scale (VAS) among patients 55 with Chronic Pain who received cannabis adjuvant to opioids vs. opioid alone in RCTs ...... 37 56 57 58 1 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 42 of 88

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Supplement Figure 5: Subgroup analysis for pain relief on a 10-cm VAS and risk of bias (high risk vs. low 4 5 risk) from 5 RCTs of Cannabis+opioids vs. placebo ...... 38 6 Supplement Figure 6: forest plot for pain relief on a 10-cm VAS among patients with Chronic Pain who 7 8 received cannabis adjuvant to opioids vs. opioid alone in observational studies with no control group . 39 9 Supplement Figure 7: forest plot for sleep disturbance on a 10 cm VAS for sleep disturbance among 10 11 patients with Chronic Pain who received cannabis adjuvant to opioids vs. opioid alone in RCTs ...... 40 12 Supplement Figure 8: Subgroup analysis for sleep disturbance a 10-cm VAS for sleep disturbance and risk 13 14 of bias (high risk vs. low risk) from 5 RCTs of Cannabis+opioids vs. placebo ...... 41 15 Supplement Figure 9: Risk difference of nausea among patients with Chronic Pain who received cannabis 16 For peer review only 17 adjuvant to opioids vs. opioid alone in RCTs ...... 42 18 Supplement Figure 10: Relative Risk of nausea among patients with Chronic Pain who received cannabis 19 20 adjuvant to opioids vs. opioid alone in RCTs ...... 42 21 Supplement Figure 11: Relative Risk of vomiting among patients with Chronic Pain who received 22 23 cannabis adjuvant to opioids vs. opioid alone in RCTs ...... 43 24 Supplement Figure 12: Risk Difference of vomiting among patients with Chronic Pain who received 25 26 cannabis adjuvant to opioids vs. opioid alone in RCTs ...... 43 27 Supplement Figure 13: Relative Risk of constipation among patients with Chronic Pain who received 28 29 cannabis adjuvant to opioids vs. opioid alone in RCTs ...... 44 30 Supplement Figure 14: Risk difference of constipation among patients with Chronic Pain who received 31 32 cannabis adjuvant to opioids vs. opioid alone in RCTs ...... 44 33 Appendix B: Reference List of Eligible studies ...... 45 http://bmjopen.bmj.com/ 34 35 Technical Appendix ...... 47 36 37 38 39 40 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 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 43 of 88 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Appendix A: Literature Search Strategies 4 5 Database: OVID Medline Epub Ahead of Print, In-Process & Other Non-Indexed 6 Citations, Ovid MEDLINE(R) Daily and Ovid MEDLINE(R) 1946 to Present. 7 8 The search terminology included all types of chronic pain AND any kinds of cannabinoids: 9 ………………………………………………………………………………………………. 10 11 12 Database: OVID Medline Epub Ahead of Print, In-Process & Other Non-Indexed Citations, 13 14 Ovid MEDLINE(R) Daily and Ovid MEDLINE(R) 1946 to Present 15 Search Strategy: 16 ------For peer review only 17 18 1 exp Analgesics, Opioid/ (111496) 19 2 opioid*.mp. (112576) 20 3 (alfentanil or alphaprodine or beta-casomorphin$ or buprenorphine or 21 carfentanil or codeine or deltorphin or dextromethorphan or dezocine or 22 dihydrocodeine or dihydromorphine or enkephalin$ or ethylketocyclazocine or 23 ethylmorphine or etorphine or fentanyl or heroin or hydrocodone or hydromorphone or 24 ketobemidone or levorphanol or lofentanil or meperidine or meptazinol or methadone or 25 methadyl acetate or morphine or nalbuphine or opium or oxycodone or oxymorphone or 26 27 pentazocine or phenazocine or phenoperidine or pirinitramide or promedol or 28 propoxyphene or remifentanil or sufentanil or tilidine or tapentadol).mp. [mp=title, 29 abstract, original title, name of substance word, subject heading word, floating 30 sub-heading word, keyword heading word, organism supplementary concept word, protocol 31 supplementary concept word, rare disease supplementary concept word, unique 32 identifier, synonyms] (150565) 33 4 or/1-3 (207118) http://bmjopen.bmj.com/ 34 5 exp Narcotics/ (119511) 35 6 (adolonta or Anpec or Ardinex or Asimadoline or Alvimopam or amadol or 36 37 biodalgic or biokanol or Codinovo or contramal or Demerol or Dicodid or 38 Dihydrocodeinone or dihydromorphinone or dihydrohydroxycodeinone or dihydrone or 39 dilaudid or dinarkon or dolsin or dolosal or dolin or dolantin or dolargan or 40 dolcontral or duramorph or duromorph or duragesic or durogesic or eucodal or 41 Fedotzine or Fentanest or Fentora or Fortral or Hycodan or Hycon or Hydrocodone or on October 1, 2021 by guest. Protected copyright. 42 Hydrocodeinonebitartrate or hydromorphon or hydroxycodeinon or isocodeine or 43 isonipecain or jutadol or laudacon or l dromoran or levodroman or levorphan or levo- 44 dromoran or levodromoran or lexir or lidol or lydol or morfin or morfine or morphia 45 46 or morphin or morphinium or morphinene or morphium or ms contin or n-methylmorphine 47 or n methylmorphine or nobligan or numorphan or oramorph or oxycodeinon or oxiconum 48 or oxycone or oxycontin or palladone or pancodine or pethidine or phentanyl or 49 prontofort or robidone or skenan or sublimaze or sulfentanyl or sulfentanil or 50 sufenta or takadol or talwin or theocodin or tramadol or tramadolhameln or tramadolor 51 or tramadura or tramagetic or tramagit or tramake or tramal or tramex or tramundin or 52 trasedal or theradol or tiral or topalgic or tradol or tradolpuren or tradonal or 53 tralgiol or tramadorsch or tramadin or tramadoc or ultram or zamudol or zumalgic or 54 zydol or zytram).mp. [mp=title, abstract, original title, name of substance word, 55 56 subject heading word, floating sub-heading word, keyword heading word, organism 57 58 3 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 44 of 88

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 supplementary concept word, protocol supplementary concept word, rare disease 4 supplementary concept word, unique identifier, synonyms] (10373) 5 6 7 or/1-6 (213683) 7 Annotation: opioid block 8 8 (Cannabis or cannabinol or cannabinoid* or cannabidiol or bhang or cannador or 9 charas or ganja or ganjah or hashish or hemp or marihuana or marijuana or nabilone or 10 cesamet or cesametic or ajulemic acid or cannabichromene or cannabielsoin or 11 cannabigerol or tetrahydrocannabinol or dronabinol or levonantradol or nabiximols or 12 palmidrol or tetrahydrocannabinolic acid or tetrahydro cannabinol or marinol or 13 tetranabinex or sativex or endocannabinoid*).mp. [mp=title, abstract, original title, 14 name of substance word, subject heading word, floating sub-heading word, keyword 15 16 heading word, organismFor supplementary peer concept review word, protocol supplementaryonly concept 17 word, rare disease supplementary concept word, unique identifier, synonyms] (52087) 18 9 Cannabis/ (8573) 19 10 exp CANNABINOIDS/ (13258) 20 11 8 or 9 or 10 (52087) 21 Annotation: cannabis block 22 12 7 and 11 (6089) 23 Annotation: opioid and cannabis 24 25 13 (chronic adj4 pain*).mp. [mp=title, abstract, original title, name of 26 substance word, subject heading word, floating sub-heading word, keyword heading 27 word, organism supplementary concept word, protocol supplementary concept word, rare 28 disease supplementary concept word, unique identifier, synonyms] (65717) 29 14 Chronic Pain/ (12620) 30 15 exp Osteoarthritis/ (59676) 31 16 osteoarthrit*.mp. (84419) 32 17 osteo-arthritis.mp. (375) 33 http://bmjopen.bmj.com/ 34 18 exp Arthritis, Rheumatoid/ (109607) 35 19 exp Neuralgia/ (19415) 36 20 Diabetic Neuropathies/ (14247) 37 21 (neuropath* adj5 pain*).mp. [mp=title, abstract, original title, name of 38 substance word, subject heading word, floating sub-heading word, keyword heading 39 word, organism supplementary concept word, protocol supplementary concept word, rare 40 disease supplementary concept word, unique identifier, synonyms] (23043) on October 1, 2021 by guest. Protected copyright. 41 22 neuralg*.mp. (26154) 42 23 zoster.mp. (20386) 43 44 24 Irritable Bowel Syndrome/ (6748) 45 25 IBS.mp. (8435) 46 26 Migraine Disorders/ (24388) 47 27 migraine.mp. (37040) 48 28 Fibromyalgia/ (8088) 49 29 fibromyalg*.mp. (11178) 50 30 complex regional pain syndromes/ or exp causalgia/ or exp reflex sympathetic 51 dystrophy/ (5426) 52 53 31 Pain, Intractable/ (6126) 54 32 Phantom Limb/ (1816) 55 33 Hyperalgesia/ (11136) 56 57 58 4 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 45 of 88 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 34 exp back pain/ or exp failed back surgery syndrome/ or exp low back pain/ 4 (37369) 5 6 35 radiculopathy.mp. (8722) 7 36 musculoskeletal pain/ or headache/ (29687) 8 37 exp Headache Disorders/ (33178) 9 38 headache*.mp. (89612) 10 39 exp Temporomandibular Joint Disorders/ (16711) 11 40 whiplash.mp. or exp whiplash injury/ (3896) 12 41 exp Cumulative Trauma Disorders/ (13326) 13 42 exp Peripheral Nervous System Diseases/dt [Drug Therapy] (14079) 14 43 Pain Measurement/de [Drug Effects] (6594) 15 16 44 (backache* or backpain*For orpeer dorsalgi* or arthralgi*review or polyarthralgi* only or 17 arthrodyni* or myalgi* or fibromyalgi* or myodyni* or neuralgi* or ischialgi* or crps 18 or rachialgi*).ab,ti. (43072) 19 45 ((noncancer* or non-cancer* or back or discogen* or chronic* or recurrent or 20 persist* or bone or musculoskelet* or muscle* or skelet* or spinal or spine or 21 vertebra* or joint* or arthritis or Intestin* or neuropath* or neck or cervical* or 22 head or facial* or complex or radicular or cervicobrachi* or orofacial or somatic or 23 non-malign* or shoulder* or knee* or hip or hips) adj3 pain).mp. (206944) 24 25 46 exp Pain/ (379991) 26 47 pain*.mp. (745044) 27 48 or/13-47 (1122771) 28 49 12 and 48 (1034) 29 30 31 Database: Embase <1974 to 2019 September 04> 32 Search Strategy: 33 http://bmjopen.bmj.com/ 34 ------35 1 exp narcotic analgesic agent/ (317763) 36 2 (opioid* or opiate*).mp. [mp=title, abstract, heading word, drug trade name, 37 original title, device manufacturer, drug manufacturer, device trade name, keyword, 38 floating subheading word, candidate term word] (188237) 39 3 (alfentanil or alphaprodine or beta-casomorphin$ or buprenorphine or 40 carfentanil or codeine or deltorphin or dextromethorphan or dezocine or on October 1, 2021 by guest. Protected copyright. 41 dihydrocodeine or dihydromorphine or enkephalin$ or ethylketocyclazocine or 42 ethylmorphine or etorphine or fentanyl or heroin or hydrocodone or hydromorphone or 43 44 ketobemidone or levorphanol or lofentanil or meperidine or meptazinol or methadone or 45 methadyl acetate or morphine or nalbuphine or opium or oxycodone or oxymorphone or 46 pentazocine or phenazocine or phenoperidine or pirinitramide or promedol or 47 propoxyphene or remifentanil or sufentanil or tilidine or tapentadol).mp. (278150) 48 4 (adolonta or Anpec or Ardinex or Asimadoline or Alvimopam or amadol or 49 biodalgic or biokanol or Codinovo or contramal or Demerol or Dicodid or 50 Dihydrocodeinone or dihydromorphinone or dihydrohydroxycodeinone or dihydrone or 51 dilaudid or dinarkon or dolsin or dolosal or dolin or dolantin or dolargan or 52 53 dolcontral or duramorph or duromorph or duragesic or durogesic or eucodal or 54 Fedotzine or Fentanest or Fentora or Fortral or Hycodan or Hycon or Hydrocodone or 55 Hydrocodeinonebitartrate or hydromorphon or hydroxycodeinon or isocodeine or 56 isonipecain or jutadol or laudacon or l dromoran or levodroman or levorphan or levo- 57 58 5 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 46 of 88

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 dromoran or levodromoran or lexir or lidol or lydol or morfin or morfine or morphia 4 or morphin or morphinium or morphinene or morphium or ms contin or n-methylmorphine 5 6 or n methylmorphine or nobligan or numorphan or oramorph or oxycodeinon or oxiconum 7 or oxycone or oxycontin or palladone or pancodine or pethidine or phentanyl or 8 prontofort or robidone or skenan or sublimaze or sulfentanyl or sulfentanil or 9 sufenta or takadol or talwin or theocodin or tramadol or tramadolhameln or tramadolor 10 or tramadura or tramagetic or tramagit or tramake or tramal or tramex or tramundin or 11 trasedal or theradol or tiral or topalgic or tradol or tradolpuren or tradonal or 12 tralgiol or tramadorsch or tramadin or tramadoc or ultram or zamudol or zumalgic or 13 zydol or zytram).mp. [mp=title, abstract, heading word, drug trade name, original 14 title, device manufacturer, drug manufacturer, device trade name, keyword, floating 15 16 subheading word, candidateFor term peer word] (50642) review only 17 5 or/1-4 (403926) 18 6 exp cannabis/ (32390) 19 7 cannabinoid/ or cannabidiol/ or cannabinoid derivative/ or cannabinol/ or 20 cannabinol derivative/ or cannabis derivative/ or delta8 tetrahydrocannabinol/ or 21 delta8 tetrahydrocannabinol derivative/ or "delta9(11) tetrahydrocannabinol"/ or 22 dronabinol/ or medical cannabis/ or nabiximols/ or tetrahydrocannabinol/ or 23 tetrahydrocannabinol derivative/ or tetrahydrocannabinolic acid/ (26180) 24 25 8 (Cannabis or cannabinol or cannabidiol or bhang or cannador or charas or ganja 26 or ganjah or hashish or hemp or marihuana or marijuana or nabilone or cesamet or 27 cesametic or ajulemic acid or cannabichromene or cannabielsoin or cannabigerol or 28 tetrahydrocannabinol or dronabinol or levonantradol or nabiximols or palmidrol or 29 tetrahydrocannabinolic acid or tetrahydro cannabinol or marinol or tetranabinex or 30 sativex or endocannabinoid*).mp. [mp=title, abstract, heading word, drug trade name, 31 original title, device manufacturer, drug manufacturer, device trade name, keyword, 32 floating subheading word, candidate term word] (69860) 33 http://bmjopen.bmj.com/ 34 9 6 or 7 or 8 (75281) 35 10 5 and 9 (16412) 36 11 (chronic adj4 pain*).mp. [mp=title, abstract, heading word, drug trade name, 37 original title, device manufacturer, drug manufacturer, device trade name, keyword, 38 floating subheading word, candidate term word] (109897) 39 12 chronic pain/ (57642) 40 13 exp osteoarthritis/ (122475) on October 1, 2021 by guest. Protected copyright. 41 14 osteoarthrit*.mp. (136019) 42 15 osteo-arthritis.mp. (424) 43 44 16 degenerative arthrit*.mp. (1563) 45 17 exp rheumatoid arthritis/ (194747) 46 18 exp neuralgia/ (99958) 47 19 diabetic neuropathy/ (22699) 48 20 (neuropath* adj5 (pain* or diabet*)).mp. (71799) 49 21 neuralg*.mp. (29200) 50 22 zoster.mp. (36684) 51 23 irritable colon/ (24792) 52 53 24 (Irritable Bowel Syndrome or IBS).mp. [mp=title, abstract, heading word, drug 54 trade name, original title, device manufacturer, drug manufacturer, device trade 55 name, keyword, floating subheading word, candidate term word] (24025) 56 25 exp migraine/ (60235) 57 58 6 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 47 of 88 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 26 migraine.mp. (66593) 4 27 fibromyalgia/ (19402) 5 6 28 fibromyalg*.mp. (20958) 7 29 reflex sympathetic dystrophy.mp. (2356) 8 30 (complex regional pain syndromes or causalgia).mp. (1275) 9 31 intractable pain/ (4701) 10 32 phantom limb.mp. or agnosia/ or phantom pain/ or amputation stump/ (7388) 11 33 hyperalgesia/ (18711) 12 34 ((noncancer* or non-cancer*or chronic* or recurrent or persist* or non- 13 malign*) adj3 pain).mp. (27031) 14 35 exp backache/ (104042) 15 16 36 radiculopathy.mp.For or exp radiculopathy/peer review(37176) only 17 37 musculoskeletal pain/ (10292) 18 38 exp arthralgia/ (58208) 19 39 headache/ (204055) 20 40 headache*.mp. (264831) 21 41 temporomandibular joint disorder/ (13308) 22 42 ((TMJ or TMJD) and pain*).mp. (3648) 23 43 whiplash.mp. or whiplash injury/ (4815) 24 25 44 exp cumulative trauma disorder/ (20089) 26 45 exp pain/ (1249315) 27 46 pain*.mp. (1280762) 28 47 or/11-46 (1963522) 29 48 10 and 47 (3115) 30 31 Search Name: cannabis pain 32 Date Run: 05/09/2019 16:12:03 33 http://bmjopen.bmj.com/ 34 Comment: 35 36 ID Search Hits 37 #1 MeSH descriptor: [Cannabis] explode all trees 293 38 #2 MeSH descriptor: [Cannabinoids] explode all trees 743 39 #3 MeSH descriptor: [Endocannabinoids] explode all trees 46 40 #4 MeSH descriptor: [Endocannabinoids] explode all trees 46 on October 1, 2021 by guest. Protected copyright. 41 #5 (Cannabis or cannabinol or cannabinoid* or cannabidiol or bhang or cannador or charas or ganja 42 or ganjah or hashish or hemp or marihuana or marijuana or nabilone or cesamet or cesametic or 43 44 ajulemic acid or cannabichromene or cannabielsoin or cannabigerol or tetrahydrocannabinol or 45 dronabinol or levonantradol or nabiximols or palmidrol or tetrahydrocannabinolic acid or tetrahydro 46 cannabinol or marinol or tetranabinex or sativex or endocannabinoid*):ti,ab,kw (Word variations have 47 been searched) 4215 48 #6 #1 or #2 or #3 or #4 or #5 4215 49 #7 MeSH descriptor: [Pain] explode all trees 45094 50 #8 (pain*):ti,ab,kw (Word variations have been searched) 164064 51 #9 #7 or #8 169846 52 53 #10 #6 and #9 578 54 #11 [mh Osteoarthritis] or [mh ^"Arthritis, Rheumatoid"] or [mh Neuralgia] or [mh ^"Diabetic 55 Neuropathies"] or [mh ^"Irritable Bowel Syndrome"] or [mh ^"Migraine Disorders"] or [mh 56 Fibromyalgia] or [mh ^"complex regional pain syndromes"] or [mh causalgia] or [mh ^"reflex 57 58 7 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 48 of 88

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 sympathetic dystrophy"] or [mh ^"pain Intractable"] or [mh ^"Phantom Limb"] or [mh Hyperalgesia] or 4 [mh ^"back pain"] or [mh ^"failed back surgery syndrome"] or [mh ^"low back pain"] or [mh 5 6 Radiculopathy] or [mh ^"musculoskeletal pain"] or [mh headache] or [mh Arthralgia] or [mh ^"Headache 7 Disorders"] or [mh ^"Temporomandibular Joint Dysfunction Syndrome"] or [mh ^"whiplash injury"] or 8 [mh ^"Cumulative Trauma Disorders"] or [mh "Peripheral Nervous System Diseases"/DT] or [mh ^"Pain 9 Measurement"/DE] 28499 10 #12 (osteoarthrit* or osteo-arthritis or arthrit* or neuropath* or neuralgi* or zoster* or migraine* 11 or headache* or fibromyalgi* or causalgia or radiculopathy* or whiplash or backache* or backpain* or 12 dorsalgi* or arthralgi* or polyarthralgi* or arthrodyni* or myalgi* or myodyni* or ischialgi* or crps or 13 rachialgi*or TMJ or TMJD or IBS or crohn* or colitis* or enteritis* or ileitis*) 104465 14 #13 (irrita* or inflam*) near/4 (bowel or colon) 7249 15 16 #14 #11 or #12 orFor #13 peer113256 review only 17 #15 #6 and #14 in Trials 353 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 October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 8 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 49 of 88 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Characteristics of eligible studies and Risk of Bias Assessment 4 5 6 7 Supplement Table 1: Detailed guidance for risk of bias assessment RCTs 8 Domain Judgment 9 10 Random allocation concealment Definitely yes (low risk): used 11 12 central allocations (e.g. computer, 13 telephone) 14 15 Probably yes (low risk): 16 For peer reviewsequentially only numbered, opaque, 17 18 sealed envelopes; studies did not 19 provide enough information about 20 concealment approach; however, it 21 22 was placebo-control trial with 23 double blinded design. 24 25 Probably no (high risk): not 26 27 enough information was provided 28 and study was not blinded. 29 30 Definitely no (high risk): used any 31 unconcealed approach of allocation 32 33 (e.g. case record number, day of http://bmjopen.bmj.com/ 34 week, health-care decision). 35 36 Blinding of patients Definitely yes (low risk): explicitly 37 38 mentioned that patients were blinded 39 40 Probably yes (low risk): a placebo- 41 controlled double-blinded trial. on October 1, 2021 by guest. Protected copyright. 42 43 Probably no (high risk): no explicit 44 45 statement about blinding status and 46 not double-blinded placebo- 47 controlled trial. 48 49 Definitely no (high risk): explicitly 50 51 mentioned that patients were not 52 blinded. 53 54 55 56 57 58 9 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 50 of 88

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Blinding of health care providers Definitely yes (low risk): explicitly 4 5 mentioned that this group was 6 blinded. 7 8 Probably yes (low risk): mentioned 9 10 that it was a double-blinded study; 11 mentioned investigator were blinded. 12 13 Probably no (high risk) 14 15 Definitely no (high risk): explicitly 16 For peer reviewmentioned only that this group was not 17 blinded. 18 19 Blinding of data collector Definitely yes (low risk): explicitly 20 mentioned that this group was 21 blinded. 22 23 Probably yes (low risk): mentioned 24 25 that it was a double-blinded study; 26 mentioned investigator were blinded. 27 28 Probably no (high risk) 29 30 Definitely no (high risk): explicitly 31 32 mentioned that this group was not 33 blinded. http://bmjopen.bmj.com/ 34 Blinding of outcome assessor Definitely yes (low risk): explicitly 35 mentioned that this group was 36 37 blinded. 38 39 Probably yes (low risk): mentioned 40 that it was a double-blinded study. 41 on October 1, 2021 by guest. Protected copyright. 42 Probably no (high risk) 43 44 Definitely no (high risk): explicitly 45 46 mentioned that this group was not 47 blinded; open-blinded or unblended 48 trial. 49 50 51 Blinding data analyst Definitely yes (low risk): explicitly 52 mentioned that this group were 53 blinded 54 55 56 57 58 10 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 51 of 88 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Probably yes (low risk): 4 5 6 Probably no (high risk): no explicit 7 statement about blinding and only 8 mentioned double-blinded. 9 10 Definitely no (high risk): explicitly 11 12 mentioned that this group was not 13 blinded; open-blinded or unblended 14 trial. 15 16 For peer review only 17 Loss to follow-up Definitely yes: the retention rate was 18 at least 90% through the study. 19 20 Probably yes (low risk): the 21 retention rate approximately 80-89% 22 23 and loss to follow-up unlikely to be 24 related to the outcome, or missing 25 outcome data were balanced across 26 27 groups. 28 29 Probably no (high risk): the 30 retention rate approximately 80- 31 89%, however its rate likely to be 32 33 related to the loss to follow-up. http://bmjopen.bmj.com/ 34 35 Definitely no (high risk): the 36 retention rate was less than 80%. 37 Sample size We also considered the sample size 38 lower than 300 for continuous as 39 40 high risk of bias and rated down on 41 the basis of imprecision in GRADE on October 1, 2021 by guest. Protected copyright. 42 assessment. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 11 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 52 of 88

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Supplement Table 2: Detailed guidance for risk of bias assessment retrospective or 4 5 prospective chart-reviews with control group 6 Domain Judgment 7 8 1) Did the study match participants for all variables Definitely yes (low risk): 9 that are associated with the outcome of interest 10 studies that adjusted based on 11 or did the statistical analysis adjust for these all important covariates (This item queries how 12 prognostic variables? including age, sex, baseline 13 confident we are that the reported association or 14 lack thereof is not due to confounding). pain, baseline opioid dose, and 15 other disabilities. 16 For peer review only 17 Probably yes (low risk): 18 studies that adjusted at a 19 20 minimum for baseline pain and 21 baseline opioid dose. 22 23 Probably no (high risk): 24 25 studies that did not provide any 26 details about analysis method. 27 28 Definitely no (high risk): 29 30 Studies that did not adjust based 31 on baseline opioid dose or 32 33 baseline pain. http://bmjopen.bmj.com/ 34 35 2) Was selection of exposed and non-exposed Definitely yes (low risk): 36 cohorts drawn from the same population? (this Studies in which selection for 37 item queries whether participants who co-used participation is not dependent 38 cannabis and opioids or used opioids alone were 39 on exposure status (cannabis drawn from the same population) 40 and opioid co-use). 41 on October 1, 2021 by guest. Protected copyright. 42 Probably yes (low risk): 43 44 studies that did not provide 45 enough information about 46 recruitment to judge whether 47 48 recruitment into the study was 49 dependent on exposure status or 50 not. 51 52 53 Probably no (high risk): NA 54 55 56 57 58 12 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 53 of 88 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Definitely no (high risk): 4 5 studies that compared cannabis 6 and opioid co-users and non- 7 users from different cohort. 8 9 10 3) Can we be confident in the assessment of Definitely yes (low risk): if 11 exposure? (this item queries how confident we are study reported some 12 about the quantification of cannabis and opioids co- ascertainment methods for 13 use). 14 cannabis use (e.g. urine 15 analysis), or study prescribed 16 For peer review onlythe specific dose of medical 17 18 cannabis to the participants. 19 20 Probably yes (low risk): self- 21 report of cannabis use. 22 23 Probably no (high risk): when 24 study did not provide any details 25 26 about assessing exposure status. 27 28 Definitely no (high risk): 29 participants self-reported 30 31 cannabis usage only at baseline, 32 or exposure status not assessed 33 during the 4-weeks follow-up at http://bmjopen.bmj.com/ 34 35 least one time, or level of 36 cannabis usage was not similar 37 among participants. For 38 39 example, some studies allowed 40 patients to select the type or 41 dose of cannabis themselves. on October 1, 2021 by guest. Protected copyright. 42 43 4) Can we be confident in the assessment of the Definitely yes (low risk): when 44 45 presence or absence of prognostic factors? patients self-reported the 46 prognostic factors. 47 48 Probably yes (low risk): when 49 the method of assessment was 50 not reported, it was considered 51 as probably yes. 52 53 54 *Note that for this item, we are 55 only concerned with the 56 measurement of the prognostic 57 58 13 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 54 of 88

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 factors that mentioned in item 4 5 number 1 as minimum adjusted 6 variables (baseline pain 7 intensity and opioid dose). 8 5) Were co-interventions similar between groups? Definitely yes (low risk): study 9 (this item queries how similar are the use of other reported that co-intervention 10 pain killers (e.g. NSAIDs) between cannabis users 11 other than study intervention and non-users. 12 were limited during the study 13 period. 14 15 Probably yes (low risk): when 16 For peer review only 17 co-intervention usage was 18 approximately balanced 19 between both intervention and 20 21 control groups. 22 23 Probably no (high risk): when 24 study did not provide enough 25 26 information about other drugs 27 that participants may use. 28 29 Definitely no (high risk): when 30 participants were allowed to use 31 32 all other co-interventions that 33 could affect the outcome of the http://bmjopen.bmj.com/ 34 study. 35 36 37 6) Was the follow up of cohorts adequate? (This Definitely yes (low risk): the 38 item queries the risk of bias associated with loss to retention rate was at least 90% 39 follow-up and missing outcome data). through the study. 40 41 Probably yes (low risk): the on October 1, 2021 by guest. Protected copyright. 42 43 retention rate approximately 80- 44 89% and loss to follow-up 45 unlikely to be related to the 46 47 outcome. 48 49 Probably no (high risk): the 50 retention rate approximately 80- 51 89%, however its rate likely to 52 53 be related to the loss to follow- 54 up. For instance, if patients were 55 required to come to clinic for 56 57 58 14 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 55 of 88 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 outcome measurement, patients 4 5 who had poorer outcomes, or on 6 the other hand, patients who 7 were feeling better, may be less 8 9 likely to attend the clinic. 10 11 Loss to follow-up did not report 12 or could not estimate. 13 14 Definitely no (high risk): loss 15 16 For peer review onlyto follow-up more than 20%. 17 18 7) Can we be confident in the assessment of Definitely yes (low risk): study 19 outcome? (This item queries our confidence in the used a validated/reliable 20 accuracy of the measurement of the outcome). 21 measurement for pain 22 assessment (e.g. VAS, NRS); 23 reported opioid dose in a 24 morphine equivalence dose by 25 26 assessing patients’ medical or 27 prescription records. 28 29 Probably yes (low risk): NA 30 31 Probably no (high risk): when 32 33 study did not provide enough http://bmjopen.bmj.com/ 34 information about the outcome 35 measurement. 36 37 38 Definitely no (high risk): study 39 used non-validated/reliable 40 instrument. 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 15 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 56 of 88

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Supplement Table 3: Detailed guidance for risk of bias assessment retrospective or 4 5 prospective chart-reviews with no control group 6 Domain Judgment 7 8 Is the source population (sampling frame) Definitely yes (low risk): 9 10 representative of the general population? participants were selected from a 11 representative sample (e.g. national 12 population registry) 13 14 Probably yes (low risk): single 15 16 For peer reviewcommunity only center, however the 17 center was the only referral center 18 that provided cannabis legally to 19 20 participants. 21 22 Probably no (high risk): based on 23 the provided information source 24 25 population could not be defined. 26 27 Definitely no (high risk): sampling 28 from one center or clinic or hospital 29 or patients selected through using 30 31 convenience sampling. 32 33 Is the assessment of the outcome accurate both at Definitely yes (low risk): study used http://bmjopen.bmj.com/ 34 baseline and at follow-up? a validated/reliable measurement for 35 pain assessment (e.g. VAS, NRS); 36 37 reported opioid dose in a morphine 38 equivalence dose by assessing 39 patients’ medical or prescription 40 41 records. on October 1, 2021 by guest. Protected copyright. 42 43 Probably yes (low risk): NA 44 45 Probably no (high risk): when 46 study did not provide enough 47 48 information about the outcome 49 measurement. 50 51 Definitely no (high risk): used of 52 53 different instruments at different 54 follow-up intervals with concern of 55 56 57 58 16 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 57 of 88 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 accuracy of responses, or used 4 5 invalidated/reliable instruments. 6 7 Is there little missing data? Definitely yes (low risk): the 8 retention rate was at least 90% 9 10 through the study. 11 12 Probably yes (low risk): the 13 retention rate approximately 80-89% 14 and loss to follow-up unlikely to be 15 16 For peer reviewrelated only to the outcome. 17 18 Probably no (high risk): the 19 retention rate approximately 80- 20 21 89%, however its rate likely to be 22 related to the loss to follow-up. For 23 instance, if patients were required to 24 come to clinic for outcome 25 26 measurement, patients who had 27 poorer outcomes, or on the other 28 hand, patients who were feeling 29 30 better, may be less likely to attend 31 the clinic. 32 33 Loss to follow-up did not report or http://bmjopen.bmj.com/ 34 35 could not estimate. 36 37 Definitely no (high risk): loss to 38 follow-up more than 20%. 39 40 41 on October 1, 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 58 of 88

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Supplement Table 4: Characteristics of Eligible studies 4 5 6 7 Barlowe et al-20191 8 9 Study design Retrospective chart review. 10 11 Participants 34 chronic painful pancreatitis patients with chronic use of 12 13 opioids enrolled in a state therapeutic cannabis program 14 were compared to 19 non-enrolled patients. 15 16 Intervention (comparison)For peermedical review cannabis added to only opioids (no cannabis). 17 18 Follow-up Cohort of patients who enrolled into the program had 19 20 received cannabis therapy with a range from 34 to 297 21 weeks. 22 23 Funding source No industry funding reported. 24 Outcome -Reduction of opioid (calculated in average daily 25 26 intravenous [IV] morphine equivalence dosages) 27 28 29 30 Bellnier et al-20182 31 32 Study design One-arm observational study (before/after). http://bmjopen.bmj.com/ 33 Participants 29 patients with chronic pain who used opioids enrolled in a 34 35 state therapeutic cannabis program. 36 Intervention (comparison) medical cannabis added to opioids (no cannabis). 37 Follow-up 13 weeks 38 Funding source Not reported. 39 Outcome -Reduction of opioid (calculated in average daily 40

intravenous [IV] morphine equivalence dosages) on October 1, 2021 by guest. Protected copyright. 41 42 -Pain Quality Assessment Scale (PQAS) paroxysmal 43 domain 44 45 46 Capano et al-20203 47 48 Study design One-arm observational study (before/after). 49 131 patients with chronic pain who used opioids enrolled in 50 Participants 51 a pain clinic cannabis therapy. 52 Intervention (comparison) medical cannabis added to opioids (no cannabis). 53 Follow-up 8 weeks 54 Funding source Industry fund reported. 55 56 57 58 18 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 59 of 88 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Outcome - Reduction of opioid use (reported as percentage of patients 4 5 who reduced their opioid use after 8 weeks). 6 - Pain disability index 7 - Pittsburgh Sleep Quality Index 8 - Pain intensity and interference index (PEG) 9 10 11 Haroutounian et al-2016 4 12 13 14 Study design One-arm observational study (before/after). 15 16 Participants For peerChronic review non-cancer pain (14only individuals had pain due to 17 cancer) with a duration of 3 months or longer, and a lack of 18 satisfactory analgesic response or intolerable adverse effects 19 20 with at least 2 analgesics from 2 different drug classes at full 21 dose (Opioid user: N=73; 35%). 22 23 Intervention (comparison) The initial recommended medical cannabis dose was 20 24 g/mo added to opioids, which could be obtained as smoked 25 26 cannabis, baked cookies or oil taking from cannabis 27 dispensary centers. Cannabis could be titrated up to 3 times 28 a day until satisfactory pain relief was gained (before using 29 cannabis). 30 Follow-up 26 weeks. 31 32 No industry funding reported. Funding source http://bmjopen.bmj.com/ 33 34 Outcome - Reduction of opioid (calculated in median daily 35 intravenous [IV] morphine equivalence dosages among 36 opioid users). 37 38 39 40 5 41 Maida et al-2008 on October 1, 2021 by guest. Protected copyright. 42 43 Study design Prospective cohort study. 44 45 Participants 47 patients with chronic cancer pain who were opioid user 46 and treated with nabilone were compared to 65 non-treated 47 48 patients. 49 50 Intervention (comparison) nabilone added to opioids (no nabilone). 51 52 Follow-up 4 weeks. 53 54 Funding source Industry funding reported. 55 56 57 58 19 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 60 of 88

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Outcome -Reduction of opioid (calculated in average daily morphine 4 5 equivalence dosages); 6 7 -Pain reduction (Edmonton Symptom Assessment System 0: 8 no pain-10: most severe pain); 9 -anxiety, nausea, depression. 10 11 12 13 6 14 Narange et al-2008 15 16 Study design For peerPhase II:review One-arm observational only study (before/after). 17 18 Participants 30 patients with chronic non-cancer pain who were taking 19 opioids for a long time. 20 21 Intervention (comparison) The starting dose was 5mg of dronabinol twice daily and 22 23 titrated up to 20 mg 3 times a day added to opioids (before 24 using dronabinol). 25 26 Follow-up 4 weeks 27 28 29 Funding source Industry funding reported. 30 Outcome -Pain reduction (VAS 0: no pain-10: most severe pain); 31 -pain interfere with sleep (Brief pain inventory) 32 33 -sleep disturbance http://bmjopen.bmj.com/ 34 35 -adverse events including anxiety, dizziness, and inability to 36 concentrate. 37 38 39 40 O’Connell et al-20197 41 on October 1, 2021 by guest. Protected copyright. 42 Study design One-arm observational study (before/after). 43 44 45 Participants 77 mixed type of chronic non-cancer pain patients who used 46 opioids (96%) or benzodiazepines. 47 48 Intervention (comparison) Medical cannabis including THC, CBD products added to 49 opioid (before using cannabis) 50 51 52 Follow-up 26 weeks 53 54 Funding source No industry funding reported. 55 Outcome - Reduction of opioid (calculated in mean daily morphine 56 equivalence dosages among opioid users). 57 58 20 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 61 of 88 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 -pain reduction (VAS 0: no pain-10: most severe pain). 4 5 6 8 7 Pritchard-2019 8 9 Study design Retrospective chart review. 10 11 Participants 22 patients who had chronic cancer-related pain and used 12 opioids with the presence of THC in their urine drug 13 14 screening were compared to 61 patients with opioid use 15 only. 16 For peer review only 17 Intervention (comparison) medical cannabis added to opioids (no cannabis). 18 19 20 Follow-up 26 weeks. 21 22 Funding source No industry funding reported. 23 Outcome -Reduction of opioid (calculated in average daily 24 intravenous [IV] morphine equivalence dosages) 25 26 27 28 9 29 Pawasarat-2020 30 31 Study design Retrospective chart review. 32 33 Participants 137 chronic cancer-related pain patients with chronic use of http://bmjopen.bmj.com/ 34 opioids enrolled in a State of New Jersey Medicinal 35 36 Marijuana Program Registry were compared to 95 non- 37 enrolled patients. 38 39 Intervention (comparison) medical cannabis added to opioids (no cannabis). 40 41 Follow-up Between 36 and 52 weeks for enrolled patients and 24 on October 1, 2021 by guest. Protected copyright. 42 43 weeks for non-enrolled patients. 44 45 Funding source No industry funding reported. 46 Outcome -Reduction of opioid (calculated in average daily 47 intravenous [IV] morphine equivalence dosages) 48 49 50 -Pain reduction (VAS 0: no pain-10: most severe pain). 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 62 of 88

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Rod-201910 4 5 6 Study design One-arm observational study (before/after). 7 8 Participants 600 of chronic pain patients who used opioids and indicated 9 they were prepared to reduce their opioid dose. 10 11 Intervention (comparison) Medical cannabis added to opioid (before using cannabis) 12 13 26 weeks 14 Follow-up 15 16 Funding source For peerNo industry review funding reported. only 17 Outcome - Reduction or cease of opioid use (reported as percentage of 18 patients who ceased or reduced their opioid use after 6 19 months). 20 21 22 11 23 Takakuwa et al-2020 24 25 Study design One-arm observational study (before/after). 26 27 Participants 61 of chronic non-cancer pain patients (low-back pain) who 28 used opioids. 29 30 Intervention (comparison) Medical cannabis added to opioid (before using cannabis) 31 32 33 Follow-up Median of 6.4 years among patients who ceased opioids http://bmjopen.bmj.com/ 34 completely 35 36 Funding source Industry funding reported. 37 Outcome - Reduction of opioid (calculated in median daily morphine 38 39 equivalence dosages among chronic and intermittent opioid 40 users). 41 on October 1, 2021 by guest. Protected copyright. 42 43 Vigil et al-2017 12 44 45 Study design Retrospective chart review. 46 47 Participants 37 habitual opioid using, severe CNCP patients enrolled in 48 49 the Medical Cannabis Program were compared to 29 non- 50 enrolled patients. 51 52 Intervention (comparison) Medical cannabis added to opioids (no cannabis). 53 54 Follow-up 52 weeks 55 56 57 58 22 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 63 of 88 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Funding source No industry funding reported. 4 5 Outcome -Cessation of opioid (defined as the absence of opioid 6 prescriptions activity during the last three months of 7 observation) 8 9 -Reduction of opioid (calculated in average daily 10 11 intravenous [IV] morphine equivalence dosages); 12 13 -Pain reduction only among cannabis users (VAS 0: no 14 pain-10: most severe pain); 15 16 For peer-Quality review of life (no effect; only good benefit; great benefit; 17 18 negative effect; and extremely negative effect of co- 19 prescription of cannabis on quality of life). 20 21 22 23 Yassin et al-201913 24 25 26 Study design One-arm observational study (before/after). 27 28 Participants 31 patients with fibromyalgia were treated for at least 12 29 months with 5 mg of oxycodone hydrochloride equivalent to 30 4.5 mg oxycodone and 2.5 mg naloxone hydrochloride twice 31 32 a day and duloxetine 30 mg once a day. 33 http://bmjopen.bmj.com/ 34 Intervention (comparison) 20 grams of smoked medical cannabis added to opioids 35 (before cannabis inhalation). 36 37 Follow-up 26 weeks 38 39 40 Funding source No industry funding reported. 41 Outcome -Pain reduction (VAS 0: no pain-10: most severe pain) on October 1, 2021 by guest. Protected copyright. 42 -Change in pain medication use in 5 categories: 43 1) increased doses, 2) stable dose through medical cannabis 44 therapy duration, 3) less than half reduction in medication 45 consumption, 4) more than half reduction in analgesic 46 47 consumption, 5) deceased analgesic consumption. 48 - Owestry Disability Index reduction (scale 0: no disability, 49 100: total disability) 50 51 52 53 54 55 56 57 58 23 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 64 of 88

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Johnson et al-201014 4 5 6 Study design Parallel, multi-center randomized double-blinded, placebo- 7 controlled trial. 8 9 Participants 177 patients with chronic cancer pain who were under 10 treatment by opioid regimen. 11 12 13 Intervention (comparison) tetrahydrocannabinol: cannabidiol (THC:CBD) extract 14 added to opioids (placebo) 15 16 Follow-up For peer2 weeks review only 17 18 Funding source Industry funding reported. 19 20 Outcome - Reduction of opioid (calculated in mean daily morphine 21 equivalence dosages) 22 -Pain reduction (VAS 0: no pain-10: most severe pain) 23 -Sleep disturbance (NRS 0: no disturbance-10: most severe 24 disturbance) 25 -Physical, emotional, role, and social functioning (QLQ- 26 C30) 27 28 -Nausea, vomiting, constipation. 29 30 31 Portenoy et al-201215 32 33 Study design Parallel, randomized double-blinded, placebo-controlled http://bmjopen.bmj.com/ 34 trial. 35 36 37 Participants 360 patients with chronic cancer pain who were under 38 treatment by opioid regimen. 39 40 Intervention (comparison) Nabiximols at a low dose (1–4 sprays/day), medium dose on October 1, 2021 by guest. Protected copyright. 41 (6–10 sprays/day), or high dose (11–16 sprays/day) added to 42 43 opioids-(placebo) 44 45 Follow-up 5 weeks 46 47 Funding source Industry funding reported. 48 Outcome - Reduction of opioid (calculated in mean daily morphine 49 50 equivalence dosages) 51 -Pain reduction (VAS 0: no pain-10: most severe pain) 52 -Sleep disturbance (NRS 0: no disturbance-10: most severe 53 disturbance) 54 - Nausea, vomiting, constipation. 55 56 57 58 24 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 65 of 88 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Fallon et al-2017-Study 116 4 5 6 Study design Parallel, multi-center randomized double-blinded, placebo- 7 controlled trial. 8 9 Participants 399 patients with chronic cancer pain who were under 10 treatment by opioid regimen. 11 12 13 Intervention (comparison) Sativex (Δ9-tetrahydrocannabinol (27 mg/mL): cannabidiol 14 15 (25 mg/mL) added to opioids (placebo) 16 For peer review only 17 Follow-up 5 weeks 18 19 Funding source Industry funding reported. 20 21 Outcome -Reduction of opioid (calculated in mean daily morphine 22 equivalence dosages) 23 -Pain reduction (VAS 0: no pain-10: most severe pain) 24 -Sleep disturbance (NRS 0: no disturbance-10: most severe 25 disturbance) 26 - Nausea, vomiting, constipation. 27 28 29 30 16 31 Fallon et al-2017-Study 2 32 33 Study design Parallel, multi-center randomized double-blinded, placebo- http://bmjopen.bmj.com/ 34 controlled trial. 35 36 Participants 206 patients with chronic cancer pain who were under 37 38 treatment by opioid regimen. 39 40 Intervention (comparison) Sativex (Δ9-tetrahydrocannabinol (27 mg/mL): cannabidiol 41 on October 1, 2021 by guest. Protected copyright. 42 (25 mg/mL)) added to opioids (placebo)-patients who 43 44 tolerated titrated dose of cannabis and showed an 45 improvement of at least 15% on pain NRS score randomized 46 into this study (randomized withdrawal design). 47 48 Follow-up 5 weeks 49 50 51 Funding source Industry funding reported. 52 Outcome -Reduction of opioid (calculated in mean daily morphine 53 equivalence dosages) 54 -Pain reduction (VAS 0: no pain-10: most severe pain) 55 56 57 58 25 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 66 of 88

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 -Sleep disturbance (NRS 0: no disturbance-10: most severe 4 5 disturbance) 6 7 8 Lichtman et al-201717 9 10 Study design Parallel, multi-center randomized double-blinded, placebo- 11 controlled trial. 12 13 14 Participants 397 patients with chronic cancer pain who were under 15 treatment by opioid regimen. 16 For peer review only 17 Intervention (comparison) Nabiximols was added to opioids and was titrated the 18 maximum allowed daily dosage of 10 sprays per day 19 20 (placebo). 21 22 Follow-up 5 weeks 23 24 Funding source Industry funding reported. 25 Outcome -Reduction of opioid (calculated in mean daily morphine 26 27 equivalence dosages) 28 -Pain reduction (NRS 0: no pain-10: most severe pain) 29 -Sleep disturbance (NRS 0: no disturbance-10: most severe 30 disturbance) 31 32 33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 26 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 67 of 88 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Supplement Table 5: Risk of bias assessment for RCTs 4 5 6 Blinding of outcomeof Blinding Blinding of patients health Blinding of health

Study follow to Loss

7 Blinding Data of data collectorsdata

8 (author-year) concealment Blinding of Allocation p assessors (≤20%) analyst 9 roviders 10 care 11

12

-

up 13 14

15 16 Johnson et al- PYes PYes PYes PYes PYes PNo Plow- For peer review only € 17 2010 risk 18 19 Portenoy et al- DYes DYes PYes PYes PYes PNo Dhigh- £ 20 2012 risk 21 Fallon et al- PYes PYes PYes PYes PYes PNo Dhigh- 22 2017 risk¥ 23 Study 1 24 Fallon et al- PYes PYes PYes PYes PYes PNo Plow- 25 2017 risk€ 26 Study 2 27 Lichtman et PYes PYes PYes PYes PYes PNo Dhigh- 28 al-2017 risk¥ 29 30 31 32 * DYes: definitely yes; DNo: definitely no; PYes: probably yes; PNo: probably no 33 DYes/PYes= low risk of bias; DNo/PNo=high risk of bias. http://bmjopen.bmj.com/ £ 34 The rate of loss to follow-up was more than 27%. 35 ¥The rate of loss to follow-up was approximately 26%. 36 €The rate of loss to follow-up was approximately less than 20% 37 38 All RCTs used intention-to-treat (ITT) analysis, which included all randomized patients who had 39 40 at least one post-randomization efficacy endpoint into the analysis. 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 27 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 68 of 88

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 4 5 Supplement Table 6: Risk of bias assessments for chart reviews with control group 6 unexposed drawn from same Were the exposed and assessmentof exposure? Are we confident in the absence of prognostic assessment presenceof the or Can we confident be the in outcome assessment? Can we confident be the in W Study up? similar? Were the co different confounders? Did the authors adjust for Overall

7 adequate follow there as 8 risk of

9 bias 10

11 -interventions 12 13

14 15

16 For peer review - only 17 18 19 Vigil 2017 DYes DNo PYes PNo PYes PNo PYes High 20 21 Maida 2008 DYes DYes PYes DYes PNo PNo PYes High 22 23 Barlowe 2019 DYes DNo PYes DYes PNo PNo PNo High 24 25 Pritchard-2020 DYes DYes PYes DYes DNo PNo PNo High 26 27 Pawasarat-2020 DYes DNo PYes DYes DYes PNo PNo High 28 29 30 31 * DYes: definitely yes; DNo: definitely no; PYes: probably yes; PNo: probably no 32 DYes/PYes= low risk of bias; DNo/PNo=high risk of bias. 33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 28 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 69 of 88 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 4 5 Supplement Table 7: Risk of bias assessments for one-arm studies with no control 6 group 7 Study Is the source Is the assessment Is there little Overall risk 8 population (sampling of the outcome missing data? of bias 9 10 frame) representative accurate both at 11 of the general baseline and at 12 population? follow-up? 13 Haroutounian et DNo DYes PNo High 14 al-2016 15 16 Narang et al-2008 For peerDNo reviewDYes onlyPYes High 17 18 19 Yassin et al-2019 DNo DYes PYes High 20 21 22 O’Connell et al- DNo DYes PYes High 23 2019 24 25 Takakuwa et al- DNo DYes PYes High 26 2020 27 28 Vigil et al-2017 DNo PNo PYes High 29 30 Bellnier-2018 DNo DYes DYes High 31 32 Capano et al-2020 DNo DYes PNo High 33 http://bmjopen.bmj.com/ 34 Rod-2019 DNo PNo PNo High 35 36 37 38 39 * DYes: definitely yes; DNo: definitely no; PYes: probably yes; PNo: probably no 40 DYes/PYes= low risk of bias; DNo/PNo=high risk of bias. 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46

47 48 49 50 51 52 53 54 55 56 57 58 29 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 70 of 88

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Table 8: Other reported outcomes in observational studies 4 5 Sleep disturbance results from two observational studies 6 7 8 9 Capano et3 al assessed the effect The mean of Pittsburgh Sleep Very-low certainty 10 of adding CBD among patients Quality Index* decreased from evidence; p 11 with chronic pain who were 12.09±4.1 at baseline to 10.3±4.3 at value=0.03 12 opioid users for at least 1 year. the end of week 8. 13 14 15 Narang et al6 also evaluated the The sleep disturbance decreased Very low certainty 16 For peer review only 17 impact of adding dronabinol significantly at the end of week 4. evidence; p-value 18 among 30 patients taking opioids <0.01 19 for chronic pain. 20 21 22 *Ranges between 0 to 21 with the higher total score (referred to as global score) indicating 23 worse sleep quality. 24 25 26 27 28 29 30 Other reported outcomes in one observational study 31 Capano et3 al reported that pain disability index1 did not show a significant reduction, from 32 38.02±15.2 at baseline to 34.1±12.4 at week 4 (P-value=0.09). 33 2 http://bmjopen.bmj.com/ 34 Pain intensity and inference index reduced from 6.5±1.9 to 5.7±2 after 8 weeks’ follow 35 up (P-value=0.006). 1 36 Ranges from 0 to 70 (The higher the index the greater the person's disability due to pain). 37 2 38 PEG ranges from 0 to 10 (The higher the worse pain and interference). 39 40 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 30 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 71 of 88 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Table 9: GRADE evidence profile of cannabis adjuvant to opioids vs. opioid alone for 4 5 physical function among patients with chronic pain from 1 RCT 6 Outcome n of Follow- Mean difference Certainty Plain- 7 participants up of evidence language 8 (studies) (GRADE) summary 9 b 10 Physical Cannabis=118, Two THC: CBD vs. Moderate Adding 14 11 functioning placebo=59 weeks placebo: -4.23 cannabis to 12 (1 RCT14) (P=0.108) opioids 13 THC vs. placebo: probably does 14 -1.25 (P=0.631) not improve 15 physical 16 For peer review only functioning. 17 a b 18 In favor of placebo; Due to imprecision. 19 20 21 Table 10: GRADE evidence profile of cannabis adjuvant to opioids vs. opioid alone for 22 emotional function among patients with chronic pain from 1 RCT 23 24 Outcome n of Follow- Mean Certainty of Plain-language 25 participants up difference evidence summary 26 (studies) (GRADE) 27 Emotional Cannabis=118, Two THC: CBD vs. Moderateb Adding 28 functioning14 placebo=59 weeks placebo: 6.73 cannabis to 29 14 30 (1 RCT ) (P=0.084) opioids 31 THC vs. probably does 32 placebo: not improve 33 5.22 (P=0.174) emotional http://bmjopen.bmj.com/ 34 functioning. 35 a In favor of cannabis; b Due to imprecision. 36 37 38 39 40 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 31 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 72 of 88

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Supplement Table 11: Summary of adverse events among included observational 4 5 studies* 6 Study Method of Adverse events reported 7 8 assessment 9 10 11 Haroutounian et al4 Self-reported. Two participants discontinued treatment 12 13 due to serious side effects. 14 15 16 Maida et al5 ForSelf peer-reported reviewAnxiety only(P=0.028), nausea (P<0.001), and 17 18 distress (P=0.021) were decreased 19 20 21 significantly among patients who used 22 23 nabilone in comparison to patients who did 24 25 not use it. 26 27 6 28 Narang et al Self-reported (29-item Phase II: Dry mouth, tiredness (both 29 30 symptom Side Effect P<0.0001), abnormal thinking, anxiety, 31 32 Checklist). facial flushing, eye irritation, headache, 33 http://bmjopen.bmj.com/ 34 35 and ringing in the ears, and drowsiness (P< 36 37 0.05) showed a significantly higher 38 39 occurrence at the 20 mg dronabinol dose 40 on October 1, 2021 by guest. Protected copyright. 41 compared with placebo. 42 43 44 -Dry mouth, difficulty speaking, 45 46 forgetfulness, confusion, dizziness, and 47 48 euphoria were more occurred in both 49 50 51 treatment group versus placebo (P= 0.01) 52 53 54 55 56 57 58 32 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 73 of 88 BMJ Open

BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 1 2 3 Vigil et al12 Self-reported. No respondents reported any serious side 4 5 6 effects from cannabis use (only 9% of 7 8 patients reported cannabis affected 9 10 negatively their concentration). 11 12 13 13 Yassin et al Self-reported Mostly mild adverse events were reported 14 15 (e.g. red eye, sore throat, increase 16 For peer review only 17 appetite); only 6 patients out of withdrew 18 19 20 due to the side effects in non-cannabis 21 22 group. 23 24 *O’Connell et al7, Barlowe et al1, Rod 2019, and Takakuwa et al11 did not report adverse events. 25 26 27 28 29 30 31 32 33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on October 1, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 33 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 74 of 88

1 2 3 Additional results tables and figures 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 10 % 11 12 Study WMD (95% CI) Weight 13 14 15 16 Fallon-2017 study1 -9.35 (-18.81, 0.12) 36.24 17 18 Fallon-2017 study2 For peer review only-7.11 (-23.92, 9.69) 19.91 19 20 Lichtman-2018 -0.34 (-8.26, 7.58) 40.91 21 22 Johnson-2010 37.90 (-33.80, 109.60) 1.61 23 24 Johnson-2010 68.30 (-11.08, 147.60) 1.32 25 26 Overall (I-squared = 40.1%, p = 0.154) -3.43 (-12.66, 5.80) 100.00 27 28 29 NOTE: Weights are from random effects analysis 30 31 -30 0 30 150 32 Favors Opioids+cannabis Favors opioid 33 34 35 36

37 Supplement Figure 1: forest plot for oral morphine equivalence dose reduction http://bmjopen.bmj.com/ 38 among patients with Chronic Pain who received cannabis adjuvant to opioids vs. 39 40 opioid alone in RCTs 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

34

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 75 of 88 BMJ Open

1 2 3 4 5 % BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 Weight 7 8 Study WMD (95% CI) (I-V) 9 10 High Risk 11 Fallon-2017 study1 -9.35 (-18.81, 0.12) 36.00 12 13 Lichtman-2018 -0.34 (-8.26, 7.58) 51.42 14 I-V Subtotal (I-squared = 51.2%, p = 0.152) -4.05 (-10.12, 2.02) 87.42 15 16 Low Risk 17 18 Fallon-2017 study2 For peer review only-7.11 (-23.92, 9.69) 11.42 19 Johnson-2010 37.90 (-33.07, 108.90) 0.64 20 Johnson-2010 68.30 (-10.92, 147.50) 0.51 21 I-V Subtotal (I-squared = 56.5%, p = 0.101) -1.74 (-17.75, 14.28) 12.58 22 23 24 Heterogeneity between groups: p = 0.791 25 I-V Overall (I-squared = 40.4%, p = 0.152) -3.76 (-9.44, 1.92) 100.00 26 D+L Overall -3.40 (-12.67, 5.86) 27 28 29 30 -30 0 30 150 31 Favors Opioids+cannabis Favors opioid 32 33 34 35 Supplement Figure 2: Subgroup analysis for opioid dose reduction and risk of 36

37 bias (high risk vs. low risk) from 4 RCTs of Cannabis+opioids vs. placebo http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

35

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 76 of 88

1 2 3 4 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 5 % 6 7 Study WMD (95% CI) Weight 8 9 10 Bellnier-2018 -105.10 (-147.68, -62.52) 11.71 11 12 Barlowe-2019 -63.70 (-176.67, 49.27) 2.90 13 14 Haroutounian-2016 -7.50 (-26.25, 11.25) 19.71 15 Maida-2008 -19.90 (-55.71, 15.91) 13.74 16 17 Pritchard-2020 88.52 (-32.80, 209.84) 2.56 18 For peer review only Pawasarat-2020 0.00 (-26.02, 26.02) 17.11 19 20 O’Connell-2019 -37.54 (-87.13, 12.05) 9.93 21 22 Vigil-2017 -12.00 (-21.92, -2.08) 22.35 23 Overall (I-squared = 70.5%, p = 0.001) -22.52 (-43.06, -1.97) 100.00 24 25 26 NOTE: Weights are from random effects analysis 27 -176 -100 0 50 28 Favors Cannabis+Opioids Favors Opioids 29 30 31 32 33 Supplement Figure 3: forest plot for oral morphine equivalence dose reduction 34 among patients with Chronic Pain who received cannabis adjuvant to opioids vs. 35 opioid alone in observational studies 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

36

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 77 of 88 BMJ Open

1 2 3 4 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 5 % 6 7 Study WMD (95% CI) Weight 8 9 10 Fallon-2017 study1 0.12 (-0.18, 0.42) 22.98 11 12 Fallon-2017 study2 -0.02 (-0.42, 0.38) 16.42 13 14 Lichtman-2018 -0.16 (-0.45, 0.12) 24.19 15 Portenoy-2012 -0.75 (-1.45, -0.04) 6.92 16 17 Portenoy-2012 -0.36 (-1.10, 0.38) 6.36 18 For peer review only 19 Portenoy-2012 -0.09 (-0.85, 0.67) 6.08 20 Johnson-2010 -0.67 (-1.27, -0.06) 8.92 21 22 Johnson-2010 -0.32 (-0.96, 0.32) 8.14 23 24 Overall (I-squared = 28.1%, p = 0.204) -0.18 (-0.38, 0.02) 100.00 25 26 NOTE: Weights are from random effects analysis 27 28 -1.5 -1 0 .5 1 29 Favors Opioids+cannabis Favors opioid 30 31 32 33 34 Supplement Figure 4: forest plot for pain relief on a 10-cm Visual Analog Scale 35 (VAS) among patients with Chronic Pain who received cannabis adjuvant to 36

37 opioids vs. opioid alone in RCTs http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

37

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 78 of 88

1 2 3 4 5 % BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 Weight 7 Study WMD (95% CI) (I-V) 8 9 High Risk 10 11 Fallon-2017 study1 0.12 (-0.18, 0.42) 27.49 12 Lichtman-2018 -0.16 (-0.45, 0.12) 30.46 13 Portenoy-2012 -0.75 (-1.45, -0.04)4.98 14 Portenoy-2012 -0.36 (-1.10, 0.38) 4.52 15 Portenoy-2012 -0.09 (-0.85, 0.67) 4.28 16 I-V Subtotal (I-squared = 33.1%, p = 0.201) -0.10 (-0.29, 0.08) 71.73 17 18 For peer review only 19 Low Risk 20 Fallon-2017 study2 -0.02 (-0.42, 0.38) 15.46 21 Johnson-2010 -0.67 (-1.27, -0.06)6.76 22 Johnson-2010 -0.32 (-0.96, 0.32) 6.04 23 I-V Subtotal (I-squared = 36.8%, p = 0.206) -0.24 (-0.54, 0.06) 28.27 24 25 Heterogeneity between groups: p = 0.440 26 27 I-V Overall (I-squared = 28.1%, p = 0.204) -0.14 (-0.30, 0.02) 100.00 28 D+L Overall -0.18 (-0.38, 0.02) 29 30 31 -1.5 -1 0 .5 1 32 Favors Opioids+cannabis Favors opioid 33 34 35 Supplement Figure 5: Subgroup analysis for pain relief on a 10-cm VAS and risk of 36

37 bias (high risk vs. low risk) from 5 RCTs of Cannabis+opioids vs. placebo http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

38

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 79 of 88 BMJ Open

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 % 7 Study WMD (95% CI) Weight 8 9 10 11 Narange-2009 -1.70 (-2.81, -0.59) 16.30 12 13 Vigil-2017 -3.30 (-4.04, -2.56) 16.77 14 15 O'Connell-2019 0.32 (-0.13, 0.77) 17.01 16 Yassin-2019 -4.80 (-5.68, -3.92) 16.61 17 18 For peer review only Maida-2008 -3.90 (-4.71, -3.09) 16.69 19 20 Bellnier-2018 -4.72 (-5.58, -3.86) 16.63 21 22 Overall (I-squared = 97.7%, p = 0.000) -3.01 (-5.02, -1.00) 100.00 23 24 25 NOTE: Weights are from random effects analysis 26 -4 -3 -2 -1 0 1 2 3 4 27 Favors Cannabis+Opioids Favors Opioids 28 29 30 31 Supplement Figure 6: forest plot for pain relief on a 10-cm VAS among patients 32 33 with Chronic Pain who received cannabis adjuvant to opioids vs. opioid alone in 34 observational studies with no control group 35 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

39

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 80 of 88

1 2 3 4 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 5 % 6 7 Study ES (95% CI) Weight 8 9 10 Fallon-2017 study1 0.06 (-0.28, 0.39) 24.61 11 Fallon-2017 study2 -0.31 (-0.67, 0.05) 21.31 12 13 Lichtman-2018 -0.34 (-0.64, -0.04) 30.69 14 15 Portenoy-2012 -0.88 (-1.68, -0.08) 4.28 16 Portenoy-2012 -0.33 (-1.13, 0.47) 4.28 17 18 Portenoy-2012 For peer review only-0.08 (-0.84, 0.68) 4.73 19 Johnson-2010 -0.31 (-1.07, 0.45) 4.78 20 21 Johnson-2010 0.02 (-0.70, 0.74) 5.33 22 Overall (I-squared = 0.0%, p = 0.449) -0.22 (-0.39, -0.06) 100.00 23 24 25 26 -3 -2 -1 0 1 2 3 27 Favors Opioids+cannabis Favors opioid 28 29 30 31 32 33 Supplement Figure 7: forest plot for sleep disturbance on a 10 cm VAS for sleep 34 disturbance among patients with Chronic Pain who received cannabis adjuvant to 35 opioids vs. opioid alone in RCTs 36

37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

40

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 81 of 88 BMJ Open

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

37 Supplement Figure 8: Subgroup analysis for sleep disturbance a 10-cm VAS for http://bmjopen.bmj.com/ 38 sleep disturbance and risk of bias (high risk vs. low risk) from 5 RCTs of 39 40 Cannabis+opioids vs. placebo 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

41

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 82 of 88

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 % 8 study RD (95% CI) Weight 9 10 11 12 Johnson-2010 0.02 (-0.06, 0.10) 17.97 13 14 Portenoy-2012 0.09 (0.00, 0.17) 16.36

15 Fallon-2017-study1 0.01 (-0.04, 0.07) 38.40 16 17 Lichtman-2018 0.05 (-0.02, 0.12) 27.26 18 For peer review only 19 Overall (I-squared = 0.0%, p = 0.477) 0.04 (0.00, 0.07) 100.00 20

21 NOTE: Weights are from random effects analysis 22 23 -.15 0 .15 .25 24 25 26 27 Supplement Figure 9: Risk difference of nausea among patients with Chronic Pain 28 29 who received cannabis adjuvant to opioids vs. opioid alone in RCTs 30 31 32 33 34 % 35 36 study RR (95% CI) Weight

37 http://bmjopen.bmj.com/ 38 39 Johnson-2010 1.25 (0.41, 3.82) 7.98 40 41 Portenoy-2012 1.67 (0.94, 2.96) 30.27 42 43 Fallon-2017-study1 1.18 (0.63, 2.23) 24.65 44

45 Lichtman-2018 1.47 (0.88, 2.47) 37.10 on October 1, 2021 by guest. Protected copyright. 46 47 Overall (I-squared = 0.0%, p = 0.875) 1.43 (1.04, 1.96) 100.00 48 49 NOTE: Weights are from random effects analysis 50 51 1 1.5 2 2.5 3 3.5 4 52 53 54 55 Supplement Figure 10: Relative Risk of nausea among patients with Chronic Pain 56 57 who received cannabis adjuvant to opioids vs. opioid alone in RCTs 58 59 60

42

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 83 of 88 BMJ Open

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 10 11 12 % 13 study RR (95% CI) Weight 14 15 16 17 Fallon-2017-study1 1.38 (0.69, 2.73) 33.90 18 For peer review only Lichtman-2018 1.22 (0.61, 2.48) 32.08 19 20 Portenoy-2012 2.04 (0.95, 4.37) 27.31 21 22 Johnson-2010 1.75 (0.38, 8.16) 6.72 23 24 Overall (I-squared = 0.0%, p = 0.791) 1.50 (1.01, 2.24) 100.00 25 26 NOTE: Weights are from random effects analysis 27 28 1 3 5 8 29 30 31 32 Supplement Figure 11: Relative Risk of vomiting among patients with Chronic 33 34 Pain who received cannabis adjuvant to opioids vs. opioid alone in RCTs 35 36

37 % http://bmjopen.bmj.com/ 38 study RD (95% CI) Weight 39 40 41 42 Fallon-2017-study1 0.02 (-0.03, 0.08) 30.00 43 Lichtman-2018 0.01 (-0.04, 0.07) 31.85 44

45 Portenoy-2012 0.08 (0.01, 0.15) 17.04 on October 1, 2021 by guest. Protected copyright. 46 47 Johnson-2010 0.03 (-0.04, 0.09) 21.11 48 Overall (I-squared = 0.0%, p = 0.498) 0.03 (0.00, 0.06) 100.00 49 50 51 NOTE: Weights are from random effects analysis

52 53 -.15 0 .15 .25 54 55 56 57 Supplement Figure 12: Risk Difference of vomiting among patients with Chronic 58 Pain who received cannabis adjuvant to opioids vs. opioid alone in RCTs 59 60

43

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 84 of 88

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 Supplement Figure 13: Relative Risk of constipation among patients with Chronic 27 28 Pain who received cannabis adjuvant to opioids vs. opioid alone in RCTs 29 30 31 32 33 % 34

35 study RD (95% CI) Weight 36

37 http://bmjopen.bmj.com/ 38 39 Lichtman-2018 -0.01 (-0.06, 0.04) 38.57 40 41 Fallon-2017-study 1 -0.02 (-0.06, 0.03) 40.16 42 43 Portenoy-2012 -0.00 (-0.07, 0.06) 21.27 44 Overall (I-squared = 0.0%, p = 0.947) -0.01 (-0.04, 0.02) 100.00 45 on October 1, 2021 by guest. Protected copyright. 46

47 NOTE: Weights are from random effects analysis 48

49 -.15 0 .15 50 51 52 53 Supplement Figure 14: Risk difference of constipation among patients with 54 Chronic Pain who received cannabis adjuvant to opioids vs. opioid alone in RCTs 55 56 57 58 59 60

44

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 85 of 88 BMJ Open

1 2 3 Appendix B: Reference List of Eligible studies 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 1. Barlowe TS, Koliani-Pace JL, Smith KD, Gordon SR, Gardner TB. Effects of Medical Cannabis 8 9 on Use of Opioids and Hospital Visits by Patients With Painful Chronic Pancreatitis. Clin 10 11 Gastroenterol Hepatol 2019;17(12):2608-9.e1. doi: 10.1016/j.cgh.2019.01.018. 12 13 2. Bellnier T, Brown GW. Preliminary evaluation of the efficacy, safety, and costs associated 14 15 with the treatment of chronic pain with medical cannabis. Ment Health Clin 2018;8(3):110-5. doi: 16 17 10.9740/mhc.2018.05.110. 18 For peer review only 19 3. Capano A, Weaver R, Burkman E. Evaluation of the effects of CBD hemp extract on opioid use 20 21 22 and quality of life indicators in chronic pain patients: a prospective cohort study. Postgrad Med 23 24 2020;132(1):56-61. doi: 10.1080/00325481.2019.1685298. 25 26 4. Haroutounian S, Ratz Y, Ginosar Y, et al. The Effect of Medicinal Cannabis on Pain and 27 28 Quality-of-Life Outcomes in Chronic Pain: A Prospective Open-label Study. Clin J Pain 29 30 2016;32(12):1036-43. doi: 10.1097/ajp.0000000000000364. 31 32 5. Maida V, Ennis M, Irani S, Corbo M, Dolzhykov M. Adjunctive nabilone in cancer pain and 33 34 symptom management: a prospective observational study using propensity scoring. J Support Oncol 35 36 2008;6(3):119-24. 37 http://bmjopen.bmj.com/ 38 6. Narang S, Gibson D, Wasan AD, et al. Efficacy of dronabinol as an adjuvant treatment for 39 40 41 chronic pain patients on opioid therapy. J Pain 2008;9(3):254-64. doi: 10.1016/j.jpain.2007.10.018. 42 43 7. O'Connell M, Sandgren M, Frantzen L, Bower E, Erickson B. Medical Cannabis: Effects on 44

45 Opioid and Benzodiazepine Requirements for Pain Control. Ann Pharmacother 2019;53(11):1081-6. on October 1, 2021 by guest. Protected copyright. 46 47 doi: 10.1177/1060028019854221. 48 49 8. Pritchard ER, Dayer L, Belz J, et al. Effect of cannabis on opioid use in patients with cancer 50 51 receiving palliative care. J Am Pharm Assoc (2003) 2020;60(1):244-7. doi: 52 53 10.1016/j.japh.2019.10.013. 54 55 9. Pawasarat IM, Schultz EM, Frisby JC, et al. The Efficacy of Medical Marijuana in the 56 57 Treatment of Cancer-Related Pain. J Palliat Med 2020;23(6):809-16. doi: 10.1089/jpm.2019.0374. 58 59 60

45

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 86 of 88

1 2 3 10. Rod K. A pilot study of a medical cannabis-opioid reduction program. Am J Psychiatry 4 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 5 Neurosci 2019;7(3):74. doi: 10.11648/j.ajpn.20190703.14. 6 7 8 11. Takakuwa KM, Hergenrather JY, Shofer FS, Schears RM. The Impact of Medical Cannabis 9 10 on Intermittent and Chronic Opioid Users with Back Pain: How Cannabis Diminished Prescription 11 12 Opioid Usage. Cannabis Cannabinoid Res 2020;5(3):263-70. doi: 10.1089/can.2019.0039. 13 14 12. Vigil JM, Stith SS, Adams IM, Reeve AP. Associations between medical cannabis and 15 16 prescription opioid use in chronic pain patients: A preliminary cohort study. PLoS One 17 18 2017;12(11):e0187795.For doi: 10.1371/journal.pone.0187795. peer review only 19 20 13. Yassin M, Oron A, Robinson D. Effect of adding medical cannabis to analgesic treatment in 21 22 patients with low back pain related to fibromyalgia: an observational cross-over single centre study. 23 24 Clin Exp Rheumatol 2019;37(Suppl 116):S13-S20. 25 26 14. Johnson JR, Burnell-Nugent M, Lossignol D, et al. Multicenter, double-blind, randomized, 27 28 29 placebo-controlled, parallel-group study of the efficacy, safety, and tolerability of THC:CBD extract 30 31 and THC extract in patients with intractable cancer-related pain. J Pain Symptom Manage 32 33 2010;39(2):167-79. doi: 10.1016/j.jpainsymman.2009.06.008. 34 35 15. Portenoy RK, Ganae-Motan ED, Allende S, et al. Nabiximols for opioid-treated cancer 36

37 patients with poorly-controlled chronic pain: a randomized, placebo-controlled, graded-dose trial. J http://bmjopen.bmj.com/ 38 39 Pain 2012;13(5):438-49. doi: 10.1016/j.jpain.2012.01.003. 40 41 16. Fallon MT, Albert Lux E, McQuade R, et al. Sativex oromucosal spray as adjunctive therapy 42 43 in advanced cancer patients with chronic pain unalleviated by optimized opioid therapy: two double- 44

45 on October 1, 2021 by guest. Protected copyright. blind, randomized, placebo-controlled phase 3 studies. Br J Pain 2017;11(3):119-33. doi: 46 47 48 10.1177/2049463717710042. 49 50 17. Lichtman AH, Lux EA, McQuade R, et al. Results of a Double-Blind, Randomized, Placebo- 51 52 Controlled Study of Nabiximols Oromucosal Spray as an Adjunctive Therapy in Advanced Cancer 53 54 Patients with Chronic Uncontrolled Pain. J Pain Symptom Manage 2018;55(2):179-88.e1. doi: 55 56 10.1016/j.jpainsymman.2017.09.001. 57 58 59 60

46

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 87 of 88 BMJ Open

1 2 3 4 5 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 Technical Appendix 9 This appendix provides additional details on two different methods of estimation, including 10 11 1) estimating the mean and standard deviation (SD) from sample size, median, and 12 interquartile range (IQR); 2) estimating missing SD (for two non-randomized studies 5,7) 13 14 using the available SD from other included studies. 15 16 17 1) Estimating the mean and standard deviation (SD) from sample size, median, and 18 For peer review only 19 IQR: 20 1) Pawasarat et al 2020 original reported data: median total morphine equivalent=45, 21 22 n=137, and IQR=135. 23 1 24 -Using Wan et al method produced: mean=60, SD=101 25 -Method recommended by Cochrane as sensitivity analysis: 26 27 28 29 30 31 32 q3-q1=IQR. This method produced SD=100. 33 34 35 2) Bellnier et al 2018 original reported data: median total morphine equivalent (before 36 adding cannabis) =79.94, range=0 to 450, median (after adding cannabis) =19.65; range 37 http://bmjopen.bmj.com/ 38 =0 to 150, n=29. 39 40 41 -Using Wan et al method produced: mean (before)=152.4, SD=111; mean (after)=47.3, 42 SD=37.0 43 44 3 45 -Using Cochrane approach (Hozo et al ): Mean (before)= 152.4, SD= 112.5; mean on October 1, 2021 by guest. Protected copyright. 46 47 (after)= 47.3, SD= 37.5 48 49 We finally included estimation by Wan et al method. The excel sheet including all 50 51 formula was provided by Wan et al in supplementary file of their article1. 52 53 54 2) Estimating missing SD using the available SD from other included studies: 55 1) Maida et al 2008 did not report SD around the mean at the end of follow-up for 56 57 pain intensity. Original reported data: mean (SD) before adding cannabis= 58 59 7.1(2.4); after adding cannabis mean=3 (missing) 60

47

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 88 of 88

1 2 3 2) Connell et al 2019 original reported data: mean (SD) before adding cannabis=6.25 4 5 (missing); mean after adding cannabis=6.57 (missing) BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from 6 7 8 9 We imputed missing SDs for these two studies from the given SDs related to other 10 five included studies using prognostic method that presented by Ma et al2: 11 12 13 14 15 16 17 18 Assume there areFor k + l trials peer altogether review where k trials are only with full given information 19 20 SEM: value for trial j (missing) with sample size: 21 22 nj: sample size for study with missing information. 23 SD (imputed) for first study= 1.51 24 25 SDs (imputed) for second study=1.76, 1.20 26 27 28 29 30 1 Wan X, Wang W, Liu J, et al. Estimating the sample mean and standard deviation from the sample size, 31 32 median, range and/or interquartile range. BMC medical research methodology 2014;14(1):135. 33 2 Ma J, Liu W, Hunter A, et al. Performing meta-analysis with incomplete statistical information in clinical 34 trials. BMC medical research methodology 2008;8(1):56. 35 36 3 Hozo, S.P., Djulbegovic, B. & Hozo, I. Estimating the mean and variance from the median, range, and the

37 http://bmjopen.bmj.com/ size of a sample. BMC Med Res Methodol 5, 13 (2005). https://doi.org/10.1186/1471-2288-5-13 38 39 40 41 42 43 44

45 on October 1, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

48

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

Page 89 of 88 BMJ Open Reported Section/topic # Checklist item 1 on page # 2 TITLE 3 4 Title 1 Identify the report as a systematic review, meta-analysis, or both. 1 5 ABSTRACT 6 7 Structured summary 2 Provide a structured summary including, as applicable: background; objectives; data sources; study eligibility criteria, 3-4 8 participants, and interventions; study appraisal and synthesis methods; results; limitations; conclusions and 9 implications of key findings; systematic review registration number. 10 INTRODUCTION 11 12 Rationale 3 DescribeFor the rationale peer for the review in thereview context of what is already only known. 5 13 Objectives 4 Provide an explicit statement of questions being addressed with reference to participants, interventions, comparisons, 5 14 outcomes, and study design (PICOS). 15 16 METHODS 17 Protocol and registration 5 Indicate if a review protocol exists, if and where it can be accessed (e.g., Web address), and, if available, provide 6 18 registration information including registration number. http://bmjopen.bmj.com/ 19 20 Eligibility criteria 6 Specify study characteristics (e.g., PICOS, length of follow-up) and report characteristics (e.g., years considered, 6 21 language, publication status) used as criteria for eligibility, giving rationale. 22 Information sources 7 Describe all information sources (e.g., databases with dates of coverage, contact with study authors to identify 6 23 additional studies) in the search and date last searched. 24 25 Search 8 Present full electronic search strategy for at least one database, including any limits used, such that it could be 6 26 repeated. on October 1, 2021 by guest. Protected copyright. 27 Study selection 9 State the process for selecting studies (i.e., screening, eligibility, included in systematic review, and, if applicable, 7 28 included in the meta-analysis). 29 Data collection process 10 Describe method of data extraction from reports (e.g., piloted forms, independently, in duplicate) and any processes 30 7 for obtaining and confirming data from investigators. 31 32 Data items 11 List and define all variables for which data were sought (e.g., PICOS, funding sources) and any assumptions and 7 33 simplifications made. 34 Risk of bias in individual 12 Describe methods used for assessing risk of bias of individual studies (including specification of whether this was 35 7-8 studies done at the study or outcome level), and how this information is to be used in any data synthesis. 36 37 Summary measures 13 State the principal summary measures (e.g., risk ratio, difference in means). 8 38 Synthesis of results 14 Describe the methods of handling data and combining results of studies, if done, including measures of consistency 39 8 (e.g., I2) for each meta-analysis. 40 41 Page 1 of 2 42 Reported 43 Section/topic # Checklist item on page # 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2020-047717 on 28 July 2021. Downloaded from

BMJ Open Page 90 of 88 Risk of bias across studies 15 Specify any assessment of risk of bias that may affect the cumulative evidence (e.g., publication bias, selective 9 reporting within studies). 1 2 Additional analyses 16 Describe methods of additional analyses (e.g., sensitivity or subgroup analyses, meta-regression), if done, indicating 9-10 3 which were pre-specified. 4 RESULTS 5 6 Study selection 17 Give numbers of studies screened, assessed for eligibility, and included in the review, with reasons for exclusions at 11 7 each stage, ideally with a flow diagram. 8 Study characteristics 18 For each study, present characteristics for which data were extracted (e.g., study size, PICOS, follow-up period) and 11 9 provide the citations. 10 11 Risk of bias within studies 19 Present data on risk of bias of each study and, if available, any outcome level assessment (see item 12). 12 12 Results of individual studies 20 For all outcomesFor considered peer (benefits orreview harms), present, for each only study: (a) simple summary data for each 12-14 13 intervention group (b) effect estimates and confidence intervals, ideally with a forest plot. 14 15 Synthesis of results 21 Present results of each meta-analysis done, including confidence intervals and measures of consistency. 12-14 16 Risk of bias across studies 22 Present results of any assessment of risk of bias across studies (see Item 15). 9 17 18 Additional analysis 23 Give results of additional analyses, if done (e.g., sensitivity or subgroup analyses, meta-regressionhttp://bmjopen.bmj.com/ [see Item 16]). 12-14 19 20 DISCUSSION 21 Summary of evidence 24 Summarize the main findings including the strength of evidence for each main outcome; consider their relevance to 15 22 key groups (e.g., healthcare providers, users, and policy makers). 23 Limitations 25 Discuss limitations at study and outcome level (e.g., risk of bias), and at review-level (e.g., incomplete retrieval of 15 24 identified research, reporting bias). 25 26 Conclusions 26 Provide a general interpretation of the results in the context of other evidence, and implications on October 1, 2021 by guest. Protected copyright. for future research. 16 27 28 FUNDING 29 Funding 27 Describe sources of funding for the systematic review and other support (e.g., supply of data); role of funders for the 17 30 systematic review. 31 32 33 Page 2 of 2 34 35 36 37 38 39 40 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60