To identify eligible studies for the living review, automated searches on multiple electronic databases are run on a biweekly basis. Distiller^® SR software is used to assist in managing abstract and full-text review, and EndNote™ software is used for reference management. All citations identified in the searches are screened for eligibility based on titles and abstracts against prespecified inclusion criteria by one reviewer, and all citations not relevant by the first reviewer undergo screening by a second reviewer. Any citation deemed potentially eligible by at least one reviewer is retrieved for full-text review. All articles selected for full-text review undergo dual independent review to determine eligibility for inclusion. Disagreements are resolved by consensus, or a third reviewer if needed. All articles excluded at the full-text stage are assigned a code indicating the reason for exclusion.

Articles excluded at the full-text review stage represent a subset of studies that were considered potentially eligible for inclusion based on titles and abstracts, but did not meet criteria upon examination of the full paper. Therefore, examination of excluded full-text articles may provide insight into the larger evidence base on PBCs for chronic pain beyond the studies eligible for inclusion in the living review, including the study designs utilized and the populations and interventions evaluated in this topic area. It should be noted that articles selected for full-text review do not necessarily represent a comprehensive set of relevant articles, as reviewers vary with regard to the threshold used to select articles for full-text review. Most articles are excluded at the title and abstract review stage and do not undergo full-text review; in accordance with PRISMA guidance,¹¹ specific reasons for exclusions at the title and abstract stage are not recorded. It is also common for articles to have more than one reason for exclusion at the full-text stage. Although we focused on what we judged as the primary reason for exclusion, many studies had multiple reasons for exclusion. For this report, we focused on studies that were excluded at the full-text review stage for ineligible population, intervention, outcome, or comparator.

A total of 82 articles were excluded at the full-text stage for ineligible population, intervention, outcome, or comparator (Appendix A). Thirty-seven studies evaluated an ineligible population, 12 studies evaluated an ineligible intervention, and 33 studies evaluated did not have a comparison group or evaluated an ineligible comparison. No study was excluded for evaluating ineligible outcomes. None of the studies excluded at the full-text stage evaluated patients with subacute pain or adolescents; the living review was expanded to include these populations in year 3.

Among the 38 studies (in 37 publications) excluded for ineligible population, 12 were excluded because they evaluated cannabis in patients with multiple sclerosis who were not clearly described as having chronic pain.^12–23 Of these, six (n=24 to 630, total N=1,985) were randomized controlled trials^{12,16,18,20,22,23} and one¹⁷ (n=191) was a post-hoc analysis of a randomized controlled trial. The living review had originally been scoped to exclude multiple sclerosis without chronic pain. Although patients with multiple sclerosis frequently have spasticity, this was considered distinct from chronic pain. As described in the second white paper, we considered expanding the scope to include multiple sclerosis with spasticity following year one of the living review.⁹ However, a Technical Expert Panel did not support expanding to multiple sclerosis with spasticity, viewing pain and spasticity as distinct conditions. In addition, potential overlap was noted with existing systematic reviews of cannabis for multiple sclerosis (with or without chronic pain).^24–27

Six studies (in 5 publications) that were excluded for evaluation of ineligible populations studied cannabis in patients with advanced cancer (patients receiving palliative care and at end-of-life were not eligible for inclusion).^28–32 In one of the studies, patients also were not required to have chronic pain.³² Four of the studies (in 3 publications) of patients with advanced cancer and pain were randomized controlled trials (n=206 to 399; total N=1,362), each evaluating nabiximols (a comparable tetrahydrocannabinol [THC] to cannabidiol [CBD] ratio plant-extracted product).^28,29,31

Eight studies that were excluded for evaluation of ineligible populations studied cannabis in patients with nonchronic (e.g., episodic) pain conditions or with conditions not necessarily associated with chronic pain (and were not required to have chronic pain to be enrolled).^33–40 The conditions were ulcerative colitis,³⁵ Huntington’s disease,³⁶ HIV infection,³³ HIV and hepatitis C virus infection,³⁸ illicit opioid use,⁴⁰ ocular pain,³⁹ functional chest pain,³⁷ and inflammatory bowel disease.³⁴ Of these, one small study³⁶ (n=26) of nabiximols for Huntington’s disease and one small study³⁷ (n=13) of dronabinol (a synthetic, high THC-to-CBD-ratio product) for functional chest pain were randomized controlled trials.

Seven studies that were excluded for evaluation of ineligible populations studied cannabis in mixed populations of chronic and nonchronic pain, without reporting results separately for patients with chronic pain.^41–47 None of these studies were randomized controlled trials. The proportion of patients with chronic pain was unreported in three studies and ranged from ~50 percent to 89 percent in the others. Of five remaining studies excluded for evaluation of ineligible population, two were animal studies,^48,49 one evaluated healthy volunteers,⁵⁰ one evaluated a population with induced/experimental pain on top of chronic pain,51 and one evaluated patients undergoing surgery.⁵²

Twelve articles were excluded at the full-text stage because they evaluated ineligible interventions.^53–64 Of these, three articles evaluated possession of a medical marijuana card, rather than use of a specific cannabis product.^53,54,63 Two of the medical marijuana card articles reported results from the same randomized controlled trial (n=186), which was not restricted to patients with pain (the study also enrolled patients with insomnia, anxiety, or depressive symptoms).^53,63 Another uncontrolled study evaluated hemp seed, a product excluded because it only contains trace amounts of THC or CBD,59 two observational studies of cannabis for chronic pain were excluded because they did not describe the cannabis products used,^58,62 and one observational study evaluated cannabis access laws.⁶⁰ Four studies did not evaluate cannabis or other eligible PBCs.^55,57,61,64 One study evaluated integrative medicine therapies,⁵⁵ two evaluated opioids,^56,57 one evaluated olorinab,⁶⁴ and one evaluated psilocybin.⁶¹ As described in the second white paper, we previously considered potential expansion of the scope of the living review to include psilocybin. Based on a literature scan and input from a Technical Expert Panel, however, the decision was made to not expand to include psilocybin because studies were not yet available on psilocybin for chronic pain and were not imminently expected. The excluded study of psilocybin was a randomized trial (n=16) but would also have been excluded for ineligible population, as it evaluated patients with cluster headaches, an episodic (nonchronic) pain condition.⁶¹

Thirty-three studies of cannabis products were excluded because they did not have an eligible comparison group.^65–97 Four studies^74,85,86,94 only compared different doses or frequencies of use and the others were uncontrolled studies (i.e., no comparison group). Among the uncontrolled studies, sample sizes ranged from 7 to 8,165 (mean 551, median 151). Most studies were conducted in patients prescribed medical cannabis for mixed chronic pain; the pain conditions and cannabis products evaluated were generally not well described.

Table 1

Characteristics of studies excluded at full-text review.