Abstract
Background:
Contemporary data are needed about the utility of cannabinoids in chronic pain.
Purpose:
To evaluate the benefits and harms of cannabinoids for chronic pain.
Data Sources:
Ovid MEDLINE, PsycINFO, EMBASE, the Cochrane Library, and Scopus to January 2022.
Study Selection:
English-language, randomized, placebo-controlled trials and cohort studies (≥1 month duration) of cannabinoids for chronic pain.
Data Extraction:
Data abstraction, risk of bias, and strength of evidence assessments were dually reviewed. Cannabinoids were categorized by THC-to-CBD ratio (high, comparable, or low) and source (synthetic, extract or purified, or whole plant).
Data Synthesis:
Eighteen randomized, placebo-controlled trials (n = 1740) and 7 cohort studies (n = 13 095) assessed cannabinoids. Studies were primarily short term (1 to 6 months); 56% enrolled patients with neuropathic pain, with 3% to 89% female patients. Synthetic products with high THC-to-CBD ratios (>98% THC) may be associated with moderate improvement in pain severity and response (≥30% improvement) and an increased risk for sedation and are probably associated with a large increased risk for dizziness. Extracted products with high THC-to-CBD ratios (range, 3:1 to 47:1) may be associated with large increased risk for study withdrawal due to adverse events and dizziness. Sublingual spray with comparable THC-to-CBD ratio (1.1:1) probably is associated with small improvement in pain severity and overall function and may be associated with large increased risk for dizziness and sedation and moderate increased risk for nausea. Evidence for other products and outcomes, including longer-term harms, were not reported or were insufficient.
Limitation:
Variation in interventions; lack of study details, including unclear availability in the United States; and inadequate evidence for some products.
Conclusion:
Oral, synthetic cannabis products with high THC-to-CBD ratios and sublingual, extracted cannabis products with comparable THC-to-CBD ratios may be associated with short-term improvements in chronic pain and increased risk for dizziness and sedation. Studies are needed on long-term outcomes and further evaluation of product formulation effects.
Primary Funding Source:
Agency for Healthcare Research and Quality, U.S. Department of Health and Human Services. (PROSPERO: CRD42021229579)
Chronic pain, typically defined as pain lasting longer than 3 to 6 months (1, 2), affects approximately 100 million persons in the United States (3) and adversely affects physical and mental functioning, productivity, and quality of life. It is often refractory to treatment and associated with substantial costs to both the patient and the health care system (4, 5).
Although opioids are often prescribed for chronic pain, a recent series of systematic reviews found that opioid medications (6), several nonopioid medications (7), and some nonpharmacologic treatments (8) have small to moderate effects on pain and function. Opioid and nonopioid medications were also found to have some frequent adverse effects and some less frequent but serious ones, which vary according to the specific drug or drug class. These findings and the ongoing opioid crisis drive a search for alternative pain treatments, ideally with similar or better benefits, but with an improved safety profile.
Treatments derived from plant sources, such as cannabis and its related compounds, are among those being considered to potentially fill this gap (9). The term cannabinoid refers to compounds that are active in cannabis, with the 2 main cannabinoid compounds being tetrahydrocannabinol (THC) and cannabidiol (CBD). In preclinical studies, THC and related compounds have demonstrated analgesic properties (10), although its psychoactive effects and addiction potential may limit its suitability as an analgesic. Cannabidiol and other cannabinoids may also have some analgesic or anti-inflammatory properties and are not believed to be psychoactive or addictive (11, 12). Given the variation in analgesic effect with THC and CBD, response may differ according to the ratio of THC to CBD in products used to treat pain.
The purpose of this systematic review is to evaluate the benefits and harms of cannabinoids to treat chronic pain, using a novel categorization scheme for the amount of THC versus CBD in cannabis products. The evidence is expected to develop more rapidly over the coming years, and this review will be updated on a quarterly basis going forward. This review was commissioned by the Agency for Healthcare Research and Quality (AHRQ).
Methods
This article focuses on evidence from placebo-controlled randomized controlled trials (RCTs) and cohort studies with a concurrent control group. The protocol for the systematic review, developed with input from an independent panel of experts, was registered in PROSPERO (CRD42021229579) and published on the AHRQ website (https://effectivehealthcare.ahrq.gov/products/plant-based-chronic-pain-treatment/protocol). We engaged a technical expert panel to develop the protocol and framework for the review. The comprehensive evidence report, interactive figures for primary results, and quarterly surveillance reports, which include additional details and studies of active comparisons, are available at the AHRQ website (13–16) (https://effectivehealthcare.ahrq.gov/products/plant-based-chronic-pain-treatment/living-review). A visual abstract and brief for clinicians is available at www.cannabisevidence.org.
Data Sources and Searches
We searched Ovid MEDLINE, PsycINFO, EMBASE, the Cochrane Library, and Scopus databases monthly (divided into biweekly alternating searches) from database inception to January 2022. Search terms included “chronic pain” and “cannabinoids” (Supplement). We posted a request for information in the Federal Register and supplemented searches by reviewing reference lists.
Study Selection
We included English-language studies of patients with chronic pain that compared cannabis products with a placebo or no treatment (that is, usual care) for at least 4 weeks of treatment or follow-up (detailed criteria are in the full report). Eligible study designs were placebo-controlled RCTs or comparative observational studies with a concurrent control (for example, cohort studies). Two investigators independently reviewed abstracts and full-text articles against prespecified eligibility criteria, using consensus for any disagreements.
Data Abstraction and Risk-of-Bias Assessment
Data extraction was completed by 1 investigator and verified by a second. Primary outcomes were measures of pain, physical or general functioning, and adverse events. Adverse events of interest were serious adverse events, adverse events leading to study withdrawal, nausea, dizziness, sedation, psychosis, development of cannabis use disorder, and cognitive deficits. Secondary outcomes were quality of life, mental health, sleep, and effect on opioid use. Two investigators independently assessed risk of bias for each study as low, moderate, or high using the Cochrane Back Pain Group’s version of the Cochrane guidance for randomized trials (17) and criteria developed by the U.S. Preventive Services Task Force for observational studies (18). Disagreements were resolved through consensus.
Data Synthesis and Analysis
We developed a novel system to categorize studies of cannabinoids according to their THC-to-CBD ratios based on input from a technical expert panel consisting of 8 persons with expertise in cannabis and pain research. Cannabinoids were categorized as high, comparable, or low THC-to-CBD ratio (Table 1). Table 1 includes additional categories of “unreported” (specific products not reported) and “not applicable” (1 study of other cannabinoids). An additional category of “unreported” was added on the basis of study reporting in some cases. The categories were further stratified on the basis of whether interventions were synthetic or derived from whole-plant cannabis. The synthetic THC analogue nabilone was categorized with synthetic THC dronabinol. Cannabinoid products can be extracted from a whole-plant source to isolate THC, CBD, or other cannabinoids. Extracted products may contain additional cannabinoids and other compounds (for example, terpenes) present in whole-plant cannabis that may or may not influence the effect of the intervention. These products can also then be purified. Purified products are pharmaceutical grade, ensure a specific concentration, and are free of contaminants (that is, they do not contain other cannabinoids, terpenes, or other contaminants). Namisol (Echo Pharmaceuticals) presented a challenge for categorization because it is a product that is purified from a plant source but chemically identical to synthetic dronabinol (19). Therefore, we grouped Namisol together with synthetic dronabinol but also did sensitivity analyses without Namisol. We categorized the duration of studies as short term (1 to 6 months), intermediate term (>6 to 12 months), or long term (>12 months).
Table 1. Organizing Principle of Cannabis-Related Studies Based on Ratios of THC to CBD
 |
After examining clinical and methodological heterogeneity to determine the appropriateness of quantitative synthesis, we conducted meta-analyses using the profile likelihood random-effects model (20). If topical products clearly were intended to have systemic effects, they were analyzed with oral and sublingual products but evaluated separately if intended to have local effects or if it was unclear if they were systemic. We analyzed studies according to the THC-to-CBD ratio category and source (synthetic vs. extracted). Heterogeneity was assessed using the I2 statistic and the Cochran Q statistic χ2 test (21). All meta-analyses were done using the metan and admetan commands in Stata/SE, version 16.1 (StataCorp). Sensitivity analyses were done by excluding studies rated as high risk of bias, excluding the trial of Namisol (22) that was grouped with synthetic THC, and by repeating analyses using the Bartlett correction to the profile likelihood method to reduce potential deviation from the null distribution when the number of studies is small (20). Although the Bartlett correction resulted in greater imprecision in pooled estimates (that is, wider CIs), the statistical significance of findings was unchanged, with 2 exceptions (dizziness and sedation with synthetic products with high THC-to-CBD ratios). Therefore, it is not discussed further, except for those analyses. Publication bias (small study effect) was assessed using funnel plots and the Egger test when there were 8 or more studies included in a meta-analysis. Consistent with other recent systematic reviews on treatments for chronic pain (6–8), the magnitude of benefit was evaluated on a scale of 0 to 10 and categorized into no effect (mean difference [MD], <0.2), small effect (MD, 0.2 to 0.5), moderate effect (MD, >0.5 to 0.8), and large effect (MD, >0.8) (Supplement Table 1). We graded the strength of evidence for each primary outcome (change in pain severity, proportion of participants with a clinically important change in pain [defined as ≥30% reduction from baseline], change in general function, and each of the specified adverse events) using the approach described in the AHRQ Methods Guide for Effectiveness and Comparative Effectiveness Reviews (23). One investigator did the initial assessments of strength of evidence, and the final ratings were determined by consensus. Meta-analysis was not done on observational studies because of methodological and clinical variability or on secondary outcomes (quality of life, depression, anxiety, sleep, and opioid use) because of infrequent reporting (see the full report for details on secondary outcomes [16]). The findings of these were narratively summarized.
Role of the Funding Source
This project was funded under contract number 75Q80120D00006 from the AHRQ, U.S. Department of Health and Human Services. Staff from the AHRQ assisted in developing the scope and key questions. A representative from the AHRQ served as a Contracting Officer’s Technical Representative and provided technical assistance during the conduct of the full evidence report and comments on draft versions of the report. The AHRQ did not directly participate in the literature search, determination of study eligibility criteria, data analysis, interpretation, or decision to submit this manuscript.
Results
Description of the Evidence
We included 18 placebo-controlled RCTs (22, 24–40) and 7 observational studies with usual care controls (41–47) (Appendix Figure 1 and Appendix Tables 1 and 2). Ten studies evaluated products with high THC-to-CBD ratios (22, 25, 31, 34–36, 38, 40, 41, 47)—6 RCTs and 1 cohort study of synthetic products with high THC-to-CBD ratios (dronabinol, including 1 of plant-purified dronabinol, or the THC analogue nabilone [22]) (22, 31, 34–36, 38, 41); 2 RCTs of plant extracts (25, 40); and 1 cohort study of whole-plant cannabis, with an established high THC concentration (47). Seven placebo-controlled RCTs evaluated products with comparable THC-to-CBD ratios, specifically nabiximols, extracted from whole-plant cannabis (Table 2) (24, 27–30, 32, 33). Two RCTs assessed products with low THC-to-CBD ratios (CBD topical oil and oral capsule) compared with placebo (37, 39), and another assessed an oral extract of the phytocannabinoid cannabidivarin compared with placebo (26). In addition, 5 observational studies evaluated any cannabis product purchased from a cannabis dispensary (patient’s choice) used to treat chronic pain compared with usual care (42–46).
Table 2. Characteristics of Included Studies Stratified by THC-to-CBD Ratio and Study Design
 |
Appendix Table 1. Randomized Controlled Trials Evidence Table
 |
Appendix Table 2. Observational Studies Evidence Table
 |
Study and patient characteristics are summarized in Table 2. Most patients who enrolled had forms of neuropathic pain, were middle aged, were White, and had mean baseline pain scores of 5 to 6 on the pain numerical rating scale of 0 to 10 (where 0 = no pain and 10 = worst pain imaginable). All of the RCTs (study sample sizes ranged from 29 to 882) were short term, with most 4 to 8 weeks in duration, and placebo controlled. One half were medium risk of bias, 30% were high risk of bias, and 20% were low risk of bias. Seven RCTs that allowed opioid use during the study period did not report on changes in opioid use during the study period (27, 29, 31, 33, 34, 48). Observational studies (study sample sizes ranged from 156 to 12 508) had longer follow-up (12 to 208 weeks) and compared initiation of cannabis products to usual care. None were low risk of bias (4 were high and 3 were moderate risk of bias) (Supplement Tables 2 and 3).
High THC-to-CBD Ratio Products
Synthetic products with high THC-to-CBD ratios (>98% THC) were associated with moderate improvement in pain severity on a scale of 0 to 10 (6 RCTs [n = 390]; MD, −1.15 [95% CI, −1.99 to −0.54]; I2 = 48%) (Figure 1). Results were similar when the trial of plant-purified dronabinol was excluded (5 RCTs; MD, −1.20 [CI, −2.21 to −0.47]; I2 = 58%). Stratified analysis by dronabinol and nabilone showed that the pooled effect estimate for nabilone (MD, −1.59 [CI, −2.49 to −0.82]; I2 = 0%) was statistically significant. The estimate for dronabinol was not statistically significant, and somewhat smaller than with nabilone, but the difference between the subgroups was not statistically significant (Appendix Figure 2 and Supplement Table 4) (22, 31, 34–36, 38). Improvement in pain severity meeting the threshold for response (≥30% improvement from baseline) was reported in a single low risk of bias RCT (n = 26) of patients with diabetic neuropathy. More patients receiving nabilone had a response than those receiving placebo (85% vs. 38%; relative risk [RR], 2.20 [CI, 1.06 to 4.55]) (35). Pooled analysis found no effect on overall function or disability with nabilone (2 RCTs [n = 40]; 0 to 10 scale; MD, −0.35 [CI, −1.90 to 0.94]; I2 = 72%) (Appendix Figure 3), and a third study (n = 13) reported that there was no difference in function between groups but did not provide data for the meta-analysis (38). Synthetic products with high THC-to-CBD ratios were associated with moderate increase in risk for sedation (2 RCTs on dronabinol and 1 RCT on nabilone [n = 335]; 19% vs. 10%; RR, 1.73 [CI, 1.03 to 4.63]; I2 = 28%). In stratified analyses, the study of nabilone (n = 33) reported a greater magnitude of effect for sedation (RR, 8.40 [CI, 1.16 to 60.84]) than the combined trials of dronabinol (n = 302; RR, 1.55 [CI, 0.84 to 3.07]; I2 = 0%), with no statistically significant subgroup differences (P = 0.105). Dronabinol was found to be associated with increased risk for dizziness (2 RCTs on dronabinol [n = 302]; 32% vs. 11%; RR, 2.74 [CI, 1.47 to 6.86]; I2 = 40%) (22, 31). The frequency of study withdrawal due to adverse events was greater with high THC-to-CBD ratio products than placebo but not statistically significant. Pooled results for adverse events are available in Appendix Figures 4, 5, 6 and 7.
* Refers to pain severity on a 0 to 10 scale at follow-up, except in the case of Zajicek and colleagues (40), which is reporting the mean change from baseline, and Wissel and colleagues (38), which is the median pain severity score at follow-up.
† Dronabinol tablet = plant-derived, purified product Namisol (Echo Pharmaceutical).
* Refers to pain severity on a 0 to 10 scale at follow-up, except in the case of Wissel and colleagues (38), which is the median pain severity score at follow-up.
† Namisol is a purified, plant-based product but is grouped with synthetic dronabinol because they are chemically identical.
* Refers to pain severity at follow-up on a 0 to 10 scale.
* Namisol is a purified, plant-based product but is grouped with synthetic dronabinol because they are chemically identical.
* Namisol is a purified, plant-based product but is grouped with synthetic dronabinol because they are chemically identical.
* Namisol is a purified, plant-based product but is grouped with synthetic dronabinol because they are chemically identical.
* Namisol is a purified, plant-based product but is grouped with synthetic dronabinol because they are chemically identical.
Reports of any adverse events, nausea, and serious adverse events and subgroup analysis of change in pain by the type of THC were not statistically significantly different between groups (Supplement Table 4). However, sensitivity analyses using the Bartlett correction for the meta-analysis model resulted in more imprecise pooled estimates for dizziness and sedation that were no longer statistically significant (Supplement Table 5). Quality-of-life findings were mixed across 2 studies of nabilone at 4 and 5 weeks (34, 35). Anxiety scores were improved more with nabilone in both studies, and depression scores were not improved with nabilone in either study.
Two placebo-controlled RCTs (n = 297) studied high THC-to-CBD ratio products extracted from whole-plant cannabis (25, 40). The products studied differed widely in both THC to CBD ratio (1.4 to 3:1 ratio vs. 48 to 60:1) and in daily dose (5 to 25 mg of THC vs. mean 4.4 mg of THC daily). In pooled analysis, pain severity was improved with these extracted products with high THC-to-CBD ratios, but the effect was not statistically significant because of heterogeneity in the degree of benefit (Figure 1). There was a high degree of heterogeneity in this combined estimate, making the finding insufficient to draw conclusions. Pain response (the proportion with ≥30% improvement in pain) was not reported. For secondary outcomes, 1 RCT (n = 17) of patients with fibromyalgia reported that physical functioning was not improved (Fibromyalgia Impact Questionnaire, subscale 0 to 10; MD, 1.75 [CI, −0.46 to 3.98]) and that quality of life was improved (Fibromyalgia Impact Questionnaire 0 to 100 scale; MD, 36.0; P = 0.005) with extracted products with high THC-to-CBD ratios compared with placebo. However, these analyses were not adjusted for potentially important differences between groups in baseline scores (25). Differences in depression and anxiety were not found between groups. In the larger study (n = 277) of patients with multiple sclerosis, extracted products with high THC-to-CBD ratios were associated with large increased risk for study withdrawal due to adverse events (13.9% vs. 5.7%; RR, 3.12 [CI, 1.54 to 6.33]) and dizziness (62.2% vs. 7.5%; RR, 8.34 [CI, 4.53 to 15.34]) compared with placebo (40). The rate of serious adverse events did not statistically significantly differ between groups (4.9% vs. 2.2%; RR, 2.19 [CI, 0.58 to 8.28]).
A prospective observational study (n = 431) of a whole-plant cannabis product determined to have 12.5% THC (amount of CBD not reported) reported on only adverse events (47). The route of administration and dosing regimen was determined by the patient (details not reported). There were no statistically significant differences at 52 weeks of follow-up for any adverse event, serious adverse events, or dizziness. Nausea (16.7% vs. 9.7%; RR, 1.72 [CI, 1.04 to 2.85]) and sedation (13.5% vs. 4.6%; RR, 2.91 [CI, 1.46 to 5.83]) were reported statistically significantly more frequently in the cannabis group. Cognitive deficits were also reported, using 2 subsets each of the Wechsler Memory Scale and the Wechsler Adult Intelligence Scale, with a nonstatistically significant difference between groups.
Comparable THC-to-CBD Ratio Products
Comparable THC-to-CBD ratio products were associated with small improvements in pain severity on a 0 to 10 scale (7 RCTs [n = 702]; MD, −0.54 [CI, −0.95 to −0.19]; I2 = 39%) (Figure 2) and overall function on a 0 to 10 scale (6 RCTs [n = 616]; MD, −0.42 [CI, −0.73 to −0.16]; I2 = 32%) (Appendix Figure 8) (24, 27–30, 32, 33). Although more patients had a clinically important improvement in pain (defined as ≥30% improvement from baseline), the difference was small and did not reach statistical significance (Appendix Figure 9). Compared with placebo, products with comparable THC-to-CBD ratios were associated with large increased risk for dizziness (6 RCTs [n = 866]; 30% vs. 8%; RR, 3.57 [CI, 2.42 to 5.60]; I2 = 0%) and sedation (6 RCTs [n = 866]; 8% vs. 1.2%; RR, 5.04 [CI, 2.10 to 11.89]; I2 = 0%) and moderate increased risk for nausea (6 RCTs [n = 866]; 13% vs. 7.5%; RR, 1.79 [CI, 1.19 to 2.77]; I2 = 0%) (Appendix Figures 10, 11 and 12). There was no effect on study withdrawal due to adverse events (Appendix Figure 13). Quality of life was not different between groups. Four of 5 studies reported statistically significantly better sleep outcomes in the comparable THC-to-CBD ratio groups versus placebo groups (24, 27, 29, 30, 33). Changes in depression and anxiety were not reported.
* Refers to pain severity at follow-up on a 0 to 10 scale.
* Refers to mean score on a 0 to 10 scale, except for Rog and colleagues (30), Nurmikko and colleagues (29), and Langford and colleagues (27), which reported mean change from baseline at follow-up.
Low THC-to-CBD Ratio Products (CBD Alone) and Other Cannabinoids
In the short term, low THC-to-CBD ratio products (CBD topical and oral) had insufficient evidence to draw conclusions based on one 4-week, high risk of bias RCT (n = 29) of patients with neuropathic pain (39). A single moderate risk of bias RCT (n = 31) of a cannabinoid other than THC and CBD (cannabidivarin) was also insufficient to draw conclusions (26).
Unreported THC-to-CBD Ratio (Patient Choice) Cannabis Products
Five observational studies (n = 12 508) compared any cannabis product (patient choice) with usual care (3 high and 2 moderate risk of bias). Three studies enrolled patients in medical cannabis programs (43, 44, 46) and 2 relied on patient self-reported use of cannabis to treat chronic pain (42, 45). The 2 studies reporting on primary pain or overall function outcomes (42, 43) provided insufficient evidence to draw conclusions because of moderate to high risk of bias and small sample sizes (Appendix Tables 2, 3 and 4). Prescription opioid use was not different between participants with chronic pain using cannabis products or usual care treatments in 4 studies (n range, 66 to 10 747) based on varying methods and outcome definitions (42, 44–46). None of the studies reported on adverse events.
Appendix Table 3. Cannabinoids to Treat Chronic Pain: SOE for Randomized Controlled Trials Versus Placebo
 |
Appendix Table 4. Cannabinoids to Treat Chronic Pain: SOE for Observational Studies—Unknown THC-to-CBD Ratio (Patient Choice)
 |
Discussion
The findings of our review are summarized in Figure 3 and are applicable to short-term treatment (1 to 6 months) in patients with chronic pain (mainly neuropathic pain) compared with placebo or usual care. The strongest evidence to date is for synthetic products with high THC-to-CBD ratios and extracted products with comparable THC-to-CBD ratios, both of which resulted in improvements in pain severity (moderate improvements for high THC-to-CBD ratio and small improvements for comparable THC-to-CBD ratio products). Small improvements in overall functioning were seen with products with comparable THC-to-CBD ratios, whereas none were seen with synthetic products with high THC-to-CBD ratios. Moderate to large increased risk for dizziness, sedation, and nausea may be associated with both types of products. Evidence for benefit with extracted, high THC-to-CBD ratio products was either insufficient or missing. Although both studies of extracted products with high THC-to-CBD ratios found statistically significant improvement in pain severity, the limitations of the individual studies, degree of heterogeneity, and marked imprecision due to limited evidence suggests that uncertainty remains about the exact magnitude and statistical significance of a possible treatment effect. Evidence for whole-plant products, CBD, and other cannabinoids was limited by serious imprecision and lack of ability to assess consistency and study methodological limitations. These studies were not designed to evaluate harm outcomes and often excluded patients who were at higher risk for harms, potentially underestimating the harms of cannabinoid treatment. Other key adverse event outcomes (psychosis, cannabis use disorder, and cognitive deficits) and outcomes on the effect on prescription opioid use were not reported or insufficient.
*k = number of studies contributing evidence to finding.
† Effect size: none (i.e., no effect/no statistically significant effect); small, moderate, or large effect (see the Methods section); potential effect assigned to findings that are not statistically significant but with small or > magnitude of effect and an SOE rating of at least low. These cases are believed to suggest an effect, but the lack of precision (inadequate sample size) resulted in a nonstatistically significant result.
‡ [+] = low, [++] = moderate, [+++] = high, or [insufficient].
§ Downgraded because of study limitations (moderate SOE) and imprecision.
|| Downgraded because of study limitations (moderate SOE).
¶ Downgraded because of imprecision.
In recent years, there have been several systematic reviews done on the use of cannabinoids to treat chronic pain (49–53). Our review is distinct to these reviews in important ways. An important strength of our approach is the scheme used for categorizing cannabis products according to the amount of THC versus CBD, providing a potential framework for future studies in this area. This classification system led our review to have conclusions that differ from other reviews. For example, a recent high-quality review found moderate-certainty evidence that noninhaled cannabinoids had small effects on pain versus placebo, categorizing cannabinoids only by route of administration. This review did not consider THC-to-CBD ratio, included products not classified as cannabinoids, included studies of fewer than 4 weeks duration, and did not stratify by duration of follow-up, potentially limiting applicability of the findings.
The evidence base on cannabis-based products for chronic pain has several important limitations that affect applicability of the findings and our ability to understand the effects of specific products in particular patients. A key limitation is unstandardized and inconsistent reporting about specific cannabis products (for example, THC-to-CBD ratio or purity of extracted products) and the lack of adequate studies of varying products to allow analysis of factors, such as formulation, route of administration, and dose or dose regimen. In all cases other than the 2 U.S Food and Drug Administration–approved drug products (dronabinol [Marinol, AbbVie Pharmaceuticals] and nabilone [Cesamet, Bausch Health]), it is unclear whether the products studied are available in the United States. For some cannabis products, such as whole-plant products, the data are sparse with imprecise estimates of effect, and studies had methodological limitations. Additional limitations include the populations (limited data on conditions other than neuropathic pain; patients with psychiatric illness or other comorbidities, such as spasticity without pain; older adults; and children) and outcomes (lack of or inadequate data on the likelihood of having clinically important improvement in pain, function, opioid use, or long-term outcomes, including important harms) studied.
Limitations of our review process may include decisions we made in the review process. We categorized nabilone as a synthetic product with a high THC-to-CBD ratio, although it is more accurately described as a synthetic cannabinoid—a chemical analogue to THC. Similarly, we grouped the plant-derived, purified dronabinol product Namisol as a synthetic, high THC-to-CBD ratio product. This product is more than 98% pure delta-9 THC, and is therefore much more similar to synthetic THC than to other plant-derived (extracted), high THC-to-CBD ratio products. To assess the possibility that these decisions affected the results, we did sensitivity analyses of nabilone and dronabinol and of synthetic versus plant-derived dronabinol. Although visual inspection of the forest plot suggests that nabilone may have a greater effect on pain than dronabinol, data were too imprecise to determine if there are differences in the effect on pain severity. With only 2 studies of dronabinol, 1 of the synthetic form and 1 of the plant-derived form (Namisol), it was not possible to determine whether the source of dronabinol affects outcomes. Although most findings were robust regardless of whether the Bartlett correction was used (Supplement Table 5), the estimates for sedation and dizziness with synthetic, high THC-to-CBD ratio products became statistically nonsignificant with this approach. However, the Bartlett correction may result in overly conservative (imprecise) estimates, particularly when there are fewer than 5 studies (54). More studies would help reduce imprecision. We were unable to assess publication bias (small sample size bias) for most outcomes because most meta-analyses included fewer than 8 studies. As in other recent systematic reviews of interventions to treat chronic pain, we grouped the magnitude of effects into small, moderate, and large effects, rather than according to other published thresholds (6, 7), and excluded non–English-language publications.
Additional studies are needed to clarify the effect size estimates and our confidence in the findings. We plan to continue to monitor the evidence on this topic on a regular basis to identify important new evidence as it emerges. The categorization scheme we used here based on the ratio of THC to CBD in products may be useful in primary research and future systematic reviews going forward.
In conclusion, oral, synthetic, high THC-to CBD ratio and sublingual, plant-extracted, comparable THC-to-CBD ratio cannabis interventions may be associated with short-term improvements in primarily neuropathic chronic pain and increased risk for dizziness and sedation. Evidence on other products was insufficient or lacking. Studies are needed on long-term outcomes and further evaluation of product formulation effects.
