Cannabis use disorder (CUD) represents a major global public health burden. Epidemiological estimates cited in the protocol report that roughly 25–30% of cannabis users develop CUD, with a 2–3% annual prevalence among past-year users, particularly in frequent or high-potency consumers. Behavioral therapies such as cognitive-behavioral therapy (CBT), motivational enhancement therapy (MET), and contingency management (CM) are widely used and have demonstrated reductions in use frequency and dependence severity versus inactive controls. Neurobiological modalities (for example, acupuncture and repetitive transcranial magnetic stimulation [rTMS]) aim to directly modulate reward and stress circuits but have been less consistently included in prior syntheses.
Conventional systematic reviews have limitations: they often lump heterogeneous NPIs into broad categories, use small localized samples, inadequately explore heterogeneity and effect modification, and deprioritize neuromodulatory interventions. This protocol proposes a unified comparative approach to synthesize evidence across psychosocial and neurobiological NPIs to better inform clinical decision-making and trial design.
The primary objective is to produce a comprehensive comparative efficacy hierarchy of non-pharmacological interventions (NPIs) for individuals with CUD by conducting a systematic review and frequentist network meta-analysis (NMA) of randomized controlled trials (RCTs). Secondary aims include evaluating acceptability and safety and assessing the certainty of evidence using the GRADE approach.
This protocol follows PRISMA-P 2015 reporting standards and is registered on PROSPERO (CRD420251173215). Pairs of reviewers will independently conduct study selection, data extraction, and risk of bias assessments. A frequentist NMA will synthesize direct and indirect comparisons among NPIs and control conditions. GRADE will be used to evaluate certainty and interventions will be categorized with a minimally contextualized decision threshold framework.
Systematic searches were conducted in PubMed, Embase, the Cochrane Central Register of Controlled Trials (CENTRAL), and CINAHL from inception to December 30, 2025. The search strategy was developed by a methodologically trained researcher and refined by a senior methodologist. Reference lists of eligible studies and relevant reviews will be screened manually. The search will be updated prior to the final meta-analysis to include newly published eligible trials.
Eligible studies are RCTs comparing any NPI against another NPI, treatment as usual (TAU), standard care, waiting list, or placebo. Participants will be classified into (1) formally diagnosed CUD cases (DSM or ICD criteria) and (2) screening-identified problematic cannabis users (for example, recruited from detoxification or emergency settings, or reporting past-month use with baseline problems). The primary analysis will synthesize all eligible populations combined; pre-specified subgroup and sensitivity analyses will assess the impact of diagnostic certainty.
Age groups are predefined as adolescents (12–17 years) and adults (≥18 years). Age will be treated as a primary effect modifier, with network meta-regressions using trial-level mean age and subgroup analyses stratified by age. Sensitivity analyses excluding adolescent trials are planned.
Primary benefit outcomes:
Primary acceptability outcome:
Primary safety outcome:
Pairs of reviewers will extract data using standardized forms. Risk of bias will be assessed independently by reviewer pairs. The minimal dataset, extraction sheets, summary outcome data, and analytic scripts will be made publicly available upon completion, with deposition in a repository such as the Open Science Framework (OSF).
A frequentist NMA will synthesize evidence across NPIs and control conditions to estimate comparative treatment effects. Analyses will integrate psychosocial and neuromodulatory modalities into a single network to derive relative rankings and pooled effect estimates. The protocol specifies using network meta-regression to explore continuous effect modifiers (for example, mean trial age) and predefined subgroup/sensitivity analyses to address heterogeneity.
Pre-specified subgroup analyses include diagnostic certainty (formal diagnosis vs screening-identified) and age category (adolescents vs adults). Sensitivity analyses will assess robustness to excluding adolescent trials and other design variations. Network meta-regressions will evaluate potential treatment-by-covariate interactions, particularly with trial-level mean age as an effect modifier.
Evidence certainty for network estimates will be assessed using the GRADE approach. Interventions will be categorized under a minimally contextualized framework using predefined decision thresholds to assist interpretation and clinical applicability of the ranking.
The protocol is registered on PROSPERO (CRD420251173215). Comprehensive search strategies were validated and screening, data extraction, quantitative NMA, and GRADE assessments were scheduled for completion by September 2026. All relevant data and analytic scripts will be deposited in a public repository (e.g., OSF) with a persistent DOI after study completion.
Funding sources and grant numbers are reported in the protocol; funders had no role in study design, analysis, or publication decisions. The authors declare no competing interests. The planned NMA aims to bridge psychosocial and neuromodulatory modalities, providing a standardized, actionable evidence hierarchy to guide tailored clinical care and future trial methodology for CUD.