Pharmacotherapies for cannabis dependence

Suzanne Nielsen; Linda Gowing; Pamela Sabioni; Bernard Le Foll

doi:10.1002/14651858.CD008940.pub3

. 2019 Jan 28;2019(1):CD008940. doi: 10.1002/14651858.CD008940.pub3

Pharmacotherapies for cannabis dependence

Suzanne Nielsen ¹, Linda Gowing ^2,^✉, Pamela Sabioni ³, Bernard Le Foll ³

Editor: Cochrane Drugs and Alcohol Group

PMCID: PMC6360924 PMID: 30687936

Abstract

Background

Globally, cannabis use is prevalent and widespread. There are currently no pharmacotherapies approved for treatment of cannabis use disorders.

This is an update of a Cochrane Review first published in the Cochrane Library in Issue 12, 2014.

Objectives

To assess the effectiveness and safety of pharmacotherapies as compared with each other, placebo or no pharmacotherapy (supportive care) for reducing symptoms of cannabis withdrawal and promoting cessation or reduction of cannabis use.

Search methods

We updated our searches of the following databases to March 2018: the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase, PsycINFO and Web of Science.

Selection criteria

Randomised controlled trials (RCTs) and quasi‐RCTs involving the use of medications to treat cannabis withdrawal or to promote cessation or reduction of cannabis use, or both, in comparison with other medications, placebo or no medication (supportive care) in people diagnosed as cannabis dependent or who were likely to be dependent.

Data collection and analysis

We used standard methodological procedures expected by Cochrane.

Main results

We included 21 RCTs involving 1755 participants: 18 studies recruited adults (mean age 22 to 41 years); three studies targeted young people (mean age 20 years). Most (75%) participants were male. The studies were at low risk of performance, detection and selective outcome reporting bias. One study was at risk of selection bias, and three studies were at risk of attrition bias.

All studies involved comparison of active medication and placebo. The medications were diverse, as were the outcomes reported, which limited the extent of analysis.

Abstinence at end of treatment was no more likely with Δ⁹‐tetrahydrocannabinol (THC) preparations than with placebo (risk ratio (RR) 0.98, 95% confidence interval (CI) 0.64 to 1.52; 305 participants; 3 studies; moderate‐quality evidence). For selective serotonin reuptake inhibitor (SSRI) antidepressants, mixed action antidepressants, anticonvulsants and mood stabilisers, buspirone and N‐acetylcysteine, there was no difference in the likelihood of abstinence at end of treatment compared to placebo (low‐ to very low‐quality evidence).

There was qualitative evidence of reduced intensity of withdrawal symptoms with THC preparations compared to placebo. For other pharmacotherapies, this outcome was either not examined, or no significant differences was reported.

Adverse effects were no more likely with THC preparations (RR 1.02, 95% CI 0.89 to 1.17; 318 participants; 3 studies) or N‐acetylcysteine (RR 0.94, 95% CI 0.71 to 1.23; 418 participants; 2 studies) compared to placebo (moderate‐quality evidence). For SSRI antidepressants, mixed action antidepressants, buspirone and N‐acetylcysteine, there was no difference in adverse effects compared to placebo (low‐ to very low‐quality evidence).

There was no difference in the likelihood of withdrawal from treatment due to adverse effects with THC preparations, SSRIs antidepressants, mixed action antidepressants, anticonvulsants and mood stabilisers, buspirone and N‐acetylcysteine compared to placebo (low‐ to very low‐quality evidence).

There was no difference in the likelihood of treatment completion with THC preparations, SSRI antidepressants, mixed action antidepressants and buspirone compared to placebo (low‐ to very low‐quality evidence) or with N‐acetylcysteine compared to placebo (RR 1.06, 95% CI 0.93 to 1.21; 418 participants; 2 studies; moderate‐quality evidence). Anticonvulsants and mood stabilisers appeared to reduce the likelihood of treatment completion (RR 0.66, 95% CI 0.47 to 0.92; 141 participants; 3 studies; low‐quality evidence).

Available evidence on gabapentin (anticonvulsant), oxytocin (neuropeptide) and atomoxetine was insufficient for estimates of effectiveness.

Authors' conclusions

There is incomplete evidence for all of the pharmacotherapies investigated, and for many outcomes the quality of the evidence was low or very low. Findings indicate that SSRI antidepressants, mixed action antidepressants, bupropion, buspirone and atomoxetine are probably of little value in the treatment of cannabis dependence. Given the limited evidence of efficacy, THC preparations should be considered still experimental, with some positive effects on withdrawal symptoms and craving. The evidence base for the anticonvulsant gabapentin, oxytocin, and N‐acetylcysteine is weak, but these medications are also worth further investigation.

Plain language summary

Medicines for the treatment of cannabis dependence

Background

Cannabis use is relatively common and widespread worldwide. Demand by cannabis users for treatment has been increasing in most regions of the world. Moves in some countries to decriminalise or legalise cannabis use is likely to result in this trend continuing. Currently there are no medicines specifically for the treatment of cannabis use. This review sought to assess the effectiveness and safety of medicines for the treatment of cannabis dependence.

Search date

We searched the scientific literature in March 2018.

Study characteristics

We identified 21 randomised controlled trials (clinical studies where people are allocated at random to one of two or more treatment groups) involving 909 participants treated with active medicines, and 846 who received placebo (a pretend treatment). Key features of dependent drug use are compulsive use, loss of control over use and withdrawal symptoms on cessation of drug use. This review included studies where participants were described as dependent or were likely to be dependent based on cannabis use occurring several days a week, or daily.

The mean age of participants in individual studies ranged from 22 to 41 years, excluding three studies that targeted young people. Most (75%) study participants were male. Most (16) of the studies were undertaken in the USA, with three occurring in Australia, one in Canada and one in Israel. The studies tested a wide range of medicines to reduce the symptoms of cannabis withdrawal and to promote cessation or reduction of cannabis use.

Four studies received study medicines from the manufacturing pharmaceutical company but none were funded by pharmaceutical companies. One study did not report funding or medicine source.

Key results

For the outcome of abstinence at the end of treatment, Δ⁹‐tetrahydrocannabinol (THC, the major constituent in cannabis) preparations were probably ineffective; antidepressants called selective serotonin reuptake inhibitors, mixed action antidepressants, a medicine called buspirone and a medicine called N‐acetylcysteine may also have been ineffective; and we are uncertain about the effect of anticonvulsants and mood stabilisers.

For the outcome of completion of the scheduled period of treatment, THC preparations, mixed action antidepressants, anticonvulsants and mood stabilisers may not have been effective, we were uncertain about the effect of SSRI antidepressants, and N‐acetylcysteine probably did not support treatment completion. The use of anticonvulsants and mood stabilisers may have increased the likelihood that people left treatment early.

THC preparations and N‐acetylcysteine were probably no more likely to cause side effects than placebo, mixed action antidepressants and buspirone may have been no more likely to cause side effects than placebo, and we were uncertain about SSRI antidepressants.

Based on current research, all medicines should be considered still experimental.

Quality of the evidence

The quality of the evidence for many of the outcomes in this review was low or very low because each medicine was investigated by a small number of studies (ranging from one to four), each study involved small numbers of participants, there was some inconsistency in the findings and there was a risk of bias due to study participants dropping out of treatment.

Summary of findings

Background

Description of the condition

Cannabis production and consumption is highly prevalent and widespread globally (World Drug Report 2017). It is estimated that 3.8% of the global adult population used cannabis in the past year (World Drug Report 2017).

Cannabis use disorders are the reason for treatment in around half the people seeking treatment for the first time at the global level (World Drug Report 2017). Cannabis is identified as the primary drug of concern for more than half of people in treatment for drug use in Africa and Oceania (World Drug Report 2017). Cannabis use within some indigenous communities in North America and Australia may be more prevalent than for their non‐indigenous counterparts (Beauvais 2004; Clough 2004).

The main psychoactive compound in all cannabis products is Δ⁹‐tetrahydrocannabinol (THC; EMCDDA Cannabis Drug Profile). Cannabis use causes significant adverse effects (Budney 2007a). The acute effects of short‐term cannabis use include impaired memory (Solowij 2008); impaired motor co‐ordination with an associated increased risk of involvement in motor vehicle accidents (Hall 2009); altered judgement; and, in high doses, paranoia and psychosis (Volkow 2014). Long‐term or heavy use of cannabis has been associated with: the development of dependence (Budney 2007a), chronic bronchitis and increased risk of chronic psychotic disorders in people with a predisposition for development of such disorders (Volkow 2014). When use is commenced early in adolescence, long‐term or heavy cannabis use has also been associated with altered brain development, poor educational outcome (Silins 2014; Silins 2015), cognitive impairment (Solowij 2008), and diminished life satisfaction and achievement (Gruber 2003).

One study using a large epidemiological survey in the USA estimated that 47.4% of males and 32.5% of females exposed to cannabis in their lifetime will develop a cannabis use disorder. For most people, the disorder would be cannabis abuse by Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM‐IV) criteria, but approximately 7.0% of males and 5.3% of females who use cannabis at some point in their life would be likely to develop cannabis dependence (Lev‐Ran 2013a). It has been estimated that, globally in 2017, more than 19 million people were cannabis dependent (Peacock 2018).

As with other drugs of dependence, the risk of developing dependency is influenced by multiple factors. However, intensive use of cannabis, that is daily or near daily use, is likely to increase the risk of cannabis dependence (EMCDDA 2004). It has been suggested that the earlier initiation of cannabis use (Copeland 2014), use of more potent forms of cannabis (e.g. the flowering heads of the female cannabis plant) and the greater use of water‐pipes may have led to an increased amount of THC consumption by some cannabis users and, therefore, possibly greater rates of cannabis dependence (Hall 2001).

The use of cannabis has consistently been associated with psychotic symptoms (Minozzi 2010), and may be associated with the earlier onset of psychotic illness in some people (Large 2011). Cannabis use has been associated with a range of mental health disorders, such as anxiety and mood disorders (Lev‐Ran 2013b). These associations are particularly pronounced with bipolar disorder, substance use disorders and specific (antisocial, dependant and histrionic) personality disorders (Lev‐Ran 2013b).

Estimates of the number of people who use cannabis and experience withdrawal are variable (Agrawal 2008; Budney 2006; Chung 2008; Copersino 2006; Cornelius 2008; Hasin 2008). Evidence regarding factors influencing the severity of cannabis withdrawal remains limited, but there is evidence that the amount of cannabis smoked is predictive of the intensity of withdrawal during abstinence from cannabis (McClure 2012). Smoking behaviour also appears to be a strong predictor for the severity of cannabis dependence (van der Pol 2014).

General acceptance of a specific cannabis withdrawal syndrome is indicated by the inclusion of diagnostic criteria for cannabis withdrawal in the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM‐5). The DSM‐5 defines cannabis withdrawal by development of three or more of the following signs and symptoms within approximately one week of cessation of heavy and prolonged cannabis use: 1. irritability, anger or aggression; 2. nervousness or anxiety; 3. sleep difficulty; 4. decreased appetite or weight loss; 5. restlessness; 6. depressed mood; and 7. at least one of the following physical symptoms causing significant discomfort: stomach pain, shakiness or tremors, sweating, fever, chills or headache (DSM‐5). Onset of symptoms is usually within 24 to 48 hours of abstinence, reaching peak intensity within the first week (Budney 2007a). Symptoms may persist for up three to four weeks (Milin 2008), although there appears to be significant individual variability. Cannabis withdrawal is not life threatening, neither is it associated with significant medical or psychiatric consequences (Budney 2003).

Demand for treatment by people who use cannabis has generally increased worldwide since the mid to late 2000s, albeit with significant regional variation. The World Drug Report gives data on treatment demand in terms of the proportion of treatment services provided for the major drugs of dependence. People who use cannabis have dominated demand for drug treatment in Africa since the mid to late 2000s with treatment rates consistently over 60%. Demand for cannabis treatment has grown significantly in some regions, more than doubling in Europe and South America and more than trebling in Oceania (World Drug Report 2017). North America as a whole was the only region to see a decrease in the contribution of cannabis to treatment demand (World Drug Report 2017), but within the USA, cannabis admissions increased by 32% between 1996 and 2006 (SAMHSA 2008). With moves to decriminalise or legalise cannabis use in some parts of the world, the trend of increasing demand for treatment is likely to continue.

Increases in the THC content of cannabis may be a factor in the increasing demand for treatment. In the USA, THC content, as detected in confiscated samples, increased from about 3% in the 1980s to 12% in 2012 (Volkow 2014). Cannabis users adjust their smoking behaviour when smoking stronger cannabis but the adjustment does not fully compensate for the increased strength (van der Pol 2014). Hence, cannabis users would be expected to be exposed to higher doses of THC as a result of the increasing potency of cannabis preparations. Cannabis users who seek treatment typically have a long history of cannabis use disorder and multiple previous attempts to quit (Copeland 2014).

Description of the intervention

There are currently no accepted pharmacotherapies for the treatment of cannabis withdrawal or cessation. The identification and development of medications to fill this gap has long been a priority among researchers (Vandrey 2009), and a number of pharmacotherapies have been proposed as possible experimental interventions to attenuate the symptoms and signs of cannabis withdrawal and to promote cessation.

These medications are diverse in nature, encompassing medications that affect cannabinoid receptor systems (e.g. preparations of THC), medications that affect dopamine pathways, medications that affect the specific symptoms of cannabis withdrawal or that have been used in managing withdrawal from other substances, and medications that affect mental health conditions, such as depression, that may be factors contributing to cannabis use.

How the intervention might work

The proposed pharmacological interventions may potentially lessen the symptoms and signs of cannabis withdrawal, including craving. The availability of effective pharmacotherapy for cannabis withdrawal may encourage people who are cannabis dependent to enter treatment, and may increase the rates of completion of withdrawal, cessation of cannabis use and entry into relapse prevention treatment.

It has been reported that the experience of cannabis withdrawal symptoms may be a significant obstacle to the achievement of abstinence by people who are cannabis dependent (Budney 2006; Copeland 2001; Hart 2005). Therefore, the effective treatment of the cannabis withdrawal syndrome may promote cessation of cannabis use and provide a first step towards abstinence and recovery.

Why it is important to do this review

As discussed above, there is increasing recognition that cannabis use and dependence is an important public health issue.

Not all people who use cannabis will need pharmacotherapies to manage withdrawal or support cessation of their use. However, it is important that effective pharmacotherapies are identified for the treatment of cannabis withdrawal, especially in intensive cannabis users who describe withdrawal symptoms on cessation. As such, with the increase in the number of pharmacotherapies tested, this review sought to establish current knowledge on the effectiveness of different medications in the treatment of cannabis dependence.

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Randomised and quasi‐randomised controlled trials that provided detailed information on the type and dose of intervention medication used and the characteristics of participants treated.

Types of participants

We included studies that involved participants diagnosed as cannabis dependent or who were likely to be dependent based on reported dose, duration and frequency of use (daily or multiple days per week).

We included studies involving participants dependent on, and withdrawing from, both cannabis and nicotine, but excluded studies involving participants dependent on and withdrawing from substances other than cannabis and nicotine.

We included studies undertaken in either inpatient or outpatient settings. We excluded studies undertaken in purely research settings, such as residential research laboratories. Some of these studies provide insight into the effect of different medications on signs and symptoms of cannabis withdrawal and are considered in the 'Discussion' section. However, such studies generally involved participants who were not seeking treatment for cannabis use and cessation of cannabis use was not the goal of the interventions provided, and the nature of outcomes assessed were generally different to those expected of treatment interventions. For these reasons, we excluded such studies from this review.

We excluded studies involving participants with diagnosed schizophrenia and cannabis use disorder. The primary therapeutic goal in these studies was management of psychotic symptoms, with consideration of the effect of different antipsychotic medications on cannabis use. This limits the application of findings of such studies to the general group of people with cannabis use disorders. Cannabis use in schizophrenia is considered by a separate Cochrane Review (McLoughlin 2014).

Types of interventions

Experimental interventions involved the administration of any medication with the aim of reducing the symptoms and signs of cannabis withdrawal or promoting cessation of cannabis use.

Comparison interventions involved the use of different pharmacotherapies, placebo or no pharmacotherapy (supportive care).

Types of outcome measures

Primary outcomes

Number of participants abstinent from cannabis at the end of treatment as determined by self‐report or urine drug screens, or both.
Intensity of withdrawal as determined by scores on withdrawal scales, the need for symptomatic medications in addition to the experimental intervention or overall assessments by clinicians and participants.
Nature, incidence and frequency of adverse effects and whether the planned medication regimen was modified in response to adverse effects.
Completion of scheduled treatment.

Secondary outcomes

Level of cannabis use at the end of treatment as measured via participant‐reported level of use or urine drug screens, or both.
Number of participants engaged in further treatment following completion of the withdrawal intervention. As discussed in the 'Background' section, treatment of the cannabis withdrawal period may be considered as the first step in treatment, therefore engagement in further relapse prevention treatment may be considered to be a valid outcome of interest.

Search methods for identification of studies

All searches included non‐English language literature. We found no studies in languages other than English.

Electronic searches

We searched:

Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library 2018, Issue 2;
MEDLINE (1946 to week 3 March 2018) via Ovid Online;
Embase Online (26 March 2018);
PsycINFO (1806 to week 3 March 2018) via Ovid Online;
Web of Science, online (26 March 2018).

We developed a search strategy to retrieve references relating to the pharmacological treatment of cannabis withdrawal. This strategy was adapted to each of the databases listed above.

For details see Appendix 1; Appendix 2; Appendix 3; Appendix 4; Appendix 5.

We also searched the following electronic sources of ongoing trials:

ANZCTR registry (January 2018);
ClinicalTrials.gov (www.clinicaltrials.gov; January 2018).

Searching other resources

We checked the reference lists of relevant review articles and retrieved studies to identify any further studies of interest that were not retrieved by the electronic search. We contacted selected researchers who were active in the area seeking information about unpublished study reports. We also checked conference proceedings likely to contain trials relevant to the review.

Data collection and analysis

Selection of studies

Two authors (LG and SN or PS) independently assessed the titles and abstracts of records retrieved from the systematic search according to the identified inclusion and exclusion criteria. All authors agreed on the inclusion and exclusion decisions. We made no attempt to blind the authors to the names of the study authors, institutions, journal of publication and results when applying eligibility criteria.

Data extraction and management

Two authors (LG and SN) extracted key information from the included studies using a data collection form to record information against the outcome measures (abstinence, intensity of withdrawal, adverse effects, completion of treatment, change in cannabis use and engagement in follow‐up treatment) and study characteristics (location, participant characteristics, interventions, study design and source of funding). We confirmed data by consultation with the other review authors. We summarised key findings of studies descriptively in the first instance and considered the capacity for quantitative meta‐analysis.

Assessment of risk of bias in included studies

We assessed the risk of bias of included studies according to the approach recommended in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). This was based on the evaluation of seven specific methodological domains (namely, sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting and other bias). For each study, we analysed the seven domains, described them as reported in the study and provided a final judgement on the likelihood of bias in terms of low, high or unclear risk of bias. We based these judgements on the criteria indicated by the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), and their applicability to the addiction field.

In general, subjective outcomes are more prone to performance and detection bias than objective outcomes. The outcome measures in this review that included a subjective component were self‐reported cannabis use (including abstinence at the end of treatment) and intensity of withdrawal. When considering the risk of bias due to blinding of participants and personnel (performance bias) and blinding of outcome assessment (detection bias), we based the judgement on the outcomes most prone to bias, and also considered the use of measures, such as urinalysis, to support the subjective assessment. We only considered incomplete outcome data for the intensity of withdrawal, change in cannabis use, and nature and incidence of adverse effects. Retention in treatment (duration of treatment) and completion of treatment are frequently primary outcome measures in addiction research. See Appendix 6 for a detailed description of the criteria we considered in the 'Risk of bias' assessment.

Details of the assessments of risk of bias are included in the Characteristics of included studies table.

Measures of treatment effect

Where possible, for dichotomous outcomes (e.g. number completing treatment), we calculated risk ratios (RR) with 95% confidence intervals (CI). There were no continuous data but the intention was to express continuous outcomes as a mean difference where there was a comparable outcome measure (e.g. time in treatment) or as a standardised mean difference where there was variability in the outcome measure (e.g. withdrawal assessment scales), with 95% CIs.

Unit of analysis issues

One included study involved three treatment arms (two different active medications and placebo) (Carpenter 2009). The active medications, compared to placebo, were included in separate comparisons thereby avoiding the unit of analysis error of double‐counting participants. Where urine drug screens were reported in studies, the unit of analysis was the number of study participants and not the number of tests performed.

Dealing with missing data

We contacted original investigators to request missing data, and, where unpublished data were available, included these in the analyses and noted in the study record. We also checked Clinicaltrials.gov where data were missing. It was also intended to use sensitivity analysis to assess the impact of different approaches to handling missing data, but there were insufficient data for this.

Assessment of heterogeneity

We assessed clinical and methodological heterogeneity by reviewing the variation between studies in terms of the characteristics of participants included, the interventions and the reported outcomes. We grouped studies for analyses by the nature of the medication used (experimental intervention).

We assessed statistical heterogeneity using the Chi² test and its P value, by visual inspection of the forest plots and the I² statistic. A P value of the Chi² test lower than 0.10 or an I² statistic of at least 50% indicated significant statistical heterogeneity.

Data synthesis

We used Review Manager 5 for statistical analyses (Review Manager 2014). In all analyses, we used a random‐effects model.

Subgroup analysis and investigation of heterogeneity

This review aimed to consider the following potential sources of heterogeneity through subgroup analyses:

patterns of cannabis use and the estimated level of THC intake (as indicated by duration and level of use, number of days of use, number of uses per day (frequency), modality of use or route of administration, age at initiation of use);
concurrent tobacco smoking;
concurrent psychiatric illness and current treatment for a psychiatric illness;
the nature of the treatment setting;
the nature of adjunct treatment.

None of these analyses were possible due to limitations of the studies that met the inclusion criteria.

Sensitivity analysis

We did not use risk of bias as a criterion for inclusion in this review. We intended to assess the impact of risk of bias through sensitivity analysis. This would have involved considering the overall estimate of effect with studies with a high risk of bias included or excluded. Limitations of data reported by the studies that met the inclusion criteria meant that sensitivity analysis was not possible. However, we discussed the risk of bias when presenting the results.

Grading of evidence

We assessed the overall quality of the evidence for the primary outcomes using the system developed by the GRADE Working Group for grading the quality of evidence (Schűnemann 2013). GRADE takes into account issues not only related to internal validity but also to external validity, such as directness, consistency, imprecision of results and publication bias. The 'Summary of findings' tables present the main findings of a review in a transparent and simple tabular format. In particular, they provide key information concerning the quality of evidence, the magnitude of effect of the interventions examined and the sum of available data on the main outcomes.

The GRADE system uses the following criteria for assigning grades of evidence.

High: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Grading of the quality of randomised controlled trials is decreased for the following reasons.

Serious (–1) or very serious (–2) study limitation for risk of bias.
Serious (–1) or very serious (–2) inconsistency between study results.
Some (–1) or major (–2) uncertainty about directness (the correspondence between the population, the intervention or the outcomes measured in the studies actually found and those under consideration in our systematic review).
Serious (–1) or very serious (–2) imprecision of the pooled estimate (–1).
Publication bias strongly suspected (–1).

Results

Description of studies

See Characteristics of included studies; Characteristics of excluded studies tables.

Results of the search

Our search strategy identified 1364 records through database searching and five additional records from other sources. After removing duplicates, there were 809 unique records. We excluded 720 records based on title and abstract and obtained 89 full‐text for further assessment. On reading the full text, we excluded 43 reports (relating to 34 studies) with reasons (see Characteristics of excluded studies table). We included 21 studies (45 reports) in the review (see Figure 1).

Included studies

Twenty‐one randomised controlled trials (45 reports) involving 1755 participants met the inclusion criteria for this review (Allsop 2014; Carpenter 2009; Cornelius 2010; Frewen 2007; Gray 2012; Gray 2017; Johnston 2014; Levin 2004; Levin 2011; Levin 2013; Levin 2016; Mason 2012; McRae‐Clark 2009; McRae‐Clark 2010; McRae‐Clark 2015; McRae‐Clark 2016; Miranda 2017; Penetar 2012; Sherman 2017; Trigo 2018; Weinstein 2014; see Characteristics of included studies table). In total, 909 participants received active medication and 846 participants received placebo. One study reported only the total number of participants (81) and we assumed group sizes of 41 and 40 (Frewen 2007). All studies offered participants some form of psychological therapy in addition to medication or placebo.

All studies involved a comparison between an active medication and placebo but medications were diverse in nature. The medications investigated, grouped according to type and mechanism of action, were:

preparations containing THC: dronabinol (Levin 2011), dronabinol plus lofexidine (Levin 2016), nabiximols (Allsop 2014; Trigo 2018);
selective serotonin reuptake inhibitor (SSRI) antidepressants: fluoxetine (Cornelius 2010), escitalopram (Weinstein 2014), vilazodone (McRae‐Clark 2016);
mixed action antidepressants (noradrenergic and serotonergic effects): nefazodone (Carpenter 2009), mirtazapine (Frewen 2007), venlafaxine (Levin 2013);
anticonvulsant and mood stabilisers: divalproex sodium (Levin 2004), gabapentin (Mason 2012), lithium (Johnston 2014), topiramate (Miranda 2017);
atypical antidepressant (dopamine reuptake inhibitor and weak noradrenaline reuptake inhibitor): bupropion (Carpenter 2009; Penetar 2012);
anxiolytic (serotonin 5‐HT_1A partial agonist): buspirone (McRae‐Clark 2009, McRae‐Clark 2015);
selective noradrenaline reuptake inhibitor: atomoxetine (McRae‐Clark 2010);
supplement promoting glutamate release and modulating N‐methyl‐D‐aspartate (NMDA) receptor, N‐acetylcysteine: (Gray 2012; Gray 2017); and
neuropeptide, oxytocin (Sherman 2017).

All except two of the studies were undertaken in outpatient settings. Allsop 2014 and Johnston 2014 were primarily studies of cannabis withdrawal, with medication administered in an inpatient (hospital) setting over six or seven days, with follow‐up interviews after discharge. The scheduled duration for outpatient studies ranged from three weeks (Penetar 2012) to 13 weeks (Carpenter 2009), with 11 to 13 weeks being most common.

Sixteen studies were undertaken in the USA, with three studies in Australia (Allsop 2014; Frewen 2007; Johnston 2014), one study in Israel (Weinstein 2014), and one in Canada (Trigo 2018). Twenty studies reported the source of funding to be (government) research grants; the funding source was unclear for one study (Frewen 2007). Five studies received medications from the manufacturing company (Allsop 2014; Levin 2016; McRae‐Clark 2010; McRae‐Clark 2016; Trigo 2018). Researchers associated with 10 studies declared past or current associations with pharmaceutical companies. Researchers associated with eight studies declared no conflict of interest. Authors of the remaining studies made no declarations.

Four studies included participants with cannabis use disorders as well as cannabis dependence, but the majority of participants met diagnostic criteria for cannabis dependence (Cornelius 2010; Levin 2013; Miranda 2017; Penetar 2012). In the other studies, all participants were cannabis dependent.

The mean age of study participants ranged from 22 to 41 years, and for 12 studies it was between 30 and 35 years; one study did not provide data on age (Penetar 2012). The target population for three studies was adolescents and young adults (Cornelius 2010; Gray 2012; Miranda 2017). The mean age of participants in these studies was 21 (Cornelius 2010), 19 (Gray 2012), and 20 years (Miranda 2017).

One study did not provide information on the gender of participants (Penetar 2012); the majority (61% to 93%) of participants in 19 studies were male. Miranda 2017 was notable in having a smaller proportion (48%) of males.

Participants in two studies had comorbid depression and cannabis use disorders (Cornelius 2010; Levin 2013), and in one study participants met diagnostic criteria for attention deficit hyperactivity disorder (ADHD) as well as cannabis dependence (McRae‐Clark 2010). Gray 2012 reported that 13.8% of participants had some psychiatric comorbidity, but 16 studies excluded people with significant or unstable psychiatric conditions. One study did not report on this aspect (Penetar 2012).

Excluded studies

We excluded 34 studies (43 reports) that were potentially relevant to the review and assessed in detail from the review (see Figure 1; Characteristics of excluded studies table). The reasons for exclusion were: study was exploratory (mostly laboratory‐based) research with participants who were not seeking treatment (13 studies); minority of participants were cannabis dependent, or dependence was unclear (six studies); no treatment comparison (six studies); comparison of antipsychotic drugs for treatment of schizophrenia with concurrent cannabis use (five studies); no treatment intervention, or no medication treatment (five studies); and cannabis use was not the main focus of the treatment intervention (three studies). Five studies were excluded for more than one reason.

Risk of bias in included studies

For summary results of the judged risk of bias across the included studies for each domain, see Figure 2 and Figure 3.

Methodological quality summary: review authors' judgements about each methodological quality item for each included study.

Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.

Allocation

We judged seven studies at unclear risk of bias due to the reporting of insufficient information on the method of random sequence generation (Carpenter 2009; Johnston 2014; Levin 2004; McRae‐Clark 2015; Penetar 2012; Sherman 2017; Weinstein 2014). We judged six studies at unclear risk of bias due to the reporting of insufficient information on the method of allocation concealment (Levin 2004; McRae‐Clark 2015; McRae‐Clark 2016; Penetar 2012; Sherman 2017; Weinstein 2014). We judged the remaining studies at low risk of allocation bias.

Blinding

We considered one study to have an unclear risk of both performance and detection bias because insufficient information was reported on the blinding of participants and personnel to make a judgement (McRae‐Clark 2016). All other studies were at low risk of performance and detection bias.

Incomplete outcome data

Completion of treatment is a primary outcome measure for this review. Hence, we only considered the risk of bias due to incomplete data for the outcomes of intensity of withdrawal, adverse effects and abstinence (or use of cannabis). We judged the risk of bias to be unclear for two studies (Frewen 2007; Mason 2012), and high for three studies (Levin 2004; Penetar 2012; Weinstein 2014).

Selective reporting

Frewen 2007 was a secondary analysis of data from a randomised controlled trial and reported some but not all findings from the main study. The full report of the study was not available and hence the risk of reporting bias was unclear. Penetar 2012 did not discuss adverse effects making it unclear whether adverse effects were systematically assessed during the study (unclear risk of reporting bias). On ClinicalTrials.gov, Sherman 2017 indicated "satisfaction with therapy" as a primary outcome but this was not reported in the publication associated with the study (high risk of reporting bias).

Other potential sources of bias

Johnston 2014 administered two‐thirds of participants a benzodiazepine for disturbed sleep. Although the use of such medications were stated to be similar in the two groups, the effect of this additional medication was unclear. We judged McRae‐Clark 2016 to have a high risk of other bias as the placebo group attended a greater proportion of scheduled visits and hence may have received more adjunct interventions.

Effects of interventions

See: Table 1; Table 2; Table 3; Table 4; Table 5; Table 6

Summary of findings for the main comparison. Δ⁹‐Tetrahydrocannabinol preparation compared to placebo for cannabis dependence.

THC preparation compared to placebo for cannabis dependence
Patient or population: cannabis dependence  Setting: inpatient or outpatient  Intervention: THC preparation  Comparison: placebo
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)
	Risk with placebo	Risk with THC preparation
Participants abstinent at end of treatment	Study population		RR 0.98  (0.64 to 1.52)	305  (3 RCTs)	⊕⊕⊕⊝  Moderate^a
	204 per 1000	200 per 1000  (131 to 310)
Participants experiencing adverse effects	Study population		RR 1.02 (0.89 to 1.17)	318  (3 RCTs)	⊕⊕⊕⊝  Moderate^a
	690 per 1000	704 per 1000  (614 to 807)
Participants withdrawn due to adverse effects	Study population		RR 2.72 (0.51 to 14.59)	318  (3 RCTs)	⊕⊕⊝⊝  Low^b
	13 per 1000	34 per 1000  (6 to 185)
Completion of scheduled treatment	Study population		RR 1.10 (0.88 to 1.37)	369  (4 RCTs)	⊕⊕⊝⊝  Low^a,c
	648 per 1000	0 per 1000  (0 to 0)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio; THC: Δ⁹‐tetrahydrocannabinol.
GRADE Working Group grades of evidence  High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

SSRI antidepressant compared to placebo for cannabis dependence
Patient or population: cannabis dependence  Setting: outpatient  Intervention: SSRI antidepressant  Comparison: placebo
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)
	Risk with placebo	Risk with SSRI antidepressant
Participants abstinent at end of treatment	Study population		RR 1.73 (0.61 to 4.89)	128  (2 RCTs)	⊕⊕⊝⊝  Low^a,b
	82 per 1000	142 per 1000  (50 to 401)
Participants experiencing adverse effects	Study population		RR 0.76 (0.57 to 1.02)	76  (1 RCT)	⊕⊝⊝⊝  Very low^a,c
	800 per 1000	608 per 1000  (456 to 816)
Participants withdrawn due to adverse effects	Study population		RR 1.71 (0.16 to 18.04)	76  (1 RCT)	⊕⊝⊝⊝  Very low^a,c
	29 per 1000	49 per 1000  (5 to 515)
Completion of scheduled treatment	Study population		RR 0.79 (0.49 to 1.27)	198  (3 RCTs)	⊕⊝⊝⊝  Very low^a,b,d
	680 per 1000	538 per 1000  (333 to 864)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio; SSRI: selective serotonin reuptake inhibitor.
GRADE Working Group grades of evidence  High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Mixed action antidepressant compared to placebo for cannabis dependence
Patient or population: cannabis dependence  Setting: outpatient  Intervention: mixed action antidepressant  Comparison: placebo
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)
	Risk with placebo	Risk with mixed action antidepressant
Participants abstinent at end of treatment	Study population		RR 0.82 (0.12 to 5.41)	179  (2 RCTs)	⊕⊕⊝⊝  Low^a,b
	250 per 1000	205 per 1000  (30 to 1000)
Participants experiencing adverse effects	Study population		RR 0.93 (0.55 to 1.55)	76  (1 RCT)	⊕⊕⊝⊝  Low^c
	450 per 1000	419 per 1000  (248 to 698)
Participants withdrawn due to adverse effects	Study population		RR 1.44 (0.11 to 18.90)	179  (2 RCTs)	⊕⊕⊝⊝  Low^a,b
	11 per 1000	16 per 1000  (1 to 205)
Completion of scheduled treatment	Study population		RR 0.93 (0.71 to 1.21)	169  (2 RCTs)	⊕⊕⊝⊝  Low^c
	573 per 1000	533 per 1000  (407 to 694)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence  High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Anticonvulsants and mood stabilisers compared to placebo for cannabis dependence
Patient or population: cannabis dependence  Setting: inpatient or outpatient  Intervention: anticonvulsants and mood stabilisers  Comparison: placebo
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)
	Risk with placebo	Risk with anticonvulsants and mood stabilisers
Participants abstinent at end of treatment	Study population		RR 1.14 (0.64 to 2.04)	48  (2 RCTs)	⊕⊝⊝⊝  Very low^a,b
	440 per 1000	502 per 1000  (282 to 898)
Participants withdrawn due to adverse effects	Study population		RR 3.67 (0.41 to 32.69)	116  (2 RCTs)	⊕⊝⊝⊝  Very low^b,c
	39 per 1000	144 per 1000  (16 to 1000)
Completion of scheduled treatment	Study population		RR 0.66 (0.47 to 0.92)	141  (3 RCTs)	⊕⊕⊝⊝  Low^b
	556 per 1000	367 per 1000  (261 to 511)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence  High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Buspirone compared to placebo for cannabis dependence
Patient or population: cannabis dependence  Setting: outpatient  Intervention: buspirone  Comparison: placebo
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)
	Risk with placebo	Risk with buspirone
Participants abstinent at end of treatment	Study population		RR 1.98 (0.62 to 6.33)	175  (1 RCT)	⊕⊕⊝⊝  Low^a
	46 per 1000	91 per 1000  (29 to 291)
Participants experiencing adverse effects	Study population		RR 1.14 (1.00 to 1.29)	225  (2 RCTs)	⊕⊕⊝⊝  Low^a
	763 per 1000	870 per 1000  (763 to 984)
Participants withdrawn due to adverse effects	Study population		RR 0.63 (0.15 to 2.60)	225  (2 RCTs)	⊕⊕⊝⊝  Low^a
	44 per 1000	28 per 1000  (7 to 114)
Completion of scheduled treatment	Study population		RR 0.96 (0.74 to 1.23)	225  (2 RCTs)	⊕⊕⊝⊝  Low^a
	526 per 1000	505 per 1000  (389 to 647)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence  High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

N‐acetylcysteine compared to placebo for cannabis dependence
Patient or population: cannabis dependence  Setting: outpatient  Intervention: N‐acetylcysteine  Comparison: placebo
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Certainty of the evidence  (GRADE)
	Risk with placebo	Risk with N‐acetylcysteine
Participants abstinent at end of treatment	Study population		RR 0.89 (0.59 to 1.35)	302  (1 RCT)	⊕⊕⊝⊝  Low^a
	242 per 1000	215 per 1000  (143 to 326)
Participants experiencing adverse effects	Study population		RR 0.94 (0.71 to 1.23)	418  (2 RCTs)	⊕⊕⊕⊝  Moderate^b
	329 per 1000	309 per 1000  (233 to 404)
Participants withdrawn due to adverse effects	Study population		RR 3.00 (0.12 to 72.15)	116  (1 RCT)	⊕⊕⊝⊝  Low^a
	0 per 1000	0 per 1000  (0 to 0)
Completion of scheduled treatment	Study population		RR 1.06 (0.93 to 1.21)	418  (2 RCTs)	⊕⊕⊕⊝  Moderate^b
	652 per 1000	691 per 1000  (607 to 789)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence  High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.  Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.  Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.  Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Date	Event	Description
15 May 2018	New citation required but conclusions have not changed	The updated search found seven additional studies.
9 April 2018	New search has been performed	New search, new studies, change in authorship

Item	Judgement	Description
1. Random sequence generation (selection bias)	Low risk	The investigators describe a random component in the sequence generation process such as: random number table; computer random number generator; coin tossing; shuffling cards or envelopes; throwing dice; drawing of lots; minimisation.
	High risk	The investigators describe a non‐random component in the sequence generation process such as: odd or even date of birth; date (or day) of admission; hospital or clinic record number; alternation; judgement of the clinician; results of a laboratory test or a series of tests; availability of the intervention.
	Unclear risk	Insufficient information about the sequence generation process to permit judgement of low or high risk.
2. Allocation concealment (selection bias)	Low risk	Investigators enrolling participants could not foresee assignment because 1 of the following, or an equivalent method, was used to conceal allocation: central allocation (including telephone, web‐based, and pharmacy‐controlled, randomisation); sequentially numbered drug containers of identical appearance; sequentially numbered, opaque, sealed envelopes.
	High risk	Investigators enrolling participants could possibly foresee assignments because 1 of the following method was used: open random allocation schedule (e.g. a list of random numbers); assignment envelopes without appropriate safeguards (e.g. if envelopes were unsealed or nonopaque or not sequentially numbered); alternation or rotation; date of birth; case record number; any other explicitly unconcealed procedure.
	Unclear risk	Insufficient information to permit judgement of low or high risk. This is usually the case if the method of concealment is not described or not described in sufficient detail to allow a definite judgement.
3. Blinding of participants and personnel (performance bias)	Low risk	No blinding or incomplete blinding, but the review authors judge that the outcome is not likely to be influenced by lack of blinding; blinding of participants and key study personnel, and unlikely that the blinding could have been broken.
	High risk	No blinding or incomplete blinding, and the outcome is likely to be influenced by lack of blinding. Blinding of participants and key study personnel attempted, but likely that the blinding could have been broken, and the outcome is likely to be influenced by lack of blinding.
	Unclear risk	Insufficient information to permit judgement of low or high risk.
4. Blinding of outcome assessment (detection bias)	Low risk	No blinding or incomplete blinding, but the review authors judge that the outcome is not likely to be influenced by lack of blinding. Blinding of outcome assessment ensured, and unlikely that the blinding could have been broken.
	High risk	No blinding or incomplete blinding of outcome assessment, and the measurement is likely to be influenced by lack of blinding. Blinding of outcome assessment, but likely that the blinding could have been broken, and the outcome measurement is likely to be influenced by lack of blinding.
	Unclear risk	Insufficient information to permit judgement of low or high risk.
5. Incomplete outcome data (attrition bias) For all outcomes except retention in treatment or drop out	Low risk	No missing outcome data. Reasons for missing outcome data unlikely to be related to true outcome (for survival data, censoring unlikely to be introducing bias). Missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups. For dichotomous outcome data, the proportion of missing outcomes compared with observed event risk not enough to have a clinically relevant impact on the intervention effect estimate. For continuous outcome data, plausible effect size (difference in means or standardised difference in means) among missing outcomes not enough to have a clinically relevant impact on observed effect size. Missing data have been imputed using appropriate methods. All randomised patients are reported/analysed in the group they were allocated to by randomisation irrespective of non‐compliance and cointerventions (intention to treat).
	High risk	Reason for missing outcome data likely to be related to true outcome, with either imbalance in numbers or reasons for missing data across intervention groups. For dichotomous outcome data, the proportion of missing outcomes compared with observed event risk enough to induce clinically relevant bias in intervention effect estimate. For continuous outcome data, plausible effect size (difference in means or standardised difference in means) among missing outcomes enough to induce clinically relevant bias in observed effect size. 'As‐treated' analysis done with substantial departure of the intervention received from that assigned at randomisation.
	Unclear risk	Insufficient information to permit judgement of low or high risk (e.g. number randomised not stated, no reasons for missing data provided; number of drop out not reported for each group).
6 Selective reporting (reporting bias)	Low risk	The study protocol is available and all of the study's prespecified (primary and secondary) outcomes that are of interest in the review have been reported in the prespecified way. The study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were prespecified (convincing text of this nature may be uncommon).
	High risk	Not all of the study's prespecified primary outcomes have been reported. 1 or more primary outcomes is reported using measurements, analysis methods or subsets of the data (e.g. subscales) that were not prespecified. 1 or more reported primary outcomes were not prespecified (unless clear justification for their reporting is provided, such as an unexpected adverse effect). 1 or more outcomes of interest in the review are reported incompletely so that they cannot be entered in a meta‐analysis. The study report fails to include results for a key outcome that would be expected to have been reported for such a study.
	Unclear risk	Insufficient information to permit judgement of low or high risk.
7. Other bias	Low risk	Potential confounding factors identified but evenly distributed between groups. Study ceased early but with no indications of selection bias. Interventions delivered consistently.
	High risk	Potential confounding factors unequally distributed between groups. Study ceased early with risk of selection bias. Differences in aspects of delivery of interventions. Mandatory treatment.
	Unclear risk	Confounding possible but not able to be assessed. Study ceased early and unable to determine possible bias. Unclear if delivery of interventions was equivalent.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Participants abstinent at end of treatment	3	305	Risk Ratio (M‐H, Random, 95% CI)	0.98 [0.64, 1.52]
2 Participants experiencing adverse effects	3	318	Risk Ratio (M‐H, Random, 95% CI)	1.02 [0.89, 1.17]
3 Participants withdrawn due to adverse effects	3	318	Risk Ratio (M‐H, Random, 95% CI)	2.72 [0.51, 14.59]
4 Completion of scheduled treatment	4	369	Risk Ratio (M‐H, Random, 95% CI)	1.10 [0.88, 1.37]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Participants abstinent at end of treatment	2	128	Risk Ratio (M‐H, Random, 95% CI)	1.73 [0.61, 4.89]
2 Participants experiencing adverse effects	1	76	Risk Ratio (M‐H, Random, 95% CI)	0.76 [0.57, 1.02]
3 Participants withdrawn due to adverse effects	1	76	Risk Ratio (M‐H, Random, 95% CI)	1.71 [0.16, 18.04]
4 Completion of scheduled treatment	3	198	Risk Ratio (M‐H, Random, 95% CI)	0.79 [0.49, 1.27]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Participants abstinent at end of treatment	2	179	Risk Ratio (M‐H, Random, 95% CI)	0.82 [0.12, 5.41]
2 Participants experiencing adverse effects	1	76	Risk Ratio (M‐H, Random, 95% CI)	0.93 [0.55, 1.55]
3 Participants withdrawn due to adverse effects	2	179	Risk Ratio (M‐H, Random, 95% CI)	1.44 [0.11, 18.90]
4 Completion of scheduled treatment	2	169	Risk Ratio (M‐H, Random, 95% CI)	0.93 [0.71, 1.21]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Participants abstinent at end of treatment	2	48	Risk Ratio (M‐H, Random, 95% CI)	1.14 [0.64, 2.04]
2 Participants withdrawn due to adverse effects	2	116	Risk Ratio (M‐H, Random, 95% CI)	3.67 [0.41, 32.69]
3 Completion of scheduled treatment	3	141	Risk Ratio (M‐H, Random, 95% CI)	0.66 [0.47, 0.92]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Participants abstinent at end of treatment	1	175	Risk Ratio (M‐H, Random, 95% CI)	1.98 [0.62, 6.33]
2 Participants experiencing adverse effects	2	225	Risk Ratio (M‐H, Random, 95% CI)	1.14 [1.00, 1.29]
3 Participants withdrawn due to adverse effects	2	225	Risk Ratio (M‐H, Random, 95% CI)	0.63 [0.15, 2.60]
4 Completion of scheduled treatment	2	225	Risk Ratio (M‐H, Random, 95% CI)	0.96 [0.74, 1.23]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Participants experiencing adverse effects	1	38	Risk Ratio (M‐H, Random, 95% CI)	1.18 [0.95, 1.46]
2 Participants withdrawn due to adverse effects	1	38	Risk Ratio (M‐H, Random, 95% CI)	3.0 [0.13, 69.31]
3 Completion of scheduled treatment	1	38	Risk Ratio (M‐H, Random, 95% CI)	1.29 [0.60, 2.74]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Participants abstinent at end of treatment	1	302	Risk Ratio (M‐H, Random, 95% CI)	0.89 [0.59, 1.35]
2 Participants experiencing adverse effects	2	418	Risk Ratio (M‐H, Random, 95% CI)	0.94 [0.71, 1.23]
3 Participants withdrawn due to adverse effects	1	116	Risk Ratio (M‐H, Random, 95% CI)	3.0 [0.12, 72.15]
4 Completion of scheduled treatment	2	418	Risk Ratio (M‐H, Random, 95% CI)	1.06 [0.93, 1.21]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Participants abstinent at end of treatment	1	16	Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
2 Participants experiencing adverse effects	1	16	Risk Ratio (M‐H, Random, 95% CI)	0.50 [0.06, 4.47]
3 Completion of scheduled treatment	1	16	Risk Ratio (M‐H, Random, 95% CI)	0.86 [0.53, 1.38]

Methods	Double‐blind, randomised, placebo‐controlled trial
Participants	Setting: inpatient (2 hospitals), New South Wales, Australia. Total duration of inpatient admission: 9 days 51 adults seeking treatment for cannabis use, dependent by DSM‐IV‐TR Group sizes: group 1, 27; group 2, 24 Groups well matched apart from differences in baseline withdrawal score and disability scale scores. Mean age 35 years 76% male 53% unemployed; 25% married or in de facto relationship On average using 23 g cannabis per day, mean duration of use 20 years; 71% also nicotine dependent Dependence on alcohol or other drugs except nicotine or caffeine and unstable medical or psychiatric conditions were exclusion criteria.
Interventions	Group 1: nabiximols (cannabis extract, Sativex®), maximum dose THC 86.4 mg, cannabidiol 80 mg; medication: 6 days, washout: 3 days Group 2: placebo Cognitive‐behavioural therapy tailored to inpatient cannabis withdrawal as adjunct intervention Follow‐up interview after 28 days. Participants compensated AUD 40 for follow‐up interviews
Outcomes	Overall withdrawal score, irritability, craving and depression reported as graphs and results of statistical analyses with imputation for missing data Number retained in treatment at all time points, median days inpatient stay Withdrawal and craving assessed with Cannabis Withdrawal Scale (19 items on 11‐point Likert scale for the previous 24 hours) Drug use by modified TLFB Change in amount of cannabis use from baseline to 28‐day follow‐up
Notes	Funding: research grant (Australian National Health and Medical Research Council), with study drugs provided by manufacturer (GW Pharmaceuticals, UK). Declaration of conflict of interest not published.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "An independent statistician generated a randomization list for each site using random block sizes in Stata, version 11.1 …"
Allocation concealment (selection bias)	Low risk	Comment: method of allocation concealment not reported, but generation of lists by independent statistician and use of matching placebos would be expected to provide good control of bias.
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Quote: "Patients, investigators, and outcome assessors were blind to treatment allocation until all research procedures were complete. Blinding was maintained by the use of a matched placebo … The success of patient blinding was formally assessed before hospital discharge."
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Quote: "Patients, investigators, and outcome assessors were blind to treatment allocation until all research procedures were complete. Blinding was maintained by the use of a matched placebo … The success of patient blinding was formally assessed before hospital discharge."
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Statistical methods used to impute missing data and assess data as missing at random.
Selective reporting (reporting bias)	Low risk	None apparent
Other bias	Low risk	None apparent

Methods	Multisite, double‐blind randomised, placebo‐controlled trial
Participants	Setting: outpatient, 6 sites within the National Drug Abuse Treatment Clinical Trials Network, USA. Scheduled duration 12 weeks Participants: 302 adults aged 18–50 years, seeking treatment, cannabis dependent by DSM‐IV‐TR and positive urine test during the initial screening visit Group sizes: group 1, 153; group 2, 149 Groups similar on demographics and baseline clinical characteristics, except employment (group 1, 42.5%; group 2, 60.4% working) and education (group 1, 28.5%; group 2, 14.8% graduated from high school) Mean age 30.3 years 71.5% male 58.3% white Mean cannabis use 26.0/30 days at baseline. Exclusion criteria: unstable psychiatric conditions, dependence on substances other than cannabis or tobacco, recent synthetic cannabinoid use
Interventions	Group 1: N‐acetylcysteine, 1200 mg twice daily Group 2: placebo All participants received contingency management and medical management
Outcomes	Odds of negative urine cannabinoid tests during treatment Cannabis urine screens by central laboratory during weekly clinic visits and at post‐treatment follow‐up Adverse effects at each study visit Adherence defined as taking ≥ 80% of prescribed medication each study week, confirmed by riboflavin level
Notes	Funding: research grants (US National Institutes of Health) 5/23 authors declared research support, or unrestricted grants from pharmaceutical companies, or the provision of advisory and consultancy services to pharmaceutical companies, none of which were involved in this trial.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Randomization, conducted centrally through the CTN Data and Statistics Center, was on a 1:1 ratio, with stratification by study site and self‐reported binary tobacco smoking status."
Allocation concealment (selection bias)	Low risk	Quote: "Randomization, conducted centrally through the CTN Data and Statistics Center;" "United States Pharmacopeia grade NAC [N‐acetylcysteine] powder was encapsulated in 600 mg quantities (two 600 mg capsules per twice‐daily dose). Matched placebo capsules were also prepared."
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Quote: "Among participants assigned to NAC, 46.5% guessed they were receiving NAC and 53.5% guessed they were receiving placebo, and the medical clinician guessed that 52% were receiving NAC and 48% were receiving placebo. Among those assigned to placebo, 53.7% guessed they were receiving NAC and 46.3% guessed they were receiving placebo, and the medical clinician guessed that 57.3% were receiving NAC and 42.7% were receiving placebo. These differences were not statistically significant, and the participant and medical clinician agreed on guesses more often than by chance (p < 0.0001)."
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	As indicated above, participant may have been able to accurately guess their group allocation, but the outcomes reported were objective in nature and these are less likely to be affected if the blind is broken.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Retention rates and data availability similar in the 2 groups Missing urine tests imputed as positive
Selective reporting (reporting bias)	Low risk	Published protocol, reported outcomes consistent with protocol
Other bias	Low risk	Riboflavin added to all capsules as a biomarker for medication adherence. Criteria defining adherence specified in advance. Analyses comparing outcomes for those meeting or not meeting these criteria were undertaken.

Methods	Randomised, double‐blind, placebo‐controlled, trial. 1‐week placebo lead‐in phase; those who used cannabis less than twice a week during this phase were not randomised.
Participants	Setting: outpatient with clinic attendance twice weekly, New York, USA. Scheduled duration 12 weeks Participants: 156 adults seeking outpatient treatment for problems related to cannabis use, dependent by DSM‐IV‐TR, using cannabis ≥ 5 days/week in prior 28 days Group sizes: group 1, 79; group 2, 77 No significant group differences in demographic or clinical characteristics at baseline Mean age 38 years 82% male 60% employed full‐time, 13% part‐time; 31% married Exclusion criteria: significant psychiatric condition and dependence on other substances except nicotine
Interventions	Group 1: oral dronabinol, commenced at 10 mg/day, titrated to 20 mg twice a day or the maximum tolerated Group 2: placebo Medication maintained to end of week 8 then tapered over 2 weeks Weekly individual therapy based on coping skills plus MET as adjunct intervention. Participants earned vouchers with value increased by USD 1.50 for each consecutive visit, with value reset for non‐attendance, and USD 10 for returning their tablet bottle and remaining medication. Maximum possible earnings were USD 570. Cash payments of USD 5–20 were made at each visit for transport costs.
Outcomes	Number achieving 2 weeks' abstinence in weeks 7 and 8 and median maximum consecutive days abstinence Number retained in study to week 8 Mean number of therapy sessions attended Number experiencing any adverse effects, requiring dose reduction, serious adverse events and number withdrawn due to adverse events Withdrawal scores reported as graph and results of statistical modelling Medication compliance Cannabis use assessed by TLFB. Urine samples tested at each clinic visit for confirmation of self‐report Withdrawal symptoms assessed twice weekly using the Withdrawal Discomfort Score (10 items, scores 0–30) Craving by MCQ with the 47‐item version completed once a month, and the 12‐item version weekly Adverse effects assessed twice a week using the Modified Systematic Assessment for Treatment and Emergent Events (SAFTEE)
Notes	Funding: research grant (NIDA) 1 author declared prior associations with pharmaceutical companies
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "participants were randomized … using a fixed block size of 4, stratified by joints used per week … and whether or not they were receiving a psychotropic medication."
Allocation concealment (selection bias)	Low risk	Quote: "A research pharmacist, who was independent of the research team, conducted the randomization."
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Quote: "Donabinol … or matching placebo … was prepared by the pharmacy … packaged in matching gelatin capsules with lactose filler and an equal amount of riboflavin. All capsules looked identical …"
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Study conducted double‐blind as indicated above. Participants may have been able to distinguish the effects of dronabinol, but use of urine screening to support self‐report would be expected to reduce risk of bias.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Quote: "All analyses were conducted on the intent‐to‐treat population." "… missing data in weeks 7 and 8 were scored as indicating cannabis use …"
Selective reporting (reporting bias)	Low risk	None apparent
Other bias	Low risk	None apparent

Methods	Randomised double‐blind, placebo‐controlled trial. 1‐week placebo lead‐in phase prior to randomisation
Participants	Setting: outpatient, New York, USA. Scheduled duration 11 weeks Participants: 122 adults seeking treatment, cannabis dependent by DSM‐IV, daily smoking in 28 days prior to study entry and THC‐positive urine sample on day of study entry Group sizes: 61 in each group Groups similar on demographic and clinical characteristics at baseline Mean age 35.2 years 68.9% male 19.7% currently married, 41.0% unemployed Exclusion criteria: severe mental illness, dependence on other drugs except nicotine
Interventions	Group 1: dronabinol 20 mg + lofexidine 0.6 mg 3 times a day; medications combined in capsule Group 2: placebo Medications were maintained until the end of week 8, were then tapered over 2 weeks and participants were monitored off medications during the last study week. All participants received weekly motivational enhancement and relapse prevention therapy.
Outcomes	Number achieving "consecutive abstinence" (defined as 21 consecutive days of abstinence based on TLFB self‐report during titration and maintenance phase) Marijuana craving assessed weekly using the modified 12‐item MCQ Marijuana withdrawal assessed using a 10‐item self‐report checklist with each item rated 0–3 for a possible total score of 30
Notes	Funding: research grant (NIDA) 2 authors declared associations with pharmaceutical companies Additional data on clinicaltrials.gov/ct2/show/results/NCT01020019
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Participants were randomized at the end of the placebo lead‐in phase using computer generated random blocks of sizes 4, 6, and 8, with a 1:1 allocation ratio stratified by joints used per week."
Allocation concealment (selection bias)	Low risk	Quote: "A Ph.D. statistician at Columbia University independent of the research team conducted the randomization and maintained the allocation sequence."
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Quote: "Participants, investigators and study staff were blind to allocation" … "Lofex–Dro or matching placebo (PBO) was prepared by the un‐blinded pharmacy …, packaged in matching gelatin capsules with lactose filler and an equal amount of riboflavin …"
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Study undertaken double‐blind as indicated above
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Quote: "The observed proportion of subjects completing the titration and maintenance medication phases of the trial were 37/61 (60.66%) in Lofex–Dro and 42/61 (68.85%) in PBO. " "There was not a significant difference in retention between treatment groups." Unclear how missing data were handled but would equally effect both groups.
Selective reporting (reporting bias)	Low risk	None apparent, reported outcomes consistent with clinical trials registration
Other bias	Low risk	None apparent

Methods	Randomised, double‐blind, placebo controlled trial
Participants	Setting: outpatient with weekly clinic visits, California, USA. Scheduled duration 12 weeks Participants: 50 adults, seeking treatment, current cannabis dependence by DSM‐IV, smoked cannabis at least once in week prior to randomisation Group sizes: 25 in each group No significant group differences on demographic or clinical variables at baseline Mean age 33.9 years 88% male 62% employed full‐time; 40% married Mean 11.6 years of daily cannabis use, smoking a mean of 11.0 g/week Exclusion criteria: abuse or dependence on substances other than cannabis or nicotine, and significant psychiatric disorders
Interventions	Group 1: oral gabapentin 300 mg/day, increasing to 1200 mg/day Group 2: placebo Abstinence‐oriented individual counselling weekly
Outcomes	Change in amount of cannabis use, frequency of use and withdrawal symptoms, as graphs and results of statistical tests Number completing treatment Cannabis use by weekly urine toxicology and self‐report by TLFB interview Withdrawal symptoms by Marijuana Withdrawal Checklist Marijuana Problems Scale completed at baseline and end of treatment
Notes	Funding: research grants (NIDA) 1 author declared past associations with pharmaceutical companies
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "subjects were randomly assigned … in a 1:1 ratio, on the basis of a computer‐generated randomization code."
Allocation concealment (selection bias)	Low risk	Quote: "The randomization code was kept by the study pharmacist, who provided subjects with a 1‐week supply of medication in a blister card package at each weekly study visit …" Comment: allocation by pharmacy would support adequate concealment of allocation.
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Quote: "Subjects, care providers, and those assessing outcomes were blinded to the identity of drug assignment. Gabapentin was purchased and over‐encapsulated to match placebo capsules."
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Quote: "Subjects, care providers, and those assessing outcomes were blinded to the identity of drug assignment. Gabapentin was purchased and over‐encapsulated to match placebo capsules."
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	High rate of dropout. Extent of missing data, and adjustments for missing data unclear
Selective reporting (reporting bias)	Low risk	None apparent
Other bias	Low risk	None apparent

PERMALINK

Pharmacotherapies for cannabis dependence

Suzanne Nielsen

Linda Gowing

Pamela Sabioni

Bernard Le Foll

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Summary of findings

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Data synthesis

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Grading of evidence

Results

Description of studies

Results of the search

1.

Included studies

Excluded studies

Risk of bias in included studies

2.

3.

Allocation

Blinding

Incomplete outcome data

Selective reporting

Other potential sources of bias

Effects of interventions

Summary of findings for the main comparison. Δ9‐Tetrahydrocannabinol preparation compared to placebo for cannabis dependence.

Summary of findings 2. Selective serotonin reuptake inhibitor antidepressant compared to placebo for cannabis dependence.

Summary of findings 3. Mixed action antidepressant compared to placebo for cannabis dependence.

Summary of findings 4. Anticonvulsants and mood stabilisers compared to placebo for cannabis dependence.

Summary of findings 5. Buspirone compared to placebo for cannabis dependence.

Summary of findings 6. N‐acetylcysteine compared to placebo for cannabis dependence.

1. Preparations containing Δ9‐tetrahydrocannabinol

Participants abstinent at end of treatment

1.1. Analysis.

Withdrawal symptoms and cravings

Participants experiencing adverse effects

1.2. Analysis.

Participants withdrawn due to adverse effects

1.3. Analysis.

Completion of scheduled treatment

1.4. Analysis.

2. Selective serotonin reuptake inhibitor antidepressants

Participants abstinent at end of treatment

2.1. Analysis.

Withdrawal symptoms and craving

Summary of findings for the main comparison. Δ⁹‐Tetrahydrocannabinol preparation compared to placebo for cannabis dependence.

1. Preparations containing Δ⁹‐tetrahydrocannabinol

Methods	Randomised placebo‐controlled trial
Participants	Setting: outpatient (no further details on site, appears to be a single site). Scheduled duration 12 weeks Participants: 175 adults, current cannabis dependence by DSM‐IV Group sizes: group 1, 88; group 2, 87 Groups similar on demographics and clinical characteristics at baseline Mean age 24 years 77% male 64% Caucasian Exclusion criteria: current dependence on other substances except caffeine and nicotine, and significant psychiatric condition
Interventions	Group 1: buspirone, up to 60 mg/day Group 2: placebo Brief MET intervention and contingency management to encourage study retention as adjunct interventions
Outcomes	Participants with ≥ 1 negative urine test during treatment Point prevalence of abstinence by urine test at week 12 Number reporting adverse events
Notes	Funding: research grant (NIDA) 2 authors report previous pharmaceutical company funding
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Method of sequence generation not described
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not described
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Quote: "Buspirone and placebo tablets were packaged in identical opaque gelatin capsules with lactose." Comment: it was likely that participants and treating clinicians were blind to group allocation.
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Double‐blind likely as indicated above, and only objective outcomes reported which are less likely to be affected by knowledge of group allocation.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Rates of attrition high, but similar in both groups. Missing data on cannabinoid urine tests counted as positive.
Selective reporting (reporting bias)	Low risk	None apparent, published outcomes accorded with prospectively registered outcomes.
Other bias	Low risk	None apparent

Methods	Randomised placebo‐controlled pilot study. Participants stratified on sex, cannabis dependence and baseline working memory function prior to randomisation
Participants	Setting: outpatient, Providence, USA. Scheduled duration 6 weeks Participants: 66 adolescents (aged 15–24 years), used cannabis at least twice weekly in prior 28 days and experiencing some clinically significant problems Group sizes: group 1, 40; group 2, 26 Groups similar on demographics (except mean age) and baseline clinical characteristics Mean age: group 1, 20.3 years, group 2, 18.8 years 48.5% male 80% met DSM‐IV‐TR criteria for cannabis abuse or dependence Exclusion criteria: significant psychiatric conditions.
Interventions	Group 1: topiramate, titrated over 4‐weeks then stabilised at 200 mg/day for 2 weeks Group 2: placebo MET biweekly for 3 sessions
Outcomes	Number completing study, days in treatment, sessions of MET attended Cannabis use (grams per use day and % days with use) Medication compliance Adverse effects
Notes	Funding: research grant (NIDA) 1 author declared associations with pharmaceutical companies.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "An investigator with no direct participant contact used a computer‐generated random allocation sequence to assign participants to treatment conditions on a 2:1 (topiramate to placebo) ratio."
Allocation concealment (selection bias)	Low risk	Quote: "An investigator with no direct participant contact used a computer‐generated random allocation sequence to assign participants to treatment conditions." "An independent compounding pharmacy provided topiramate and placebo capsules, which were identical in appearance. Capsules were prepackaged in 7‐day blister packaging cards consecutively numbered according to a computer‐generated randomization schedule to ensure that the researchers who enrolled and assessed participants were blind to treatment assignments."
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Quote: "An independent compounding pharmacy provided topiramate and placebo capsules, which were identical in appearance;" "Participants and study personnel in direct contact with participants were blind to treatment assignments." "Counselors were blind to participants' medication condition and did not conduct any research assessments with participants."
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Quote: "An independent compounding pharmacy provided topiramate and placebo capsules, which were identical in appearance;" "Participants and study personnel in direct contact with participants were blind to treatment assignments."
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Quote: "Of the 66 participants, 39 youth (59.1 percent) completed the trial. As expected, attrition occurred disproportionally in the topiramate condition (52.5 percent) compared with the placebo condition (23.1 percent)." Comment: attrition is an outcome for this review. Missing data were imputed in analyses of cannabis use outcomes.
Selective reporting (reporting bias)	Low risk	None apparent
Other bias	Low risk	None apparent

Study	Reason for exclusion
Adams 2018	Participants recruited from opioid treatment programme; all receiving medication‐assisted treatment. Cross‐over design, not randomised controlled trial, comparing medication‐assisted treatment with no medication, or with varenicline. Total 7 participants, 4 of whom were cannabis dependent, 3 met criteria for cannabis abuse.
Akerele 2007	Participants diagnosed with abuse or dependence on marijuana or cocaine. Data reported separately for cocaine and marijuana use, but it was not possible to extract data just for those dependent on marijuana. All participants were diagnosed with schizophrenia; the management of substance use in the context of schizophrenia was the main focus of the study.
Brown 2013	Secondary analysis of data from a randomised controlled trial comparing 2 behavioural interventions. No use of medications.
Budney 2007b	Laboratory study involving non‐treatment seeking cannabis users. Not all users were cannabis dependent, and participants were not trying to reduce their cannabis use.
Cooper 2013	Laboratory study involving marijuana smokers who were not seeking treatment. Investigation of research model of withdrawal and relapse rather than treatment intervention.
Cornelius 1999	Randomised controlled trial comparing fluoxetine and placebo for treatment of alcohol dependence with comorbid major depression. Effect on subgroup with diagnosed marijuana abuse considered as secondary analysis.
Cornelius 2008	Reported cannabis withdrawal symptoms in participants entering 2 separate trials of fluoxetine. No treatment intervention for cannabis dependence.
Cornelius 2015	Open‐label study of mirtazapine and motivation therapy for people with major depressive disorder and substance use disorder. Most study participants had used cannabis, but unclear how many were dependent. No treatment comparison.
Findling 2009	Randomised controlled trial comparing fluoxetine and placebo for treatment of depressive symptoms in adolescents with comorbid substance use disorder. Cannabis use reported by 88.2% of participants (41.2% dependent). The emphasis of this study was on the amelioration of depression. Outcome data not reported separately for the subset of cannabis‐dependent participants.
Geller 1998	Randomised controlled trial comparing lithium and placebo for treatment of adolescents with bipolar disorder and comorbid substance use disorder. Majority of participants were polydrug users; 2/25 were dependent on cannabis only.
Gillman 2006	Reported the use of nitrous oxide for treatment of withdrawal associated with the smoking of methaqualone combined with cannabis. Unclear how many participants were cannabis dependent. All participants received placebo then analgesic nitrous oxide. Effectiveness assessed only in terms of improvement in withdrawal symptoms.
Gray 2010	Open‐label single group study investigating the effectiveness of N‐acetylcysteine in promoting cessation of cannabis use. No treatment comparison.
Haney 2001	Comparison of bupropion and placebo in terms of effect on mood when administered in conjunction with active or placebo cannabis cigarettes. Laboratory study that aimed to assess the therapeutic potential of bupropion, but not a treatment intervention.
Haney 2003a	Laboratory study comparing the effect of nefazodone 450 mg/day and placebo on the acute effects of cannabis, and on cannabis withdrawal symptoms. The study aimed to assess the therapeutic potential of nefazodone in cannabis withdrawal but was not a treatment intervention.
Haney 2003b	Investigation of mechanism of effects of cannabis through comparison of naltrexone and methadone, administered prior to oral THC, and different doses of oral THC administered in combination with naltrexone or placebo. No treatment intervention.
Haney 2004	2 separate laboratory‐based studies, 1 assessing THC and 1 assessing divalproex, compared to placebo, in terms of effects on cannabis withdrawal. Studies aimed to assess the therapeutic potential of THC and divalproex but were not treatment interventions.
Haney 2008	Laboratory study investigating the effect of lofexidine and THC (separately and in combination) compared with placebo on cannabis withdrawal symptoms and a model of cannabis relapse. The study aimed to test the therapeutic potential of lofexidine in cannabis withdrawal but was not a treatment intervention.
Haney 2010	Controlled laboratory study investigating the effects of baclofen or mirtazapine on cannabis smoking, craving and withdrawal. Exploratory study of the potential therapeutic value of baclofen and mirtazapine, but not a treatment intervention.
Haney 2013	Laboratory study assessing effect of nabilone on marijuana withdrawal symptoms, and laboratory measure of relapse. Study aimed to test the therapeutic potential of nabilone but was not a treatment intervention.
Haney 2015	Laboratory study comparing impact of naltrexone and placebo on effects of active or inactive cannabis. Participants not seeking treatment.
Haney 2016	Laboratory dose‐ranging study of effects of cannabidiol. Participants not seeking treatment.
Herrmann 2016	Laboratory study investigating effect of zolpidem and nabilone (separately and in combination) compared with placebo on marijuana withdrawal symptoms and a model of marijuana relapse. The study aimed to test the therapeutic potential of zolpidem in marijuana smokers but was not a treatment intervention.
Nanjayya 2010	Open‐label study investigating the use of baclofen for the treatment of cannabis dependence. No treatment comparison.
Notzon 2018	Single group study of injectable naltrexone for treatment of cannabis dependence. No treatment comparison.
Rubio 2006	Comparison of antipsychotic drugs for treatment of schizophrenia, with consideration of effects on cannabis use.
Schnell 2014	Comparison of antipsychotic drugs for treatment of schizophrenia, with consideration of effects on cannabis use.
Sevy 2011	Comparison of antipsychotic drugs for treatment of schizophrenia, with consideration of effects on cannabis use.
Sugarman 2011	Controlled study assessing the safety of modafinil in combination with THC. While the study contributed to assessment of the therapeutic potential of modafinil, this study did not involve a treatment intervention. Participants were occasional cannabis users (people who were heavy users or dependent were excluded).
Trigo 2016	Laboratory study assessing effects of Sativex on cannabis withdrawal and craving. Not a treatment intervention. Preliminary study to Trigo 2018.
Van Nimwegen 2008	Randomised controlled trial comparing olanzapine and risperidone for treatment of schizophrenia. Majority of participants were not using cannabis and cannabis dependence was not assessed.
Vandrey 2011	Cross‐over study comparing zolpidem and placebo during short (3‐day) periods of abstinence from cannabis in terms of sleep parameters. Not a full treatment intervention for cannabis dependence.
Vandrey 2013	Comparison of dronabinol and placebo in terms of effect on cannabis withdrawal and subjective effects of smoked cannabis, but without providing a treatment intervention for cannabis dependence.
Vandrey 2016	Laboratory study of effect of zolpidem on sleep during cannabis withdrawal. Not a full treatment intervention.
Winstock 2009	An open‐label study investigating the use of lithium carbonate for the management of cannabis withdrawal. No treatment comparison.

Trial name or title	Randomised controlled trial (RCT) of cannabinoid replacement therapy (nabiximols) for the management of treatment‐resistant cannabis dependent patients
Methods	Randomised controlled trial
Participants	142 cannabis‐dependent adults
Interventions	Experimental: nabiximols Comparison: placebo
Outcomes	Non‐prescribed cannabis (self‐reported cannabis use days, urine toxicology)
Starting date	2016
Contact information	Nicholas Lintzeris, University of Sydney
Notes	Recruitment complete, protocol published and results pending (requested)

Trial name or title	Safety and efficacy of a FAAH‐inhibitor to treat cannabis withdrawal
Methods	Randomised controlled trial
Participants	120 cannabis‐dependent participants
Interventions	Experimental: FAAH inhibitor PF‐04457845 Comparison: placebo
Outcomes	Withdrawal score, cannabis use, craving, relapse rates
Starting date	June 2012
Contact information	Deepak C D'Souza, Yale University
Notes

Trial name or title	Gabapentin treatment of cannabis dependence
Methods	Randomised controlled trial
Participants	150 cannabis‐dependent participants
Interventions	Experimental: gabapentin Comparison: placebo
Outcomes	Negative urine drug screens at 12 weeks' follow‐up
Starting date	2009
Contact information	Barbara J Mason
Notes

Trial name or title	Combination of dronabinol and clonidine for cannabis dependence in patients with schizophrenia
Methods	Randomised controlled trial
Participants	18 cannabis‐dependent participants with schizophrenia
Interventions	Experimental: dronabinol and clonidine Comparison: placebo
Outcomes	Change in cannabis use
Starting date	May 2012
Contact information	William M Hurley‐Welljams‐Dorof, McLean Hospital, USA
Notes

Trial name or title	Cannabidiol, a novel intervention for cannabis use problems?
Methods	Randomised controlled trial
Participants	96–168 young people who want to quit cannabis
Interventions	Experimental: cannabidiol Comparison: placebo
Outcomes	Stage 1: identification of most effective dose
Starting date	March 2014
Contact information	Clinical Psychopharmacology Unit, UK
Notes

Trial name or title	Hormones and reduction in co‐users of marijuana and nicotine
Methods	Randomised controlled trial
Participants	100 marijuana‐dependent participants
Interventions	Experimental: progesterone Comparison: placebo
Outcomes	Change in marijuana use by timeline follow‐back method
Starting date	2015
Contact information	Sharon Allen, University of Minnesota
Notes