Types of studies

Randomized controlled trials (RCTs) comparing MBIs for SUDs with other treatments or no intervention were eligible for inclusion in the review. Trials employing a cross‐over design were also eligible, using data from the first active treatment stage only to encounter the risk of carry‐over effects.

Types of participants

RCTs with patients suffering from SUDs including individuals with alcohol, prescription‐, illicit‐ and poly‐substance use disorders were considered as eligible for the review. There was no limitations on age or other participant characteristics. Besides accepted SUD diagnostic criteria including DSM‐III (APA 1980), DSM‐ III‐R (APA 1987), DSM‐IV‐TR (APA 2000), DSM‐V (APA 2013) and ICD‐10 (WHO 1992; WHO 2010), we also included studies in which SUD was not formally diagnosed, acknowledging that diagnostic systems are not consequently used in primary research.

Types of interventions

In accordance with the definition by Bishop and colleagues (Bishop 2004) of mindfulness, we consider as mindfulness‐based all approaches which promote a) an individual's attention towards the immediate present moment experience and b) an open and accepting orientation irrespective of the applied technique. In order to isolate the effects of mindfulness meditation practice specifically, we excluded interventions that involve solely the concept of mindfulness and do not include formal instruction in mindfulness meditation practice e.g. Acceptance and Commitment Therapy (ACT), Dialectical Behaviour Therapy (DBT). This definition of MBIs is in keeping with that proposed by Crane 2017 and widely implemented in the meta‐analytic literature (see Goldberg 2021).

Accordingly, the following experimental interventions were included in the review:

• ancient Buddhist meditations such as Vipassana meditation and Zen meditation;

• other mindfulness meditation;

• modern standardized group‐based meditation practices including mindfulness‐based stress reduction (MBSR), mindfulness‐based cognitive therapy (MBCT) and mindfulness‐based relapse prevention (MBRP) and mindfulness training.

Any type of manually‐based and face‐to‐face treatment delivery including individual therapy and group session format were considered. Media‐ and CD‐supported interventions were accepted as complementary formats of treatment support (e.g. home‐practice formats), but not as exclusive modes of treatment delivery.

All comparators were eligible, with the exception of comparisons with other MBIs or similar mind‐body approaches. Comparisons could include standard care, further psychotherapeutic, psychosocial or pharmacological interventions or instructions, waiting list or no treatment. Comparisons were categorized into no treatment (which included standard care when both the mindfulness and non‐mindfulness arms received this) and other treatment (which included standard care when only the comparison condition received this).

Types of outcome measures

The study endpoints of the primary outcomes were considered essential to determine the effectiveness of MBIs, while secondary outcomes have only complementary value in the interpretation of results. If a study met the inclusion criteria, but did not provide necessary information for estimating effect sizes, while such data were also not available from the authors, the study was excluded from the meta‐analysis, but included in the qualitative analyses.

Primary and secondary outcomes were selected with regard to clinical relevance and conceptual considerations. With the "rate of continuous abstinence", the "per cent days with substance use" limited to non‐abstinent individuals and "consumed amount" limited to days with substance use, primary efficacy outcomes of the review assess conceptually‐distinct achievements in substance use control (Keller 1972); including an individual's ability to a) achieve and maintain continuous abstinence; and b) their ability to refrain from substance use on individual days; and c) to stop substance use once started. Evaluation of adverse effects, serious adverse effects, and treatment acceptability was included to allow evaluation of the safety and acceptability of MBIs relative to controls.

Electronic searches

The Cochrane Drugs and Alcohol Group Information Specialist conducted systematic searches in the following databases for randomized controlled trials and controlled clinical trials without language, publication year or publication status restrictions:

• Cochrane Drugs and Alcohol Group's Specialised Register of Trials (searched on 26 April 2021);

• The Cochrane Central Register of Controlled Trials (CENTRAL; 2021, Issue 3);

• PubMed (January 1966 to 26 April 2021);

• Embase (OVID) (January 1974 to 26 April 2021);

• CINAHL (Cumulative Index to Nursing and Allied Health Literature) (EbscoHOST) (1982 to 26 April 2021);

• Web of Knowledge, Web of Science (1990 to 26 April 2021);

• PsycINFO (OVID) (1806 to 26 April 2021).

The Information Specialist modeled subject strategies for databases on the search strategy designed for PubMed. Where appropriate, they were combined with subject strategy adaptations of the highly sensitive search strategy designed by the Cochrane Collaboration for identifying randomized controlled trials and controlled clinical trials (as described in the Cochrane Handbook for Systematic Reviews of Interventions Chapter 6, Lefebvre 2011). Search strategies for major databases are provided in Appendix 1; Appendix 2; Appendix 3; Appendix 4; Appendix 5; Appendix 6.

We searched the following trials registries on 26 April 2021:

• the ISRCTN registry (www.isrctn.com);

• ClinicalTrials.gov (clinicaltrials.gov).

Searching other resources

Key informants, primary authors and review authors were contacted with the request to indicate further studies of potential relevance. For this purpose, reference lists with identified studies and criteria of inclusion and exclusion of the review were provided. Finally, handsearching of reference lists of included studies and current reviews was conducted to complete the searches.

Selection of studies

Two review authors independently assessed the eligibility and relevance of trials on the basis of their abstracts retrieved from the electronic searches. For studies that met the inclusion criteria according to the abstract information, we obtained full‐text versions for closer inspection. Full‐text versions were also obtained if review authors differed in their judgement. Again, the relevance and eligibility of studies on the basis of full‐text versions was independently assessed by two review authors. In case of disagreements, the eligibility will be discussed with an additional consultant. The process of study identification and its results are outlined as flow diagrams according to the PRISMA statement (Moher 2009).

Data extraction and management

The review authors had full access to details about authors, institutions, and journals at all times. Information related to the study design and setting, the study participants, the interventions and comparators as well as the outcomes and methods for their assessment was abstracted from the original reports and entered into the study tables. The following information was extracted in detail:.

1. Study design and setting: design, principle of analysis, setting, study sites, country

2. Study sample: sample size, diagnosis, specific characteristics, age, gender

3. Interventions: description of the type of experimental and control intervention, treatment duration, treatment adherence

4. Outcomes: outcomes, methods of measurement, time points for assessment, compliance

Two review authors independently extracted all relevant outcome data onto pre‐specified data extraction forms and compared data value by value. In case of disagreements, the following sequential procedures were undertaken in descending order.

1. Comparison of published and extracted information to identify transcription and comprehension errors

2. Explanation of the coding decisions by each review author, followed by consensus discussion and arbitration

Any disagreement was discussed with an additional expert, and, when necessary, the authors of the studies were contacted for further information. Finally, after comparisons and corrections are concluded, we entered data into the Review Manager software (RevMan 2008).

Assessment of risk of bias in included studies

The risk of bias assessment for RCTs and CCTs in this review was performed using the criteria recommended by the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). The recommended approach for assessing risk of bias in studies included in Cochrane Reviews is a two‐part tool, addressing seven specific domains, namely sequence generation and allocation concealment (selection bias), blinding of participants and providers (performance bias), blinding of outcome assessor (detection bias), incomplete outcome data (attrition bias), selective outcome reporting (reporting bias), and other sources of bias. The first part of the tool involves describing what was reported to have happened in the study. The second part of the tool involves assigning a judgement relating to the risk of bias for that entry, in terms of low, high or unclear risk. To make these judgements we used the criteria indicated by the Cochrane Handbook for Systematic Reviews of Interventions adapted to the addiction field. The criteria considered as constitutive for the rating of bias risks are outlined in Appendix 7.

The domains of sequence generation and allocation concealment (avoidance of selection bias) were addressed in the tool by a single entry for each study. We planned to consider blinding of participants, personnel and outcome assessor (avoidance of performance bias and detection bias) separately for objective outcomes (e.g. dropout, substance use measured by urine‐analysis, participants who relapsed at the end of follow‐up,participants engaged in further treatments), and subjective outcomes (e.g. duration and severity of signs and symptoms of withdrawal, patient self‐reported use of substance, side effects, social functioning as integration at school or at work, family relationship). Incomplete outcome data (avoidance of attrition bias) were considered for all outcomes except for the dropout from the treatment, which is very often the primary outcome measure in trials on addiction. We considered the equivalence of baseline characteristics an additional indication of selection bias.

Measures of treatment effect

We measured treatment effects for dichotomous effectiveness outcomes (abstinence rate, retention rate) with a risk ratio (RR). For continuous outcomes (days with substance use, consumed amount per day, craving intensity), we planned to asses treatment effects using the mean differences (MD) for outcomes measured on the same scale.We used the standardized mean difference (SMD) for outcomes measured on different scales. We calculated all treatment effects within 95% confidence intervals (CIs). When effects on binary outcomes reached statistical significance, we calculated the number needed to treat for an additional beneficial outcome (NNTB). A P value of 0.05 and below was chosen to indicate statistical significance of effects.

Unit of analysis issues

Only individually‐randomized trials with the individual participants constituting the unit of analysis were included in the review. To control unit of analysis errors in studies with multiple treatment groups of the same type (e.g. multiple alternative treatment comparisons; Bowen 2014), we combined interventions to create single‐pair comparisons.

Dealing with missing data

Outcome statistics were included as intent‐to‐treat (ITT) analyses. Sample sizes for continuous outcomes which were not explicitly provided in the trial publication were imputed by the size of treatment‐received samples or ‐ if not available ‐ by the size of the randomized sample. An exception was if samples were from analyses explicitly specified as completer analyses, which exclusively reported on patients who completed the trial. For differences in means, missing serious adverse events (SEs) were obtained from standard deviations (SDs), CIs or t values and P values. If only the medians were provided in the trial publications, the outcome statistics were not be included in the meta‐analyses, but we considered the information on the significance of effects in the qualitative discussion of results.

Assessment of heterogeneity

We quantified inconsistency across studies with the I² statistic (Higgins 2003), using threshold values for substantial heterogeneity as outlined by Deeks 2001. The Tau² statistic was additionally considered to provide an estimate of between‐study variance (Rücker 2008) independent of the sample size. A value of P < .10 was considered as significant statistical heterogeneity.

Assessment of reporting biases

When there were more than 10 included studies, we graphically illustrated the risk of publication bias with the funnel plot method (Light 1984; Egger 1997).

Data synthesis

For synthesizing outcome measures, we used a random‐effects model (DerSimonian 1986), with study effects being weighted using the Mantel‐Haenszel approach (Mantel 1959). For outcomes with low effect heterogeneity (I² < 30%), we applied a fixed‐effect model within the scope of sensitivity analyses (see Sensitivity analysis).

Subgroup analysis and investigation of heterogeneity

Inconsistency across studies was quantified as described above (see Assessment of heterogeneity).

Sensitivity analysis

When the number of included studies was sufficient (> 10), we conducted sensitivity analyses to determine the influence of the following variables on the primary outcomes:

1. the underlying statistical model, by comparing effect sizes for low heterogeneity outcomes (I² < 30%) based on random‐effects models versus fixed‐effect models;

2. the method of measurement, by comparing effect sizes measured by breathalyzer or laboratory tests versus self‐reported data on alcohol use.

Summary of findings and assessment of the certainty of the evidence

Findings were presented as summarized narrative and by summary of findings tables (GRADE); and the certainty of evidence was assessed with the GRADE approach for each outcome individually.

We created two summary of findings tables using the following outcomes: continuous abstinence, percentage days with substance use, consumed amount, craving intensity, treatment acceptability (attrition). We used the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness and publication bias) to assess the quality of a body of evidence as it relates to the studies which contribute data to the meta‐analyses for the prespecified outcomes (Atkins 2004). We used methods and recommendations described in Chapter 14 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2019) using GRADEproGDT software (GRADEpro GDT 2015). We justified all decisions to down‐grade or up‐grade the certainty of studies using footnotes, and we made comments to aid the reader's understanding of the review where necessary.

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.