Sex differences in outcomes of methadone maintenance treatment for opioid addiction: a systematic review protocol

The objective of this review is to summarize the current status of literature regarding sex differences in methadone treatment outcomes by systematically reporting the available research to date. Specifically, we aim to:

This systematic review will include completed randomized controlled trials (RCTs) and observational studies of methadone treatment outcomes in men and women. Included studies will focus primarily on sex differences, as opposed to studies on separate populations of men or women. Included studies must also have been conducted in the context of methadone treatment for opioid addiction. Studies including patients that are undergoing a substitute opioid therapy other than methadone (that is, buprenorphine/naloxone, naltrexone) or using methadone for the purpose of detoxification (not maintenance) will be excluded. Studies investigating patient subpopulations such as pregnant women or incarcerated individuals will be excluded as they are too specific to represent the overall population of opioid-dependent individuals and may not allow for the application and generalizability of our findings to community samples. Sex differences in these populations may also be influenced by their environment, leading to a high potential for confounding and bias in the outcomes studied. Patients that are using methadone for the treatment of a condition other than opioid addiction (that is, chronic pain) will also be excluded. Participants shall include both men and women who are receiving methadone treatment for a diagnosis of opioid dependence. No other limitations will be applied (including age or ethnicity) as our intent is to retrieve all articles on sex differences in methadone treatment without restrictions based on population characteristics. The primary outcome of this review will be the presence of sex differences in methadone treatment response, defined as abstinence from illicit opioid use and measured through self-report and/or urinalysis. Sex differences in treatment outcomes will also be assessed with respect to three life domains: drug use-related behavior, health status, and sociobehavioral functioning. These outcomes include treatment retention/duration, remission status post-treatment, polysubstance abuse, methadone dose, drug-related adverse events, health status, psychological status, mortality, criminal activity, high risk sexual behavior, social support/relations, and employment. A complete list of how these outcomes are described, defined, and measured in the literature is available in Table 1.

We shall identify all studies relevant to this review with no language or time restraints. We will search the PubMed/MEDLINE, EMBASE, PsycINFO, and CINAHL databases for relevant articles. Relevant search terms and their medical subject heading (MeSH) equivalents will be used in varying combinations; refer to Table 2 for the complete search strategy. In order to maximize the number of relevant articles retrieved, treatment outcomes will not be included in the search. We will use a wide search to include titles, abstracts, and keyword fields to avoid missing important articles whose title may not reflect the content of the article. Articles will be excluded by limiting the search to humans. We will also manually review reference lists of included studies for studies that may have been missed in the initial search. Grey literature will not be reviewed as we are looking for complete published data only.

Two independent raters (MB and BBD) will screen all citations and abstracts retrieved using the search strategy and identify all eligible articles. Articles that meet the pre-determined criteria will be included for full-text review. Disagreements at any phase of the review process will be resolved by discussion or, in the case where a consensus is not reached, a third independent rater (ZS) will determine eligibility. Ineligible studies will be excluded from the review and reasons for exclusion will be recorded. Inter-rater agreement will be calculated using the Kappa statistic [48] for each phase of screening. Authors will be contacted directly if further data clarification is needed.

The two authors (MB and BBD) will independently extract data from the studies using a pre-established pilot-tested data extraction form (see Additional file 1). Information obtained will include the author and year of publication, city and country of publication, title of article, journal name, study design, and description of sample population, including total number of men and women study participants, mean age (total and men versus women), and ethnicity. Primary and secondary outcomes, outcome measures, statistical analyses, results, and conclusions will also be recorded. In the case of missing or incomplete data, authors will be contacted for further details. Data will be combined to produce a summary estimate in a meta-analysis if the extracted data allows it.

Two authors (MB and BBD) will independently assess the risk of bias of included studies using the Newcastle-Ottawa Scale (NOS) [49] for observational studies and the Cochrane Collaboration’s tool [50] for assessing risk of bias in RCTs. For observational studies, two authors (MB and BBD) will independently assess the risk of bias of each included study using an adapted version of a modified NOS, specific to the context of this review. This will include seven questions spread across four domains of evaluation; methods for selecting study participants (selection bias), methods to control for confounding (performance bias), statistical methods (detection bias), and methods for measuring exposure and outcome variables (information bias). Risk of bias is measured on a scale of 0 (high risk of bias) to 3 (low risk of bias) and a specific description with examples of both high and low bias is provided. Items regarding selection of participants (representativeness of sample) and ascertainment of outcome (objective versus subjective measures) were retained, while other items relating to the comparability of groups and adequate follow-up for cohort and case-control studies were removed as these were not directly applicable to our topic of interest. We also introduced categories that emphasize statistical methods, confounding effects, and reporting of data to ensure that bias in methodology is assessed. These scales will be used to measure the risk of bias on a per study basis or categorized by domain to develop a general conclusion about the sources of bias in the studies included in this review (see Additional file 2). Cochrane’s tool for assessing risk of bias in RCTs includes seven domains; random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other bias. Each of these domains will be evaluated according to high or low risk of bias and will also be assessed on a per study or per domain basis. If a meta-analysis is possible, we will use the Grading of Recommendations, Assessment, Development, and Evaluation framework to rate the quality of evidence through investigation of risk of bias, imprecision (random error), inconsistency, indirectness, and publication bias. We will then summarize the evidence for individual outcomes in summary of findings tables, which will allow for assessment of our confidence in the estimates.

The results of this systematic review will be reported in a narrative and informative manner; we will discuss issues of study design and statistical analysis methods of the included studies to determine which studies are most informative and reliable. Where possible, we will assess the studies in a combined statistical manner using meta-analysis. The Kappa statistic will be used to measure level of agreement between independent raters. For dichotomous outcomes, we will compute pooled odds ratios using the Mantel-Haenszel random effects model, in which the model is able to estimate between study variation through an evaluation of each study’s final results and a Mantel-Haenszel fixed effect meta-analysis result.

For the summary estimates, we will employ a random effects model, which assumes variation between studies and their respective effect sizes. The nature of observational studies in this population is highly variable, therefore heterogeneity will be accounted for and will allow us to develop aggregate estimates. We will assess the participants, methods, and results of included studies for heterogeneity, which will allow us to determine whether results can be compared across studies. Possible sources of heterogeneity include age groups, study design, methodology, and definition of outcome. In case of heterogeneity, subgroup analyses according to these different categories will be performed. Included studies will be presented in the form of a forest plot. We will use Review Manager 5.1 software (The Cochrane Collaboration, London, UK) for all statistical analysis and results will be presented using 95% confidence intervals.

We will report the systematic review according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses guidelines [51]. A flow diagram will be used to summarize the selection process of studies at each phase and summary tables will be used to report study characteristics and presence of sex differences per methadone outcome. Publication bias will also be examined and assessed using Egger’s plot.