How do you conduct a meta-analysis?

1 Define the research question

The first step of a meta-analysis is to define a clear and focused research question that guides your search and selection of studies. Your question should specify the population, intervention, comparison, outcome, and setting (PICOS) of interest, as well as the type and design of studies you are looking for. For example, you might ask: How effective is cognitive behavioral therapy (CBT) compared to medication for treating depression in adults in clinical settings? A well-defined question will help you narrow down your scope and avoid irrelevant or heterogeneous studies.

2 Search and select the studies

The next step is to search and select the studies that meet your inclusion and exclusion criteria. You should use multiple sources and databases, such as PubMed, PsycINFO, Web of Science, and Cochrane Library, to find relevant studies. You should also use keywords, filters, and Boolean operators to refine your search. For example, you might use the following search string: (cognitive behavioral therapy OR CBT) AND (depression OR depressive disorder) AND (medication OR pharmacotherapy) AND (randomized controlled trial OR RCT). You should document your search strategy and results, as well as the reasons for excluding any studies.

3 Extract and code the data

The third step is to extract and code the data from the selected studies. You should use a standardized form or spreadsheet to record the relevant information, such as the authors, year, sample size, characteristics, intervention, comparison, outcome, effect size, and quality of each study. You should also code the studies according to any variables or moderators that might influence the effect size, such as the type, duration, or intensity of the intervention or comparison. You should check the accuracy and consistency of your data extraction and coding, and resolve any discrepancies or missing data.

4 Analyze and synthesize the data

The fourth step is to analyze and synthesize the data using statistical software or tools, such as R, Stata, or RevMan. You should choose the appropriate model and method for your meta-analysis, depending on the type and distribution of the effect sizes, the heterogeneity and publication bias of the studies, and the presence and level of moderators. You should also perform sensitivity and subgroup analyses to test the robustness and validity of your results. You should report the summary effect size and confidence interval, as well as the measures of heterogeneity, publication bias, and moderator effects.

5 Present and interpret the results

The fifth step is to present and interpret the results of your meta-analysis in a clear and comprehensive way. You should use tables, graphs, and forest plots to display the characteristics, effect sizes, and confidence intervals of the individual and pooled studies. You should also use funnel plots, meta-regression, or subgroup analysis to show the sources and impact of heterogeneity, publication bias, or moderator effects. You should explain the main findings and implications of your meta-analysis, as well as the limitations, strengths, and gaps of the existing literature.

6 Write and publish the report

The final step is to write and publish the report of your meta-analysis following the standards and guidelines of your field and journal. You should use a structured format, such as the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) checklist and flow diagram, to organize and report the essential information and steps of your meta-analysis. You should also use a transparent and reproducible language, such as the Cochrane Handbook or the Meta-Analysis Reporting Standards (MARS), to describe and justify your methods and results. You should also acknowledge the sources of funding, support, and conflict of interest of your meta-analysis.

7 Here’s what else to consider

This is a space to share examples, stories, or insights that don’t fit into any of the previous sections. What else would you like to add?