Narrative Review & Meta-Analysis Guide

Narrative Review & Meta-Analysis - A Step-by-Step Guidance

• A narrative review and meta-analysis is a method for evaluating a topic's current knowledge
• It's a step-by-step process involving several key elements, including different types of literature reviews
• Different types of knowledge synthesis, such as realist reviews, scoping reviews, rapid reviews, systematic reviews, meta-narrative reviews, state-of-the-art reviews, umbrella reviews, critical reviews, and meta-analysis

What is Literature?

• Literature includes formal and informal sources, such as books, journals, conference papers, theses, dissertations, bibliographies, maps, internet resources, indexes/abstracts, audio-visual material (CDs/DVDs), electronic databases, government reports, magazines, newspapers, grey literature, and interviews/unpublished research.

Types of Literature

• Primary literature includes reports, theses, emails, conference reports, company reports, some government publications, and unpublished manuscript sources.
• Secondary literature includes newspapers, books, journals, internet sources (some government publications).
• Tertiary literature includes indexes, abstracts, catalogues, encyclopedias, dictionaries, bibliographies, and citation indexes.

Research Loop

• Library research helps determine what is currently known about the research area.
• Hypothesis formation involves identifying gaps in current knowledge and developing new hypotheses.
• Research design focuses on how to collect appropriate data to test the hypothesis.
• Data collection involves gathering the necessary data based on the research design.
• Data analysis uses statistical analysis to interpret the collected data.
• Draw conclusions based on the results of the data, and how they change the known knowledge.

Types of Knowledge Synthesis

• Types of knowledge synthesis include realist synthesis, realist review, scoping review, rapid review, meta-narrative reviews, state-of-the-art review, umbrella review, critical review, systematic review, and meta-analysis.

Level of Evidence

• Different types of studies have varying levels of evidence.
• The levels of evidence range from the strongest (systematic reviews and meta-analysis of randomized controlled trials) to the weakest (expert opinion).
• The types of studies include randomized controlled trials (RCTs), individual RCTs, and cohort studies and other studies.

Review of Literature

• A review of literature is a structured body of text that analyzes critical points of current knowledge within a particular topic/field of study or related research.
• It involves a summary of existing published literature by experts, and an interpretation of the research findings.

Review of Literature - Purposes

• Define and limit the research problem
• Develop familiarity with topics.
• Limit research to subtopics.
• Place studies in a historical perspective.
• Avoid unintentional replications.
• Select methods/measures that are appropriate and relevant to the problem.
• Select promising methods.

Review of Literature - Designs

• Narrative Review: A selective review that broadly covers a specific topic (not strict methods)
• Systematic Review: Uses strict search strategies to identify relevant research; methodologically appraises and synthesizes findings.
• Meta-Analysis: A quantitative combination of results from studies in a systematic literature review.

Review of Literature - Types of Articles

• Reviews can be systematic or narrative.
• Meta-analyses are a specific type of systematic review that combines data quantitatively.

Steps in a Narrative Review

• Identifying a topic
• Searching and finding literature
• Evaluating literature
• Reading literature critically
• Analyzing literature
• Synthesizing literature
• Writing and presenting the review.

Searching and Finding Literature

• Collect bibliographic information, abstracts, and full texts.
• Use formal and informal sources (print and online).

Evaluating Literature

• Evaluate relevance based on the index of book, chapter or section headings, abstract, introduction, conclusion, references or bibliography.
• Evaluate reliability based on audience, authority, bias, currency, and scope.

Critical Reading

• Distinction between passive and active reading
• Previewing the text
• Reading the text.
• Taking notes
• Responding critically.

Analyzing Literature

• Analyze different definitions of key terms.
• Analyze the methodologies used.
• Examine if there is enough supporting evidence.
• Compare results to those of similar similar studies.

Synthesizing Literature

• Relate readings to the topic and purpose
• Identify major trends and patterns
• Reassemble notes using organizational tools
• Create a detailed topic outline

Synthesizing Literature (2)

• Identify relationships among studies, inconsistencies, and gaps in knowledge in the literature.
• Note landmark studies and their replication.
• Discuss how studies advance or illustrate theoretical beliefs.
• Note areas needing more research.

Writing and Presenting Literature Review

• Organize a literature review using a funnel-shaped structure.
• Broaden topics into subtopics and then narrower subtopics related to the primary research questions.

How to Organize Studies

• Chronological: Organize studies by their publication date.
• **Thematic:**Organize studies based on various themes.
• Methodological: Organize studies based on research methods used.

Making Links Between Studies

• Identify agreements among authors, and disagreements.

Summary Table

• Providing a table allows a quick overview of a large amount of information.
• A summary table can contain information about author, type of study, sample, study design, data collection, and key findings.

A Good Literature Review

• A focused topic within an appropriate scope
• A concise review presents ideas economically.
• A systematic review presents topics logically, making sure links are clear between topics.
• A well-developed summary is integrative, outlining the commonalities and differences within studies.
• A current literature review will focus on cutting-edge research within the topic.

Meta-Analysis

• A quantitative method for systematically combining studies to arrive at conclusions.

Function of Meta-Analysis

• Identify heterogeneity within studies.
• Increase statistical power/precision of the study
• Develop, refine, and test hypotheses.
• Calculate sample size for future studies.
• Identify data gaps
• Reduce subjectivity in study comparisons.

Advantages of Meta-Analysis

• Focus and increase clinical impact of trials.
• Strengthen conclusions
• Generalizability of results by incorporating more data points
• Improve precision and accuracy of estimates
• Increase statistical power to detect an effect
• Identify inconsistencies of results
• Apply hypothesis testing for summary estimates
• Incorporate moderators for variations between studies
• Investigate/identify publication bias

Disadvantages of Meta-Analysis

• May discourage large definitive trials.
• Increase tendency to mix different trials
• Potential tension between meta-analysts and study authors, creating bias
• Summary data of multiple small studies may not necessarily predict results, or correlate with results from a larger, single study.
• Sources of bias are not always controlled
• Badly designed studies will still produce problematic statistics

How a Meta-Analysis Works

• Collects similar studies from previous research.
• Calculate 'effect sizes' to reflect effect magnitude.
• Calculates 'precision' of the study effects by using confidence intervals
• Assign weights to studies based on their size, quality, or design.
• Test null hypothesis using P-value.
• Create summary effect of studies with mean, confidence interval, weight, P-value, ES heterogeneity, random effect, and publication bias.

When Can We Do Meta-Analysis?

• Examine similar constructs and relationships in the studies
• Results are in a comparable form (effect sizes)
• Have appropriate objectives for the study
• Have a good understanding of the population of the study

Steps in Meta-Analysis

• Define the research question.
• Perform the literature search.
• Select the studies.
• Extract the data.
• Analyze the data.
• Examine heterogeneity.
• Assess publication bias.
• Report the results.

Define Research Question Example

• Does vitamin E decrease the risk of death in coronary artery disease patients?
• Do patients digesting carotenoids decrease their likelihood of developing lung cancer?

Literature Searching Example

• Use a methodological plan.
• Search through popular databases like Pubmed, Medline, Embase, Cochrane Review, or a trials register.
• Conduct comprehensive additional searches, consider other databases and strategies.
• Personal references, and email searches are also good approaches.
• Google Scholar searches are also beneficial.

How to Search for Literature (Example)

• Use Medical Subject Headings (MeSH) terms to appropriately formulate your questions.
• Find appropriate text within the article, abstract, or title for additional searches.
• Combine search terms using boolean logic.

Boolean Logic Example

• AND Zone covers areas where both terms exist within an article
• OR Zone includes either of the two terms mentioned.

Example Research Question

• Does streptokinase protect against death in acute myocardial infection?

Select Studies

• Identify studies with high-quality evidence (high evidence and good quality).
• Use eligibility and inclusion/exclusion criteria
• Use previously developed guidelines like QUOROM to refine study selection criteria.
• Excludes irrelevant studies like simple case studies and case reports.
• Use a flow diagram to visualize the study selection.

Extract Data

• Use spreadsheet software (Excel or Open Office)
• Format data columns to include study name, author, publication years, number of participants, and outcome results for each study.

Analyze Data

• Data Combine all data from studies
• Calculate effect size (Odds Ratio, RR)
• Calculate confidence interval (with 95%)
• Examine heterogeneity (check if studies are alike)
• Draw two graphs
• Determine if studies have differences/heterogeneity
• Use statistical packages, both paid and free

Heterogeneity

• Develop statistical heterogeneity tests.
• Perform visual inspection using a Forest plot.
• Use meta-regression to examine and interpret heterogeneity.
• Review units of regression.
• Review dependent and independent variables.

Publication Bias

• Examine possible publication bias (studies with positive results are more likely to be published, which gives a biased result).
• Use funnel plots.
• Consider other approaches to address or mitigate this issue.

Report Results/Discussion

• Report an abstract including background, method, conclusions, and significance.
• Report method section with sources, restrictions, and inclusion/exclusion criteria.
• Report results (flow of studies, characteristics, quantitative synthesis of results).
• Report discussion highlighting results with funnel plots, forest plots, and summaries.
• Report limitations as appropriate.