Skip to Main Content
Monash Health Library
Click here to chat with a librarian

Systematic Review Guide

This stage involves combining and then evaluating the data that you have extracted from your included studies.

The methods used during this stage depend on whether you have extracted quantitative data, qualitative data, or both. 

  • Quantitative data - methods include meta-analysis, narrative synthesis, or other forms of Synthesis Without Meta-analysis (SWiM).
  • Qualitative data - methods include meta-summary, thematic synthesis, framework synthesis, meta-ethnography and meta-aggregation.

A meta-analysis is a statistical technique used to combine the results of multiple independent studies that address the same or similar research question.

Why is it important to do a meta-analysis?

  • Increases statistical power by combining data from multiple studies
  • Reduces random error compared to individual studies
  • Identifies subgroup effects (e.g., by age, dose, setting)

When do I do a meta-analysis?

A meta-analysis can only be done when your included studies meet the below criteria:

  • Multiple studies on the same research question
  • Quantitative data available (e.g. means, risk ratios)
  • Comparable studies (similar populations, interventions, outcomes)
  • Acceptable study quality
  • Consistent outcome measures
  • Sufficient data reported to calculate effect sizes

How do I do a meta-analysis?

  1. Choose effect size metric
    1. e.g., Mean difference, Odds ratio, Risk ratio, Standardized mean difference.
  2. Perform statistical analysis
    1. Use software like SPSS, PSPP, or Meta-Essentials.
  3. Assess heterogeneity
    1. Calculate I² statistic to see how much variation exists between studies.
  4. Create a forest plot
    1. Visualize individual study results and the pooled estimate.
  5. Report results
    1. Include effect sizes, confidence intervals, heterogeneity, and limitations.

Recommended resources:

Synthesis Without Meta-analysis (SWiM) refers to methods used to combine and summarize results from multiple studies when statistical meta-analysis is not possible or appropriate.

Why is it important to do SWiM?

  • Allows synthesis when studies are too different (heterogeneous) or lack data for meta-analysis.
  • Encourages clear reporting of how evidence was grouped, summarized, and interpreted.
  • Replaces vague narrative summaries with more systematic, reproducible approaches.
  • Prevents inappropriate pooling of data

When do I do use SWiM?

  • Studies are too heterogeneous (different populations, interventions, outcomes).
  • Insufficient or inconsistent data for pooling.
  • Outcomes are not quantitative or not reported in a usable format.
  • Studies use different outcome measures or time points.

How do I do SWiM?

There are various SWiM methods, as outlined below. Ensure you use the SWiM Reporting Guideline in your publication to describe your chosen method.

  • Vote counting (e.g. number of studies showing positive vs. negative effect)
  • Thematic grouping of studies by outcome or population
  • Tabulation of study characteristics and results
  • Narrative synthesis describing patterns or trends
  • Visual displays like harvest plots or summary tables
Example
O’Cathain, A., et al. (2022). Health literacy interventions for reducing the use of primary and emergency services for minor health problems: a systematic review. National Institute for Health and Care Research. 

Recommended resources:

A narrative synthesis in a systematic review is a method of summarising and explaining findings from multiple studies using words and text rather than statistical analysis. It involves identifying patterns, relationships, and themes across studies to provide an integrated interpretation of the evidence when meta-analysis is not possible or appropriate.

Why is it important to do a narrative synthesis?

  • Allows inclusion of studies with different designs, measures, or outcomes that can’t be statistically combined.
  • Explains how and why interventions or phenomena work (or don’t) across different settings.
  • Identifies consistencies, contradictions, and gaps in the evidence.

When do I do use narrative synthesis?

  • Meta-analysis is not possible
  • The review includes qualitative or mixed-methods studies that can’t be statistically combined.
  • Data are incomplete or inconsistently reported, preventing quantitative pooling.
  • The aim is to explore context, mechanisms, or implementation, not just effectiveness.
  • You want to describe and interpret patterns in study findings rather than calculate an overall effect size.
  • Heterogeneity is high, making a statistical summary misleading or inappropriate.

How do I do a narrative synthesis?

  1. Map and tabulate the studies.
    1. Extract key study details (population, intervention, comparator, outcomes, setting, design) into a table so you can see patterns at a glance.
  2. Develop a preliminary synthesis
    1. Group studies by meaningful characteristics and produce short summaries for each group.
  3. Explore relationships within and between studies - look for patterns or modifiers
  4. Use visual aids like summary tables or conceptual diagrams.
  5. Synthesise into conclusions and implications
  6. Be transparent about methods
  7. Report limitations
Example
Llewellyn-Beardsley, J., et al. (2019). Characteristics of mental health recovery narratives: Systematic review and narrative synthesis. PloS one, 14(3), e0214678. 

Recommended resources:

Cochrane Training. (2020). Definition and use of ‘narrative synthesis’.

>A synthesis of qualitative data in a systematic review is a structured process of bringing together and interpreting findings from multiple qualitative studies to identify common themes, patterns, and explanations. It aims to provide a deeper understanding of people’s experiences, perspectives, and the contexts influencing an issue, rather than measuring effect size.

Why is it important to synthesise qualitative data?

  • Captures experiences and perspectives that quantitative data alone cannot show.
  • Explains how and why interventions or phenomena work (or don’t) in real-world contexts.
  • Identifies barriers and facilitators to implementation, adoption, or behaviour change.
  • Supports patient-centred and context-sensitive decision-making in policy and practice.

When do I synthesise qualitative data?

There are multiple methods for data synthesis of qualitative data:

You may need to use qualitative software to conduct your synthesis.

  • NVivo software for qualitative data analysis is available on select PCs in Clayton, Dandenong, and Casey libraries - look for the signed PC at each library. 
  • Dedoose, Atlas.ti, and MAXQDA are paid alternatives to NVivo.
Example
Macdonald, D., et al. (2023). Experiences of women who have planned unassisted home births in high-resource countries: a qualitative systematic review. JBI evidence synthesis, 21(9), 1732–1763. 

Recommended resources:

After synthesis, you must assess the certainty of the evidence for each individual outcome in your review. 

This process shares characteristics with the critical appraisal stage, but is a separate and more robust process which is used to rate the body of evidence at the outcome level rather than the study level. Your certainty of evidence assessments will influence how readers apply the findings of your review to their own setting or clinical practice. It transforms your findings from evidence to recommendations.

Why is it important to assess the certainty of evidence?

  • Helps decision-makers judge whether evidence is strong enough to inform policy or practice.
  • Distinguishes between well-supported and uncertain results, preventing overinterpretation.
  • Highlights research gaps where evidence is weak or inconsistent.
  • Improves transparency by clearly explaining how conclusions were reached.

When do I assess the certainty of evidence?

  1. Choose a framework – GRADE for quantitative, CERQual for qualitative evidence.
  2. Start with baseline rating – e.g., RCTs = high, observational = low (GRADE).
  3. Assess key domains – risk of bias, inconsistency, indirectness, imprecision, publication bias (GRADE); methodological limitations, coherence, adequacy, relevance (CERQual).
  4. Adjust certainty by downgrading or upgrading based on the domain assessment.
  5. Clearly justify ratings in summary tables and use in conclusions.

Review teams with 3 members or less can use the free version of GRADEpro GDT, an online tool, to conduct the assessment and generate a summary of findings (SoF) table. 

Examples
GRADE Example:
Loots, E., et al. (2021). Interventions to Improve Medication Adherence in Patients with Schizophrenia or Bipolar Disorders: A Systematic Review and Meta-Analysis. International journal of environmental research and public health, 18(19), 10213. 

CERQual Example:
Cooper, S., et al. (2019). Factors that influence parents' and informal caregivers' acceptance of routine childhood vaccination: a qualitative evidence synthesis. The Cochrane Database of Systematic Reviews, 2019(2), CD013265. 

Recommended resources:

The Monash Centre for Health Research and Implementation (MCHRI) offers clinical staff at Monash Health one free biostatistics consultation, a total of up to 2 hours of statistical support. Additional support is thereafter offered for a fee. See the link below for more information.

MCHRI Biostatistics Consulting Service