Manuscript Writing

Systematic Reviews and Meta-Analyses: A Step-by-Step Guide for Researchers

This step by step guide walks you through how to conduct a high quality systematic review and meta-analysis.

Systematic reviews and meta-analyses (SRMAs) are a common study design in clinical research. They help synthesize the best available evidence to answer specific research questions, guide clinical decision-making, and inform guidelines and policy. However, conducting a high-quality SRMA is challenging and requires a systematic approach.

In this comprehensive guide, we’ll walk you through the exact steps of a SRMA, from forming your research question to publishing your findings and sharing them with the world. You can also download a free SRMA template to help save some time.

Whether you’re a first-time researcher or an experienced academic, this guide will serve as a checklist to ensure your SRMA is rigorous, reproducible, and impactful.

Step 1: Define a Clear PICO Research Question

Your systematic review begins with a focused research question. Use the PICO framework:

For outcomes, ensure you choose one primary outcome and consider any secondary outcomes you want to explore. Consider involving a patient partner to help choose outcomes that are meaningful to patients.

Also, as a general rule, SRMAs are often plagued by high heterogeneity: focusing your population as narrow as possible helps reduce this from the start. 

Step 2: Check if a Recent Systematic Review Already Exists

Before investing time, confirm:

  • Has a recent systematic review already answered your question?
  • Has new primary research been published that would make an update valuable?
  • Is there controversy in the existing literature?

For systematic reviews there is a saying ‘GIGO’: garbage in and garbage out. If there is no reason to perform another SRMA, they simply create noise in the literature as opposed to helping estimate truth (which is the purpose of SRMAs).

Step 3: Develop Your Protocol for Systematic Review and Meta-Analysis

Your protocol is the detailed plan for your systematic review. It should specify:

  • Objectives
  • Eligibility criteria
  • Databases and search strategy
  • Screening and selection processes
  • Data extraction methods
  • Risk of bias assessment tools
  • Approach to determining the certainty of evidence (e.g. GRADE)
  • Planned statistical analysis

My top recommendation for SRMAs (or any research study) is to spend more time than you think necessary on the protocol. This ensures that the research is being done in a quality and reproducible way.

Here are a few considerations for the above steps:

  1. What type of study will you include? Pooling randomized and non-randomized studies is possible, however, can be very challenging. If there are enough randomized studies answering your PICO consider restricting to randomized studies. If pooling observational studies, beware of challenges with heterogeneity, wide confidence intervals, and different measures of associations (e.g. HR vs. OR vs. RR).
  2. Database and Search Strategy: As a general rule, you should search every possible database from inception until present. Restricting to English language is acceptable, however, if you know of studies being published on the topic in another language consider including. You should include a ‘Gray Literature’ search as part of your SRMA. Identify conferences where you know (or suspect) new research pertaining to your PICO are being published. Include these in your search (e.g. do the last 3 years of the top 5 relevant conferences)
  3. Screening. Use an existing software (e.g. Covidence, Rayyan, DistillerSR) to screen your titles/abstracts, full texts. These platforms are all now integrating AI to help improve your process. 
  4. Risk of Bias . Each SRMA type (e.g. observational, interventional, diagnostic accuracy, prognostic) has a specific risk of bias tool that is used to determine potential bias in the primary studies (e.g. QUADAS-2, . These are available or linked through the Cochrane Methods website.

Step 4: Build Your Team and Get Expert Feedback

Systematic reviews benefit from collaboration:

  • Engage librarians, clinical experts, methodologists, and statisticians
  • Secure team members for duplicate screening and data extraction
  • Seek feedback to strengthen your protocol before finalization

Step 5: Design a Comprehensive Search Strategy

Work with an information specialist or librarian to build your systematic search strategy:

  • Search major databases (e.g., PubMed, EMBASE, Cochrane Library)
  • Consider grey literature and preprints
  • Use both keywords and controlled vocabulary (e.g., MeSH terms)

Step 6: Finalize Your Protocol

Incorporate feedback from your team and finalize your systematic review methodology. Your protocol will guide your work and keep you aligned with best practices.

Step 7: Register Your Systematic Review Protocol (PROSPERO or OSF)

Registering your protocol in platforms like PROSPERO or OSF demonstrates transparency and helps avoid duplication in research.

Tip: Most journals now require protocol registration as a condition of publication for SRMAs. 

Step 8: Execute Your Search and Organize References

Run your search across databases. Use dedicated SRMA software like Covidence, Rayyan, DistillerSR to efficiently screen abstracts and full-texts in duplicate.

Step 9: Screen Titles and Abstracts (Duplicate Screening)

Screen all titles and abstracts in duplicate using your eligibility criteria. At this stage, it is acceptable to have all disagreements automatically be sent to full-text review.

Duplicate screening reduces errors and increases accuracy.

Step 10: Screen Full Texts (Duplicate Screening)

Next, conduct full-text screening for studies that pass the initial review. Again, screen in duplicate for rigor. If there are disagreements, consider consensus discussion as a way of resolving these. Remember to record reasons for exclusion of studies at the full text level.

Step 11: Extract Data from Included Studies

Use a standardized data extraction form to capture:

  • Study design
  • Population details
  • Interventions and comparators
  • Outcome measures
  • Study results

Preferably, extract data in duplicate for accuracy.

Step 12: Assess Risk of Bias

Assess the quality of each study using appropriate risk of bias tools:

  • ROB 2.0 for randomized controlled trials
  • ROBINS-I for observational studies
  • QUADAS-2 for diagnostic accuracy studies
  • QUIPS for prognostic studies

Risk of bias assessments are critical for interpreting your findings.

Step 13: Conduct the Meta-Analysis (If Appropriate)

If your data are sufficiently homogeneous to consider meta-analyzing:

  • Perform meta-analysis using random-effects models
  • Assess heterogeneity (I² statistic, chi-squared test)
  • Conduct sensitivity or subgroup analyses as pre-specified. Note: the ICEMAN tool helps researchers figure out whether findings from subgroup analyses are credible are not. These should be pre-specified.

I use RevMan to help complete the actual meta-analysis. It is intuitive, fairly straightforward to use, and often free (or low cost).

Step 14: Grade the Certainty of Evidence (GRADE)

Use the GRADE approach to evaluate the overall certainty of evidence:

  • High, moderate, low, or very low
  • Assess risk of bias, inconsistency, indirectness, imprecision, and publication bias

The software GRADEpro is a fantastic free software automatically generate GRADE certainty of evidence for your SRMA. Note, GRADEing should be done on a per-outcome basis.

GRADEing the Certainty of Evidence on a per-outcome bases

Step 15: Write Your Manuscript (PRISMA-Compliant)

Follow PRISMA guidelines to structure your manuscript and to ensure important reporting guideline items are included (this increases transparency for readers). Make sure to include all the relevant sections in your manuscript. These (and more) are included in this free SRMA template.

  • Abstract
  • Introduction
  • Methods
  • Results
  • Discussion
  • Conclusions

Include a PRISMA flow diagram to document your study selection process.

Step 16: Submit Your Manuscript

Select an appropriate journal for your systematic review. Consider:

  • Journal scope and audience
  • Open access options
  • Impact factor

If you find reformatting your manuscript for each journal a huge pain, consider using a tool like Resub to help automate this.

Step 17: Mobilize Knowledge After Publication

Once published, share your findings widely:

  • Promote via social media (Twitter, LinkedIn)
  • Present at conferences and webinars
  • Create infographics and visual abstracts
  • Write blog posts or policy briefs

Effective knowledge translation increases the impact of your work.

Final Thoughts: Mastering the Systematic Review Process

A systematic review can be an important contribution to your research field; however, this is heavily dependent on the rigor of the methodology. By following these structured steps, you ensure your work is thorough, transparent, and ready to inform clinical practice. 

Want to save some time writing your SRMA? Download this free SRMA template to help get you started.