TL;DR
- A systematic review is a research methodology built on a predefined protocol, comprehensive search, and explicit inclusion criteria. A meta-analysis is a statistical technique that pools effect sizes across studies. A meta-analysis can sit inside a systematic review; a systematic review does not require one.
- Choose narrative synthesis when studies are too heterogeneous to combine. Choose pairwise meta-analysis when two or more studies measure the same outcome comparably. Choose network meta-analysis when you need to compare three or more treatments that have never been tested head-to-head.
- Both approaches follow the PRISMA 2020 reporting guideline and should be registered prospectively, ideally in PROSPERO, before data extraction begins.
If you are new to evidence synthesis, the difference between a systematic review and a meta-analysis can feel like splitting hairs. The two terms get used interchangeably in casual conversation, in cover letters, and occasionally in manuscript titles that reviewers then flag for correction. They describe different things, and the distinction changes how you plan a project, register a protocol, and describe your work to a journal.
This guide separates the methodology from the statistics, walks through when each applies, and shows how the reporting standards (PRISMA 2020, GRADE, RoB 2) tie the two together in practice.
In Simple Words
A systematic review is the process, a documented, repeatable way of finding and judging every relevant study. A meta-analysis is one possible output of that process, a single combined number, calculated when the included studies are similar enough to average together. You can have a systematic review with no meta-analysis. A meta-analysis with no systematic review behind it is generally considered weak evidence, because nobody checked whether the studies feeding into it were selected fairly.
Quick Definitions
Systematic Review: A comprehensive, transparent literature review that follows a predefined protocol to identify, evaluate, and synthesize all relevant studies on a specific question.
Meta-Analysis: A statistical technique that combines quantitative results from multiple studies to produce a pooled estimate of effect.
The relationship is one of containment, not equivalence. A meta-analysis is a statistical component that can be included within a systematic review, but a systematic review does not require a meta-analysis to be valid or publishable.
Systematic Review: The Research Methodology
A systematic review is a method for conducting research, not a statistical procedure. The PRISMA 2020 statement, published in the BMJ by Page and colleagues, replaced the 2009 version with a 27-item checklist covering how reviewers should identify, select, appraise, and synthesize studies, reflecting a decade of methodological advances since the original guideline.
Key Features
- Predefined protocol: research question, search strategy, and analysis plan documented before screening starts
- Comprehensive search: multiple databases, gray literature, and hand-searching of reference lists
- Explicit inclusion and exclusion criteria: clear rules applied consistently across all candidate studies
- Formal risk-of-bias assessment: a structured tool, not an informal judgment call
- Transparent reporting: full documentation of every screening and exclusion decision
- Narrative or quantitative synthesis: the findings can be summarized in prose, pooled statistically, or both
Protocol registration matters more than most first-time reviewers expect. A 2019 analysis of 182 orthodontic systematic reviews found that only 37 reviews, 20.3 percent of the total, had been registered in PROSPERO, and registered reviews scored measurably higher on the AMSTAR quality tool than unregistered ones. Registering a protocol in advance, through PROSPERO or a PRISMA-P protocol, guards against changing your inclusion criteria after seeing which studies support your hypothesis.
When a Narrative Systematic Review Is the Right Call
Not every systematic review needs statistical pooling. A narrative synthesis, without meta-analysis, is the appropriate output when studies are too heterogeneous to combine, when outcomes are measured on different scales across studies, when study designs vary too widely for a single pooled number to mean anything, or when the question is exploratory and better served by a qualitative map of the evidence than a single effect estimate.
Meta-Analysis: The Statistical Technique
Meta-analysis is a statistical method for combining data, not a research design in its own right. It takes effect sizes already extracted through a systematic process and produces a single pooled estimate, weighted so that larger and more precise studies contribute proportionally more to the result.
Key Features
- Quantitative pooling: combines effect sizes (odds ratios, mean differences, hazard ratios) across studies
- Weighted averaging: larger, more precise studies carry more weight in the pooled estimate
- Forest plots: visual display of each study’s effect alongside the pooled estimate
- Heterogeneity testing: I-squared and the Q statistic quantify how much studies disagree with each other
- Subgroup analysis: explores whether patient population, dose, or study design explains variation in results
- Publication bias testing: funnel plots and Egger’s test check whether small negative studies are missing from the literature
Meta-analysis makes sense once studies address the same question, measure outcomes similarly enough to standardize, and number at least two, though more studies generally produce a more stable pooled estimate. Heterogeneity does not have to be zero. It has to be low enough to explain, or explicitly modeled through subgroup analysis and meta-regression.
Editverse has supported evidence synthesis teams across specialties long enough to see the same submission pattern repeat. The most common desk-rejection trigger is a manuscript titled as a systematic review whose methods section never names a search strategy, never mentions dual independent screening, and never applies a formal risk-of-bias tool. Reviewers catch the mismatch between title and methodology within the first read-through, and it is one of the easiest problems to prevent by following PRISMA 2020 from the protocol stage onward.
Side-by-Side Comparison
| Feature | Systematic Review | Meta-Analysis |
|---|---|---|
| Type | Research methodology | Statistical technique |
| Output | Narrative synthesis | Pooled effect estimate |
| Data type | Qualitative and quantitative | Quantitative only |
| Reporting guideline | PRISMA 2020 (27-item checklist) | PRISMA 2020, same checklist |
| Risk-of-bias tool | RoB 2 (trials) or ROBINS-I (non-randomized) | Same tools, feeds the weighting |
| Visual output | PRISMA flow diagram, evidence tables | Forest plots, funnel plots |
| Certainty rating | Optional, often narrative | GRADE (high, moderate, low, very low) |
| Can stand alone? | Yes | Technically yes, but rarely advisable without one |
| Minimum studies | Can report zero eligible studies | At least two studies |
The Relationship Between Systematic Review and Meta-Analysis
A systematic review without meta-analysis is common and valid. You comprehensively identify and assess studies, then summarize findings narratively when statistical pooling would not be meaningful.
A systematic review with meta-analysis represents the stronger evidence base when the data permits it. You complete every step above, then add a statistically pooled effect estimate with a quantified measure of precision and heterogeneity.
A meta-analysis without a systematic review behind it is the combination reviewers and editors treat with the most skepticism. Pooling data without a documented, reproducible identification process risks missing relevant studies or introducing selection bias into which effect sizes get averaged.
Reporting Standards: PRISMA 2020, RoB 2, and GRADE
The PRISMA 2020 statement covers both narrative systematic reviews and meta-analyses under a single 27-item checklist, alongside a companion abstract checklist and revised flow diagrams for original and updated reviews. Protocol-stage reporting is covered separately by PRISMA-P, which documents the search strategy and analysis plan before any screening begins, guarding against post-hoc changes to eligibility criteria.
Risk of bias in the individual studies feeds directly into how much weight a meta-analysis gives each result. For randomized trials, the RoB 2 tool published by Sterne and colleagues assesses bias across five domains, from randomization process to selection of the reported result, replacing the older, less structured Cochrane risk-of-bias tool.
Once effect estimates are pooled, GRADE rates how much confidence to place in them. The GRADE framework assigns one of four certainty ratings, high, moderate, low, or very low, to a body of evidence rather than to any single study. Randomized trials start as high-certainty evidence and observational studies start low, before being adjusted up or down for factors including imprecision, inconsistency, and publication bias, as Guyatt and colleagues set out in the framework’s founding BMJ paper.
Network Meta-Analysis: Comparing Treatments That Were Never Tested Head-to-Head
Network meta-analysis (NMA) extends pairwise meta-analysis to compare three or more treatments at once, including pairs that no trial has directly compared. As Puerto Nino and Brignardello-Petersen describe in their methods overview, NMA integrates direct evidence from head-to-head trials with indirect evidence routed through a common comparator, such as placebo, to estimate how two treatments would compare even without a trial that tested them against each other.
NMA requires a connected network of comparisons, meaning every treatment must link back to the network through at least one shared comparator. It produces treatment rankings in addition to pooled effect estimates, which is why guideline panels increasingly request it when multiple competing therapies exist for the same condition and direct comparisons remain sparse.
Which Should You Choose?
Choose a Narrative Systematic Review When:
- Your question is broad, exploratory, or maps a research landscape (a scoping review)
- Studies are too diverse in design or outcome measures to combine meaningfully
- Qualitative synthesis captures more of the answer than a single pooled number would
Choose Systematic Review with Meta-Analysis When:
- Your question is focused and the outcome is quantitative
- Studies measure the same or standardizable outcomes
- You need a precise pooled effect estimate with a confidence interval
- Guideline developers or policymakers are the intended audience
Choose Systematic Review with Network Meta-Analysis When:
- Three or more treatments compete for the same indication
- Direct head-to-head trials are missing for some treatment pairs
- A clinical guideline panel needs a ranked comparison across all options, not just one pair
Common Mistakes to Avoid
Calling a literature review a systematic review without following a documented protocol is the single most frequent error we see cross an editor’s desk. A close second is running a meta-analysis on studies with substantial unexplained heterogeneity (I-squared above 75 percent, with no subgroup structure to account for it) and presenting the pooled estimate as though it settles the question. A third is skipping prospective registration entirely, then adjusting inclusion criteria mid-review once the search results are in, which is exactly the selective-reporting risk that PROSPERO registration and PRISMA-P protocols exist to prevent.
Publication and Citation Impact
Both systematic reviews and meta-analyses rank among the most cited publication types in biomedicine. Meta-analyses often out-cite narrative systematic reviews on the same topic because they resolve a debate into one number, and journals, guideline panels, and subsequent reviews all tend to cite the paper that provided the definitive pooled estimate rather than the one that summarized findings narratively. That citation advantage is a downstream effect of methodological rigor, not a shortcut. A poorly conducted meta-analysis with high, unexplained heterogeneity is less citable and more likely to draw a published rebuttal.
For h-index growth, both approaches are strong choices. The decision should be driven by which method actually answers your research question, not by which one is perceived as more prestigious.
Frequently Asked Questions
Can a systematic review exist without a meta-analysis?
Yes. A narrative systematic review is a complete, publishable, PRISMA-compliant piece of research on its own. Statistical pooling is added only when the included studies are similar enough to combine meaningfully.
Can you run a meta-analysis without doing a full systematic review first?
Technically yes, but reviewers and editors treat standalone meta-analyses with more scrutiny, since there is no documented, reproducible process confirming that every relevant study was identified and fairly assessed for inclusion.
How many studies do you need for a meta-analysis?
Two is the statistical minimum, but most methodologists recommend at least three to five before the pooled estimate becomes stable and heterogeneity statistics become interpretable.
What is the difference between a systematic review and a scoping review?
A systematic review answers a focused question and typically includes a risk-of-bias assessment. A scoping review maps the breadth of literature on a broader topic, often to identify gaps, and does not always apply the same rigor to quality appraisal.
Does a meta-analysis require PROSPERO registration?
Registration is not legally required but is expected by most high-impact journals and by Cochrane. Registered protocols are associated with measurably higher quality scores on tools like AMSTAR, and registration timestamps your methods before you have seen the results.
What is a network meta-analysis and when is it needed?
Network meta-analysis compares three or more treatments simultaneously, combining direct trial evidence with indirect comparisons through shared comparators. It is needed when guideline panels must rank multiple competing therapies that have rarely or never been tested against each other directly.
Getting Expert Help
Whether you are planning a narrative systematic review, a pairwise meta-analysis, or a network meta-analysis, methodological rigor at the protocol stage determines whether the finished manuscript survives peer review. Our team supports researchers across each of the following stages.
- Protocol development and PROSPERO registration
- Literature search strategy design and dual-reviewer screening
- Data extraction and RoB 2 / GRADE quality assessment
- Statistical analysis, including pairwise and network meta-analysis
- Manuscript writing, PRISMA 2020 compliance checks, and journal submission
Ready to start your evidence synthesis project? Explore our systematic review and meta-analysis service, read our complete guide to meta-analysis methods, or see our worked forest-plot and heterogeneity examples. For related methodology, see our guides to the PRISMA flow diagram, intention-to-treat versus per-protocol analysis, study registration for non-trial designs, and how journal citation metrics factor into where you submit. Or contact us for a free consultation.
Professional Systematic Review Service by Editverse
PhD-Level Researchers
8-16 Week Delivery
Cochrane Methodology
References
- Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. https://doi.org/10.1136/bmj.n71
- Shamseer L, Moher D, Clarke M, et al. Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015: elaboration and explanation. BMJ. 2015;350:g7647. https://doi.org/10.1136/bmj.g7647
- Sterne JAC, Savović J, Page MJ, et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ. 2019;366:l4898. https://doi.org/10.1136/bmj.l4898
- Balshem H, Helfand M, Schünemann HJ, et al. GRADE guidelines: 3. Rating the quality of evidence. J Clin Epidemiol. 2011;64(4):401-406. https://doi.org/10.1016/j.jclinepi.2010.07.015
- Guyatt GH, Oxman AD, Vist GE, et al. GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ. 2008;336(7650):924-926. https://doi.org/10.1136/bmj.39489.470347.AD
- Sideri S, Papageorgiou SN, Eliades T. Registration in PROSPERO of systematic review protocols was associated with increased review quality. J Clin Epidemiol. 2019;100:103-110. https://doi.org/10.1016/j.jclinepi.2018.01.003
- Puerto Nino AK, Brignardello-Petersen R. How to read a network meta-analysis. Eur Urol Focus. 2023. https://doi.org/10.1016/j.euf.2023.10.018