SynthGap

Guides to planning a meta-analysis or systematic review

A good systematic review or meta-analysis topic passes three tests: it is novel, it is valid, and it is impactful. These guides explain each test and how to apply it before you spend months on a review, how to plan the review once you have chosen, and where AI tools help and where they need a person.

Choose a topic5

How to find a novel, valid and impactful meta-analysis topicA practical method for choosing a meta-analysis topic that is new, methodologically sound and worth doing: three tests, five steps and a checklist.What is a research gap, and how do you find one?A research gap is a question existing evidence cannot yet answer. The main types, how to confirm a gap is real, and how to turn it into a review topic.Choosing a meta-analysis project as a medical student or residentHow to choose a realistic meta-analysis or systematic review project as a student or resident: scope, timeline, supervisor, and avoiding a topic already done.When is an old systematic review worth updating?How to decide whether to update an out-of-date systematic review: signals for updating, checking what has changed, and planning an update or living review.Using AI to find a systematic review topic: what to trust and what to checkWhat AI tools can and cannot do when you look for a review topic, a checklist for verifying their suggestions, and how to report AI use in your methods.

Test your topic6

Is my meta-analysis topic novel?What counts as a novel systematic review topic, what does not, and how to decide whether to skip, update, extend or proceed when a review already exists.Is my meta-analysis topic valid and feasible?How to test whether a meta-analysis question has enough comparable, good-quality studies to answer it, with red flags that mean you should stop.Is my meta-analysis topic impactful?How to judge whether a review topic adds to the literature and answers an important question: decision uncertainty, burden and guideline gaps.How to check whether a systematic review already existsA step-by-step method to check whether a systematic review on your question already exists, in PubMed, Cochrane, PROSPERO and Epistemonikos.Literature review finder: how to find reviews already written on a topicWhat a literature review finder does, where to find existing literature and systematic reviews on a topic, how to read one for gaps, and where SynthGap fits.How many studies do you need for a meta-analysis?There is no single minimum, but there are practical thresholds: two studies to pool, about five for heterogeneity, ten for funnel plots and subgroup analysis.

Plan the review9

Planning to do a meta-analysis? Decide these ten things firstPlanning to do a meta-analysis? Ten decisions to make before you start, in order: topic, existing reviews, feasibility, team, time, protocol and tools.Systematic review, meta-analysis, scoping review or umbrella review: which one?A systematic review is a method; a meta-analysis is a statistical step that may be part of it. The differences, and when to choose a scoping or umbrella review.How to write a PICO question for a meta-analysisA step-by-step guide to writing a PICO question for a systematic review or meta-analysis, with worked examples for treatment, risk, diagnosis and prevalence.How long does a meta-analysis take, and what does it cost?How long a meta-analysis takes, what drives the time, how many people it needs and what it costs, with figures from studies of completed reviews.Can you do a meta-analysis alone? The team and roles you needWho you need for a meta-analysis: a content expert, an information specialist, two independent reviewers and a statistician, and what to do if you work alone.Meta-analysis protocol checklist: what to decide before you searchWhat goes in a meta-analysis protocol before you search: question, eligibility, search, outcomes, bias, synthesis plan and amendments, as a checklist.How to register a PROSPERO protocolWhat a PROSPERO protocol is, which reviews PROSPERO accepts, what to write before you register one, and what to do if your review is not accepted.Meta-analysis software: how to choose between R, RevMan, Stata and the restWhich software to use for a meta-analysis? RevMan, R, Stata, JASP, jamovi and Comprehensive Meta-Analysis compared by skills needed, cost and range of methods.Ten meta-analysis mistakes, and how to avoid themThe ten commonest ways a meta-analysis goes wrong, from a loose question to double-counted trials and over-claimed conclusions, and how to prevent each.

Use AI tools8

AI tools for meta-analysis: what they can do at each stageAI tools for meta-analysis, stage by stage: what each can do, where it goes wrong, what to check, where the statistics belong and where SynthGap fits.AI tools for systematic reviews: where they help and where a person must decideWhich AI tools help with a systematic review, in three tiers of trust, what the evidence says about screening, search and extraction, and how to add one safely.Can ChatGPT do a meta-analysis?Can ChatGPT or another chatbot do a meta-analysis? What it does well, where it invents numbers and references, and a safe way to use it.AI screening for systematic reviews: how it works and how to check itHow AI screening works for systematic reviews (ranking, classifiers, language models), how to test it on your own records and what to report.Can AI write a systematic review search strategy?ChatGPT-style tools can draft Boolean searches for a systematic review, but they miss studies and invent subject headings. How to use them and what to check.AI data extraction for meta-analysis: what it gets right and the errors to catchCan AI extract data for a meta-analysis? What it gets right, the numeric mistakes to expect (SD versus SE, wrong arm) and how to check every number.How to choose an AI tool for a systematic review or meta-analysis: twelve questionsTwelve questions to ask before you use an AI tool in a systematic review or meta-analysis: evidence, transparency, control, data handling, cost and reporting.How to report AI use in a systematic review or meta-analysisWhat to declare when AI helped with a systematic review or meta-analysis: the RAISE reporting points, PRISMA 2020 items 8, 9 and 11, and wording to adapt.

Let SynthGap run the first pass

Enter an area. SynthGap proposes candidate topics and checks them against existing systematic reviews. Start on the Free plan: no card needed, and every run shows every topic it finds.

Sign up free

Looking for ideas by specialty? Browse topic guides for cardiology, oncology, neurology and more.

General methodological guidance, not medical advice.