Written and maintained by CASRAI Editorial Board
Last updated
The Cochrane Handbook for Systematic Reviews of Interventions runs to 26 numbered chapters across four Parts, plus five roman-numeral chapters on the review process itself — it is a reference manual, not a document meant to be read start to finish. Most people arrive at it mid-task with one specific question (how do I choose an effect measure, what tool covers missing outcome data, which chapter explains network meta-analysis) and lose time paging through a chapter that turns out to be adjacent to, not actually about, what they needed. This guide is a direct map: task or question on one side, the exact Cochrane Handbook chapter that answers it on the other, plus a link to CASRAI’s own coverage of that specific method where one exists.
How the Handbook is structured
Version 6 of the Handbook (maintained at cochrane.org/authors/handbooks-and-manuals/handbook/current) is organized into four Parts. Part 1 uses roman numerals and covers the review process itself — planning, reporting, updating. Parts 2 through 4 use arabic numbers and run sequentially, roughly following the order a review team actually works through a review, from formulating the question to handling non-randomized evidence.
| Part | Covers | Chapters |
|---|---|---|
| Part 1 — About Cochrane Reviews | The review process: planning, reporting, updating, overviews of reviews | I–V |
| Part 2 — Core methods | The methodology every review needs: scoping, searching, data collection, effect measures, risk of bias, meta-analysis, GRADE | 1–15 |
| Part 3 — Specific perspectives in reviews | Cross-cutting considerations that apply on top of the core methods: equity, complex interventions, patient-reported outcomes, adverse effects, economic evidence, qualitative evidence | 16–21 |
| Part 4 — Other topics | Extensions beyond the standard randomized-trial synthesis: living reviews, non-randomized studies, individual participant data | 22–26 |
Task-to-chapter map: Part 1 — the review process
| Your question | Chapter | What it covers |
|---|---|---|
| How do I plan and register a review before doing any of the actual work? | Chapter II — Planning a Cochrane Review | Protocol development, the case for a systematic review, assembling a review team. See CASRAI’s PRISMA-P protocol template guide and how to register on PROSPERO. |
| What does my written-up review need to include to be reportable? | Chapter III — Reporting the review | Structuring the final report so it’s transparent and reproducible. Pairs directly with CASRAI’s PRISMA 2020 checklist and flow-diagram guide, which is the reporting standard most journals actually require alongside Handbook-compliant conduct. |
| My review is a few years old — how do I update it, and who counts as an author on the update? | Chapter IV — Updating a review | When and how to re-run searches and re-analyze for a review update. CASRAI’s guide to Cochrane’s authorship rules covers the byline-order question this chapter raises but doesn’t fully resolve on its own. |
| I want to summarize several existing systematic reviews on a topic, not run a fresh one | Chapter V — Overviews of Reviews | The methodology for an overview (sometimes called an “umbrella review”): synthesizing multiple existing systematic reviews rather than primary studies. See CASRAI’s umbrella review definition. |
Task-to-chapter map: Part 2 — core methods
| Your question | Chapter | What it covers |
|---|---|---|
| How do I define my review question and its scope before writing eligibility criteria? | Chapter 2 — Determining the scope of the review and the questions it will address | Framing the review question, typically via PICO. See CASRAI’s PICO framework guide and how to write a PICOT question. |
| What exactly counts as an “eligible” study, and how do I group studies that measure similar-but-not-identical things? | Chapter 3 — Defining the criteria for including studies and how they will be grouped for the synthesis | Inclusion/exclusion criteria and the “lumping vs. splitting” decision for how comparisons are structured. |
| Where do I search, and how do I build a search strategy that won’t get flagged as insufficiently sensitive? | Chapter 4 — Searching for and selecting studies | Database selection, search-strategy construction, study-selection workflow. See CASRAI’s searching the Cochrane Library guide (CENTRAL vs. CDSR), PubMed-to-Embase translation, grey-literature searching, and Boolean search operators. |
| How do I extract data from included studies without introducing transcription errors? | Chapter 5 — Collecting data | Data-collection forms, piloting extraction, dual independent extraction and reconciliation. |
| Should I use a risk ratio, mean difference, or standardized mean difference — and what does Cochrane actually mean by “SMD”? | Chapter 6 — Choosing effect measures and computing estimates of effect | Effect-measure selection by outcome type. See CASRAI’s SMD and Hedges’ g guide — the Handbook’s own definition of SMD is specifically Hedges’ (adjusted) g, not raw Cohen’s d, which most generic explainers of the term miss. |
| Before I get to formal risk-of-bias tools, how should I think about bias and conflicts of interest generally? | Chapter 7 — Considering bias and conflicts of interest among the included studies | Framing for the bias-assessment chapters that follow; conflicts of interest among study authors as a distinct concern from methodological bias. |
| How do I formally assess risk of bias in a randomized trial (RoB 2)? | Chapter 8 — Assessing risk of bias in a randomized trial | The RoB 2 tool: five domains, signalling questions, traffic-light output. See CASRAI’s RoB 2 / ROBINS-I traffic-light guide. |
| How do I make sure I’m comparing studies that are actually comparable before I pool anything? | Chapter 9 — Summarizing study characteristics and preparing for synthesis | Characteristics tables, checking studies address a similar-enough PICO before synthesis proceeds. |
| How do I actually run the meta-analysis — fixed-effect or random-effects, and how do I read I²? | Chapter 10 — Analysing data and undertaking meta-analyses | Statistical combination of results. See CASRAI’s fixed-effect vs. random-effects guide, heterogeneity (I², τ², prediction intervals) guide, meta-regression guide, and how to read a forest plot. |
| I have three or more interventions to compare at once, not just two | Chapter 11 — Undertaking network meta-analyses | Combining direct and indirect evidence across multiple interventions in one model. See CASRAI’s network meta-analysis definition. |
| My studies are too different to pool statistically — what do I do instead? | Chapter 12 — Synthesizing and presenting findings using other methods | Structured narrative synthesis when meta-analysis isn’t appropriate. See CASRAI’s narrative synthesis and SWiM guide. |
| Some of my studies have missing outcome data, or I’m worried about publication bias skewing my pooled result | Chapter 13 — Assessing risk of bias due to missing results in a synthesis | Missing-results bias as a distinct mechanism from within-study risk of bias. See CASRAI’s publication bias detection guide and funnel plot definition — note the Handbook is explicit that funnel-plot asymmetry is a generic small-study-effects signal, not proof of publication bias on its own. |
| How do I build a Summary of Findings table and grade how certain my pooled result actually is? | Chapter 14 — Completing ‘Summary of findings’ tables and grading the certainty of the evidence | GRADE methodology: four certainty levels across five domains. See CASRAI’s GRADE (evidence certainty rating) definition. |
| I have my results — how do I write conclusions that don’t overstate what the evidence supports? | Chapter 15 — Interpreting results and drawing conclusions | Translating statistical results into clinically/practically meaningful interpretation, distinguishing statistical from clinical significance. |
Task-to-chapter map: Part 3 — specific perspectives
These chapters layer on top of the Part 2 core methods rather than replacing them — use them when your review has a specific angle the core chapters don’t fully address.
| Your question | Chapter |
|---|---|
| My review needs to address health equity or a specific population subgroup | Chapter 16 — Equity and specific populations |
| The intervention I’m reviewing is multi-component or context-dependent, not a single simple treatment | Chapter 17 — Intervention complexity |
| I need to synthesize patient-reported outcome measures across studies that used different instruments | Chapter 18 — Patient-reported outcomes |
| My review needs to systematically capture harms, not just efficacy | Chapter 19 — Adverse effects |
| I need to incorporate cost-effectiveness or other economic evidence | Chapter 20 — Economic evidence |
| My review needs to synthesize qualitative evidence alongside or instead of quantitative results | Chapter 21 — Qualitative evidence |
Task-to-chapter map: Part 4 — other topics
| Your question | Chapter | What it covers |
|---|---|---|
| I want my review to stay current as new trials publish, rather than being a one-time snapshot | Chapter 22 — Prospective approaches to accumulating evidence | Living systematic reviews and other prospective evidence-accumulation methods. See CASRAI’s living systematic review definition. |
| My included studies use cluster randomization, crossover design, or another randomized-trial variant | Chapter 23 — Including variants on randomized trials | Adjustments needed for non-standard randomized designs before they can be pooled with parallel-group trials. |
| I need to include non-randomized studies of an intervention’s effects, not just RCTs | Chapter 24 — Including non-randomized studies on intervention effects | When and how NRSI evidence is incorporated alongside or instead of randomized evidence. |
| How do I assess risk of bias in a non-randomized study — ROBINS-I or ROBINS-E? | Chapter 25 — Assessing risk of bias in a non-randomized study | See CASRAI’s ROBINS-I guide (non-randomized studies of interventions) and ROBINS-E guide (non-randomized studies of exposures) — distinct tools for a distinct study type, not interchangeable with RoB 2. |
| I have raw participant-level data from the original trials, not just published aggregate results | Chapter 26 — Individual participant data | IPD meta-analysis: a gold-standard variant enabling consistent outcome definitions and handling of missing data across studies, at the cost of needing cooperation from original investigators. |
What this guide doesn’t replace
This is a navigational map, not a substitute for reading the chapter you land on. The Handbook itself is the authority on methodological detail (cochrane.org/authors/handbooks-and-manuals/handbook/current); this guide exists to save the time of finding the right entry point. If your review is registered with a specific Cochrane Review Group, that group’s editorial guidance takes precedence over any general reading of a Handbook chapter where the two appear to differ — check with your editor before treating this map as the final word on a disputed methodological choice.
Frequently asked questions
Do I need to read the whole Handbook before starting a review?
No. Most review teams read Chapter II (planning) in full early on, then return to the specific Part 2 chapters relevant to decisions they’re actively making — search strategy when building the search, the effect-measure chapter when choosing how to report results, and so on. Treating it as a reference manual rather than a cover-to-cover read is the intended use.
Is the chapter numbering the same across every Handbook version?
No. This guide reflects the current Version 6 numbering as published at cochrane.org/authors/handbooks-and-manuals/handbook/current. Older citations to “Handbook 5.1” chapter numbers do not map one-to-one onto Version 6 — Version 6 added and reorganized several chapters, most notably splitting risk-of-bias content by study design (Chapters 8 and 25) and adding the current network meta-analysis and missing-results chapters (11 and 13). Always check which version a citation refers to before assuming a chapter number carries over.
My review isn’t a Cochrane review — does the Handbook still apply?
Yes, for methodology. The Handbook is written for Cochrane Reviews specifically (Cochrane Review Group editorial process, CDSR publication), but its Part 2 core-methods chapters are the de facto methodological reference most non-Cochrane systematic reviews of interventions also follow, and PRISMA 2020 reporting checklists assume Handbook-consistent conduct. What doesn’t carry over is the Cochrane-specific editorial process itself (Chapters II–V) and the Cochrane-branded reporting format.
Where do PRISMA and the Cochrane Handbook fit together?
They answer different questions. The Handbook tells you how to conduct a systematic review; PRISMA 2020 tells you what to report once you’ve conducted it. A review can follow Handbook methodology and still fail a PRISMA-based reporting check if the write-up omits required detail, and vice versa — a well-reported review can still reflect Handbook-inconsistent methods if a reviewer doesn’t check the underlying conduct, not just the checklist.
