How Long Does a Systematic Review Take?

Why Duration Varies

A systematic review's timeline is shaped by several factors working together rather than any single one: the breadth of the research question, how many databases need to be searched, how many records that search returns, the size and availability of the screening team, how easily full-text articles can be obtained, how complex the data extraction is, whether a formal risk-of-bias assessment is required for every included study, and whether the included studies are similar enough to support a meta-analysis. Two reviews on similarly sized topics can take very different amounts of time if one team works part-time around other commitments and the other is fully dedicated to the project.

Main Stages

A systematic review generally moves through the same broad stages regardless of topic: protocol development and registration, designing and running the search strategy, title and abstract screening, full-text assessment, data extraction, risk-of-bias assessment, evidence synthesis (narrative or meta-analysis), and writing up the results following a reporting standard such as PRISMA. Some of these stages can overlap — for example, data extraction often begins on the studies confirmed first, before full-text screening of the rest is finished — but each one still needs its own realistic time allowance.

An Approximate Planning Framework

The range below is an illustrative planning example, not a statistic drawn from a specific study of review timelines — treat it as a starting point to adjust based on your own scope, team size, and the number of records your search returns.

As one hypothetical illustration: a systematic review might realistically span somewhere between about 6 months and 18 months from protocol registration to a submission-ready manuscript, with a narrower, well-resourced review sometimes finishing faster, and a broad, multi-database review with a large screening volume often taking longer. This range exists only to make the stage-by-stage factors below concrete — build your own plan around those factors, not around this range.

Protocol Development

Drafting and registering a protocol — including the research question, search strategy, and eligibility criteria — typically takes a few weeks, depending on how much discussion and refinement the research question needs and whether registration (for example, through PROSPERO) is required before searching begins. Rushing this stage tends to cost more time later, since a poorly defined question or unclear eligibility criteria usually surfaces later as inconsistent screening decisions.

Search Strategy

Designing and running the search itself — building the search string for each database, testing it, and exporting results — is usually one of the faster stages, but its output determines how long everything downstream takes: a broad search across many databases can return anywhere from a few hundred to tens of thousands of records, and that record count is the single biggest driver of how long screening will take.

Screening

Title and abstract screening is usually the most time-intensive single stage, since methodological guidelines recommend two reviewers screening independently to reduce bias, and every record needs a decision. A review with a few hundred records might complete this stage in a couple of weeks; one with tens of thousands of records, especially with a small screening team, can take several months. Screening speed generally scales with team size more directly than any other stage.

Full-Text Assessment

Once title and abstract screening narrows the pool down, the remaining records need their full text retrieved and assessed against the eligibility criteria in detail. This stage is usually faster than initial screening simply because there are fewer records left, but it can be slowed significantly if full texts aren't readily accessible — for example, articles behind a paywall your institution doesn't subscribe to, or older studies that were never digitized.

Data Extraction

Extracting data from each included study into a structured, standardized form takes longer the more variables are being captured and the more inconsistently the included studies report their own methods and results. Piloting the extraction form on a handful of studies first — a step worth the time it takes — usually prevents having to redo extraction later because the form didn't capture something needed.

Risk-of-Bias Assessment

Assessing risk of bias for every included study, using a tool appropriate to the study design, adds time roughly in proportion to how many studies were included and how detailed the chosen tool is. Like screening and extraction, this is typically done by two reviewers independently and then reconciled, which takes longer than a single-reviewer assessment but is the recommended approach for reducing individual bias in the ratings themselves.

Synthesis and Meta-Analysis

If the included studies are similar enough to pool, running a meta-analysis — checking data compatibility, calculating pooled estimates, assessing heterogeneity, and producing forest plots — adds meaningful time on top of a narrative synthesis alone. A narrative synthesis of a modest number of studies can often be drafted in a few weeks; a full meta-analysis, particularly one exploring several subgroups or sensitivity analyses, can extend that by additional weeks or more depending on how many outcomes are being pooled.

Writing and PRISMA Reporting

Writing the manuscript itself — including a completed PRISMA checklist and flow diagram — typically overlaps with the later stages of extraction and synthesis rather than starting only once everything else is finished. Reviews that document their process thoroughly as they go (search dates, exact record counts at each stage, extraction decisions) generally write up faster than reviews that have to reconstruct that information retroactively.

What Usually Causes Delays

The most common sources of delay aren't usually the stages themselves but the gaps around them: waiting on full-text access requests, screening or extraction disagreements that need repeated discussion, a protocol that turns out to be too broad once real search results come in, team members with limited availability, and revision cycles after supervisor, co-author, or peer reviewer feedback. Building in time for at least one round of revisions, rather than assuming the first draft will be final, tends to make the overall timeline more realistic.

Ways to Improve Efficiency Without Sacrificing Rigor

Piloting the screening and extraction forms on a small sample before applying them to the full set catches ambiguities early, when they're cheap to fix. Using reference-management and screening software to deduplicate records and track decisions reduces manual error and rework. Requesting full-text access to likely-relevant articles early, rather than waiting until screening formally reaches them, avoids a late bottleneck. None of these steps skip a required methodological stage — they reduce wasted time around the steps that still have to happen properly.