What is study centric data management?
Traditionally, review authors have entered results data separately for each analysis, so the same study data often gets typed in multiple times. Study centric data management changes this: results are entered and stored once, at the study level, and can then be reused across as many analyses as needed, including subgroup and sensitivity analyses.
This has some practical knock-on effects. Combining arms, changing reference arms, and log-scale transformations are handled automatically. Data has better provenance, because it's entered once rather than retyped for every analysis, so there's less risk of transcription errors. And study data can be imported "in one go," rather than assembled analysis by analysis.
In addition to being a convenient way to enter data, this is a step toward evidence that is structured, reusable, and easier to share across reviews, in line with the interoperability principles of the Evidence Synthesis Infrastructure Collaborative (ESIC). The goal is a future where "study data is extracted once and used to answer many different questions."
Why this matters for your review
Working this way early pays off later. Defining your PICO and your analyses before you extract data means you're forced to think systematically about how your review will be structured, which tends to produce more focused reviews and a more efficient analysis stage. And because results live at the study level, correcting errors and mistakes can be done in one place.
How Covidence supports study centric data
Covidence's data extraction is designed to map directly onto RevMan's study centric structure, so the transfer from extraction to analysis is streamlined.
When you extract data in Covidence, you're capturing study characteristics, study arms, results, and (if you're using Risk of Bias 1) risk of bias assessments, all organised at the study level from the outset.
Once your included studies have gone through consensus and been assigned study IDs, Covidence can export this as four CSV files: study information, study arms, study results, and risk of bias.
Together with a RevMan .ris file for your included and excluded references, these four files import directly into RevMan's study centric data structure. Learn how to export your data to RevMan here.
From there:
New studies in the CSVs are created as included studies in RevMan, complete with characteristics, results, and (Risk of Bias 1) assessments if available. Studies already in RevMan get their characteristics, results, and risk of bias data updated automatically.
Because the CSV step sits between Covidence and RevMan, it also gives you a checkpoint: a chance to tidy up, edit, or remove data before it lands in your review. Even if you're not using RevMan, the CSV exports provide a clean, structured record of your extracted data. For example, the new sheet per section (Excel) export organises the data into separate worksheets for each extraction section, making it much easier to explore, interpret, analyse, and work with your extracted data outside of Covidence.
Getting the most out of it
A few tips to make the process smoother:
Define your PICO and your analyses in RevMan before you extract data in Covidence. Thinking through your review criteria early means your extraction template lines up with what RevMan expects, and it avoids the "file contains inconsistent data" message that comes from mismatched intervention or outcome names.
Define your eligibility criteria early and keep it up to date in Covidence to create a single source of truth for your team from screening through to data extraction.
Before you begin extracting data, use this opportunity to define your PICO and analyses in RevMan. Then check that your RevMan criteria match your Covidence data extraction template to avoid "file contains inconsistent data" messages caused by mismatched intervention or outcome names.
Current limits
Study centric data keeps expanding, but it isn't a fit for everything yet, and it's worth knowing where the edges are (as of July 2026):
Risk of Bias 2 assessments can't currently be exported from Covidence; RoB2 can still be enabled and completed directly in RevMan. More guidance on this can be found here.
Subgrouping by variants of an outcome (for example, different pain scales measuring the same thing), or creating subgroups within a study (such as adults versus children as separate subgroup results), aren't supported by study centric analyses. These cases call for manual-input analysis instead, RevMan supports both approaches side by side in the same review.
If you added custom effect measures into Covidence and exported the study custom results CSV file, you can add the data by creating a manual-input analysis, as the study custom results CSV can’t be uploaded via the import data feature. Manual analyses can now be converted to automatic (study-centric data) analyses in RevMan. Read how to create and set up manual-input analyses.
Core Benefit
None of this takes away from the core benefit: for the great majority of intervention review analyses, study centric data, extracted once in Covidence and imported straight into RevMan, means less duplicate data entry, fewer transcription errors, and a review that's easier to update down the line.
Related articles
How to export your data to RevMan
FAQ: What does the process of importing data into RevMan look like?
Study centric data management (Cochrane RevMan Knowledge Base)
Making our evidence more usable – interoperability in practice (Cochrane)