How Do You Handle Multiple Papers From the Same Study or Dataset?

What If Several Papers Actually Come From the Same Research?

You find three papers that all appear eligible for your review. Their titles are different. They were published in different journals. One examines the primary outcome, another reports a follow-up, and a third analyzes a secondary outcome.

Then you notice something familiar: several authors overlap, the sample sizes are similar, participants were recruited from the same institutions, and the study dates seem to match.

Are these three studies, or three papers reporting one study?

This distinction matters because systematic reviews generally treat the study, not the individual publication, as the primary unit of interest. A single study may generate a protocol, conference abstract, primary results paper, follow-up article, secondary analysis, qualitative companion paper, or other report. If those publications are mistakenly counted as independent studies, the same participants or underlying evidence can influence the review more than once.

The correct approach is therefore not to discard all but one paper. It is to determine which reports belong to the same underlying study, link them together, and use the relevant information across those reports without double-counting the study.

Treat the Study as the Unit, and the Papers as Reports About It

One Study Can Produce Many Publications

Start by Distinguishing a Study From a Report

In everyday academic conversation, researchers often use “paper” and “study” almost interchangeably. During evidence synthesis, that shortcut can become problematic.

A study is the underlying research investigation. A report is one source through which information about that study becomes available.

Cochrane explicitly identifies studies rather than reports as the primary units of interest in systematic reviews. Searches, however, usually retrieve reports. This creates an additional task during study selection: identifying which retrieved reports belong to the same underlying study.

That distinction also explains why deciding which papers belong in a review eventually becomes more complicated than simply counting eligible articles.

Why Would One Study Produce Several Papers?

There are many legitimate reasons.

A large research project may answer several questions that cannot reasonably be reported in one article. Different publications may address:

• the study protocol or design;
• primary outcomes;
• secondary outcomes;
• different participant subgroups;
• different measurement instruments;
• short-term and long-term follow-up;
• qualitative findings embedded within a larger project;
• process or implementation evaluation;
• economic analysis;
• adverse events;
• secondary analyses; or
• additional research questions using the same dataset.

Multiple publications are therefore not inherently suspicious. A rich longitudinal study may legitimately support a substantial publication program.

The methodological problem arises when the relationship among those papers is not recognized and the same underlying evidence is treated as though it came from independent studies.

Multiple Reports Are Not the Same as Duplicate Database Records

This distinction is particularly important during screening.

If you search several databases, the exact same journal article may appear multiple times. Those are duplicate records of one report and can usually be removed during deduplication.

Multiple reports from the same study are different. They may have different titles, years, journals, authors, outcomes, and even sample sizes.

The neighboring problem of duplicate publications and the bias they can create therefore overlaps with, but is not identical to, the broader problem of multiple legitimate reports from one study.

How Can You Tell Whether Two Papers Come From the Same Study?

Sometimes the relationship is obvious. Both papers state the trial registration number or explicitly identify themselves as analyses from the same project.

Other cases require detective work.

Cochrane recommends comparing characteristics such as:

• trial or study registration numbers;
• author names;
• institutions and study locations;
• intervention details;
• sample sizes and baseline characteristics;
• recruitment dates;
• study duration; and
• other distinctive study characteristics.

No single clue is always decisive.

Author lists can change across publications. Sample size can decrease at follow-up. Recruitment may occur at several sites. A secondary analysis may use only a subset of the original participants. Conversely, two studies conducted by the same authors at the same institution may genuinely be independent.

Look for a pattern of matching characteristics rather than relying on one superficial similarity.

Registration Numbers and Study Identifiers Are Especially Useful

When available, trial registration numbers and other unique study identifiers can make report linkage much easier.

For example, two papers carrying the same ClinicalTrials.gov identifier strongly indicate that they relate to the same registered study, even if their titles, authorship order, or outcomes differ.

Registries can also help identify reports you have not yet found. A registry entry may link to publications, identify the study's primary and secondary outcomes, or reveal that what appears to be a standalone paper is actually a secondary analysis from a larger trial.

Identifiers are not universally available, particularly for older research and study designs that are not routinely registered. They should therefore be used as strong evidence when present rather than as a requirement for establishing that reports are related.

Author Overlap Is Helpful but Not Conclusive

Shared authorship is another useful clue.

If two papers involve similar samples, the same institution, matching recruitment dates, and several common authors, they may well originate from the same study.

But author lists can vary considerably. A statistician may appear only on one analysis. A doctoral student may lead a secondary paper. A multicenter study may generate publications with different subsets of investigators.

The reverse problem also occurs. A research group may conduct several similar studies over successive years, producing papers with almost identical author lists and settings.

Shared authors therefore raise the possibility of overlap. They do not prove it.

Sample Size Differences Do Not Necessarily Mean Different Studies

This can be especially confusing.

One paper reports 500 participants. Another apparently related article reports 436. A third analyzes 212 participants.

Are these separate studies?

Possibly, but several other explanations exist. The first report may describe everyone enrolled, the second only participants with complete follow-up data, and the third a subgroup eligible for a particular analysis.

Cochrane specifically notes that participant numbers may differ across publications from the same study. Recruitment dates and baseline characteristics can help determine whether changing sample sizes reflect different stages or subsets of one study rather than independent research.

Do Not Choose One Paper and Throw the Others Away

Once you determine that several papers describe one study, the tempting solution is to keep the “main paper” and delete the rest.

That can lose valuable information.

Cochrane requires multiple reports from the same study to be collated and specifically cautions against discarding secondary reports because they may contain useful information about study design, conduct, outcomes, or other characteristics.

One article may provide the clearest description of participant recruitment. Another may contain the outcome needed for your review. A follow-up paper may provide the only long-term results. A protocol may explain allocation procedures more clearly than the results article.

The study should therefore have one identity in your review, but that identity may draw information from several sources.

You May Still Need to Identify a Primary Report

Although information can come from several reports, it is often useful to identify one publication as the primary or main report for the study.

Cochrane requires review authors to choose and justify which report is used as the principal source for study results when multiple reports exist. The most appropriate report is not necessarily the earliest publication or the paper in the highest-ranked journal.

The primary report might be the one that:

• provides the most complete description of the main study;
• reports the outcomes most relevant to the review;
• contains the most complete participant information;
• corresponds to the prespecified primary analysis; or
• provides the clearest and most complete results for the relevant time point.

Secondary reports should remain linked to that study because they may supplement or clarify the primary report.

What If the Reports Disagree?

Multiple reports from the same study do not always tell the same story.

Sample sizes may differ. Outcome values may be reported differently. A later paper may use a different analysis. One report may label an outcome as primary while another emphasizes something else.

These discrepancies should not be resolved by automatically choosing whichever result is most favorable or easiest to extract.

First determine whether the apparent conflict has a legitimate explanation:

• different follow-up periods;
• different analytic populations;
• updated or corrected data;
• adjusted versus unadjusted analyses;
• subgroup versus full-sample results;
• different outcome definitions; or
• different stages of recruitment.

If the discrepancy remains unresolved, check protocols, registrations, corrections, supplementary materials, or other reports. Contacting investigators may also be appropriate.

PRISMA 2020 asks systematic reviewers to report decision rules used when selecting data from multiple reports of the same study and any steps taken to resolve inconsistencies across reports.

One Dataset Can Produce Several Genuine Research Questions

The issue becomes more subtle when researchers publish multiple analyses from the same dataset.

Suppose a national student survey contains responses from 20,000 participants. One paper examines academic stress, another investigates digital literacy, and a third analyzes generative AI use. All three use the same underlying dataset but address different questions.

These are distinct papers and may contain genuinely different analyses. Whether they should be treated as one “study” for every purpose depends on the structure of your review and the specific evidence being synthesized.

The critical concern is statistical and evidential independence. If two papers contribute different outcomes from the same participants, that may be entirely appropriate. If they contribute estimates to the same synthesis as though they came from independent samples, the shared participants can create dependency and effectively give that dataset more influence than intended.

Overlapping Samples Are Harder Than Identical Samples

Sometimes two papers do not use exactly the same participants, but the samples overlap.

For example, one article may analyze the first three waves of a longitudinal cohort while another analyzes waves two through five. Or two papers may draw participants from the same large registry during overlapping recruitment periods.

This creates a dependency problem rather than a simple duplicate-publication problem.

You may need to determine:

• how much participant overlap exists;
• whether the same outcomes and time points are being analyzed;
• whether one report contains a subset of another;
• whether estimates can legitimately enter the same synthesis; and
• whether statistical methods are needed to account for dependence.

The appropriate solution depends on the synthesis method. The central principle remains that the same observations should not be treated as independent simply because they appear in separate publications.

Multiple Time Points Do Not Automatically Create Multiple Studies

A longitudinal study may produce separate publications for six-month, one-year, and five-year outcomes.

These are still reports from the same underlying study, although each may provide evidence relevant to different time frames in your review.

Cochrane guidance on outcome multiplicity recommends prespecifying how multiple eligible time points or measures will be handled. Selecting among them after seeing which result is most favorable can introduce bias.

If your review distinguishes short-, medium-, and long-term outcomes, different reports from the same study may legitimately contribute to different time frames. What you should not do is count the same study as three independent studies merely because three publications exist.

Multiple Reports Can Improve Critical Appraisal

Linking reports is not only about preventing double-counting.

Additional publications can reveal information that changes your understanding of methodological quality or risk of bias.

A brief primary article may provide little detail about allocation procedures, while the protocol describes them clearly. A follow-up paper may reveal attrition that was not obvious in an earlier report. A registry entry may show that outcomes were prespecified differently from how they were presented in the published article.

For this reason, critical appraisal should consider relevant information across reports of the same study rather than treating each article as an isolated methodological object.

Study-Level Organization Makes the Review Easier to Manage

Once you recognize multiple reports, create a study-level identifier.

For example:

Study ID: Santos 2024 AI Feedback Trial

Under that study record, you might link:

• Santos et al. 2023 protocol;
• Santos et al. 2024 primary outcomes;
• Reyes et al. 2024 qualitative process evaluation; and
• Santos et al. 2025 twelve-month follow-up.

The exact naming convention does not matter as much as maintaining the relationship. Your screening, extraction, appraisal, and synthesis records should make clear that these publications originate from one research project.

For large reviews, systematic review software may support study-level grouping. A spreadsheet can also work if it contains separate identifiers for reports and underlying studies.

PRISMA Distinguishes Studies From Reports for a Reason

PRISMA 2020 explicitly distinguishes the number of studies included from the number of reports describing those studies.

The flow diagram can therefore show, for example, that 42 studies were included but those studies were represented by 57 reports.

That is not a contradiction. It communicates that some studies generated more than one relevant report.

This distinction becomes particularly important when preparing a PRISMA flow diagram for study selection. Counting every report as a separate study would misrepresent the evidence base.

Do Not Confuse Multiple Reports With Multiple Independent Studies in One Paper

The reverse situation can also occur: one paper may report more than one independent study.

An article might contain “Study 1” and “Study 2,” each with different participants and procedures. A paper might report two separate experiments. A multi-cohort article may include independent samples that satisfy your eligibility criteria separately.

In such cases, the publication count is smaller than the study count.

This is another reason paper-level counting is unreliable. The relationship between reports and studies can be one-to-one, many-to-one, or occasionally one-to-many.

The correct unit depends on what was actually done, not on how many PDF files you downloaded.

Four Papers, but Only One Underlying Study

Now suppose Paper C uses only 25 participants from the original sample but reports qualitative experiences not available anywhere else. It should not be discarded merely because the participants already appear in the larger trial. Its qualitative findings may provide genuinely different evidence. The important point is to recognize the relationship and handle the resulting dependence and study structure appropriately.

Common Mistakes When the Same Study Produces Several Papers

Build Your Review Around Studies, Then Attach the Papers to Them

When two or more papers look suspiciously similar, do not immediately decide that one is a duplicate. First determine the relationship among them.

A practical data-management structure is to assign both a study ID and a report ID. The study ID groups all material from the same underlying research, while each report ID identifies the individual publication or source.

For example:

Study ID: AI-FEEDBACK-01

Reports: AI-FEEDBACK-01-A, AI-FEEDBACK-01-B, AI-FEEDBACK-01-C

This simple distinction can prevent a surprisingly large amount of confusion once screening, extraction, critical appraisal, and synthesis begin.

It also makes later documentation easier because inclusion and exclusion decisions ultimately need to be tracked at the appropriate study level, even though searches initially identify records and reports.

Before Treating Two Papers as Independent Studies

Questions About Multiple Papers From the Same Study or Dataset

Count the Research Once, but Use All the Relevant Reports

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.