01 · The Question
What If Several Papers Are Actually Reporting the Same Evidence?
You conduct a literature review and find six papers supporting an intervention. They have different titles, appear in different journals, and were published in different years. Six supportive studies sounds considerably more convincing than one.
But what if three of those papers came from the same underlying trial?
A single study can generate multiple journal articles, conference abstracts, subgroup reports, follow-up analyses, or publications describing different outcomes. That is not necessarily improper. The methodological problem begins when overlapping reports are mistakenly treated as independent studies.
The same participants can then influence the apparent evidence more than once.
03 · What You Need to Know
How One Study Can Become Several Apparently Independent Pieces of Evidence
A study and a publication are not the same unit
This distinction is fundamental to evidence synthesis. A study is the underlying investigation involving a particular design, sample, intervention or exposure, and set of observations. A publication is one report arising from that research.
One study may produce several publications. Conversely, one publication may sometimes report more than one study.
Study
The underlying investigation and participants or observations that generated the data.
Report or publication
A particular document communicating some or all of the study's methods, outcomes, analyses, or follow-up findings.
Cochrane therefore recommends linking multiple reports of the same study rather than treating each report as a separate study. It specifically warns that duplicate publication can introduce substantial bias when studies are inadvertently included more than once in a meta-analysis.
Duplicate publication does not always mean an identical article
The easiest duplicate to identify would be two nearly identical papers containing the same participants, outcomes, and results. Real cases can be much harder.
Cochrane notes that duplicate publication ranges from identical manuscripts to reports containing different outcomes or different follow-up periods from the same study. Participant numbers can also differ among publications.
For example, one trial might generate an initial efficacy paper, a safety paper, a subgroup analysis, a long-term follow-up, and a conference abstract. These documents may all provide legitimate additional information. They still originate partly or entirely from the same underlying participants.
| Publication pattern |
What may differ |
Risk for the reviewer |
| Near-identical duplicate |
Little beyond journal or formatting |
The same study is obviously counted twice if duplication is missed |
| Different outcomes |
Each paper emphasizes different endpoints |
The reports may look like separate studies despite sharing participants |
| Different follow-up periods |
Short-term versus long-term results |
Related observations may be mistaken for independent samples |
| Subgroup publication |
Only part of the original sample appears |
Participant overlap may be difficult to recognize |
| Expanded or reduced sample |
Participant numbers differ across reports |
Partial overlap may be mistaken for complete independence |
Why duplicate evidence is especially dangerous in meta-analysis
A conventional meta-analysis assumes that each included study contributes an appropriate amount of independent information. If the same participants appear in two supposedly separate studies, that assumption can be violated.
The duplicated evidence effectively receives additional weight. If the overlapping reports point in the same direction, the pooled estimate may shift toward that result and its apparent precision may also be misleading.
A study of duplicate publications in Korean meta-analyses illustrates the problem. Among 86 meta-analyses examined, six included duplicate publications. Reanalyzing the data with duplicated reports increased the mean effect size and the fail-safe number. The authors concluded that even a relatively small number of duplicated articles could affect meta-analytic results.
Those estimates describe that particular sample and should not be generalized as the prevalence of duplication in all disciplines. They demonstrate the mechanism: duplicated observations can change a quantitative synthesis.
Duplicate publication can also change the apparent number of studies
The problem is not confined to pooled effect estimates. Even a narrative review can be distorted if multiple papers from one study are described as independent confirmations.
Imagine one large trial generating four supportive papers while four separate trials produce less favorable results. Counting publications gives the appearance of four supportive versus four less supportive pieces of evidence. Counting underlying studies reveals one supportive study versus four others.
The papers themselves may contain useful distinct analyses. What changes is the interpretation of independence and replication.
Multiple publication can itself be associated with study findings
There is an additional complication. Studies producing statistically significant or larger effects may be more likely to generate multiple publications. Cochrane describes this as multiple or duplicate publication bias.
This means positive evidence can receive two visibility advantages. A favorable study may generate several reports, making it easier to locate, and those reports may then be mistaken for multiple independent studies.
Classic discussions of bias in meta-analysis similarly note that positive trials can be reported more than once, increasing their probability of being found and potentially included.
Several papers from one study are not automatically unethical duplication
Researchers can legitimately publish multiple reports from a substantial study. Different papers may address distinct outcomes, follow-up periods, qualitative components, methodological questions, or secondary analyses that cannot reasonably fit into one article.
The critical requirements are transparency and appropriate cross-referencing. Readers should be able to determine that the publications arise from the same underlying study and understand how their samples and analyses overlap.
From the reviewer's perspective, the central question is therefore not simply, "Are these papers duplicates?" It is, "Which observations in these papers come from the same underlying study, and how should they contribute to my synthesis?"
Detecting overlapping reports may require detective work
Cochrane explicitly notes that identifying multiple reports from one study can require detective work. Useful clues include trial registration numbers, author names, sample sizes, recruitment dates, study locations, intervention descriptions, baseline characteristics, and other study identifiers.
No single clue is foolproof. Author lists can change between papers. Sample sizes may differ because of attrition or subgroup analyses. Study titles may be completely different.
The task is therefore to compare combinations of characteristics and determine whether apparently separate reports plausibly share the same participants.
Watch Out
Different titles, journals, publication years, or author lists do not guarantee independent studies. When recruitment periods, locations, interventions, participant characteristics, and study identifiers overlap substantially, investigate before counting the reports separately.
Duplicate publication differs from selective outcome reporting
These mechanisms can look similar because one study may distribute different outcomes across several publications.
Selective outcome reporting concerns whether particular measured outcomes are reported depending on their results. Duplicate publication concerns multiple reports arising from the same or overlapping underlying data.
One study can experience both. Favorable outcomes might be emphasized across several publications while unfavorable outcomes remain incompletely reported.
Duplicate publications should usually be linked, not simply discarded
When several reports describe one study, the solution is not necessarily to choose one paper and throw the others away. Different reports may contain complementary information needed for risk-of-bias assessment, participant characteristics, outcomes, or follow-up.
Cochrane recommends collating multiple reports so that the study, rather than the individual report, becomes the unit of interest.
You may therefore use information from several publications while ensuring that overlapping participants do not contribute multiple independent observations to the same analysis.
06 · What This Means for You
Count Studies, Not Papers
The practical principle is simple but consequential: determine the underlying study structure before deciding how many independent pieces of evidence you have.
A simple decision framework
If two reports have the same trial or registration identifier
Treat them as reports from the same study unless there is clear evidence otherwise.
If authors, recruitment dates, locations, interventions, and participant characteristics overlap
Investigate whether the publications share participants before including them as independent studies.
If multiple reports contain complementary information
Link the reports and use the relevant information from each while keeping the underlying study as the unit of evidence.
If participant overlap cannot be determined
Document the uncertainty and consider how alternative assumptions about overlap could affect the synthesis.
For systematic reviews, create study-level records that link every report believed to arise from the same investigation. Registration identifiers are particularly useful where they exist, but older studies and many non-clinical fields may lack them.
For narrative reviews, be equally careful with wording. "Four papers reported favorable findings" does not necessarily mean "four independent studies found the effect." Sometimes that distinction changes the entire impression of replication.
07 · A Quick Checklist
How to Detect and Handle Overlapping Publications
Before treating two reports as independent studies, check:
Whether the reports share a trial registration number, project identifier, grant number, or other study identifier.
Whether author names overlap, while remembering that different author lists do not rule out duplication.
Whether recruitment dates, study locations, interventions, eligibility criteria, and baseline characteristics are similar.
Whether differences in sample size can be explained by follow-up, attrition, subgroup selection, or expanded recruitment.
Whether one report cites another publication from the same study or research program.
Whether overlapping participants would be counted more than once in the same quantitative synthesis.
Whether multiple reports have been linked under a single study record while preserving useful information from each.
