What Is an Overlapping Study Population, and Why Does It Matter?

What Does It Mean When Study Populations Overlap?

Imagine that your systematic review contains six papers. At first glance, you have six separate pieces of evidence. Then you discover that two papers drew patients from the same national registry during overlapping years, while another paper analyzed a subgroup from one of those populations.

You may still have six publications and potentially several distinct analyses, but you no longer have six completely independent groups of participants.

This is the problem of overlapping study populations. It matters because many forms of evidence synthesis depend, explicitly or implicitly, on understanding which observations are independent. If the same people repeatedly enter the evidence base without that dependence being recognized, a particular population can exert more influence than its actual size warrants.

Overlap Means Some People Contribute to More Than One Study or Analysis

How Overlapping Populations Arise and Why They Complicate Evidence Synthesis

Overlap Exists on a Continuum

Two samples do not have to be identical to overlap. At one extreme, two papers may analyze exactly the same participants. At the other, only a relatively small subset may be shared.

The methodological consequences depend on how much overlap exists, where it occurs, what effect estimates are being combined, and how the synthesis handles dependence. Merely labeling overlap as present or absent can therefore conceal important differences.

Multiple Papers From One Study Are an Obvious Source

A single investigation can generate a primary results paper, secondary outcome paper, subgroup analysis, harms report, long-term follow-up, and other publications. Cochrane guidance explicitly treats the study rather than the individual report as the primary unit of interest in systematic reviews and requires multiple reports of the same study to be collated.

If every publication is mistakenly entered as a separate study, the same participant pool may be represented repeatedly. This is why establishing whether multiple papers represent one study or several pieces of evidence is more than a bibliographic housekeeping exercise.

Overlap Can Also Occur Across Apparently Separate Studies

The problem is broader than duplicate publication. Researchers may conduct separate analyses using the same cohort, registry, electronic health record system, administrative database, biobank, household survey, or other data resource.

Consider two observational papers using the same national health database. One examines records from 2015 through 2020 and another from 2018 through 2023. If their eligibility criteria are similar, some individuals may appear in both analytic cohorts even though the publications address different research questions and are presented as separate studies.

Methodological literature has specifically identified reuse of the same database and overlapping analytic timeframes as routes through which participants can be double-counted in evidence synthesis.

A Follow-Up Sample May Overlap Without Being Identical

Longitudinal research naturally creates related populations. Suppose 800 people enter a cohort at baseline and 600 remain at a three-year follow-up. A baseline paper and the follow-up paper do not contain identical analytic samples, but the 600 retained participants are not new independent observations.

The same issue occurs when a follow-up paper uses only part of the original sample. The changing sample size should not obscure participant continuity.

Why Double-Counting Is Different From Simply Having More Evidence

If two genuinely independent studies each enroll 500 people, together they provide observations from 1,000 distinct participants. If two papers each analyze the same 500 people, there are still only 500 distinct participants behind those analyses.

This does not make the second analysis useless. It may address a different outcome, follow-up period, exposure, subgroup, or research question. The problem arises when dependence is ignored and repeated observations are treated as though they came from entirely new people.

Overlap Can Violate an Independence Assumption

Conventional meta-analytic models commonly treat study effect estimates as statistically independent. When two estimates are calculated partly or wholly from the same participants, they may be correlated.

Cochrane guidance similarly notes that multiple effects calculated from the same participants are statistically dependent and that multiplicity may arise from multiple outcomes, analyses, or reports of the same study. Appropriate handling may involve selecting an eligible effect according to prespecified rules or using analytical methods that account for dependence.

The important point is not that every instance of overlap invalidates a meta-analysis. Rather, unrecognized overlap can make an independence assumption inappropriate.

Overlap Can Give Some Populations Too Much Influence

Suppose one regional registry generates eight eligible publications while another region is represented by one study. If the eight registry analyses contain many of the same patients and all are treated as independent, the first population may effectively receive repeated representation.

This can affect weighting and precision and can also shape the substantive impression of the literature. A research area may look larger and more diverse than it really is because publication count is being mistaken for population diversity.

This is one reason to examine whether the same dataset is generating many apparently independent papers and, at a broader level, to prevent a prolific dataset from dominating your understanding of the literature.

Not Every Overlap Requires the Same Response

How overlap should be handled depends on what you are synthesizing. Two papers using the same participants may report different outcomes, in which case both may be needed to characterize the study. Two correlated estimates of the same outcome may require a choice between estimates or an analytical approach capable of accommodating dependence.

Shared controls in multi-arm studies create another form of dependence. Repeated measurements and multiple follow-up periods can do the same. The methodological response should therefore be tied to the particular estimand and synthesis rather than to a blanket rule that one of every two overlapping papers must be discarded.

Sometimes Overlap Can Be Suspected but Not Quantified

Real-world datasets create a particularly awkward problem. Papers may report the database and observation years without identifying which individual records are shared. You may therefore know that overlap is possible or highly plausible without knowing its exact magnitude.

That uncertainty should be documented rather than converted into an arbitrary estimate. Depending on its importance, sensitivity analyses or alternative synthesis decisions may help assess whether conclusions depend on potentially overlapping studies.

How Overlap Can Make Three Studies Look More Independent Than They Are

Common Misunderstandings About Overlapping Populations

Map Overlap Before Deciding How to Handle It

The first task is identification, not deletion. Map the parent studies, datasets, participant pools, recruitment periods, subgroups, and follow-up waves represented by your included reports. Then determine which effect estimates relevant to a particular synthesis could share participants.

Whatever approach you choose should be transparent and, where possible, prespecified. The objective is not to force every complicated literature into perfectly discrete boxes. It is to ensure that the apparent quantity of evidence is not confused with the number of independent participants behind it.

How to Check for Overlapping Study Populations

Questions About Overlapping Participants in Research

Publication Count Is Not Participant Count

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.