How Do You Use Companion Papers Without Double-Counting Participants?

How Can You Keep Useful Companion Papers Without Counting the Same People Twice?

You identify a primary study report and then discover several related papers. One provides adverse-event data, another reports a secondary outcome, and a third contains long-term follow-up results. They all contain information your review could use.

Discarding the companion papers would lose potentially important evidence. Treating every paper as a separate independent study creates the opposite problem: the same participants may be represented repeatedly.

The solution is to separate two tasks that are easy to conflate. Use publications as sources of information, but organize that information around the underlying study and participant population. Cochrane explicitly recommends collating multiple reports of the same study rather than treating each report as another study.

Link the Reports First, Then Extract at the Study Level

How to Separate Useful Information From Duplicate Participant Contributions

Start With the Study, Not the Paper

Systematic reviews commonly retrieve reports rather than neatly packaged studies. One study may be represented by a journal article, conference abstract, protocol, registry entry, secondary analysis, follow-up report, and other sources. Cochrane therefore treats the study, rather than each individual report, as the principal unit of interest.

Once you have established that several publications arise from the same investigation, assign them to a common study identifier. Each publication should retain its own report identifier so that you can trace exactly where information came from.

This simple distinction prevents a common data-management problem: allowing the structure of your bibliography to dictate the structure of your evidence.

Companion Papers May Contain Information the Main Paper Does Not

A secondary report should not be discarded without examination. Cochrane notes that a single source rarely contains complete information about a study and specifically cautions that secondary reports may contain valuable information about study design, conduct, outcomes, and results.

A companion paper might provide:

• an outcome omitted or abbreviated in the main report;
• adverse events or harms;
• longer-term follow-up;
• details of intervention implementation;
• a subgroup or secondary analysis;
• additional information about methods or participant flow;
• results for another measurement instrument; or
• information that helps explain an apparent discrepancy elsewhere.

Using this information does not mean counting another study. You are enriching the record of the same underlying investigation.

Ask Whether the Same Participants Enter the Same Synthesis More Than Once

Double-counting becomes particularly consequential when two estimates based on the same participants are entered into one analysis as though they were independent.

Suppose a randomized trial of 500 participants produces two papers. Paper A reports depression using Scale X, while Paper B reports depression using Scale Y in the same participants at the same follow-up. If both effect estimates are entered into a conventional meta-analysis as if they came from two independent studies, the underlying trial may receive unintended extra influence.

By contrast, if Paper A reports depression and Paper B reports an eligible adverse-event outcome, both reports may contribute useful information to different outcome syntheses without duplicating participants within either synthesis.

Do Not Confuse Data Extraction With Meta-Analytic Inclusion

You can extract information from a report without placing every number it contains into a meta-analysis. These are separate decisions.

For example, a subgroup paper may clarify treatment adherence or provide a relevant descriptive characteristic. You can use that information in the study record even if its subgroup effect estimate is not eligible for the primary quantitative synthesis.

This distinction is useful because an overly aggressive response to duplication can lead reviewers to discard informative reports merely to keep the dataset tidy. Methodological neatness is desirable, of course, but preferably not at the expense of actual information.

Map Which Participants Contribute to Which Result

When companion reports are complex, create a small outcome-by-report map. Record the participants, outcome, measurement instrument, time point, and analytic population behind each potentially eligible result.

This makes it easier to see whether two apparently different results are actually alternative measurements of the same outcome in the same people.

Prespecify Rules for Multiple Eligible Results

A study can report several eligible results for one outcome because of different instruments, subgroups, time points, or statistical summaries. Cochrane recommends specifying in the protocol how such multiplicity will be handled, rather than choosing among results after seeing which estimate is most favorable or convenient.

Depending on the review question, a protocol might specify preferred measurement instruments, clinically meaningful time windows, preferred analysis populations, or another defensible hierarchy. More complex synthesis methods may sometimes accommodate correlated estimates rather than requiring selection of only one, but the method must actually account for the dependence.

Different Follow-Up Times Require an Outcome-Time Framework

Companion papers frequently report later follow-up. Suppose the primary report provides a three-month outcome and another paper reports the same participants at 12 months. Both results may be relevant if your protocol defines short-term and long-term outcome windows.

They should not, however, silently become two independent studies. They are repeated observations from the same underlying investigation.

Subgroup Papers Require Participant-Level Awareness

A companion analysis might use only women, participants above a particular age, one study site, or participants with a particular baseline condition. If those people are already represented in a full-sample estimate, adding the subgroup estimate to the same analysis can count them again.

This is why you need to know whether two papers use the same participants and whether the samples constitute an overlapping study population.

Do Not Assume the Primary Report Is Correct Whenever Reports Disagree

Companion papers can expose inconsistencies. Sample sizes, event counts, outcome definitions, or effect estimates may differ across reports. Cochrane recommends planning how discrepancies across sources will be resolved and, where necessary, contacting investigators.

Sometimes the discrepancy is legitimate: the papers use different follow-up periods or analysis populations. Sometimes a later publication corrects an earlier value. Occasionally the available sources do not permit reconciliation. The appropriate response is investigation and documentation, not automatic deference to whichever publication you labeled primary.

Using Four Papers From One Trial Without Turning Them Into Four Trials

Common Mistakes When Working With Companion Publications

Track Reports Separately but Synthesize Participants Deliberately

A practical workflow is to maintain one study record with multiple linked report records. For every extracted result, preserve its source report, participant population, outcome, time point, and analysis type. Cochrane specifically recommends recording both study and report identifiers in data-collection systems when studies have multiple reports.

The aim is not to minimize the number of papers you use. It is to preserve the maximum relevant information while maintaining an accurate account of how many studies and participants actually generated it.

Before Using Data From a Companion Paper

Questions About Companion Papers and Double-Counting

Use All Relevant Reports, but Count Participant Contributions Carefully

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.