One Acupressure Study, Several Reports: Avoid Double-Counting

Separate a document count from a study count

A protocol describes a plan. A primary results paper reports selected results. A follow-up paper may report a later visit, and a registry record may contain additional information about the same trial. Four documents can therefore describe one study.

Cochrane’s Chapter 4 distinguishes studies from reports and explains that each study may have several articles, abstracts or other records. Its guidance requires linking reports, then deciding what information they contribute. [1] This article adapts that distinction to an ordinary reading log; it is not a completed systematic review.

“I saved four records” and “I found four independent studies” are different claims. Do not upgrade the former to the latter because files have different titles or publication years.

Match several identity fields, not one convenient clue

Cochrane lists these kinds of criteria and notes that participant numbers can differ across publications. [1] A count mismatch deserves investigation; it does not automatically prove that studies are separate. Conversely, matching counts do not prove that two papers share participants.

Record the explicit linking evidence and any contradiction. If a link remains uncertain, keep a provisional group rather than forcing a match. Cochrane notes that unresolved details may require information from the authors or investigators; an unavailable answer stays unavailable in this reader’s log.

Reconcile a complete fictional seven-record ledger

The seven records resolve to two corroborated study groups, A and B, plus one unresolved report. Five records belong to A, one to B and one remains unassigned. The known enrolled counts for the two explicitly distinct cohorts total 60 + 60 = 120 in this invented example.

A’s 48 later observations are drawn from its original 60; adding them would count overlapping people twice. U-REPORT’s 48 cannot be added to a unique-participant total because overlap is unknown. The honest total is “120 enrolled across the two confirmed distinct example cohorts, with one unresolved report,” not 168 or 216 unique people.

The example explicitly supplies cohort distinction for B. Real records often do not; without that evidence, even different trial IDs are not a blanket assurance that people cannot overlap.

Download the complete fictional seven-record example (TXT).

Retain secondary reports without counting the result twice

A later or secondary report may clarify a time point, subgroup, outcome definition or study conduct. Cochrane says to collate reports and not discard secondary ones merely because another report is designated primary. When reports conflict, the selected result source and reason need justification. [1]

For each result you note, attach the report, measure, unit, comparison, time point and analysed group. A’s enrolled 60 and later observed 48 do not share a denominator definition. Keep both labels rather than substituting the larger number to make follow-up look complete.

Do not treat different visits from one cohort as independent studies merely because their numbers differ. Do not splice incompatible rows into one stronger-looking result. Where the relationship or correction is uncertain, record the discrepancy rather than averaging it away.

Do not turn a duplicate record into a misconduct claim

A second copy of the same PDF can be a duplicate item in your folder. A legitimate protocol and follow-up paper can be different reports of one study. NLM’s duplicate-publication indexing is a separate bibliographic policy; NLM does not routinely check the originality of all publications. [2]

Use descriptive wording: “same document,” “linked report of the same study,” “explicitly distinct cohort” or “unresolved.” Those labels identify the evidence you have. They do not accuse a research team of misconduct or determine whether the research is clinically useful.

After grouping, use the outcome-reading guide for actual results and the notice-checking guide for editorial status. The grouping ledger ends with traceable records, not a treatment recommendation.

Use the optional document record

Open the original TXT record.

STUDY / REPORT GROUPING — OPTIONAL BLANK RECORD
Local study group label (not a real registry ID):
Document title / exact ID / role:
Trial IDs and explicit linking statement:
Authors / setting / recruitment period:
Intervention and comparator labels:
Enrolled / observed / analysed count: preserve each definition
Corroborated same cohort / explicit distinct cohort / unresolved:
Report supplying a particular outcome / time / denominator:
Other reports retained and conflicting information:
Unknown overlap or missing linking material:
Count documents, studies and unique people separately.