How to Find the Original Study Behind a Claim in a Review Article

A review article compresses dozens of primary studies into a few sentences. When you need to cite, replicate, or challenge one of those sentences, the review is a signpost, not the destination. The task is to trace the reference back to the original study, confirm that the study is what the review says it is, and then compare the review's wording with what the primary report actually found.

This guide gives a repeatable sequence for that work, with diagnostic checkpoints, a worked example, a reusable checklist, and clear rules for when to stop or escalate. It is written for researchers preparing a manuscript, responding to peer review, or assembling an evidence summary where every claim needs an auditable trail.

A review's sentence is a secondary claim. It has already passed through at least one author's reading, selection, and paraphrase. The reference list points to a work, but a reference entry alone does not tell you whether the cited item is an empirical report, a commentary, a letter, or another review. Before you can compare wording, you have to establish what kind of document you are holding and whether it is the study the review describes.

Two habits prevent most errors. First, treat the reference as a pointer to be resolved, not as a verified fact. Second, keep a written trail of each resolution step so a collaborator or reviewer can retrace your path.

Start by isolating the exact sentence or clause that carries the claim. Copy it verbatim into your working notes, along with the reference number or author-year marker attached to it. Note whether the sentence cites one source or several; multi-source sentences often blend findings and need to be split before tracing.

Checkpoint: You should have a single quoted claim, its anchor reference, and the page or section where it appears. If the sentence cites multiple works, split it into one claim per source before continuing.

Take the DOI from the reference and resolve it through doi.org to the current destination registered for that identifier. Resolution helps you locate the work, but it does not by itself confirm identity: the bibliographic fields and the source text still need to be checked against the review's description.

If the reference has no DOI, search by author and title. Google Scholar documents searching by author and title, date sorting and limits, Cited by and Related articles, citation export, and ways to locate alternative versions of a work. Use those features to find a stable record, then confirm the title, authors, venue, and year match the reference entry.

For a faster starting point, you can run the title or author through WisPaper search to locate the paper record, then resolve the DOI for the authoritative destination.

Checkpoint: You have a resolved record with title, authors, venue, year, and identifier. If the resolved record's title or author list does not match the reference, stop and treat this as a possible citation error before reading further.

This is the step most often skipped. A cited item may be an editorial, a commentary, a perspective piece, a letter, or a review, rather than a study that collected or analyzed data. Read the abstract and the opening section with one question in mind: does this document report its own methods and results, or does it discuss someone else's?

Signals that you are holding a commentary or secondary piece include a short format, an absence of a methods section, and language that refers to other studies rather than presenting new data. Signals of an empirical report include a described sample or dataset, an analysis procedure, and a results section with the study's own findings.

Checkpoint: Record a one-line classification: empirical report, commentary, review, or other. If the review cited a commentary as if it were primary evidence, note that discrepancy; it changes how you interpret the claim.

Identity verification means confirming that the study the review describes is the study you have in hand. Compare four fields side by side: the population or sample, the intervention or exposure, the outcome, and the setting or timeframe. A review sentence may describe a study accurately in general terms while shifting one of these details.

Use publisher-deposited metadata to cross-check. Crossref exposes titles, authors, venues, dates, identifiers, licenses, funding, ORCID and ROR identifiers, abstracts when deposited, and post-publication updates from trusted sources. That metadata helps you confirm authorship and venue, and deposited abstracts can help you check whether the study's stated scope matches the review's description.

Checkpoint: All four identity fields match, or you have documented a specific mismatch. A mismatch in population or outcome is a stop signal for the claim as written.

Now read the primary study's results and discussion directly. Put the review sentence and the primary finding side by side and compare them on three dimensions: direction (did the effect go the way the review says), strength (does the review overstate or soften the result), and scope (does the review generalize beyond the studied population or condition).

Pay attention to hedged language. A primary report may say an association was observed in one subgroup; a review may render that as a general effect. Your job is to record the difference precisely, not to decide which author is right.

Checkpoint: Write a comparison note with three lines: direction match, strength match, scope match. Any line marked as a mismatch becomes a candidate correction for your manuscript or a question for the review's authors.

Suppose your manuscript draft contains this sentence, attributed to a recent review: "Intervention X improved outcomes across adult populations." You are preparing a revision and a reviewer has asked you to support it.

1. Extract. You copy the sentence and its anchor reference into your notes.
2. Resolve. You resolve the DOI to the review and confirm the title and authors match. You then locate the specific passage and its own reference, which points to a primary study.
3. Classify. You resolve that primary reference and read its abstract. It reports its own sample and analysis, so you classify it as an empirical report.
4. Verify identity. You compare population, intervention, outcome, and setting. The primary study enrolled adults in one clinical setting, which is consistent with the review's phrasing but narrower than "across adult populations."
5. Compare wording. The primary result reports an improvement in the studied setting. The review's "across adult populations" extends the scope beyond what the primary study examined.

Your revision note records: direction matches, strength broadly matches, scope does not. You then either narrow your sentence to match the primary study or add a second citation that genuinely covers the broader population. The trail — claim, resolved identifiers, classification, identity fields, and comparison note — stays in your project files so a co-author can audit it.

This table works as a per-claim log. For a manuscript with many claims, one row per claim keeps the audit trail readable.

A trace is only useful if someone else can follow it. Store the resolved identifier, the classification, the identity fields, and the comparison note together, ideally next to the claim in your project files. When several co-authors are tracing claims in parallel, a shared workspace prevents duplicated effort and inconsistent standards.

You can organize the papers you resolve in My Library so the primary studies and the review sit together, and group the work under Research Projects when the tracing belongs to a larger manuscript or grant task. If you need to ask questions across the papers you have collected, Library QA lets you query papers uploaded to My Library or added from search results.

For longer tracing tasks that span many claims, Scholar Agent fits literature review and related research workflows. When your reference list itself needs screening, TrueCite checks BibTeX against real academic databases to flag hallucinated references — a useful first pass before you trace individual claims.

Use these rules to decide when a trace is complete and when it needs another person.

• Stop and correct when the resolved record's title or authors do not match the reference. Treat it as a citation error until proven otherwise.
• Stop and reclassify when the cited item turns out to be a commentary or review rather than an empirical report. The claim's support base has changed.
• Stop and narrow when the primary study's population, outcome, or setting differs from the review's description. Do not carry the broader wording forward.
• Escalate to a co-author or the review's authors when the primary result is ambiguous, when the review cites a source you cannot resolve, or when direction and scope both mismatch.
• Escalate to a librarian or metadata specialist when identifier resolution leads to a record whose metadata conflicts with the reference in ways you cannot reconcile.