01 · The Question
Does Every Included Study Need to Become a Citation-Chaining Starting Point?
You have completed your database searches, screened the results, and identified a substantial set of included studies. Citation chaining seems like the obvious next step. The practical problem is scale: if you have 60 included papers, should you examine the references and citing papers of all 60?
There is a legitimate reason to consider doing so. Citation searching can identify relevant studies that keyword-based searches miss because it follows relationships among publications rather than relying on the terminology used in titles, abstracts, and indexing. But turning every included paper into a citation-chaining seed can also produce substantial duplication, especially when the papers belong to the same closely connected literature.
The question, then, is not simply whether citation chaining is useful. It is whether every included paper needs to be chained, and whether a selective approach can sometimes achieve the search objective without unnecessary work.
03 · What You Need to Know
Think About Coverage, Not Simply the Number of Papers You Chain
Why citation-chain included studies in the first place?
Database searching and citation searching retrieve literature in different ways. A conventional database strategy generally depends on words, phrases, controlled vocabulary, fields, indexing, and database coverage. Citation searching instead uses the citation relationships surrounding a known paper.
This distinction matters because an eligible study may use terminology that your search did not anticipate. It may be poorly indexed, published in an unexpected disciplinary venue, or described using an older conceptual vocabulary. Citation searching can therefore serve as a complementary discovery method rather than merely repeating the database search.
For this reason, citation searching is commonly used after researchers have identified relevant or eligible studies. Cochrane guidance, for example, describes citation searching on key articles as an additional method alongside database searching and notes that citation searches conducted from included studies may reveal additional relevant studies.
Backward and forward chaining answer different discovery questions
When you move backward or forward through citations , you are exploring different portions of the literature.
Approach
What you inspect
What it may help uncover
Backward citation searching
References cited by the seed paper
Earlier studies, methodological precedents, conceptual origins, and older terminology
Forward citation searching
Later publications that cite the seed paper
Subsequent studies, replications, extensions, applications, and later work using related concepts
That difference also means that chaining several papers is not necessarily redundant. Two included studies published in different periods, disciplines, countries, or research traditions may connect to substantially different citation neighborhoods.
Why chaining every paper can increase coverage
The strongest argument for chaining every included study is straightforward: you do not know in advance which paper will reveal the missing study.
A paper with relatively little apparent importance may cite an obscure earlier study that the major papers ignored. Another may be the only included paper cited by a later eligible study. Citation networks are uneven, so selecting seeds solely by citation count, journal prestige, or familiarity can systematically favor already visible parts of the literature.
Comprehensive chaining also makes the procedure easier to describe. Instead of making discretionary decisions about which studies deserved chaining, you can state that all eligible studies meeting a specified criterion were subjected to backward citation searching, forward citation searching, or both.
Why chaining every paper can also produce diminishing returns
Included studies are often highly interconnected. Ten papers from the same research tradition may cite many of the same foundational studies and be cited by many of the same later publications. Chaining all ten can therefore generate a large volume of duplicate records without proportionately increasing discovery.
This becomes especially noticeable in mature, densely connected literatures. If successive seed papers repeatedly lead to records you have already screened, the marginal yield of additional chaining may become small. That is a different issue from deciding how far citation chaining should continue once a chaining process has begun, but the two decisions are closely related.
There is no universal numerical threshold at which the process becomes inefficient. The additional yield of reference checking varies considerably among reviews and topics. A Cochrane methodological review found evidence that reference-list checking can identify additional eligible studies, but also concluded that the underlying evidence about its effectiveness was limited and heterogeneous. Consequently, a rule such as “chain the ten most important papers” or “stop after twenty seeds” would be difficult to defend as a general methodological standard.
A systematic review may justify a different answer from a narrative review
The purpose of the review matters. A systematic review designed to identify all studies satisfying explicit eligibility criteria places a high value on sensitivity and reproducibility. Missing an obscure eligible study may matter even if it does not alter the eventual conclusion. In that setting, comprehensive citation searching of eligible studies may be justified, particularly when database retrieval is difficult.
A scoping, mapping, qualitative, realist, or narrative synthesis may operate under different search principles depending on its methodology and objectives. Some reviews emphasize exhaustive identification, whereas others emphasize conceptual coverage, theoretical development, information power, or iterative exploration. The citation-chaining strategy should therefore follow the methodological logic of the review rather than being copied automatically from another review type.
Watch Out
Do not use “efficiency” as an after-the-fact justification for selectively chaining whichever papers looked interesting. If you use a targeted approach, define a defensible basis for choosing the seed papers and report that basis transparently.
If you chain selectively, diversity among seed papers matters
A selective strategy is stronger when the seeds represent distinct portions of the relevant literature rather than merely its most prominent papers. Depending on the review question, useful variation might include publication period, terminology, methodology, population, geographical setting, discipline, intervention, theoretical tradition, or research group.
This is why identifying the best starting papers for citation chaining is not the same as simply choosing the papers with the highest citation counts. A highly cited paper can be an excellent seed, but prominence is not equivalent to coverage.
Be careful about citation-network bias
Citation chaining follows relationships created by researchers. Those relationships are not neutral maps of everything relevant to your question. Researchers disproportionately cite work they know, work published in visible journals, work from their intellectual communities, and literature framed in compatible ways.
If all of your seeds belong to the same intellectual lineage, repeated chaining may keep returning material from that lineage. This is one reason to consider whether citation chaining is keeping you inside one academic network . Citation searching should normally complement rather than automatically replace independent searching based on concepts, databases, authors, and other appropriate discovery routes.
Document citation searching as a search method
If citation chaining contributes to study identification, treat it as part of the search methodology. Record enough information to explain what was done: which papers were used as seeds, whether backward or forward searching was performed, which citation indexes or platforms were used where relevant, when the searches were conducted, and how records were screened.
This matters especially when selection was purposive. A reader should be able to understand why some papers were chained and others were not. Transparent reporting does not require pretending that every search decision was mechanical. It requires making consequential decisions visible.
04 · A Practical Example
What Happens When You Have 48 Included Papers?
Hypothetical Example
A review with several clusters of included studies
Suppose you are conducting an evidence synthesis and database searching produces 48 eligible papers. They are not 48 independent pieces of literature. Twenty-two come from a closely connected research community, 12 use a different theoretical framework, eight come from an older literature using different terminology, and six represent newer work in another discipline.
Decision First determine whether your review protocol or methodological framework calls for comprehensive citation searching of eligible studies. If it does, the clustering of the papers is not by itself a reason to omit some seeds.
If comprehensive chaining is required Perform the specified backward and/or forward citation searches across all eligible seed studies, deduplicate the retrieved records, and screen them using the same relevant eligibility criteria.
If a targeted strategy is methodologically acceptable Select seeds across the different clusters rather than taking only the most cited or most recent papers. Include papers capable of exposing the older terminology and the separate disciplinary literature.
Evaluate the yield Track whether successive seeds continue to produce previously unseen potentially eligible records. Repeated retrieval of already screened material provides useful information about diminishing marginal yield, although it does not by itself prove that no unseen studies exist.
Report the procedure Explain whether all included studies or a defined subset were chained, how any subset was selected, which directions were searched, and what tools or citation indexes were used.
The important methodological distinction is between a planned selective strategy and an arbitrary one. “We chained papers representing each major literature identified during screening” can be examined and critiqued. “We chained the papers that seemed important” leaves much more of the search process hidden.
06 · What This Means for You
Choose the Scope of Citation Chaining Before Convenience Chooses It for You
Start by asking what failure would matter in your review. If the review claims comprehensive identification of all eligible studies, the threshold for omitting eligible seed papers should be high. If the synthesis is explicitly purposive or iterative, selecting strategically diverse seeds may be defensible and considerably more efficient.
A simple decision framework
If your protocol, review standard, or methodology specifies citation searching of all eligible studies
Follow that procedure and document it rather than reducing the seed set for convenience.
If exhaustive identification of eligible studies is central to the review
Consider comprehensive citation searching, particularly when terminology, indexing, or database coverage makes studies difficult to retrieve.
If your methodology permits purposive or iterative searching
Choose seeds for coverage and diversity, not merely citation count or familiarity.
If successive seeds produce mostly duplicate or already screened records
Assess the marginal yield and whether additional chaining remains justified under your review method.
If one seed suddenly exposes a distinct conceptual literature
Whatever strategy you choose, preserve an audit trail. Citation searching is easiest to defend when another researcher can see the logic behind the seed set and understand how the resulting records entered the screening process.
07 · A Quick Checklist
Before Deciding Which Included Papers to Citation-Chain
Before starting citation chaining, check:
Does your protocol, reporting standard, or review methodology specify how citation searching should be conducted?
Is comprehensive identification of all eligible studies an explicit objective of the review?
Are important studies likely to be difficult to retrieve because of terminology, indexing, age, or disciplinary boundaries?
If you are selecting seeds, do they represent meaningfully different parts of the literature rather than only the most visible papers?
Have you decided whether backward searching, forward searching, or both are appropriate?
Are you recording which seed papers and citation-searching tools or databases were used?
Are you tracking unique records and additional eligible studies rather than judging productivity from the raw number of citations retrieved?
Could your selected seeds be keeping the search inside one research group, discipline, theoretical tradition, or citation network?
Can you explain and report why every included paper was chained, or why only a subset was used?
11 · Cite this Guide
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.
Recommended (Field Guide)
APA
MLA
Chicago
Copy Citation
