Citation chasing and snowballing are supplementary search methods that find studies by following the connections between papers rather than by matching search terms. Backward citation chasing scans the reference list of a relevant study to find earlier work it cites; forward citation chasing finds later studies that cite it. Together they catch eligible studies that a term-based database search missed because of unusual wording or weak indexing.
Why a perfect database search still misses studies
Even a peer-reviewed search strategy depends on authors and indexers using words you anticipated. A genuinely relevant study described in idiosyncratic language, or indexed under the wrong headings, can slip through every Boolean line. Citation methods sidestep that weakness entirely: they follow how researchers actually cite one another, which does not depend on vocabulary. That is why they so often surface studies the main search overlooked, and why they are a standard part of running a systematic literature search.
Backward citation chasing
Backward chasing means reading the reference lists of your already-included studies and any relevant existing reviews. Because a study can only cite work that came before it, backward chasing reaches into the older literature. Reference lists of prior systematic reviews are especially productive, since someone has already gathered the relevant older studies in one place.
Forward citation chasing
Forward chasing uses a citation index to find every later paper that cited a known relevant study. This reaches into the newer literature, including recent work that may not yet be fully indexed in the bibliographic databases you searched. Forward chasing is particularly useful for catching studies published after your main search was run.
Snowballing as an iterative cycle
Snowballing turns single passes into a cycle. When chasing the references and citations of your seed studies turns up a new eligible study, that study’s own references and citations are chased in turn. You keep iterating until a round produces no new includes, at which point the snowball has stopped growing. This is the same logic behind tools that automate the process, but the principle holds whether you do it by hand or with software.
Where citation methods fit in the workflow
Citation chasing is a supplement to, never a replacement for, structured database searching and grey literature searching. Records it surfaces pass through the same title and abstract screening against your eligibility criteria as any other record.
Choosing seeds and tools for citation chasing
The yield of citation chasing depends heavily on what you feed it. The most productive seeds are your already-included studies, the reference lists of recent systematic reviews on the same question, and any landmark paper everyone in the field cites. A useful sequence is:
- Run the structured database search and complete title and abstract screening first, so your seed set is the studies that genuinely passed eligibility.
- Backward-chase each seed’s reference list and the reference lists of any relevant prior reviews to reach the older literature.
- Forward-chase each seed through a citation index to reach studies published after your main search ran.
- Screen everything new against the same criteria, then snowball from any fresh includes.
Forward chasing needs a citation index, so the tool matters: multidisciplinary indexes such as Web of Science or Scopus track who cited a paper, and several free citation tools now automate the backward-and-forward pass across a whole seed set at once. Whatever the tool, the principle is the same: you are following how researchers cite one another, which does not depend on the wording your Boolean lines guessed at.
Knowing when the snowball has stopped
Snowballing has a natural stopping point: the round in which chasing a fresh set of includes produces no new eligible studies. Reaching that saturation is the signal that the citation network around your question has been exhausted, and it is worth recording how many rounds it took, because a reviewer who reports reaching saturation gives stronger evidence of completeness than one who simply “checked some references”. The same logic guards against the opposite failure of stopping too early, after a single pass, and leaving connected studies unchased.
Common citation-chasing mistakes
Three errors recur. The first is using citation chasing as the main search rather than a supplement, which cannot give the reproducible coverage a review needs. The second is chasing only backward, which reaches older work but misses the recent citing studies forward chasing would catch. The third is not recording the seeds and direction, which leaves this part of the search unreproducible. Avoiding these keeps citation methods doing what they do best, recovering the eligible studies that slipped past a term-based search, as part of the wider systematic search workflow.
Reporting what citation searching found
Because citation methods are less reproducible than a Boolean search, the record matters. Note which studies you used as seeds, which direction you chased, the tool or index used, and the date, following documenting and reporting your search. Studies found this way are counted into the PRISMA flow diagram under the other-methods category, keeping the route by which every study entered the review fully transparent.