How and Where to Find Research Papers for Literature Reviews: Literature Search Strategies and Examples

Key Takeaways:

A clear research question, not a random keyword search, is the foundation of an effective literature search.
Boolean operators, controlled vocabulary, and citation tracking help you find more relevant papers in less time.
Screening for duplicates and checking source credibility protect the quality of your review.
R Discovery combines database aggregation, AI-based recommendations, and citation tracking in 1 free app.

Table of Contents

Glossary of Key Terms

Use this table as a quick reference before you read the rest of the guide.

Term	Definition
Boolean operators	Words such as AND, OR, and NOT used to combine or exclude search terms.
Controlled vocabulary	A standardized set of subject terms, such as MeSH, used by a database to index articles.
PICO framework	A method for structuring a research question around population, intervention, comparison, and outcome.
SPIDER framework	A method for structuring qualitative research questions around sample, phenomenon of interest, design, evaluation, and research type.
Backward citation search	Reviewing the reference list of a relevant paper to find earlier related work.
Forward citation search	Finding newer papers that have cited a relevant paper since its publication.
Grey literature	Research such as theses, reports, and conference papers not published in traditional journals.
Preprint	A research manuscript shared publicly before formal peer review.
PRISMA	A reporting checklist used to document the methods and results of a systematic review.
Predatory journal	A publication that charges fees but skips genuine peer review.
Deduplication	The process of removing repeated search results found across multiple databases.
Systematic review	A structured, documented review that follows a predefined, reproducible search protocol.
Narrative review	A less formal review that summarizes literature without a fixed search protocol.

What Is a Literature Search Strategy?

A literature search strategy is a planned, repeatable process for finding, screening, and organizing published research on a topic. It combines a clear research question, defined keywords, chosen databases, and documented search terms so the results stay relevant and traceable.

The volume of published research grows every year across nearly every field, and preprint servers add even more material outside traditional journals. Without a plan, researchers can spend hours scanning results that turn out to be irrelevant, outdated, or unreliable. A defined search strategy narrows this gap and makes the process easy to repeat for later updates or related projects.

How Do You Define Your Research Question First?

Start by breaking your topic into its core concepts using a structured framework, such as PICO or SPIDER. This step turns a broad interest into specific, searchable terms before you open a single database.

Using the PICO Framework

PICO is widely used in health and clinical research to structure a focused, answerable question.

Element	Stands For	Example
P	Population or problem	Adults with type 2 diabetes
I	Intervention	Low-carbohydrate diet
C	Comparison	Standard dietary advice
O	Outcome	Change in HbA1c levels

Using the SPIDER Framework for Qualitative Research

SPIDER works well for qualitative or mixed-methods questions, where PICO’s clinical focus does not fit as neatly.

Element	Stands For
S	Sample
PI	Phenomenon of Interest
D	Design
E	Evaluation
R	Research type

How to Build a Strong Search String

Once your concepts are defined, translate each one into keywords and combine them using search logic. This is the step most researchers skip, and it is the one that saves the most time later.

Boolean Operators and Truncation

These simple tools narrow or broaden a search without changing your underlying concepts.

Operator	Function	Example
AND	Narrows results to papers containing all terms	diabetes AND diet
OR	Broadens results to papers containing any term	diabetes OR “blood sugar”
NOT	Excludes a term from results	diabetes NOT gestational
*	Truncation, finds word variations	educat* finds education, educator, educating

Quotation marks around a phrase, such as “climate change,” tell most databases to search that exact phrase instead of the words separately.

Controlled Vocabulary vs Free-Text Terms

Free-text keywords are flexible but can miss papers that use different wording for the same idea. Controlled vocabulary, such as PubMed’s MeSH terms, indexes papers under standardized subject headings, which improves precision once you know which terms apply.

Approach	Best Used When
Free-text keywords	Topic is new, emerging, or the database lacks a subject thesaurus
Controlled vocabulary	Database offers standardized subject headings, such as MeSH or Emtree

Steps to Build a Search String

Start with your framework terms. Pull keywords directly from your PICO, PCC, or SPIDER elements so every concept in your question is represented.
List synonyms for each concept. Combine them with OR inside parentheses, for example (diabetes OR “blood sugar” OR hyperglycemia), so wording differences across papers don’t cost you results.
Combine concepts with AND. Use AND only between different concepts, never between synonyms of the same concept, or you’ll accidentally narrow your search too far.
Use truncation carefully. A wildcard such as educat* saves time but can pull in unrelated words, so check the result list before trusting it.
Add controlled vocabulary where available. Layer MeSH or Emtree terms alongside free-text keywords for extra precision, especially in health databases.
Use phrase searching for fixed terms. Quotation marks around multi-word phrases, such as “machine learning,” stop the database from searching the words separately.
Test and refine. Run the string, scan the first 20-30 results, and adjust terms that return too much noise or too few hits.
Save the final string. Copy it into your documentation table so the search can be rerun or updated later.

Where Should You Search for Literature?

Search more than 1 source. Combine subject-specific databases, multidisciplinary aggregators, and citation indexes, since no single database indexes every relevant journal or preprint.

Academic Databases and Aggregators

Subject-specific databases, such as PubMed or IEEE Xplore, for deep coverage in 1 field.
Multidisciplinary databases, such as Scopus or Web of Science, for cross-field coverage.
Google Scholar, for broad, quick searches across sources.
AI-based research aggregators, such as R Discovery, which pull from 250M+ articles across trusted sources, including CrossRef, Unpaywall, PubMed, PubMed Central, and OpenAlex, into a single feed.

Grey Literature and Preprints

Theses, conference papers, government reports, and preprints from servers such as arXiv, bioRxiv, and medRxiv often appear months before a peer-reviewed version. Including them gives a more complete picture, especially in fast-moving fields, though they should be flagged as not yet peer-reviewed when you cite them.

Which Databases to Choose, and in What Order to Search Them

The right database order depends on your level of study and how deep your review needs to go. Searching in a logical sequence, broad to specific, saves time and reduces the chance of missing key papers.

For undergraduate students

The priority in undergrad assignments like essays is speed and accessibility over exhaustive coverage.

Start with Google Scholar. It’s free, familiar, and gives a quick sense of how much has been written on your topic, plus easy access to full text through your university’s proxy links.
Move to your library’s discovery layer (the single search box most university library homepages offer). It searches multiple subscribed databases at once and filters out predatory or low-quality sources better than open web search.
Add 1 subject-specific database relevant to your field, such as PsycINFO for psychology or IEEE Xplore for engineering, once you have a working set of keywords from steps 1 and 2.
Use an aggregator like R Discovery to keep a running feed on your topic without repeating the same manual search each week.

For graduate students

Coverage and reproducibility matter more in a thesis or dissertation, so the order shifts toward rigor.

Start with 1 or 2 subject-specific databases in your field (PubMed for health sciences, Scopus or Web of Science for broad multidisciplinary coverage) since these give the most precise, citation-tracked results.
Add a second multidisciplinary database to catch papers your primary field database might not index, and cross-check overlap between the two.
Search grey literature and preprint servers separately (institutional repositories, arXiv, bioRxiv, ProQuest Dissertations) since these rarely show up in standard database searches.
Run Google Scholar last, specifically for forward citation searching and catching anything the structured databases missed.
Use R Discovery throughout, not just at the end, since its cross-source aggregation and citation tracking can shorten steps 1-3 into a single ongoing feed instead of separate manual searches.

Database Type	Best For	Search Order Tip
Google Scholar	Quick orientation, forward citation search	Undergrad: first. Grad: last
Library discovery layer	Multi-database access, filtered quality	Undergrad: second
Subject-specific database	Precision, controlled vocabulary	Grad: first
Multidisciplinary database	Cross-field coverage	Grad: second
Grey literature/preprints	Completeness, emerging work	Grad: third
R Discovery	Ongoing aggregation, daily updates	Both: alongside or after manual search

A general rule for both levels: search from broad-and-fast to narrow-and-rigorous, and always keep 1 tool running in the background (an aggregator or saved alert) so new literature keeps surfacing after your initial search ends.

Common Filters in Databases and When to Apply Them

Most databases offer built-in filters that narrow results after you run your search string. Filters are fast, but applying the wrong one at the wrong stage can quietly exclude relevant papers, so it helps to know what each filter actually does before switching it on.

Publication date

Restricts results to a chosen year range. Apply this early if your review has a defined scope, such as “last 10 years,” but apply it last if you’re unsure of the right cutoff, since narrowing too early can hide foundational papers you’d want to cite for context.

Peer review or “scholarly/academic journals” only

Excludes magazines, trade publications, and grey literature. Useful for systematic and rapid reviews where peer review is a core inclusion criterion, but turn this off deliberately for scoping, integrative, or realist reviews, where grey literature and non-peer-reviewed sources often carry real value.

Study design or article type

Filters such as “randomized controlled trial,” “review,” or “clinical trial” are built mainly for clinical, quantitative research. Apply these only when your review explicitly requires a single design type. For integrative or mixed-methods reviews, leave this off, since it will silently exclude qualitative and case-study literature.

Language

Restricts to 1 or more languages, most often English. Apply this if your team can’t realistically screen or translate non-English papers, but note it as a limitation in your methods rather than treating it as a neutral default.

Full text availability

Limits results to papers your institution has full-text access to. This is convenient but risky to apply during the search stage, since it excludes relevant papers you might still obtain through interlibrary loan or an author request. Apply it during retrieval, not during the search itself.

Subject or discipline filters

Narrows results to a specific field, useful in large multidisciplinary databases like Scopus or Web of Science when your keywords return results from unrelated disciplines. Apply this cautiously in integrative or realist reviews, where relevant explanatory literature often sits outside your core discipline.

Age group, sex, or population filters

Common in health databases such as PubMed. Apply only if your inclusion criteria specify a population subgroup; otherwise, this can exclude papers that studied a mixed sample but still report relevant subgroup data.

Filter	Apply When	Avoid or Delay When
Publication date	Scope has a defined time window	Establishing foundational context
Peer review only	Systematic or rapid review	Scoping, integrative, or realist review
Study design	Single design type required	Reviewing mixed-methods literature
Language	Team can’t screen other languages	Comprehensive global coverage needed
Full text availability	During retrieval stage	During the search stage
Subject/discipline	Results return unrelated fields	Cross-disciplinary review types

General rule

Apply restrictive filters as late as possible, after you’ve seen your full result set, so you can judge what you’d actually be excluding before you commit to it.

How to Find and Search Grey Literature

Grey literature is research that isn’t published through traditional journals, including theses, conference papers, government and NGO reports, policy briefs, working papers, and clinical trial registries. It matters because it often surfaces evidence months or years before a peer-reviewed version appears, or it may never be formalized into a journal article at all, especially for niche topics, negative results, or emerging fields.

Where to look, by type:

Grey Literature Type	Where to Search
Theses and dissertations	ProQuest Dissertations and Theses, university repositories
Conference papers	Conference proceedings databases, society websites
Government and policy reports	Government agency websites, OpenGrey, WorldCat
Preprints	arXiv, bioRxiv, medRxiv, SSRN
Clinical trials	ClinicalTrials.gov, WHO ICTRP
NGO and think-tank reports	Organization websites, Google with site-specific searches

Search strategy for grey literature

Grey literature isn’t indexed with controlled vocabulary or consistent metadata, so Boolean logic and MeSH terms won’t get you far. Instead:

Use targeted Google searches with site restrictions, such as searching your keywords limited to a government domain, to surface reports that generic search engines bury.
Check organization websites directly. Many government agencies, NGOs, and professional associations publish reports that never reach a database at all, so browsing their publications page is often more productive than searching for them.
Search conference proceedings separately. Major conferences in your field usually archive past proceedings on their own site, searchable by year or keyword.
Use a grey literature checklist. Resources like the CADTH Grey Matters checklist walk through source types systematically, which helps you avoid missing a category entirely.

Document it like everything else

Record which grey literature sources you searched, the search terms used, and the date, in the same documentation table you use for databases. This matters even more for grey literature, since these sources change or disappear from the web more often than journal archives do.

Know when it matters most

Grey literature carries more weight in scoping and mapping reviews, health technology assessments, and fast-moving fields where formal publication lags behind real-world developments. For a narrow, clinical systematic review, it may add less value relative to the time it takes to search, so scope your effort to your review type.

A caution on quality

Grey literature skips peer review, so evaluate it more carefully: check the author’s or organization’s credibility, look for a clear methodology section, and note in your write-up that a source wasn’t peer-reviewed.

Practical Tip

R Discovery’s aggregation already pulls in preprints alongside peer-reviewed content from its trusted sources, which covers 1 major grey literature category automatically, though government reports, theses, and conference proceedings still need to be searched separately through the sources above.

How Do You Track Citations Effectively?

Track citations in 2 directions. Look backward at what a relevant paper cites, and forward at newer papers that cite it, so you catch both the foundational work and the latest developments.

Backward Citation Search

Once you find a highly relevant paper, scan its reference list for earlier studies on the same question. This is one of the fastest ways to locate foundational research that your keyword search may have missed.

Forward Citation Search

Forward citation search finds papers published after your key source that cite it, showing how the topic has developed since. Databases such as Google Scholar and Web of Science, along with citation-mapping tools, support this, and R Discovery’s tangential topic matching surfaces related newer work automatically.

How Should You Screen and Remove Duplicate Results?

Screen results in 2 passes: first remove duplicates found across databases, then review titles and abstracts against your inclusion criteria before reading full text.

Deduplicate by DOI or exact title match before screening begins.
Read only the title and abstract in the first pass; exclude clearly irrelevant papers.
Apply your inclusion and exclusion criteria consistently, and note the reason for each exclusion.
Keep a running count of papers found, duplicates removed, and papers included, for later reference.

How to Set Inclusion and Exclusion Criteria for a Literature Search

Inclusion and exclusion criteria are the rules that decide which papers make it into your review and which get filtered out. Setting them before you start screening, not while you’re screening, is what keeps the process consistent and defensible.

Base your criteria on your framework

Whatever structure you used to define your question (PICO, PCC, or SPIDER) should map directly onto your criteria. Each element of the framework becomes a filter: population, intervention or concept, context, and study type all give you a natural starting list rather than an arbitrary one.

Common criteria categories to define:

Population or sample: age range, condition, demographic, or setting you want represented.
Study design: randomized controlled trials only, or a broader mix including observational studies, depending on your review type.
Publication date range: for example, papers published in the last 10 years, unless your topic requires historical coverage.
Language: English-only is common for undergrad and many grad projects, but note it as a limitation if you exclude non-English papers.
Publication status: peer-reviewed only, or peer-reviewed plus grey literature and preprints, depending on your review type.
Outcome or concept relevance: the paper must report on, or discuss, the specific outcome or concept your question addresses.

Write criteria as explicit, testable statements, not vague themes. “Studies must report a quantitative outcome measure for glycemic control” is testable. “Studies about diabetes management” is not, since almost every result will technically qualify.

Criterion Type	Include	Exclude
Study design	Randomized controlled trials	Case reports, opinion pieces
Population	Adults aged 18 and older	Pediatric-only studies
Publication date	2016-2026	Studies published before 2016
Language	English	Non-English, no translation available

Pilot-test your criteria before full screening

Apply them to a sample of 20-30 papers first, and check whether 2 reviewers (if you have them) agree on borderline cases. Disagreement at this stage usually means a criterion is too vague and needs rewording before you apply it to the full result set.

Keep criteria stable, but log any changes

If you discover partway through screening that a criterion needs adjusting, such as widening the date range because early results are too thin, that’s fine. Just record the change, the date, and the reason, so your search stays reproducible and transparent rather than looking inconsistent.

Practical Tip

Tools like R Discovery let you apply topic filters and give instant feedback on recommended papers, which can help you pilot-test how narrow or broad a criterion feels before you commit to it across a full, larger search.

How Do You Know a Source Is Reliable?

Check whether the journal is indexed in a recognized database, whether it describes a transparent peer-review process, and whether its editorial board has verifiable, real affiliations before you trust its content.

Confirm the journal is indexed in a database such as Scopus, Web of Science, or PubMed.
Look for a clearly described, transparent peer-review process on the journal’s site.
Verify editorial board members and their institutional affiliations independently.
Be cautious of unusually fast acceptance timelines or aggressive fee-based advertising/promotions on the journal homepage.
Cross-check the journal name against known predatory-journal watchlists.

Documenting Your Search for Reproducibility

If you are conducting a systematic review, document your process using a framework such as PRISMA. This record lets other researchers, or you, repeat the exact search later and update it as new literature is published.

Database	Search String Used	Date Searched	Results Found
PubMed	diabetes AND diet AND “blood sugar”	15 Jul 2026	412
Scopus	diabetes AND diet AND HbA1c	15 Jul 2026	287
R Discovery	Topic feed: type 2 diabetes, diet	15 Jul 2026	96

Staying Updated After Your Initial Search

A literature search is not a 1-time task. New papers publish daily, so build a habit of ongoing monitoring.

Alerts and Newsletters

Most databases and publishers let you save a search and receive email alerts when new matching papers appear. Subscribing to publisher newsletters adds another layer of coverage for your specific field.

Reference Managers

Tools such as Paperpal, Mendeley, Zotero, and EndNote let you save papers, generate citations, and organize reading lists in 1 place. Most support direct import from your browser and export to common citation styles.

How Frequently Do I Need to Re-Run My Literature Search?

It depends on your project’s timeline and the pace of publishing in your field. As a general guideline:

Short projects (a few weeks to a couple months): Re-run your search once, right before you finalize your review, to catch anything published since your initial search.
Systematic reviews and theses (6-12 months): Re-run every 3-6 months, and always do a final re-run within 3 months of submission, since journals often require an updated search before publication.
Fast-moving fields (AI, genomics, infectious disease): Consider monthly or biweekly checks, since relevant papers can appear faster than your writing timeline.
Living or continuously updated reviews: Set up ongoing alerts instead of manual re-runs, so new matches surface automatically.

Practical tip: save your original search string exactly as documented, and use saved-search alerts or a daily-recommendation tool like R Discovery so new matches surface between manual re-runs instead of requiring you to remember to check.

Which Tool Can Simplify All These Steps?

R Discovery is a free research discovery app that combines database aggregation, AI-based recommendations, and citation tracking, so you do not need to repeat these steps manually across multiple platforms.

Set up your feed once by selecting topics of interest, and the app recommends its top 3 matched papers every day, drawing from over 57M scholarly articles across 5M+ topics and 32,000+ journals.

Daily recommendation of the top 3 articles matched to your topics of interest.
Tangential topic matches, so you catch related developments outside your core keywords.
Smart summaries that let you review a paper’s key points before reading the full text.
Import your existing reading library from Mendeley, Zotero, or ORCID.
Search filters, instant feedback, and 1-click bookmarking to refine your feed over time.
Reading-history tracking, so you can revisit patterns in what you have already read.

R Discovery Prime, the paid tier, adds audio playback, reading in your own language, peer collaboration, and reference-manager auto-sync, starting at $72 a year after a free 7-day trial.

Search Strategies for a Scoping Review or Mapping Review

A scoping review or mapping review asks a different question than a systematic review. Instead of narrowing to a small set of studies that answer 1 specific question, it aims to map the breadth of literature on a topic: what has been studied, which populations and methods appear, where the evidence gaps sit. This changes how you build and run the search.

Start broader, not narrower

A systematic review search is built to be precise, often trading some sensitivity for a manageable, focused result set. A scoping review search does the opposite. You want high sensitivity, even at the cost of extra irrelevant hits, because the goal is comprehensive coverage of a field, not a filtered answer set. Use fewer restrictive AND combinations and lean more on OR to capture variation in terminology across sub-fields.

Use a framework built for breadth, not PICO

PICO assumes a comparison and an outcome, which scoping reviews often do not have. The Joanna Briggs Institute recommends PCC instead: Population, Concept, Context. This keeps your question open enough to capture diverse study designs, including qualitative work, case studies, and grey literature, without forcing an intervention-outcome structure onto a topic that does not have one.

Element	Stands For	Example
P	Population	Older adults living independently
C	Concept	Use of wearable health monitors
C	Context	Home care settings, high-income countries

Expect to search more sources, not fewer

Because scoping reviews aim to map a field rather than answer a clinical question, grey literature, conference proceedings, theses, and policy documents matter more here than in most systematic reviews. Government and NGO websites, institutional repositories, and preprint servers often carry the earliest or most practical evidence on emerging topics, and skipping them can leave real gaps unmapped.

Document everything, even more rigorously

Scoping reviews follow the PRISMA-ScR extension, which requires you to report your full search strategy, including every database, the exact search string, filters applied, and the date searched, in enough detail that someone else can rerun it. Because the search is broader and less selective than a systematic review, this documentation is what keeps the process defensible and transparent rather than arbitrary.

Screening looks different too

Expect a much larger initial results set, often in the thousands, since sensitivity was prioritized over precision. Title and abstract screening usually uses broad inclusion criteria tied to your PCC elements rather than tight methodological filters, and it is common to involve 2 independent reviewers at this stage given the volume and the higher chance of disagreement over what counts as “in scope.”

Practical Tip

This is exactly where an aggregator like R Discovery earns its keep. Its cross-database coverage (CrossRef, Unpaywall, PubMed, PubMed Central, OpenAlex, and major publishers) reduces the manual burden of running the same broad search across 5 or 6 separate platforms, and its tangential topic matching can surface adjacent sub-fields you might not have thought to search for directly, which is often exactly what a mapping review is trying to find.

Search Strategies for a Rapid Review

A rapid review answers a real question, often for policy or clinical decision-making, under a tight deadline, typically weeks rather than months. The search strategy has to trade some comprehensiveness for speed without becoming careless, since the output still needs to be defensible.

Limit your database count deliberately

Instead of searching 5 or 6 sources, most rapid reviews restrict to 2 or 3 high-yield databases most likely to cover your topic, such as PubMed plus 1 multidisciplinary database like Scopus. This is a documented methodological choice, not a shortcut you hide, so state it clearly as a limitation in your methods.

Apply date and language limits early

Rapid reviews commonly restrict results to the last 5-10 years and to English-language publications from the start, rather than screening broadly and filtering later. This cuts volume where it matters least for most fast-turnaround questions.

Use a single reviewer with spot-checks, not full dual screening

Where systematic reviews require 2 independent reviewers for every paper, rapid reviews often use 1 reviewer for the full set and a second reviewer to check a sample, commonly 10-20%, for consistency. Document this trade-off explicitly.

Skip exhaustive grey literature searching

A rapid review typically checks 1 or 2 key grey literature sources relevant to the topic (a government registry or a major preprint server) rather than working through a full grey literature checklist.

Reuse existing evidence syntheses where possible

Before building a search from scratch, check for a recent systematic review or scoping review on the same or an adjacent question. If one exists, you can update its search rather than starting fresh, which is often the single biggest time-saver in a rapid review.

Use forward citation search aggressively

Rather than an exhaustive backward and forward search, prioritize forward citation search on 2 or 3 landmark papers to catch the most recent developments quickly, since this tends to have a higher hit rate per minute spent than expanding keyword combinations further.

Standard Systematic Review Search	Rapid Review Adaptation
5 or more databases	2 or 3 high-yield databases
No default date limit	5-10 year limit applied upfront
Dual independent screening	Single reviewer, sample spot-check
Full grey literature checklist	1 or 2 targeted grey literature sources
Full backward and forward citation search	Forward search on key landmark papers only

Document every shortcut

The credibility of a rapid review rests on transparency about what was scoped out, so your methods section should state the database count, date limits, screening approach, and citation search scope as explicit decisions, not omissions.

Practical Tip

R Discovery’s daily-recommendation feed and cross-source aggregation suit rapid reviews well, since a single topic feed set up early can substitute for part of the manual multi-database search, freeing time for the screening and synthesis stages where a tight deadline is felt most.

Search Strategies for an Integrative Review

An integrative review combines findings from studies using different methodologies, quantitative, qualitative, and mixed-methods, into a single synthesis. This breadth of study design is the main thing that shapes how you search, since a search string built only around quantitative terms will miss half your relevant literature.

Build separate keyword branches for different methodologies

Instead of 1 search string, plan for overlapping strings that each capture a study-design family, then combine or run them separately depending on the database’s capabilities.

Study Type Branch	Example Keywords to Add
Quantitative	“randomized,” “cohort,” “survey,” “regression”
Qualitative	“interview,” “focus group,” “thematic analysis,” “grounded theory”
Mixed methods	“mixed methods,” “triangulation,” “sequential design”

Don’t over-restrict by study design filters

Many databases offer a “study type” filter built for clinical quantitative research, and applying it by default will silently exclude qualitative and mixed-methods studies. Search without this filter first, then sort by design during screening instead.

Search theoretical and conceptual literature alongside empirical studies

Integrative reviews often aim to build or refine a concept or theoretical framework, so searches should include conceptual and discussion papers, not only empirical study reports. Add keywords like “framework,” “concept analysis,” or “theoretical model” as a separate search layer.

Expect a 2-stage inclusion process

Because study designs vary so widely, most integrative reviews evaluate quality using different criteria per design family (a checklist for RCTs, a different one for qualitative studies) so plan your criteria before screening, matching each included paper’s design to its appropriate quality-appraisal tool.

Use citation searching to bridge disciplines

Integrative reviews often sit at the intersection of 2 or more fields, so forward and backward citation searches are especially useful for catching relevant work published in a neighboring field’s journals that your core keyword search might miss entirely.

Grey literature matters, selectively

Include grey literature where it’s likely to contain qualitative or practice-based evidence not captured in journals, such as clinical practice guidelines or professional association reports, but this is usually a secondary rather than primary source layer here.

Document design-specific decisions

Your search documentation should note not just databases and strings but which study-design branches were searched, and any design-specific filters applied or deliberately avoided, since this is where an integrative review’s methodology is most likely to be questioned.

Practical Tip

R Discovery’s tangential topic matching is particularly useful here, since integrative reviews benefit from catching adjacent-field literature that a narrow keyword search would miss, and its aggregation across a wide range of publishers reduces the risk of a methodology-based blind spot in your source coverage.

Search Strategies for a Realist Review

A realist review asks a different kind of question entirely: not “does this intervention work,” but “what works, for whom, under what circumstances, and why.” This changes the search strategy at its foundation, since you’re searching for explanatory mechanisms and contextual factors, not just outcome data.

Expect iterative, not linear, searching

Unlike a systematic review’s fixed, upfront search string, a realist review search evolves as your program theory develops. You typically run an initial broad search, refine your understanding of the mechanisms at play, then run further targeted searches based on what the first round revealed. Plan for multiple search cycles, not 1 comprehensive pass.

Start with a program theory, not a PICO question

Before searching, realist reviews develop an initial rough theory of how and why an intervention is believed to work. This theory, not a fixed population-intervention-outcome structure, guides your first search terms, and it will be refined as evidence comes in.

Search for mechanism and context terms, not just outcomes

Add keywords describing how or why something happens, alongside your topic terms: “mechanism,” “context,” “implementation,” “barriers,” “facilitators.” A standard outcome-focused search string will systematically miss the explanatory literature a realist review needs most.

Purposive sampling replaces exhaustive searching at later stages

After the initial broad search, later search cycles are often purposive, meaning you deliberately seek out specific papers likely to explain a mechanism or context you’ve identified as important, rather than exhaustively screening every result in a database.

Include theoretical and disciplinary literature outside your core field

Realist reviews frequently draw on social science, implementation science, and organizational theory literature to explain mechanisms, even when the core topic is clinical or educational. Broaden your database selection accordingly rather than staying within 1 discipline’s journals.

Document your search as a process, not a single table

Because the search evolves, your methods section should describe each search cycle: what prompted it, what terms were used, and what it contributed to the developing program theory, rather than a single static search-string table.

Search Stage	Purpose
Initial scoping search	Build a rough program theory
Structured background search	Test and refine theory with broader evidence
Purposive/theoretical search	Fill specific gaps in mechanism or context understanding

Practical Tip

Because realist reviews search iteratively and across disciplines, an aggregator like R Discovery, with its tangential topic matching and broad multi-publisher coverage, works well for the purposive later-stage searches, where you’re hunting for a specific explanatory paper rather than running a single fixed query.

Frequently Asked Questions

What is the best way to search for research papers online?

The best approach combines a defined research question, a structured keyword and Boolean search string, and more than 1 database or aggregator. Adding an AI-based tool such as R Discovery reduces manual searching by matching new papers to your interests automatically.

How many databases should I search for a literature review?

Most guidance recommends searching at least 2 to 3 databases, since no single source indexes every journal. A subject-specific database, a multidisciplinary database, and a broad tool such as Google Scholar or R Discovery together give solid coverage.

What is the difference between a systematic review and a narrative review?

A systematic review follows a predefined, documented search protocol, often reported using PRISMA, so it can be reproduced and audited. A narrative review summarizes literature more informally, without a fixed, repeatable search process.

How do I know if a journal is predatory?

Check that the journal is indexed in a recognized database, review its peer-review process, verify its editorial board’s affiliations, and cross-check the title against known predatory-journal watchlists. Unusually fast acceptance and aggressive fee emails are common warning signs.

Can I use Google Scholar for a systematic literature review?

Google Scholar is useful for a broad, quick search, but its lack of advanced filters and controlled vocabulary makes it unreliable as your only source for a systematic review. Pair it with subject-specific databases and document your full search string for reproducibility.

How do I keep track of citations while doing a literature search?

Use a reference manager, such as Mendeley or Zotero, to save papers as you find them, and run both backward and forward citation searches on your key sources. R Discovery’s citation and tangential topic tracking can automate part of this process.

What is the fastest way to stay updated on new research in my field?

Set up saved-search alerts on your core databases, subscribe to relevant publisher newsletters, and use a daily-recommendation tool such as R Discovery, which surfaces its top 3 matched papers for you automatically each day.

How do I find controlled vocabulary?

Controlled vocabulary lives inside the database you’re searching, not in a separate external list, so the first place to look is the database’s own thesaurus or subject-headings tool.

PubMed: Use the MeSH (Medical Subject Headings) database directly, or type a keyword into PubMed’s search box and check the “MeSH Terms” field in a relevant result’s full record to see which official term it was indexed under.
Embase: Look for the Emtree thesaurus, accessible from the database’s search interface, which functions similarly to MeSH but with broader pharmacological and clinical coverage.
PsycINFO: Use the APA Thesaurus of Psychological Index Terms, built into the database’s advanced search screen.
ERIC (education research): Use the ERIC Thesaurus, searchable from the database homepage.
CINAHL (nursing and allied health): Use CINAHL Subject Headings, also built into the advanced search interface.

A simple shortcut: find 1 paper you already know is highly relevant, open its full record in the database, and look for a field labeled “Subject Headings,” “MeSH Terms,” or “Descriptors.” Whatever terms appear there are the controlled vocabulary the database uses for that topic, and you can reuse them directly in your search string.

Most university library databases also list their thesaurus under an “Advanced Search” or “Subject Terms” tab, so a student who isn’t sure whether a database has one should check there before assuming it doesn’t.

What Do I Do If I Can’t Access the Full-Text Version of a Paper During Literature Search?

Start with legitimate, no-cost channels before assuming the paper is unreachable.

Check your institution’s library first. Most universities subscribe to journals your personal search won’t show as accessible by default. Search the paper’s title directly in your library’s discovery system or link resolver, even if the original database marked it as unavailable.
Look for an open-access or repository copy. Tools like Unpaywall, CORE, or OpenDOAR search for legally hosted free versions, including publisher open-access copies and author-deposited manuscripts in institutional or subject repositories. R Discovery also draws on Unpaywall and similar sources, so a paper flagged there is confirmed as legally accessible.
Check the author’s own webpage or ORCID profile. Many researchers legally post a preprint or accepted manuscript version of their published work on a personal or institutional site.
Use interlibrary loan (ILL). If your library doesn’t subscribe to the journal, ILL lets you request a copy from another institution, usually at no cost to you, though it can take a few days.
Email the corresponding author directly. Requesting a copy for personal research use is a long-standing, accepted academic practice, and most authors are glad to share their own work.
Check if it’s a preprint. If the published version is paywalled, an earlier preprint version may be freely available on arXiv, bioRxiv, SSRN, or a similar server, though you should note any differences between preprint and final published versions in your review.
Use your library’s document delivery or pay-per-view budget, if your institution offers one, rather than seeking the paper through unauthorized channels.

What to avoid entirely

Do not use shadow libraries, pirated PDF sites, or browser extensions that scrape paywalled content without authorization. These sources carry real legal risk, and papers obtained this way should not be cited or relied upon in academic work, since you can’t verify their integrity or whether the document you received is even the actual, original source.

If you genuinely can’t access it after trying the above, it’s acceptable to note in your review that a relevant paper could not be obtained in full text, and rely on the abstract only where appropriate, clearly flagging this limitation rather than working around it improperly.

Practical Tip

R Discovery’s aggregation from Unpaywall and other trusted open-access sources means a meaningful share of papers in your feed already link to a legal, free full-text copy, reducing how often you hit this problem in the first place.

What Can I Do If I’ve Applied the Wrong Filters During My Literature Search?

Re-run the search with the filter removed or corrected, then compare the new result count against your original. Most databases let you adjust filters without rebuilding your entire search string, so this is usually a quick fix rather than a full restart.

Check your saved search history first. Most databases, including PubMed and Scopus, keep a session or account-based search history, so you can revisit and edit the exact search rather than retyping it.
Re-screen only the newly added results. If removing a filter expands your result set, you don’t need to re-screen everything, just the papers that weren’t in your original set.
Update your documentation immediately. Note the correction, the date, and the reason in your search log, since an accurate record matters more than a perfect first attempt.
Treat it as a normal part of the process. Filter corrections are common and expected, not a sign your search was invalid, as long as you document the change transparently.

What Notes Should I Be Taking During My Literature Search?

Good notes turn a messy search into a reproducible, defensible process. At minimum, track the following as you go, rather than trying to reconstruct it afterward.

Search log: database name, exact search string used, filters applied, date searched, and number of results returned for every search you run, including revised ones.
Screening decisions: for each paper you exclude past the title and abstract stage, note the specific reason, such as “wrong population” or “no full text available,” rather than a vague “not relevant.”
Inclusion and exclusion criteria changes: if you adjust a criterion partway through, log the original version, the revised version, the date, and why the change was made.
Full-text access issues: which papers you couldn’t obtain, what you tried (library, ILL, author email), and how you resolved or flagged the gap.
Duplicate records: how many duplicates you removed and the method used, such as DOI matching or reference manager auto-detection.
Citation search sources: which key papers you ran backward and forward citation searches on, and what new papers each one surfaced.
Grey literature sources checked: which organizations, registries, or repositories you searched separately from your main databases, even if they returned nothing.
Reviewer disagreements, if you have a second screener, including how conflicts were resolved.

Note Type	Why It Matters
Search log	Makes the search reproducible and updatable
Screening reasons	Supports transparency and audit of your inclusion decisions
Criteria changes	Shows your process was principled, not arbitrary
Access issues	Documents limitations honestly

Practical tip

Keep this in a single running spreadsheet or table from day 1, rather than a separate document per database, so it stays easy to reference and easy to paste into your methods section or PRISMA flow diagram later.

This article was originally published on January 3, 2024, and updated on July 15, 2026.