Searching Veterinary Literature: Databases, Filters, and Search Strings

By Dr. Zubair Khalid, DVM, MS, PhD · 2026-08-11

Key Takeaways

• Comprehensive veterinary literature searches necessitate utilizing multiple primary databases, including PubMed, CAB Abstracts, Web of Science, and Scopus, to ensure broad coverage of biomedical, agricultural, and multidisciplinary research.
• Structured search strings must integrate controlled vocabulary (e.g., MeSH, CAB Thesaurus) with free-text synonyms, Boolean operators (AND, OR, NOT), phrase searching, and truncation, while carefully considering database-specific syntax and field tags (e.g., [tiab] in PubMed).
• Filters, such as publication type, date, species, or language, should be applied judiciously and always documented with justification to avoid introducing bias and ensure reproducibility, aligning with reporting standards like PRISMA and ARRIVE.
• Reproducibility demands meticulous documentation of each search, recording database names, platform, search dates, full search strings, and all applied filters, which is crucial for peer review and updating searches.
• Common search failures include overly narrow vocabulary, missed synonyms, inconsistent truncation, unrecorded filters, and concept drift, which can be mitigated by iterative testing, examining known relevant records, and maintaining a versioned search log.
• For systematic reviews, a minimum of three databases is recommended, supplemented by grey literature sources and reference list screening, to capture the dispersed nature of veterinary evidence, particularly for production animal medicine and wildlife health.

Veterinary research questions increasingly demand reproducible, transparent literature searches, whether for a systematic review, a clinical guideline, or a grant application. This article provides a practical framework for constructing and executing searches across the databases most relevant to veterinary science. It is written for veterinary researchers who need to move beyond simple keyword queries and toward structured search strategies that can be documented, peer-reviewed, and updated.

The procedural guidance that follows covers database selection, controlled vocabulary, search string construction, filters, and documentation. It does not cover critical appraisal of the retrieved records. The methods described align with the reporting expectations found in published systematic reviews, where authors routinely specify their databases, search dates, and inclusion criteria in the methods section. For example, reviews in fields as diverse as canine quality of life assessment and pulp regeneration consistently document their search strategies in this manner, and the same discipline applies to veterinary work.

At a Glance

The Logic of Structured Searching

A literature search is an experiment in information retrieval. Its outcome depends on the interaction between the query structure, the database's indexing system, and the vocabulary used by the original authors. A search that works in one database may fail in another because each platform applies different indexing rules and search syntax.

The core principle is recall versus precision. Recall is the proportion of relevant records retrieved from the total universe of relevant records. Precision is the proportion of retrieved records that are actually relevant. A search designed for a systematic review prioritizes recall, accepting that many irrelevant records will be retrieved. A search designed for a quick clinical update prioritizes precision. The researcher must decide which balance is appropriate for the question at hand before writing the first search line.

Controlled vocabularies are the backbone of reproducible searching. PubMed uses Medical Subject Headings (MeSH), a hierarchical thesaurus that assigns standardized terms to articles. CAB Abstracts uses the CAB Thesaurus, which includes agricultural and veterinary terminology that MeSH may lack. Searching with controlled vocabulary terms alone risks missing recently published articles that have not yet been indexed, so a well-constructed search combines controlled terms with free-text synonyms.

Database Selection and Coverage

No single database covers all veterinary literature. PubMed indexes biomedical journals, including many veterinary titles, but its coverage of agricultural science, production animal medicine, and some regional journals is incomplete. CAB Abstracts, produced by CAB International, has historically offered the most comprehensive veterinary coverage, particularly for livestock, wildlife, and parasitology. Web of Science and Scopus provide broad multidisciplinary coverage with powerful citation analysis tools, but their veterinary-specific indexing is less granular.

The choice of databases should follow the research question. A question about companion animal oncology may be well served by PubMed alone. A question about food safety or production animal disease transmission may require CAB Abstracts and regional databases. Systematic reviews typically search multiple databases to reduce the risk of missing relevant studies. The evidence context for a review of dietary protein and bone health, for instance, involved searching PubMed, Ovid Medline, and Agricola, reflecting the need to capture both biomedical and agricultural literature.

Controlled Vocabulary and Free-Text Searching

MeSH terms are applied by human indexers, which introduces a lag between publication and indexing. Free-text searching, using keywords in titles and abstracts, captures articles before indexing and catches terms that indexers may have assigned differently. The two approaches are complementary.

When constructing a free-text search, consider synonyms, British and American spellings, singular and plural forms, and abbreviations. For example, a search on canine osteoarthritis should include "osteoarthritis," "degenerative joint disease," and "DJD." Truncation, using an asterisk or other platform-specific symbol, captures word variants. The truncation symbol differs between platforms, and using the wrong symbol is a common source of error.

Field tags restrict searches to specific parts of the record, such as title, abstract, or author. A search restricted to title and abstract fields improves precision but may miss relevant records where the key term appears only in the full text or in the indexing terms. The decision to use field tags should be made deliberately and documented.

Filters and Limits

Filters restrict search results by publication type, date, species, language, or study design. PubMed offers a range of built-in filters, including publication type and age group. Systematic review searches often apply a publication type filter for randomized controlled trials or a date filter to capture a specific period.

Filters must be applied with caution. A filter that excludes non-English language articles may introduce language bias. A filter restricted to clinical trials will miss observational studies that may be the only available evidence for a given question. The decision to apply a filter should be justified in the search documentation. The ARRIVE guidelines, which specify the minimum information required for transparent animal research reporting, do not prescribe search filters, but they reinforce the principle that methods must be described in sufficient detail for replication.

Documenting the Search

Reproducibility requires complete documentation. For each database, record the platform, the date the search was run, the full search string, any filters applied, and the number of records retrieved. This documentation should be included in the methods section of the final report or stored as supplementary material.

The PRISMA guidelines, referenced in the evidence context of multiple systematic reviews, provide a framework for reporting the search process, including the number of records identified, screened, and included. Following this structure ensures that the search can be evaluated and replicated by other researchers.

Constructing the Search String

The search string is the executable expression of your research question. A well-formed string combines controlled vocabulary terms, free-text synonyms, Boolean operators, and field tags into a single query that the database can parse. The structure is identical across platforms, but the syntax differs.

Start with the PICO or PEO framework you refined during the planning phase. Each conceptual element becomes a separate block of terms. Within a block, join synonyms with OR. Between blocks, join with AND. This modular construction allows you to test each element independently and troubleshoot poor results.

For a question about postoperative analgesia in dogs undergoing ovariohysterectomy, the blocks might look like this:

(dog OR canine OR dogs)
AND (ovariohysterectomy OR spay OR "ovariectomy" OR "neutering")
AND (analgesia OR analgesic OR pain OR "pain management")

The first block captures the species, the second the procedure, the third the intervention or outcome. Each block uses both controlled vocabulary and free-text terms because no single term set captures all relevant records. The systematic review on dietary protein and bone health used this same logic across PubMed, Ovid Medline, and Agricola, combining subject headings with text words for each concept Wallace and Frankenfeld, 2017.

Boolean Operators and Nesting

AND narrows, OR broadens, NOT excludes. Use NOT sparingly because it can silently remove relevant records when the excluded term appears in an unexpected context. For example, excluding "cat" from a canine search will also exclude records about canine patients that mention feline comorbidities.

Parentheses control the order of operations. Most databases process NOT before AND before OR unless parentheses override this sequence. When in doubt, use parentheses around every OR cluster. This is not stylistic preference, it is the difference between retrieving 200 records and 2,000 irrelevant ones.

Phrase Searching and Truncation

Quotation marks force exact phrase matching. Use them for multi-word concepts such as "canine parvovirus" or "intervertebral disc disease". Be aware that phrase searching can miss records where the words appear in a different order or with intervening terms.

Truncation uses a wildcard symbol, usually an asterisk, to capture word variants. "Anesthe*" retrieves anesthesia, anesthesia, anesthetic, and anesthetic in databases that support this syntax. The limitation is that truncation also retrieves unintended variants such as "anesthesiologist". Review truncated results carefully. Some databases, notably CAB Abstracts on certain platforms, use different wildcard symbols, so check the platform help file before relying on truncation.

Database-Specific Syntax and Field Tags

Each database uses its own command language. PubMed uses [tiab] for title and abstract searching and [mh] for MeSH headings. CAB Abstracts on Web of Science uses TS= for topic searching and AB= for abstract fields. Ovid platforms use.ti,ab. and.sh. for subject headings.

Field tags restrict the search to specific parts of the record. Title and abstract searching is the default for most veterinary questions because it balances recall and precision. Searching only the title improves precision but misses records where the key concept appears only in the abstract or subject headings. The uveitis miRNA systematic review used a deliberately simple PubMed search with keywords in the central database, which was appropriate for a narrowly defined molecular topic with a small literature base Pockar et al., 2019.

PubMed

PubMed offers the richest set of filters for veterinary work. The MeSH database includes veterinary-specific headings such as "Dog Diseases" and "Cat Diseases" that automatically include narrower terms when exploded. The "species" filter under the Animals subset is useful but incomplete, so combine it with explicit species terms in your search string.

The systematic review on orthodontic retainers searched MEDLINE via OVID, PubMed, and the Cochrane Central Register of Controlled Trials, using a defined electronic and gray literature strategy Al-Moghrabi et al., 2016. This multi-database approach is standard for systematic reviews because single-database searching misses a substantial fraction of relevant records.

CAB Abstracts

CAB Abstracts is the most comprehensive database for veterinary and agricultural literature. Its controlled vocabulary, the CAB Thesaurus, includes terms for animal breeds, production systems, and diseases that MeSH does not cover. The database indexes conference proceedings, technical reports, and grey literature that PubMed excludes.

The canine quality of life rapid review searched both CAB Abstracts and PubMed, a combination that captured peer-reviewed instruments and the broader veterinary literature Belshaw et al., 2015. CAB Abstracts is particularly strong for production animal medicine, wildlife health, and parasitology.

Web of Science

Web of Science provides citation searching, which identifies records that cite a key paper. This forward citation chasing is valuable for finding newer work that builds on a foundational study. The pulp regeneration systematic review searched PubMed, EMBASE, and Web of Science, with the last search performed on 1 August 2021 Tirez and Pedano, 2022.

Web of Science does not use a controlled vocabulary, so searches rely entirely on free-text terms. This increases recall but reduces precision. Use the topic field (TS=) which searches title, abstract, author keywords, and KeyWords Plus.

Comparison of Core Databases

The choice of database changes with the question. A question about dairy herd mastitis control will find more relevant records in CAB Abstracts than in PubMed. A question about canine oncology clinical trials will find more in PubMed. A question about the history of veterinary education may require Web of Science for citation chaining plus CAB Abstracts for historical coverage.

Example Search Strings

The following examples illustrate complete search strings for common veterinary question types. Adapt the terms to your specific question and verify the syntax against the database help documentation.

Clinical Question: Canine Atopic Dermatitis Treatment

PubMed:

("Dermatitis, Atopic"[Mesh] OR "atopic dermatitis"[tiab] OR "atopic eczema"[tiab])
AND (dog[tiab] OR canine[tiab] OR "Dog Diseases"[Mesh])
AND (treatment[tiab] OR therapy[tiab] OR management[tiab] OR "Drug Therapy"[Mesh])

CAB Abstracts:

TS=("atopic dermatitis" OR "atopic eczema")
AND TS=(dog OR canine OR "Canis familiaris")
AND TS=(treatment OR therapy OR management)

Production Animal Question: Bovine Respiratory Disease

PubMed:

("Bovine Respiratory Disease Complex"[Mesh] OR "bovine respiratory disease"[tiab] OR BRD[tiab])
AND (cattle[tiab] OR bovine[tiab] OR calf[tiab] OR calves[tiab] OR "Cattle Diseases"[Mesh])
AND (prevention[tiab] OR control[tiab] OR vaccination[tiab] OR "Vaccination"[Mesh])

CAB Abstracts:

TS=("bovine respiratory disease" OR BRD OR "shipping fever")
AND TS=(cattle OR bovine OR calf OR calves OR "Bos taurus")
AND TS=(prevention OR control OR vaccination OR vaccine)

Wildlife or Population Question: Avian Influenza Surveillance

PubMed:

("Influenza in Birds"[Mesh] OR "avian influenza"[tiab] OR "bird flu"[tiab])
AND (surveillance[tiab] OR monitoring[tiab] OR "Population Surveillance"[Mesh])
AND (wild bird*[tiab] OR poultry[tiab] OR waterfowl[tiab])

CAB Abstracts:

TS=("avian influenza" OR "bird flu" OR "H5N1" OR "HPAI")
AND TS=(surveillance OR monitoring OR "active surveillance")
AND TS=("wild bird*" OR poultry OR waterfowl OR "Anseriformes")

Testing and Iterating the Search

Run each block separately before combining. This diagnostic step reveals which block is underperforming. If the species block returns too few records, check whether the database uses different terminology for the species, such as "Canis familiaris" in CAB Abstracts versus "dog" in PubMed.

Check a known relevant record. Retrieve it and examine which search terms it contains. The subject headings assigned to that record will show you the controlled vocabulary terms you may have missed. The systematic review on dog sled racing health screened 117 studies from Google Scholar and PubMed, then scrutinized reference lists to identify additional records Calogiuri and Weydahl, 2017. This reference list checking is a standard validation step.

Adjust the search iteratively. Add new synonyms, remove terms that retrieve irrelevant records, and test again. Document every change in your search log. The final search string should be reproducible by another researcher, which is the standard expected for systematic reviews and increasingly for narrative reviews.

Reporting the Search

The search strategy must be reported in sufficient detail for replication. State the database, the platform, the exact date of the search, and the complete search string. The ARRIVE guidelines for animal research require transparent reporting of methods, and the EQUATOR Network provides reporting checklists for systematic reviews that specify search reporting standards ARRIVE guidelines and EQUATOR reporting guidelines.

Include the number of records retrieved from each database and the number after deduplication. State any language restrictions and the justification for them. The pulp regeneration review excluded non-English articles, which is a common but potentially biasing decision Tirez and Pedano, 2022. Report this limitation explicitly.

Save the complete search history from each database session. Most platforms allow you to export the search history as a text file. This file becomes part of your research documentation and should be retained alongside the screening records.

Recognized Complications and Failure Modes

Structured searching fails in predictable patterns. The most consequential failure is the silent miss: a search that returns plausible results but omits a relevant subset of the literature. This occurs when the search string relies on a single vocabulary stream, such as MeSH terms alone, without complementary free-text terms. A search for "canine lymphoma" using only the MeSH term captures indexed articles but misses recent publications not yet indexed or those using novel terminology. Detect this early by running the search in a second database and comparing result sets. Discrepant retrieval between databases signals vocabulary or indexing gaps instead of database error.

A second failure mode is the uncontrolled filter. Applying a species filter, a publication date limit, or a language restriction without documenting the rationale narrows the search in ways that are difficult to reconstruct or defend. Filters applied at the interface level, instead of within the search string, are especially prone to being forgotten or inconsistently applied across databases. Record every filter as part of the search string itself, using the database's field tags, so the search is reproducible.

The third common failure is concept drift. As the search is iterated, terms are added to improve recall, and the query gradually shifts away from the original question. The result set grows, but its precision falls, and the searcher cannot identify which terms introduced the irrelevant records. Guard against this by maintaining a versioned log of each search iteration with the number of results retrieved at each step. When precision drops sharply between iterations, revert to the previous version and add terms one at a time.

Common Errors and Corrective Action

Less experienced searchers frequently overuse phrase searching. Quotation marks around multi-word terms such as "bovine respiratory disease" exclude records where the words appear in a different order or with intervening terms. Phrase searching is appropriate for established compound terms but should be combined with adjacency operators or free-text variants. Corrective action: run the phrase search and the equivalent Boolean search, then compare yields.

A second recurring error is the failure to explode controlled vocabulary terms. In PubMed, a MeSH term searched without explosion retrieves only records indexed to that exact term, missing all narrower terms beneath it in the hierarchy. The default in most interfaces is to explode, but this setting can be changed inadvertently. Verify the explode setting before running the search and document it in the search record.

Students also tend to conflate sensitivity and precision. A search that retrieves 2,000 records is not automatically better than one that retrieves 200. The appropriate balance depends on the question type. A systematic review requires maximum sensitivity, accepting low precision, while a quick clinical update requires the reverse. The PRISMA-style search strategies reported in published systematic reviews consistently document the sensitivity-precision trade-off in their methods sections, and reviewing these can calibrate expectations.

Limitations of the Evidence Base

The veterinary literature is indexed less completely than the human medical literature. CAB Abstracts provides strong coverage of production animal and wildlife topics, while PubMed is stronger for companion animal and translational research, but neither database indexes all relevant veterinary journals. Grey literature, including conference proceedings, theses, and government reports, is inconsistently captured. The search strategy used in a systematic review of dog sled racing health illustrates this point: the authors supplemented database searching with reference list scrutiny to capture studies missed by the primary search.

Expert opinion still differs on several practical points. Whether to search Google Scholar as a primary database or only as a supplementary tool remains contested. Google Scholar offers broad coverage and citation tracking but lacks the controlled vocabulary and export functionality of the major bibliographic databases. Some reviewers exclude it entirely on reproducibility grounds, while others use it for grey literature detection. Similarly, the role of the EQUATOR Network reporting guidelines in search reporting is settled for human medicine but less consistently applied in veterinary systematic reviews, where reporting standards remain variable.

Escalation and Referral

Most search problems resolve with iterative adjustment, but some situations warrant escalation. If a search for a systematic review or a clinical guideline continues to miss known relevant studies after multiple iterations, consult a librarian or information specialist with veterinary database expertise. These professionals can identify indexing quirks, construct complex search strings, and advise on database-specific syntax that is not documented in vendor help files.

Laboratory involvement is indicated when the search question involves diagnostic test accuracy or biomarker discovery. The systematic review of miRNA biomarkers in uveitis demonstrates that search strategies for molecular biomarkers require careful attention to nomenclature, since gene and protein names vary across databases and over time. A clinical pathologist or molecular biologist can clarify terminology before the search is run.

Regulatory reporting obligations arise when the search is conducted as part of an adverse event investigation, a pharmacovigilance review, or a notifiable disease investigation. The WOAH terrestrial animal health standards specify surveillance and reporting requirements that may mandate specific search documentation. In these contexts, retain the complete search record, including date stamps and database versions, because the search itself may become part of a regulatory file.

Frequently Asked Questions

How do I choose between PubMed and CAB Abstracts when my institutional access limits me to one database?

PubMed excels for biomedical and translational questions, with strong coverage of comparative medicine and zoonoses. CAB Abstracts provides superior coverage of production animal medicine, parasitology, and international veterinary literature. If you can access only one, select based on your clinical domain. For companion animal oncology or genetics, PubMed is usually sufficient. For herd health, food safety, or exotic species, CAB Abstracts is preferable. When neither database fully covers your question, supplement with Google Scholar for grey literature and regional publications, then screen reference lists of included studies. The EQUATOR Network's reporting guideline library can help you identify which study types your search should target before you begin.

What is the minimum number of databases I should search for a systematic review in veterinary medicine?

Search at least three databases for a systematic review. The PRISMA-based search strategies used in published veterinary systematic reviews commonly include PubMed, Embase or CAB Abstracts, and a third source such as Web of Science. Veterinary-specific evidence is dispersed across biomedical, agricultural, and zoological databases, so a single database will miss substantial relevant literature. For rapid reviews, two databases plus reference list screening may suffice, but you must state this limitation in your methods. If your review spans wildlife or exotic species, add a specialist database such as Zoological Record. Document your database selection rationale in the review protocol.

How should I document my search when collaborating with a librarian who will run the final searches?

Provide the librarian with your research question in PICO or PECO format, your provisional search string, and your inclusion criteria. Specify which databases you have already searched and the date of those searches. Ask the librarian to translate your string into each database's syntax, run the searches, and export results with full search histories. Request that the librarian document the exact date, database platform, and interface version, since these affect reproducibility. The ARRIVE guidelines for reporting animal research emphasize transparent methods, and your search documentation should allow another researcher to replicate your results exactly. Agree on a citation manager format before export to avoid duplicate handling.

How do I adapt a search strategy designed for dogs to a production species such as cattle?

Replace species-specific terms in both controlled vocabulary and free-text fields. For dogs, you might use MeSH terms such as "Dog Diseases" and free-text terms like "canine" or "dog." For cattle, use "Cattle Diseases" and terms including "bovine," "cattle," "cow," and "beef cattle" or "dairy cattle" as appropriate. Add production system terms such as "feedlot" or "pasture-based" if relevant. Be aware that production animal literature is more likely to appear in CAB Abstracts than PubMed. Consider adding breed-specific terms only if your question requires them, since breeds vary by region. The MSD Veterinary Manual can help you identify species-specific terminology and disease names before constructing your search.

What should I do when a search returns zero results for a well-recognized clinical condition?

First verify your spelling and controlled vocabulary terms. Search for the condition by its synonyms, including historical names and regional variations. Check whether the condition is indexed under a broader category, such as a class of diseases instead of a specific entity. Remove filters one at a time, starting with date limits and language restrictions. Search the reference lists of review articles on related topics. If you still find nothing, consider whether the condition is primarily described in non-indexed sources such as conference proceedings or practice bulletins. The AVMA practice resources may point to professional guidance documents that are not indexed in bibliographic databases.

How do I explain the limitations of my literature search to a referring veterinarian or client?

Be direct about what your search can and cannot establish. State the databases searched, the date range, and the inclusion criteria. Explain that a systematic search identifies available evidence but does not guarantee that evidence exists for every clinical question. For example, a search on a rare condition may yield only case reports, which cannot establish treatment efficacy. Acknowledge that absence of evidence differs from evidence of absence. The WOAH terrestrial animal health standards can provide internationally recognized definitions for disease surveillance and reporting that may be relevant when discussing population-level questions. Offer to share your search strategy so the client or referring veterinarian can assess its scope.

Related Clinical & Scientific Guides

• Conducting Systematic Reviews of Veterinary Diagnostic Test Accuracy
• Bias in Veterinary Research: Types, Sources, and Mitigation
• Cluster Randomized Trials in Veterinary Research: Design and Analysis

References and Further Reading

• Dietary Protein Intake above the Current RDA and Bone Health: A Systematic Review and Meta-Analysis.. 2017.
• The effects of fixed and removable orthodontic retainers: a systematic review.. 2016.
• Health challenges in long-distance dog sled racing: A systematic review of literature.. 2017.
• Regeneration of the Pulp Tissue: Cell Homing versus Cell Transplantation Approach: A Systematic Review.. 2022.
• Quality of life assessment in domestic dogs: An evidence-based rapid review.. 2015.
• MiRNA as biomarker for uveitis - A systematic review of the literature.. 2019.
• ARRIVE Guidelines 2.0 for Reporting Animal Research. PLOS Biology, 2020.
• EQUATOR Network Reporting Guidelines. EQUATOR Network.
• MSD Veterinary Manual, Professional Edition. MSD Veterinary Manual.

Related Articles

• Applying Competing Risks Analysis in Veterinary Research
• Appraising Diagnostic Accuracy Studies in Veterinary Medicine
• The ARRIVE Guidelines for Animal Research: Implementation and Compliance
• Assessing Risk of Bias in Veterinary Randomized Trials
• Bias in Veterinary Research: Types, Sources, and Mitigation

This article is educational professional reference material for veterinary audiences. It is not a substitute for veterinary diagnosis, individual clinical judgment, current product labeling, or applicable regulatory requirements.