Search Strategy for Systematic Review in PubMed: A High Sensitivity Checklist for a Comprehensive Search

Systematic reviews live or die by the quality of their literature search. If your goal is to make a clinical decision quickly, you can afford a narrow, highly specific query. But if you’re conducting a systematic review, the standard is different: you need a high sensitivity strategy designed to capture all potentially relevant studies—even if it means screening more results. This checklist walks you through building a comprehensive search in PubMed so you minimize the risk of missing key evidence.

Why “High Sensitivity” Matters in Systematic Reviews

A systematic review aims to summarize the totality of evidence. Missing even a few relevant papers can bias conclusions, affect meta-analysis results, and undermine credibility. That’s why systematic review searches prioritize:

• Recall (sensitivity): retrieving as many relevant records as possible
• Transparency: documenting what you searched, when, and how
• Reproducibility: allowing others to replicate the search strategy

In practice, a high sensitivity search often uses broader synonyms, controlled vocabulary (MeSH), and fewer restrictive filters than a typical clinical search.

Before You Search: Prepare a Clear Research Question

High sensitivity doesn’t mean searching randomly. Begin with a structured research question. Many reviewers use PICO:

• Population
• Intervention/Exposure
• Comparator
• Outcomes

For some topics, alternative frameworks (PEO, SPIDER) may be better, but the principle remains: you need well-defined concepts that can be translated into search terms.

The Comprehensive PubMed Checklist (High Sensitivity Search)

Use this step-by-step checklist to build a robust search strategy for systematic review work in PubMed.

1) Identify Your Core Concepts (Usually 2–4)

Break the question into concepts you can search independently. In many systematic reviews, you’ll focus primarily on the population and intervention/exposure. Outcomes are sometimes excluded from the search to improve sensitivity (because outcomes are described inconsistently across papers).

Tip: If you include too many concepts, you may unintentionally exclude relevant papers that don’t mention every concept in the title/abstract.

2) Collect Synonyms, Variants, and Related Terms

For each concept, gather:

• Synonyms (e.g., “myocardial infarction” and “heart attack”)
• Spelling variants (American/British English)
• Acronyms and full forms
• Older terminology (terms may change over time)
• Broader and narrower terms when appropriate

High sensitivity strategies typically use long OR lists per concept to maximize recall.

3) Map Terms to MeSH (Controlled Vocabulary)

PubMed uses MeSH (Medical Subject Headings) to index articles. MeSH helps catch papers that use different wording. For each concept:

• Find the appropriate MeSH term(s)
• Check entry terms (built-in synonyms)
• Review the MeSH scope note to confirm fit

High sensitivity best practice: use both MeSH terms and free-text terms (title/abstract keywords). Not all records are indexed immediately, and some citations (e.g., very recent entries) may lack MeSH.

4) Combine MeSH and Free-Text for Each Concept

For each concept, combine synonyms with OR. Example structure (illustrative):

• Concept A: (MeSH term) OR (title/abstract synonyms)
• Concept B: (MeSH term) OR (title/abstract synonyms)

Then combine concepts with AND to narrow to the intersection of topics.

5) Use Field Tags Thoughtfully (Without Over-Restricting)

Common high sensitivity field tags include:

• [Mesh] for MeSH headings
• [tiab] for terms in title/abstract
• [tw] for text words (broader than tiab in some contexts)

Tip: Overusing narrow fields can reduce sensitivity. If you aren’t sure, keep the approach inclusive and test results.

6) Pay Attention to Phrase Searching and Truncation

PubMed handles phrases and term mapping in specific ways. To stay sensitive:

• Use quotes for exact phrases only when needed (too many quotes can reduce recall)
• Use truncation (*) cautiously to capture word variants (e.g., “random*”)
• Check whether PubMed’s automatic term mapping already covers your intended variations

The key is to test and confirm you’re retrieving known relevant articles.

7) Avoid Unnecessary Limits Early (Language, Date, Humans)

Systematic review standards often recommend avoiding restrictive filters unless justified in the protocol. Common pitfalls include:

• Applying language limits without rationale (can introduce bias)
• Adding date limits prematurely (may exclude foundational studies)
• Using “Humans” or “Adult” filters too early (may miss mixed-population indexing)

If limits are necessary, document them clearly and apply them consistently across databases.

8) Be Careful with Study Design Filters (Use Validated Filters When Needed)

Sometimes a systematic review targets specific study types (e.g., RCTs). In that case:

• Use validated PubMed search filters when possible
• Prefer sensitive versions of filters if missing studies is a concern
• Test the filter against a set of known relevant studies (a “gold set”)

Remember: study design terms are often inconsistently reported, so filters can reduce sensitivity.

9) Test Your Strategy Against “Known” Key Papers

A practical quality check is to identify a handful of sentinel articles (from experts, prior reviews, or preliminary searches) and confirm your search retrieves them. If not, revise:

• Add missing synonyms
• Adjust MeSH terms
• Relax overly strict fields or phrases

This step helps ensure your comprehensive search in PubMed performs as intended.

10) Document Everything for Transparency and Reproducibility

For a systematic review, your search methods should be auditable. Record:

• The full PubMed query string
• Date searched
• Any limits/filters used
• Number of results retrieved

Consider saving the search in PubMed and exporting the strategy to your review’s appendix.

Common Mistakes That Reduce Sensitivity in PubMed

• Relying only on MeSH: misses unindexed and newly published records
• Including outcomes in the main query: can exclude relevant studies that don’t mention outcomes in the abstract
• Too many narrow phrases in quotes: reduces variability and recall
• Premature filters: language/date/population filters can remove important evidence
• Not iterating: high sensitivity searches require testing and refinement

FAQ: High Sensitivity Searching for Systematic Reviews in PubMed

Should I use PubMed filters like “Clinical Trial” for a systematic review?

Only if your protocol requires specific study designs and you’re using a validated, sensitivity-oriented filter. Otherwise, it’s often better to keep the search broad and screen results during study selection.

Do I need both MeSH and keywords?

Yes, in most systematic reviews. MeSH improves consistency across terminology, while free-text keywords capture records that aren’t indexed yet or use unexpected wording.

How do I know if my search is “comprehensive” enough?

No search is perfect, but you can increase confidence by (1) using both MeSH and free text, (2) including robust synonym lists, (3) testing against known key studies, and (4) documenting and peer-reviewing the search strategy.

Conclusion

Building a search strategy for systematic review work in PubMed is fundamentally different from everyday clinical searching. A high sensitivity approach prioritizes recall, uses both MeSH and free-text terms, avoids unnecessary limits, and relies on iterative testing to ensure you don’t miss critical evidence. Use the checklist above to create a transparent, reproducible, comprehensive search in PubMed—and strengthen the foundation of your systematic review.