how to search pubmed for medical information

How to Search PubMed as a Patient


Someone told to read up on migraine in children can learn how to search PubMed for medical information in stages: start with plain keywords, add filters only when results overwhelm, then read abstracts before chasing full text.

Quick answer: type two or three key words into the PubMed box, join them with AND or OR, narrow with the sidebar filters, and judge each result from its abstract. Use the Free Full Text option when you want to read the whole article. Each step below adds effort only when the previous one has not been enough.

This page is one part of the wider guide on how to find reliable health information online. PubMed suits readers who want the original research behind a headline or a clinician’s advice.

What you are actually searching

PubMed is a free database provided by the US government, according to the University of Connecticut library’s PubMed guide. The names PubMed, MEDLINE and PubMed Central are often blurred together, but they describe different things.

Name What it is Planning use
PubMed The free search tool and database you type into Start here for every search
MEDLINE A dataset of over 36 million citations from the biomedical literature, which makes up most of PubMed Explains why results are mostly citations and abstracts, not full articles
PubMed Central An archive of full text journal articles that PubMed also includes Go here when you want to read a whole article

The descriptions follow the University of Connecticut library guide to PubMed.

Step one: search with plain keywords

The simplest search is a few ordinary words. The Belmont University library’s basic search page explains that PubMed looks for the keywords you enter in the article title, abstract and subject headings.

A general search page under a magnifying glass; PubMed’s own search box works the same way, taking a few keywords typed in one line.
  1. Name the condition in the words a clinician would use, such as “migraine”.
  2. Add the one detail that matters most to your situation, such as “children”.
  3. Press search and scan the titles before opening anything.

Keep the first search short. A whole sentence gives PubMed more to misread, and a short pair of keywords is easier to adjust when the results look wrong.

Step two: combine terms and let MeSH help

If plain keywords return too much or too little, join them deliberately. Put quotation marks around a phrase, and use AND, OR and NOT between terms. The University of Connecticut guide warns that commands are case sensitive, so these words must be typed in capitals.

  • AND narrows. “migraine” AND “children” keeps only results that match both ideas.
  • OR widens. “children” OR “adolescents” catches articles that use either word for the same group.
  • NOT excludes. Use it sparingly, because it can remove useful articles along with the unwanted ones.

How MeSH terms work

MeSH is a controlled vocabulary. According to the University of Connecticut guide, PubMed’s creators, the National Library of Medicine, select only one MeSH for each medical topic. Searching by that MeSH term finds the articles on the topic even if they used different wording.

Patients rarely need to type MeSH terms by hand. The same guide notes that PubMed’s default search uses Automatic Term Mapping to make an educated guess at what you mean, which often links everyday words to MeSH headings behind the scenes. If a result looks off-topic, a different everyday word is usually the quickest fix.

A notebook beside the screen suits this stage: jot down each wording tried and which gave useful results, so the same searches are not repeated.

Step three: narrow with filters

Filters are for results that are still too many after better keywords. The University of Connecticut guide says the sidebar limits restrict results by publication dates, age group, species, article types and language.

  • Publication dates. Helpful for favouring newer work when a condition’s treatment has changed.
  • Age group. Useful when the condition behaves differently in children and adults, as in the migraine example.
  • Species. Keeps human studies in view and sets aside animal research.
  • Article types. Lets you choose the kind of paper you want to read, such as a trial report or a review.
  • Language. Removes articles you cannot read.

The guide also describes field tags, which are placed after a term to say where PubMed should look, for example in the title or the abstract. Searching a term only in titles is a stricter way to find articles that are mainly about it.

Apply a single filter at a time and check how the list changes. Stacking several at once can hide a useful article without any sign of what was removed. If a trial result makes you consider taking part in research yourself, read the questions to ask before joining a clinical trial before deciding.

Step four: read the abstract

Opening a result shows the title, authors and abstract. The abstract is a summary of the paper, and it is also the text PubMed searches along with the title and subject headings. That means an abstract is often enough to tell whether an article deserves more of your time.

Dense text on a page stands in for an abstract here; reading it slowly, line by line, beats skimming for a headline result.

Read the abstract from a patient’s viewpoint with these questions in mind:

  • Who was studied? Compare the people described with your own age and circumstances.
  • What was done? Note whether the paper tests a treatment, reviews other studies or describes a single case.
  • What did it conclude? Treat one abstract as one piece of evidence, not as a verdict on your care.

An abstract leaves out detail, so bring promising findings to your clinician rather than acting on them alone. For how to weigh the evidence in the full paper, see how to read a medical journal article.

Step five: get the full article for free

Many results show only the citation and abstract. PubMed Central is an archive of full text journal articles, and PubMed includes it, so some results link straight through to a readable article. The University of Connecticut guide points to a simple method of finding free full text: the Free Full Text option among the sidebar filters.

Full text means the complete paper, which is far longer than the abstract that most results show.

Switch the filter on after the search is already giving relevant results, not before. Applying it early can leave a short list that misses the best articles. When a paper you want is not free, its abstract can still guide a conversation with your clinician.

Checkpoints and when to stop

Stage Action Planning use
Keywords Try a condition plus one key detail Move on when the titles are broadly on topic
Combining Use AND, OR and quotation marks Adjust wording when results are off-topic, rather than adding filters
Filters Limit by age group, article type or dates Use only when the list is still too long
Abstract Check who was studied and what was concluded Stop on articles that do not match your situation
Full text Use the Free Full Text option Read in full only the few papers that matter

The stages draw on the University of Connecticut PubMed guide and the Belmont University basic search page.

Stop searching when new queries keep returning the same articles. At that point, write down the two or three papers that seem most relevant and take them to your clinician instead of searching further.

Frequently asked questions

Is PubMed free to use?

Yes. PubMed is a free database provided by the US government.

Does PubMed search the text inside articles?

No. According to Belmont University’s library, PubMed does not search the full text of an article, only the title, abstract and subject headings. A word that appears only in the body of a paper may not match.

How do I look for a specific author?

Enter the last name followed by initials, without periods or commas, as the Belmont University guide describes.

Do I need to know MeSH terms before I start?

No. Everyday words work for a first search, since PubMed’s default search makes an educated guess at the matching subject headings.