| 
  • If you are citizen of an European Union member nation, you may not use this service unless you are at least 16 years old.

  • You already know Dokkio is an AI-powered assistant to organize & manage your digital files & messages. Very soon, Dokkio will support Outlook as well as One Drive. Check it out today!

View
 

How to search PubMed (part 1)

Page history last edited by jon.brassey@... 16 years ago

This is an ambitious chapter.  I've decided to call it part 1, as there may well be other parts to this...

 

PubMed is huge and it can be difficult to use (click here to go to the homepage).  Below is not exhaustive, just a few hints and tips to improve your searching.  A few of my suggestions will no doubt upset some purists, but then I doubt they're reading this!

 

The simple way to search PubMed is to simply add the keywords you're interested in.  Go to PubMed (click here, but open in a new window).  I'll try the search prostate cancer.  Simply add prostate cancer into the search box and press 'Go'.  I get 64,747 results - way too many.  So what can we do to make that more managable?

 

  • Restrict to the last 5 years (or 1 year or ten years).  My view is that if a paper is important it should be incorporated into secondary reviews within 5 years.
  • I only want to see abstracts.
  • I'm interested in humans.

 

So how do I restrict the results?  I click on the tab 'Limits' (under the search box) and you're confronted with a menu of restrictions, I go and tick the box 'abstracts', I select 5 years from the drop-down 'Published in the last:' and finally tick the 'Humans' box.  If you're a bit confused click here to see how the Limits screen should look.  After applying all those changes I scroll to the bottom of the page and press 'Go'.

 

The result - 17,797 - a significant reduction but still way too many so we need to further clarify.

 

At the moment the search looks for prostate cancer, but that might be a passing reference, PubMed doesn't differentiate from a simply text search.  So a few further tips:

 

Restrict to title words. Go back into 'Limits' and scroll to the bottom limit 'Tag terms' and change the 'Default tag:' to 'Title' and press 'Go'.  This then only returns documents with prostate cancer in the title (as well as having the other restrictions highlighted above).  By restricting to title only we get down to 9,236 records

 

Use MeSH.  MeSH is strange.  MeSH is a controlled vocabulary.  In reality this means that it only has one term for a given condition, intervention etc.  In the case of prostate cancer the MeSH term is prostatic neoplasms, so helping to confuse!  So how do we know the appropriate MeSH term?  We follow the link on the left of the PubMed screen to 'MeSH Database' (click here).  Follow the link, type in prostate cancer and press 'Go' and the first term it returns is - thankfully - prostatic neoplasms. Click on this link and you get more information (If you're confused click here to see what I'm looking at).  So how can we use this to reduce the results.  I use it for two mains ways. 

 

Firstly, I tick the box 'Restrict Search to Major Topic headings only', this tells PubMed that we only want records where prostate cancer is a major topic.  I see it as being similar to restricting the records to title only.  After ticking the box I go to the right-hand drop-down menu below the search box and select 'Search box with AND'.  A new section appears which should have "Prostatic Neoplasms"[Majr], if that is the case press the 'Search PubMed' button.  This time you get 48,685, but if you still have the limits (abstracts, 5 years and humans) you'll get 13,965 results. 

 

Secondly, I go to the prostatic neoplasms MeSH screen (click here to see what I'm looking at) and I'm now looking at the load of tick boxes under the title 'subheading'.  This is where I can focus the search, it gives me lots of related terms.  I can look through the list and I may decide that I'm only really interested in the genetics of prostate cancer, so I tick the box genetics and go to the right-hand drop-down menu below the search box and select 'Search box with AND'. A new section appears which should have "Prostatic Neoplasms/genetics"[Mesh]. However, it makes sense to also press the 'Restrict Search to Major Topic headings only' (see the paragraph above for details) before selecting 'Search box with AND'.  Once you've done this you press 'Search PubMed' and (with the normal restrictions (5 years, human, abstracts) I get 1,424 results (with 143 reviews).  This is still too many - so what now?

 

Think about other terms of interest.  You've told the system that you're interested in the genetics of prostate cancer, but anything else?  Are you interested in a link with breast cancer, twin studies, prognosis.  I decide that I want to see if there is any link between breast cancer and prostate cancer.  So in the search box I type breast cancer, press 'Go'.  This returns loads of results, but we haven't combined the results yet.  We now press the 'History' tab (below the search box) and we get a list of our previous searches (click here to see what I see).  In my search you'll see that the 'Limits' box is ticked, the "Prostatic Neoplasms/genetics"[Majr] has the term #70 next to it (that's just in my example, yours will have a different number associated with it).  while by breast cancer search has the term #71 next to it.  To combine them is relatively easy, in the search box type #70 and #71 and press 'Go' - you now get a pretty managable 69 results.

 

I appreciate the above example is constructed, but the methods are useful.  The main thing is to figure out how to get a managable number of results.  This means using the tools available in PubMed (limits, MeSH etc) as well as being clear exactly what you're interested in (so that you can combine terms). 

 

Other tips that might prove useful are:

 

  • Related articles - if you find a document that you're really interested in you can view the related articles, these are articles that are lingistically similar to the article you're viewing.  These related articles can be very useful and are often different from the results found in your main search.
  • Clinical Queries - these can be very useful and are dependent on special restrictions (or filters, or hedges) devised by the clever people at the Health Information Research Unit at McMaster University.  These are special filters designed to pick out articles that are pertinent to the particular filter you use.  If you go to the Clinical Queries are (the link is on the left-hand side of the screen, or follow this link).  I most frequently use the therapy filter and this is designed to look for randomised controlled trials (generally the methodology of choice to test a therapeutic intervention).  So, I'm interested in the use of clopidogrel in stroke.  So I enter clopidogrel and stroke into the search box and press 'Go'.  I've kept my usual restrictions (five years, humans, abstracts) and this returns 55 results - pretty managable.  These can further be restricted using the methods above (restrict one or both terms to appearing in the title, using major MeSH headings etc.

 

One thing for sure is that PubMed is not easy.  Even the best information specialists using lots of experience and time will miss pertinent results.  The skills and experience mean they miss less than inexperienced searches, but it'd be the height of arrogance to suggest that they do a foolproof search! 

 

The above is a collection of thoughts which will be added to in time.  However, if you read this and struggle, or have suggestions PLEASE let me know (use the 'Contact the owner' link).

Comments (0)

You don't have permission to comment on this page.