tags: PubMed, PubMed Central, medical informatics, bioinformatics, finding scientific articles
I meant for this to be a three part series, but in part II, I learned that one more experiment had to be done. I had to know if the articles I found in PubMed Central were the same articles that I found in PubMed.
Part I and part III cover the background and my favorite method. Now, we're going to find out if my favorite method is really enough.
In other words, I had this kind of problem (shown in the diagram) and I just had to know which case was correct:
![i-b7c3f0f45492563ca28f4f124758029a-pubmed-and-pmc.gif](http://scienceblogs.com/digitalbio/wp-content/blogs.dir/460/files/2012/04/i-b7c3f0f45492563ca28f4f124758029a-pubmed-and-pmc.gif)
The method:
To test this, I did a PubMed search with term "cancer," as before, and limited the search to free, full, text.
Then, I clicked the Preview/Index tab, opened the Filter field, and selected either the pubmed pmc free filter or the pubmed pmc filter. (Both filters are shown in the image below.)
![i-5bc63bb0334fe87b6f3620a4ae21a22d-filter.gif](http://scienceblogs.com/digitalbio/wp-content/blogs.dir/460/files/2012/04/i-5bc63bb0334fe87b6f3620a4ae21a22d-filter.gif)
Then, I clicked the AND button to add that term to my query. (Using the AND, OR, or NOT buttons works wonderfully, because everything is properly formattted with quotes and brackets.)
My results:
In part II, I found 220,219 articles on cancer in PubMed and 171,702 articles in PubMed Central. In today's experiment, I found that only 52,160 articles (a little more than a third, were shared between the two databases).
![i-3f6fa639df6dc86aa841bf2bf1a69e68-history.gif](http://scienceblogs.com/digitalbio/wp-content/blogs.dir/460/files/2012/04/i-3f6fa639df6dc86aa841bf2bf1a69e68-history.gif)
In other words, this diagram shows the correct situation.
![i-3595c7ede28005bc6a396b223b0940fb-shared_articles.gif](http://scienceblogs.com/digitalbio/wp-content/blogs.dir/460/files/2012/04/i-3595c7ede28005bc6a396b223b0940fb-shared_articles.gif)
What's the take home message?
PubMed Central contains articles that are not available in PubMed (with limits). So, to get as many articles as you can, you do need to search both databases. And, if that doesn't work, my commenters (here and here) have left a number of excellent suggestions!
Read the whole series:
- part I A day in the life of an English physician,
- part II Comparing different methods,
- part III My new favorite method,
- part IV One last experiment
- Log in to post comments
Actually, my understanding is that PubMed Central *is* a strict subset of PubMed.
But, when you search PubMed Central, you are searching the fulltext, and so you find some articles you don't find with a search of PubMed abstracts
When you search PubMed, on the other hand, you will find additional articles which are not present in PubMed Central, even though they may be free on publishers sites (unfortunately some publishers, while making content free on their own site, do not allow PubMed Central to host a copy of it).
You're quite right that the ultimate end result is that you do need to separately search both databases in order to get as complete as possible a list of free articles - which is a bit of a pain.
Hi Matt,
Thanks for explaining why the searches give different results.
I thought that PubMed Central was a subset of PubMed as well, at least until I did the experiment. PubMed Central has lots of editorials - like from the British Medical Journal - that I just didn't see when I searched PubMed.