Hello,

There is a new PubMed search interface at

http://web.ncbi.nlm.nih.gov/entrez/query/static/clinical.shtml

It applies search filters, based on the work of Haynes RB, chief of the
Health Information Research Unit at McMaster.
http://www.medicalpost.com/mpcontent/article.jsp?content=20040523_111944_5200
A four-year, $500,000 US ($685,000 Cdn) project. Seems a bit expensive to me.

[I agree. It looks like it should have cost well under $10K. This is typical
government inflation. But then the individuals involved have no need to be
concerned since it does not come from their pockets and there is always more
money available to steal (tax) from the taxpayers. --Paul]


I have tried it and what it does is to apply complex boolean operators

e.g. for strategy=therapy type of search

(search terms here) AND (randomized controlled trial[Publication Type] OR
(randomized[Title/Abstract] AND controlled[Title/Abstract] AND
trial[Title/Abstract]))

At least it looks interesting to learn search strategies. Time and use will
tell if this interface is more useful than the regular one.

[Thanks, Riso, for making this this tool known. Filters have advantages - for
instance, reducing 500 results to 50. But one must use them very carefully or
risk the possibility of losing worthwhile information. I'll let Paul comment
further on this aspect. **Kitty]


Regards,

Riso

[I fail to see anything gained here. Boolean operators have always been
available in PubMed searches. I use AND and OR often and when I need to exclude
some things I also use NOT. Parentheses are used to nest these operators. The
major ability lacking in PubMed searches is the lack of a general string search.
You can only search for phrases (strings of more than one word) when they are
common and already entered in the MESH search phrases list. For uncommon phrases
a search does not work. I complained to PubMed about this lack many years ago
but they have not changed. All other major Internet search engines that I am
aware of do allow string searches which are done by simply placing quotes around
the string.

Examining the interface you have given suggests that it is merely intended as a
time saver for busy clinicians to get easily to the heart of what they want
without having to learn or enter the terms and boolean operators themselves.
OTOH, one wonders just how much off the beaten track, non-mainstream, but
important results will be filtered out and remain unseen because of the use of
such time saving devices. Personally, I have no use for such things which shield
me from the natural complexity of the real world. --Paul]