Posts Tagged ‘search filters’

Articles

Clinical queries & custom filters in PubMed

In Evidence-Based Medicine,How to work better,Web 2.0 & all that,Website reviews on July 7, 2010 by Alan Lovell Tagged: , , , , , , ,

PubMed has a new clinical queries page, apparently. To be honest I used the old one so infrequently that it could have changed six months ago and I never would have noticed. I tend to use PubMed for quick and not so dirty searches of the literature but if I’m doing a “real search” I use Ovid because of its slight value-add functionality plus it has our company’s access to Embase. Perhaps because of this access to Ovid I’ve never really paid all that much attention to the development of PubMed which, particularly over the last six months or so, seems to be whizzing ahead.

Anyway, the clinical queries page is quite fun. You put in your term and get results for “clinical study categories”, “systematic reviews” (not really so much “systematic reviews”, more “aggregate research” or “tertiary research” or similar), and “medical genetics”; you can then click “see all” for, well, seeing all, and there are drop down menus for whether you want therapy or etiology etc, and broad or narrow filters. It’s easiest just to play with it. I like the fact that at the bottom of each list if you click the word “filter” it will show you the actual search string being used to filter your results (for therapy or etiology, broad or narrow etc) and the sensitivity/specificity scores of  said filters (precision would be helpful too) along with the reference to the original paper in which the filters were based. All nice and transparent, and helpful if you wish to translate e.g. the prognosis filter to use in another database.

Of course though many of us want to add our own favourite filters. I always had a rough idea you could do this but had never bothered to really look into it, but it’s an absolute breeze. Fortunately I don’t have to describe how to add your own filters as Laika has already done such a good job of it (with screenshots and everything), and you can now add up to 15 of your own favourite search strings. Not sure what to add? Your friends at CRD/InterTASC will help you out. Once you get started you’ll  be having such a whale of a time that you’ll be looking for excuses to do quick and, as mentioned before, not so very dirty searches in PubMed for all and sundry – dragging people in from the corridor – that sort of stuff. I’m sure you’ve been doing this for years, but it’s all new to me.

I’m really beginning to like PubMed. A big thank you to the US taxpayer.

Articles

LILACS–worth another look?

In Evidence-Based Librarianship on November 7, 2008 by Danielle Tagged: , , , , , , ,

Whiting et al have done an elegant little study on database coverage for studies of test accuracy.  Not surprisingly, they found that Medline had 80% of studies cited by 8 preselected systematic reviews on this topic.  Surprisingly, a database called LILACS (Latin American and Caribbean Literature on the Health Sciences) turned up 6 studies that were not found anywhere else.  I do not know the total number of studies they looked up, but this seems significant.  It is significant to me, as I have never had the opportunity to search LILACS and I’m now wondering what has been overlooked.

Their searches missed 8 studies due to their not being indexed in any databases, and another 22 studies due to the fallibility of the search strategies they had used.

So if you ever get that anxiety-provoking question, ‘Why weren’t these studies picked up in your search?’, then consider the possibility that you may not have really needed that study type filter.  Indexing still has a ways to go in the area of study types.  Most of the time, I simply do not use them.

Or, consider that some papers are simply not indexed in electronic databases.  Hard for many folks to believe, but true.

The day that scientific literature becomes easy to locate will be the day we all start thinking alike and using the same words–I don’t think it will come.  What do you think?