The discussion below has 3 themes:
1. Authors need to apply guidelines when reporting studies
2. The developers of the search interface for PubMed need to add proximity (adjacency) searching
3. Combinations of terms from filter discussions are provided
1. Authors need to apply guidelines when reporting studies
I came across a systematic review last week that none of these searches would have found which is incredibly frustrating. I wish PubMed software could handle proximity searching! Almost as good, I wish authors would be consistent in what they call systematic reviews. For example, I found these:
systematic evidence reviews
systematic narrative review
systematic research review
systematic 25-year review
systematic scoping review
systematic, critical review
systematic qualitative review
systematic evidence review
Are they systematic reviews? Or did the authors know what they should call them? These were are found in the titles of articles in which I did a proximity search, then excluded everything that my very broad SR search filter finds.
2. The developers of the search interface for PubMed need to add proximity (adjacency) searching
Here's one example that the search filters don't find, even my far-too-broad search that I put together for PubMed.
Screening and treatment for lipid disorders in children and adolescents: systematic evidence review for the US Preventive Services Task Force. [Review] [266 refs]
Haney EM. Huffman LH. Bougatsos C. Freeman M. Steiner RD. Nelson HD.
Pediatrics. 120(1):e189-214, 2007 Jul.
The phrase "systematic review" appear anywhere in the abstract.
In Ovid, I can correct very easily by adding something like:
(systematic adj3 review*).ti,ab.
It's possible to do it in Ebsco as well:
Systematic n3 review*
Unfortunately, there is no way to do proximity searching in PubMed. The closest is: (Systematic[ti] AND review[ti])
This ensures that both words are in the title. However, what I don't like is that it is not possible to search for both words in the abstract. Why not do:
(Systematic[tiab] AND review[tiab])
There's nothing that puts each word in the same field.
3. Combinations of terms from filter discussions are provided
Using (Systematic[tiab] AND review[tiab]) increases the number of hits by over 9,000 from one of my other filters:
Method A: Most broad: 93,900 hits (uses (systematic[tiab] AND review*[tiab]))
Meta-Analysis[ptyp] OR "systematic literature"[tiab] OR (systematic[tiab] AND review*[tiab]) OR meta-analys*[tiab] OR "cochrane database syst rev"[Journal] OR "cochrane database of systematic reviews online"[Journal] OR "research synthesis"[tiab] OR "research integration"[tiab] OR "medline"[tiab] OR pubmed[tiab] OR psyclit[tiab] OR psycinfo[tiab] OR cinahl[tiab] OR "data synthesis"[tiab]
Method B: limits search for each term systematic & review to just the title: 85,433 hits (uses (Systematic[ti] AND review[ti])):
Meta-Analysis[ptyp] OR "systematic literature"[tiab] OR (systematic[ti] AND review*[ti]) OR "systematic review*"[tiab] OR meta-analys*[tiab] OR "cochrane database syst rev"[Journal] OR "cochrane database of systematic reviews online"[Journal] OR "research synthesis"[tiab] OR "research integration"[tiab] OR "medline"[tiab] OR pubmed[tiab] OR psyclit[tiab] OR psycinfo[tiab] OR cinahl[tiab] OR "data synthesis"[tiab]
Method C: My original search with some modifications from other conversations:84,757 hits
Meta-Analysis[ptyp] OR "systematic literature"[tiab] OR "systematic review*"[tiab] OR meta-analys*[tiab] OR "cochrane database syst rev"[Journal] OR "cochrane database of systematic reviews online"[Journal] OR "research synthesis"[tiab] OR "research integration"[tiab] OR "medline"[tiab] OR pubmed[tiab] OR psyclit[tiab] OR psycinfo[tiab] OR cinahl[tiab] OR "data synthesis"[tiab]
Method D: Narrower in that the db names such as Medline, PubMed, etc. and older/alternative SR terms are removed: 59,247 hits
Meta-Analysis[ptyp] OR "systematic literature"[tiab] OR (systematic[ti] AND review*[ti]) OR "systematic review*"[tiab] OR meta-analys*[tiab] OR "cochrane database syst rev"[Journal] OR "cochrane database of systematic reviews online"[Journal]
Limit: Use as an exclusion to reduce each set by 300+:
NOT ((case[ti] AND report[ti]) OR editorial[ptyp] OR comment[ptyp] OR letter[ptyp] OR newspaper article [ptyp])
When I searched using the filter sent by Brian Alper, I got 60,000+ hits. I compared my results with what his didn't find (33,500+) and, I admit, mine finds a lot of junk. But I did find some articles that looked like SRs in the descriptions of the studies.
I guess I've been burned enough by indexing and poorly written abstracts that I'm willing to go with a larger number of hits when trying to find the world of SRs and MAs. After all, this is just one component of my overall search strategy. I have students (and faculty) who need to know for sure that an SR hasn't been reported in the literature. Combining other concepts generally gets the results to below a hundred.
What I will probably do is stick with Method B, knowing that I can do the adjacency search in Ovid Medline to get to those articles missed by PubMed.
I must say this has been a very interesting conversation. I've made several changes to my original search filter and I've certainly had to rethink some of the tricks I use.
Helena
Helena M. VonVille, MLS, MPH
Library Director
University of Texas School of Public Health Library
Houston, TX 77030
[log in to unmask]
713.500.9131 (office)
713.500.9125 (fax)
-----Original Message-----
From: Evidence based health (EBH) [mailto:[log in to unmask]] On Behalf Of Michael Power
Sent: Monday, July 26, 2010 9:26 AM
To: [log in to unmask]
Subject: Re: The new PubMed Clinical Queries Interface
Thanks Brian
Your filter does exclude a lot of "noise". I will compare your and Helena's filters and let you know how they perform.
Best wishes
Michael
-----Original Message-----
From: Brian Alper MD [mailto:[log in to unmask]]
Sent: 25 July 2010 18:25
To: Michael Power; [log in to unmask]
Subject: RE: The new PubMed Clinical Queries Interface
Michael
If you are testing filters for finding systematic reviews here is a
filter we have used for finding systematic reviews for DynaMed. It
would be interesting to see how it works in a formal analysis.
((meta analysis[ptyp] OR meta-analysis[tiab] OR meta-analysis[mh] OR
(systematic[tiab] AND review[tiab]) NOT ((case[ti] AND report[ti]) OR
editorial[ptyp] OR comment[ptyp] OR letter[ptyp] OR newspaper article
[ptyp])))
You can tell by the filter that we put efforts into excluding much of
the noise that came from some other systematic review filters. In our
limited testing the exclusions did not appear to lose clinically
important content.
Brian S. Alper, MD, MSPH
Editor-in-Chief, DynaMed (www.ebscohost.com/dynamed)
-----Original Message-----
From: Evidence based health (EBH)
[mailto:[log in to unmask]] On Behalf Of Michael Power
Sent: Wednesday, July 14, 2010 12:03 PM
To: [log in to unmask]
Subject: Re: The new PubMed Clinical Queries Interface
Thanks Helena
The "MA as topic" back is back in my strategy for SRs as a result of
your explanation!
Unfortunately it does not compensate for bad indexing by Medline, or for
authors who call their article a systematic review/meta-analysis, when
it is plainly unsystematic. That's why we need people to apply some
intelligence and experience to searching and filtering.
We do have a filter for searching for guidelines, which was developed
after a lot of careful searching for and experimenting with different
filters. But, I do not use it because it does not work very well. This
is perhaps not surprising, as the indexing is very poor. For example,
guidelines such as those developed by NICE are not indexed by Medline -
except indirectly when a summary is published in a journal.
Our information specialists have just gone home, so you will have to
wait until tomorrow before I can send it to you. Because the filter is
so poor, our information specialists supplement the automated search for
guidelines with quite a detailed strategy for human searching and
filtering of a number of bibliographic databases and guideline
repositories.
If anyone has a better Medline search filter than ours, we would of
course be delighted to prove yours superior.
Best wishes
Michael
-----Original Message-----
From: Vonville, Helena M [mailto:[log in to unmask]]
Sent: 14 July 2010 15:51
To: Michael Power
Subject: RE: The new PubMed Clinical Queries Interface
Probably the MA as topic is probably overkill, especially for a quick
and dirty search. However, I often work with students who are trying to
develop topics for their theses or dissertations. One of the benefits
of using that term is that you pick up comments and analyses from other
authors that refer back to the original SR/MA. Sometimes these comments
serve as a stepping off point for students so the overkill approach
tends to work well for them. And they appreciate my thoroughness which
then encourages them to be as thorough as possible.
As to your other comment-- I agree that a search for guidelines is
critical. In the course I co-teach, the students are strongly
encouraged to look for guidelines. During one of the lectures in our
SR/MA course, the other instructor goes into http://www.guidelines.gov
and compares two or three sets of guidelines for the same
intervention/treatment/condition. This provides the students with an
opportunity to see the difference between evidence-based guidelines and
"because the experts say so" guidelines.
I probably should develop a filter for guidelines as well. Do you have
one that you use now?
Thanks so much for your comments.
Helena
Helena M. VonVille, MLS, MPH
Library Director
University of Texas School of Public Health Library
Houston, TX 77030
[log in to unmask]
713.500.9131 (office)
713.500.9125 (fax)
-----Original Message-----
From: Evidence based health (EBH)
[mailto:[log in to unmask]] On Behalf Of Michael Power
Sent: Wednesday, July 14, 2010 12:35 AM
To: [log in to unmask]
Subject: Re: The new PubMed Clinical Queries Interface
Thanks Helena, I will try your filter out.
I have two comments though.
You probably do not need "meta-analysis as topic"[MeSH]" in your search
strategy.
Search strategies for systematic reviews will miss the systematic
reviews that have been done as part of a health technology appraisal or
evidence-based guideline or economic analysis. These reviews may be the
most current and most rigorously conducted ones, and therefore the ones
that you would most want to not miss. Therefore, a search for systematic
reviews must begin with a search for evidence-based guidelines.
Michael
-----Original Message-----
From: Vonville, Helena M [mailto:[log in to unmask]]
Sent: 13 July 2010 20:06
Subject: Re: The new PubMed Clinical Queries Interface
There is a flaw in the Montori search in that the approach is far too
narrow. Try this in PubMed:
Medline[tiab] OR (systematic[tiab] AND review[tiab]) OR
meta-analysis[ptyp]
"cochrane database syst rev"[Journal] OR "cochrane database of
systematic reviews online"[Journal]
#2 NOT #1
I retrieved over 1,000 Cochrane reviews from searching by journal title
that weren't retrieved with the Montori filter.
I modified with a filter I've been working on and came up with this:
"meta-analysis as topic"[MeSH] OR Meta-Analysis[ptyp] OR "systematic
literature"[tiab] OR "systematic review*"[tiab] OR meta-analys*[tiab] OR
"cochrane database syst rev"[Journal] OR "cochrane database of
systematic reviews online"[Journal] OR "research synthesis"[tiab] OR
"research integration"[tiab] OR "medline"[tiab] OR "data
synthesis"[tiab]
I compared it against the Montori filter and found a fairly substantial
number of meta-analyses that the Montori filter missed.
I have not compared my filter against the PubMed filter. At least, not
lately. I've opted to take an overly-broad approach with the filters
I've developed. The filter above is not too broad, though, but it is a
modified version of the full SR/MA filter. A quick scan of the results
shows a high percentage of SRs and MAs.
Helena
Helena M. VonVille, MLS, MPH
Library Director
University of Texas School of Public Health Library
Houston, TX 77030
[log in to unmask]
713.500.9131 (office)
713.500.9125 (fax)
-----Original Message-----
From: Evidence based health (EBH)
[mailto:[log in to unmask]] On Behalf Of Paul Glasziou
Sent: Wednesday, June 30, 2010 1:08 PM
To: [log in to unmask]
Subject: Re: The new PubMed Clinical Queries Interface
Good point Tanya - and nice idea to add the filters to MyNCBI.
I never use the PubMed systematic reviews filter - its pretty useless
(unlike the Therapy, Diagnosis, Prognosis, Etiology, Clinical Prediction
Rule ones - which are all great!).
So in MyNCBI I have saved one of the Montori filters*:
Medline[tiab] OR (systematic[tiab] AND review[tiab]) OR
meta-analysis[ptyp]
If I am logged in to MyNCBI then this filter shows up whatever search I
do! (Including the Clinical Queries filters :-)
But this does not seem to work with the new interface.
Best wishes
Paul Glasziou
* Optimal search strategies for retrieving systematic reviews from
Medline: analytical survey.
Montori VM, Wilczynski NL, Morgan D, Haynes RB; Hedges Team.
BMJ. 2005 Jan 8;330(7482):68. Epub 2004 Dec 24
Feddern-Bekcan, Tanya wrote:
>
> Thank you for the heads up! Is it just me, or is there no drop-down
> box or checkbox for the Systematic Reviews option?
>
> Since one can have up to 15 Filters for their My NCBI account in
> PubMed, I'm thinking of just adding 4 of the clinical queries and
> Systematic Reviews as filters. It'd be a lot quicker than having to go
> to re-run the search on the Clinical Queries page. Is anyone else
> doing the same thing?
>
> Take care,
>
>
> Tanya
>
> Tanya Feddern-Bekcan, MLIS, AHIP, MOT, OTR/L
> (http://www.reocities.com/nqiya/libraryarticles.html) formerly Tanya
> Feddern
> 305.243.3999 - [log in to unmask] - 305.325.9670 (fax)
> EBM Theme Director, Head of Education, & Occupational Therapist
> Department of Health Informatics, Louis Calder Memorial Library at the
> University of Miami Miller School of Medicine
>
> They do random drug checks; why don't they do random hand swabs to see
> which unhygienic healthcare providers are killing their patients by
> spreading deadly infection?
>
> "A library without a librarian is a reading room."-- Jenny Garcia of
> the University of Wyoming, MLS, AHIP
>
>
------------------------------------------------------------------------
>
> *From:* Evidence based health (EBH)
> [mailto:[log in to unmask]] *On Behalf Of *Steve
Clancy
> *Sent:* Wednesday, June 30, 2010 1:04 PM
> *To:* [log in to unmask]
> *Subject:* The new PubMed Clinical Queries Interface
>
> Howdy.
>
> Has anyone had a chance to work with the new PubMed Clinical Queries
> interface (http://www.ncbi.nlm.nih.gov/sites/pubmedutils/clinical)?
>
> What is your opinion?
>
> --steve
>
> --
> Steve Clancy, MLS
> Medical Librarian
> Science Library, Univ. of Calif., Irvine CA. U.S.A.
> 949-824-7309 * sclancy AT uci.edu
> http://www.faculty.uci.edu/profile.cfm?faculty_id=5109
>
------------------------------------------------------------------------
-
> "We don't see things as they are...We see things as we are." - Anais
Nin
|