Monday, March 10, 2008

Search challenge 1: strategies

In the comments on Search Challenge #1, readers proposed several great search strategies and included a few example articles too. In addition to reader-developed strategies, one commenter also included the strategy used in a Cochrane review titled "Support surfaces for pressure ulcer prevention," which fairly comprehensively addresses multiple synonyms for each of the concepts in the question (e.g. multiple words for beds and mattresses, including product trade names).

The Cochrane strategy published in the systematic review is focused on searching the Cochrane Controlled Trials Register via Ovid, which allows proximity operators (e.g. "next") to aid keyword searching, an option not available in PubMed, so creating a workable strategy for PubMed is probably a good supplement to using such an approach.

The systematic review also references the broader search strategies in multiple databases used by the Cochrane Wounds Group, which include strategies Ovid-focused strategies for Medline, CINAHL, the British Nursing Index, and EMBASE. This strategy also notes other databases and resources that may be useful in identifying relevant literature for wound care and prevention topics.

In addition to these strategies, I found a fairly simple PubMed search to be useful in identifying quite a few of the clinical trials and other primary studies of bedding- and bed- related strategies for reducing the incidence of pressure ulcers:
(beds[majr] OR bedding and linens[majr]) AND (pressure ulcer[majr] OR pneumonia[mh] OR complications[sh] OR complications[tiab] OR adverse events[tiab]) AND humans[mh] NOT (case reports[pt] OR letter[pt])
I'd originally just included the MeSH term "beds", then added the "bedding and linens" term when I realized that my initial strategy was excluding some of the big RCTs of mattress supplemental products that were included in the Cochrane review (I'm probably not alone in using the references of articles I retrieve as a "quality check" on my search strategy). In reading some of the abstract and articles, it seemed as if pneumonia was the other main complication that bed-focused techniques were used to prevent, so I tried broadening the search strategy to include pneumonia plus other (as yet unknown to me) potential complications, which added about 100 additional citations to the retrieval:
(beds[majr] OR bedding and linens[majr]) AND (pressure ulcer[majr] OR pneumonia[mh] OR complications[sh] OR complications[tiab] OR adverse events[tiab]) AND humans[mh] NOT (case reports[pt] OR letter[pt])
Then limiting to English language articles further reduced the set to about 600 citations, which is still a fairly large retrieval.

Some of the "false drops" and less relevant/useful items I noticed in the retrieval for this strategy:
- netting over the bed to prevent malaria
- elevating the head of a standard hospital bed to prevent pneumonia
- one to two page educational briefs in nursing- and rehabilitation- focused journals
- pillow selection
- equipment used during an intraoperative period or during prehospital transport
- cooling blankets
- simulation studies

A few questions for our hypothetical requestor that might help us refine the search strategy:
- are they interested only in mattresses and beds, or would other devices (e.g. mattress overlays) be of interest too?
- are they interested in the long-term care literature (e.g. nursing homes, chronic care) in addition to the acute-stay inpatient literature?

This search also made me think about an issue that I've heard often in colleague discussions - how many references are you comfortable browsing through to develop a bibliography for a search? I don't have any problem paging through a few hundred results for a topic like this, when so many of the citations retrieved by the search seem relevant, but I know I've heard colleagues say that they would try to develop a much more narrow strategy as an initial approach, then broaden if it seems necessary. Your thoughts on this and the strategies above and in the comments?

We'll be posting Search Challenge #2 later today...

Labels:

1 Comments:

Blogger Martin said...

I think the strategy depends a lot an the situation: if you want to answer a clinical question rapidly, I go for (systematic) reviews first than do a narrow search as an update to the review.
If the task is to do a systematic review, the strategy must be much more detailed and refined.
Maybe you could define the timeframe for the next challenge...

3/11/2008 3:56 AM  

Post a Comment

<< Home