Monday, March 17, 2008

Search challenge 2: strategies

Search challenge 2 looks at identifying articles that discuss prognosis, with traumatic brain injury as one example "disease" for which one might want to find prognosis information.

The Clinical Queries filters in PubMed, developed by Haynes et al, give us two options for finding prognostic studies:

- the sensitive/broad search hedge: (incidence[MeSH:noexp] OR mortality[MeSH Terms] OR follow up studies[MeSH:noexp] OR prognos*[Text Word] OR predict*[Text Word] OR course*[Text Word])

- the specific/narrow hedge: (prognos*[Title/Abstract] OR (first[Title/Abstract] AND episode[Title/Abstract]) OR cohort[Title/Abstract])

For the brain injuries search, I tried a search like this:
brain injuries[majr] AND ("prognosis"[MeSH Terms] OR "Recovery of Function"[mh] OR outcome[tiab] OR outcomes[tiab] OR recovery[tiab] OR reverse[tiab] OR reversible[tiab] OR death[tiab] OR improvement[tiab] OR improve[tiab] OR mortality[sh] OR improved[tiab] OR prognosis[tiab] OR prognostic[tiab])

but it was a little too broad - I think partly because the term "outcomes" is so common in research but doesn't always mean clinically-relevant outcomes, so then I tried to narrow this part by thinking of what outcomes are important in trauma, and refined the search to this:
brain injuries[majr:noexp] AND ("prognosis"[MeSH Terms] OR "Recovery of Function"[mh] OR mortality[tiab] OR prognosis[tiab] OR prognostic[tiab] OR survival[tiab] OR Glasgow outcome scale[tiab] OR functional[tiab]) AND humans[mh] AND eng[la] NOT (case reports[pt] OR letter[pt] OR comment[pt] OR news[pt] OR editorial[pt])

which returned about 1200 hits from the last 5 years, which is still a lot.

When you start browsing through the list, a lot of them do look at outcomes, but there are a myriad of outcomes - For this search, I think a good next step would be to continue with the requester and figure out what exactly they mean by outcomes - more immediate outcomes like in-hospital mortality, or longer term things like mortality at 30 days, functional outcome, lost work, other morbidity (e.g. posttraumatic seizures). Understandign this would be useful in refining the search or at least informing how we look at the retrieval.

Other thoughts for refining this search?



Post a Comment

Links to this post:

Create a Link

<< Home