Trip Database Blog

As part of the preparation to include the big 5 medical journals (NEJM, JAMA, Lancet, BMJ and Annals of Internal Medicine) I’ve been doing some preliminary work. As feared the Lancet still does not allow direct linking to their abstracts on the the Lancet website. I can see no good reason for that – they make you register and login to view them. I think this makes them unique!

So what are the options for inclusion in TRIP? The most obvious is to link to their PubMed records. Hardly ideal. The one advantage is that this could be automated via the PubMed XML tools. But then do I include the other ‘big 5’ via this method or link to the abstracts on their respective sites? It invariably looks better going to the sites, but it is more work….

Further to my post of the 18th April….

I shared my views of adding the big 4 journals (JAMA, NEJM, Lancet, BMJ) with Paul Glasziou from the Centre for Evidence-Based Medicine. Paul carried out a large evaluation of the TRIP Database in 2005 and is someone I have huge respect for. Paul’s view was that it might be better to add Annals of Internal Medicine as well. The reason for this is that these 5 journals (plus Cochrane who we already have on the site) account for over 50% of the articles seclected as part of the Evidence-Based Medicine journal (which Paul edits).

So I’m getting increasingly drawn to the idea of adding these ‘big 5’, irrespective of the workload. However, I may, to begin with just add articles from 2000.

I came across a nice site, Visual Thesaurus, which is unfortunately not free-access! But offers a few trial goes. It reminded me of the TouchGraph feature in HubMed. I’m not sure how useful these things are, but they interest me.

I’m still keen on introducing a relationship function on the TRIP Database. I see it working like Amazon’s ‘People who bought this….’. So it’s be something like ‘People who viewed this article also looked at these’. However, as with the visualisation features above I’m not sure how useful it’d be!

Below are the top ten papers reached via the TRIP Database for March 2006.

1) Abdominal ultrasound in the diagnosis of childhood appendicitis (BestBets)
2) Geriatric Care and Treatment (SBU)
3) Osteoporosis (Patient UK)
4) Routine antenatal care for healthy pregnant women (NICE/RCOG)
5) If a patient with osteoporosis is unable to take calcium supplements, are bisphosphonates as effective as when a patient is able to take calcium supplements? (NLH Q&A Service)
6) Day Case Laproscopic Cholecystectomy (British Association of Day Surgery)
7) Evaluation of the patient with sore throat, earache, and sinusitis: an evidence based approach (DARE)
8) Burkitt Lymphoma (eMedicine)
9) Acute Pain Management (National Health and Medical Research Council)
10) Effective Medical Treatment of Opiate Addiction (NIH Consensus Statement)

I was pleased to see a NLH Q&A Service answer in the top ten – shows other people are, at least, interested in them. In March over 3,000 visits to NLH Q&A Service answers were ‘facilitated’ by the TRIP Database.

An interesting site spotted by Ben (NHS eLibraries blog) – LactMed. Its a site dedicated to drugs to which might have an effect of drugs taken during lactation. I’m sure it’s very useful although one search for simvastatin returned a record whose first sentence was “No relevant published information exists on the use of lovastatin during breastfeeding“

A UK related site is the UKMI (UK Medicines Information) site Drugs in lactation

Does it have a role in TRIP? Dean Giustini’s re-named blog serves up more food for thought. He highlights 5 main sources of medical information:

1) Medline – core
2) Big four (BMJ, JAMA, NEJM, Lancet) – inner core
3) High impact factor journals
4) Embase – alternative core
5) EBM material.

TRIP definately does 5 – that’s its main purpose. We also have created an automatic search of PubMed that covers 1, 2 and 3. I don’t think we can do much about 4.

However, the medline coverage is not ‘core’ to TRIP, it’s bolted on as a separate search. Can we integrate both systems? A few years back we had a separate category called ‘primary research’ where we incorporated 25(ish) journals, big 4, AMA, BMJPG journals and select others. The criticism we received related to potential publication bias. ‘Hooking up’ to PubMed allowed us to improve the coverage and save ourselves much time.

Is it time to consider re-integrating some primary research articles into the main search? I’m thinking of how it might work to incorporate the ‘inner core’ of medical journals. It would be relatively straightforward. However, it raises the main issue pre-occupying the current development workload at TRIP – weighting and the search algorithm. Where would a 2006 journal article appears relative to a 2003 Cochrane SR, a 2004 clinical guideline or a 2005 clinical question?

Not sure how applicable these results are to medical search….

“The study found that 62 percent of search users click on a link within the first results’ page, up from 60 percent in 2004 and 48 percent four years ago. Additionally, nearly all searchers–90 percent–click on a link within the first three pages, up from 87 percent two years ago and 81 percent in 2002.”

“When researchers asked users at what point they either revised their queries or moved to another search engine, 16 percent replied that they did so after reviewing the first few entries, and 25 percent said after the first page. An additional 27 percent changed either their queries or search engines after the first two pages, while 20 percent did so after three pages; just 12 percent stayed with the search terms and/or engine beyond three pages.”

See here for more details.

WLA (or WLAS or MAS or MLAS?) has been released (click here). Disappointed to see it doesn’t cover medicine as yet. It’s only available as a beta-version and covers “computer science, physics, electrical engineering, and related subject areas.”

The TRIP Database moving to free access appears part of a larger trend in freeing-up content across the internet. See this BBC news article (click here)