Trip Database Blog

Liberating the literature


March 2007


It’s been a while since I posted on Q&A. The ATTRACT service is approaching question number 2,500 while the NLH Q&A Service has recently gone over the 5,000 barrier. Both services continue to churn out solid, thought-provoking answers which gain huge satisfaction scores with the clinicans who use the service.

But where next with Q&A – can it stay as it is? With regard the NLH service, it will invariably change. The NLH has limited resources and need to better use exisiting ones. So what might that mean. The general idea will be to create a central triage area that will receive all the questions. These questions will then be dealt with by a number of different resources. These resources will include the NLH’s specialist libraries and possibly health librarians and even consultants. In addition a small in-house team would be well placed to ‘cherry pick’ the easier questions; questions where specialist knowledge is not necessary.

Another innovation is to work much more closely with CKS (formerly PRODIGY) to ensure their outputs are even more cllosely matched to clinically significant knowledge gaps.

Also, the NLH need to get a half-decent search engine….

These are ‘predictable’ innovations, in that they appear to be natural progressions.

Another innovation I would like to see is some mechanism of capturing questions answered in other ways. Over the world, clinicians answer thousands of questions for themselves (or with the help of the colleagues). With the rise of the internet a large number of these clinicians are (or can be) connected. If these answered questions could be shared – that would be very powerful indeed. We’re keen to develop this idea and will form a part of Gwagle (going to alpha next week – fingers crossed). However, that’s not to say it wouldn’t be an integral part of any NHS/NLH Q&A ‘service’.

The various Q&A services I’m involved with have an archive of 7,500 answers to clinical questions. These have been arrived at through a systematic review of the literature (using a broader definition of systematic than is usually associated with systematic reviews). If answers could be shared we might arrive at an archive of 50,000+. And that might move from the re-use rate of answers from the current 7-10% to over 50%. Answers start to look much cheaper and clinical information support much more instant.

On a more obvious note, if clinicians having up-to-date information support is so important, why not create an NHS Direct for clinicians?

Potential dangers associated with web 2.0

BBC News site report on a fake professor who was pretty senior in Wikipedia click here.

TRIP Improvements

It’s taken a fair bit of R&D to arrive at this point (and we’re still not sure on one point) but we’ve finalised our latest batch of improvements:

  • Results page re-design. There is too much wasted space on the current TRIP results page and I wanted to make better use of it. This is especially important for two reasons. Firstly, we’ll be moving our Google Ads to there to improve performance (given our search stats we’ve made little money using that method). Secondly, we may be incorporating snippets into our search results (more below).
  • Conclusions. Many of our publications (e.g. Cochrane, DARE, BMJ Updates, JAMA) have clear conclusions. We’ve created a system to grab and display these (via a rollover) without having to visit the actual paper.
  • Improved advanced search. Currently, the search doesn’t work, from a flow perspective, as I would like. Therefore, we’ll improve how that works, making it much more user friendly.
  • PDFs. Highlight which results are PDFs
  • RSS feeds. When doing a search you’ll be able to drag the RSS button to your RSS reader and new search results, for that search, will be highlighted.
  • Specialist search engines. We’ll be creating a significant number of specialist search engines. These will be in specialities such as cardiology, mental health, nursing etc. If you use a specialist search you’ll be able to just search the content of the top 10-20 journals in that area.
  • Synonyms. These are currently being reviewed, edited and overhauled to further enhance them.

The one uncertainty is still snippets. Snippets are the small summaries that you see in general search engines. To date our attempts to create adequate ones have been unsuccessful. However, we’ve got one more trick to try! Why are we so interested in snippets? We feel that they can better improve document selection, improving the search experience.

These improvements will be released, by May. However, some (namely the conclusions) will be released earlier.


Interesting interview with one of the people responsible for Microsoft’s QnA service click here.


We get a large number of questions on individual drugs. Unfortunately, most of the robust resources ‘hide behind’ a firewall and require an expensive subscription to obtain the information. It’s a bit like the Cochrane situation (see previous post) in that (I imagine) the taxpayers have paid for the initial research and development and then have to pay for re-accessing the same information.

I accept that many of these resources require significant input regarding appraising, searching, editing etc. So what’s the solution, perhaps via another great resource highlighted on Ben’s blog. This time it’s an open-access, wiki-based, drug resource called PubDrug. It again highlights what can be done with decent tools (wiki) and a bunch of skilled and enthusiastic individuals. Currently, it has 7 monographs. I hope this expands rapidly.


Ben has introduced me to another new blog, this time for Andrew McAfee. I’ve been discussing Gwagle with Ben for a while now and Andrew McAfee (Associate Professor at Harvard Business School) appears to be doing some very important analysis and thinking. He devised the SLATES pneumonic, which has been handily re-written (so I can understand it) by another blog:

Search: Find what you need, enhanced by emergent description (see tags, below)

Links: More to the point, link relationships or link ranking algorithms

Authoring: Ease of content creation – spare me the angle brackets, make it bone simple

Tags: What do my colleagues call this? I bet it works better than what the IT department calls it

Extensions: If you thought X was [good interesting important useful], you might, by extension find Y similarly so

Signals: tell me something has changed

The purpose of SLATES is “it helps expose how platforms and channels can be brought closer and made more effective in ways that enhance productivity and effectiveness. Yes, we’ve seen most of these elements before. It is in the ease of recombination that they change how things work.”

February Search stats – guess what?

For those of you with enough willpower to have read this blog for any length of time will realise that, at the start of each month, I tend to give the search stats for the previous month. I imagine I’ll stop when the stats stop going up! But, for February (a short month) we were searched 433,260 times (up from January’s 365,855) – see graph below.

At long last we’ve got a half-decent web-analysis package attached. From that we can also see, in February, we had over 3 million page views. No wonder our web-company is moaning about our band width usage!

Blog at

Up ↑