Jon Brassey on the TRIP database

Jon BrasseySearching for clinical information isn’t easy, I know, I’ve been doing it on an almost daily basis for over ten years. I’d like to think that I’m pretty good at searching across a number of clinical databases, I even helped create one – the free TRIP Database.

But search engines can be a real pain. 

As an aside, if someone was to start from scratch and design a product to meet the need to answer clinical questions would anyone suggest a product that returned 10-20 links to articles that might have an answer?

Back to the present…Search engines typically return a number of links that they believe will contain the information the searcher requires. Behind this obvious comment are many complications. One such complication is the fact that most people aren’t great searchers. The most popular searches on the TRIP Database are single terms such as diabetes, hypertension, asthma.  These return large numbers of results that are unfocussed, covering many different aspects of the condition. Granted, some people will want to look at anything on diabetes, but many will have a more specific reason e.g. what’s the evidence for metformin in diabetics? Or how often should you screen for retinopathy in diabetics? Searching, using single terms in situations like this, is going to make the whole experience frustrating. Most studies of the mainstream search engines show that the majority of people will only look at the first page of results and only a very small number will get as far as page 4. 

So, it’s really, really important that search engines get the top ten results right.

If you’re in a developing country (for example, a rural district hospital in Africa or South Asia) the complications mount up further as there’s a good chance you’re hampered by a lack of money to purchase the latest drugs and expensive diagnostic tools.  This can have a big effect on searching for evidence? The TRIP Database covers a range of content from secondary review articles such as Cochrane Systematic Reviews, primary research such as the BMJ through to eTextbooks.  Most content stems from organisations based in the resource rich world and reflect this.  If 10% of the content of the TRIP Database is suitable for resource poor settings that means there is additional ‘noise’ from the 90% of articles that aren’t! So, when a clinician (or patient) from a resource poor setting is looking for evidence they are faced with the additional challenge of finding evidence suitable for  that setting. This is one of the key problems expressed by health professionals in Africa and Asia on the HIFA2015 email discussion forum (Healthcare Information For All by 2015; It’s like trying to drink water from a fire hydrant.

Unfortunately, there doesn’t seem to be a clear way of identifying material suitable for the resource poor environments.  People can be pretty good at identifying such material but the trick is to harness humans in the process. Through funding from BUPA Giving, the TRIP Database has devised a new initiative to help identify evidence suitable for low resource settings.  This will use the power of crowdsourcing – using ‘crowds’ of people to solve problems.  

Users of the TRIP Database can now see under each result a new link, ‘Developing World?’ If a user feels an article is suitable they can click on the link. If a single document gets two clicks (from separate people) it will be ‘accepted’ and form the basis of a new sub-set of data.

At the same time as the link appears under each result a tick box will be created on the results page. This will allow users to restrict results to those suitable for low resource settings. For example, a user might have searched for ‘diabetes’ and been overwhelmed by the results (50,000+ results including 2,200+ systematic reviews and 1,000+ guidelines) but can then opt to see only those articles suitable for their environment.

The most obvious problem with this is that this valuable function needs time to develop. Initially, the database of approved articles will be zero (a user will go from 50,000 results to none – hardly an improvement). However, over time users will click and the database will grow. We’re confident that we’ll soon have a powerful tool to help clinicians in low resource setting identify, and therefore use, high quality clinical evidence. This initiative launches today and we ask all interested people to visit us at