Just attended a very interesting seminar given by Professor Enrico Coiera from the Centre for Health Informatics at UNSW.

Enrico and his team have done some very well-grounded, practical research, development and evaluation around the use of decision support and information access tools for clinicians. He gave some interesting statistics of the dimension of the problems faced in the clinical domain:

• Clinicians do not pursue answers for up to 70% of the questions that arise during routine care, partly because they are time-poor, but also because they have poor query formulation skills.
• The number of scientific articles doubles at 1 to 15 year intervals – growth is exponential.
• A new article is added to the medical literature every 26 seconds
• Clinicians have more questions than they look for answers – up to 6 per patient encounter. They only pursue answers in about a third of cases, and spend about 2 minutes searching for an answer.

Another problem is that clinicians’ knowledge decays with years since graduation, and research in learning theory suggests that it is in-context, on the job learning/training that most effectively combats this knowledge decay.

Suffice to say, there are real and growing problems in the clinical domain regarding access to and use of information for diagnosing and treating patients. Firstly, there’s a classic case of information overload, as evidenced by some of the statistics quoted above. Secondly clinicians can’t simply ignore this massive stream of new information, since their own knowledge is insufficient and decays over time. Finally, given the clinical focus, the problems are compounded by their context, where the cost of wrong information or a lack of the right information can mean the difference between life and death.

Clearly, there is a reasonable amount of evidence that suggests there is a place for in-context information access and decision support tools that can fit in with clinicians’ workflow and socio-technical environment, to give them a better ability to frame questions and retrieve evidence-based answers from a variety of authoritative medical data sources. This proposition has a neat and interesting correlation with our own work in information delivery, where we’re very concerned with providing only relevant information to the right people in a form that suits their particular task and interaction context. Enrico has done quite extensive evaluation on the effectiveness of their fairly simple, information retrieval-based decision support tools, and found that they can make a real difference to the correctness and confidence of clinical decision making.

I’m heartened by a few aspects of Enrico’s talk in particular:

• There are other people out there cares about contextual delivery of information to help people achieve their tasks
• Even some simple knowledge about users, their tasks and their information needs can provide a significant benefit over non-contextualised information retrieval and delivery
• There are people dedicating time and resources to address the inescapable (and scary!) fact that about 17% of hospital admissions result in an “adverse event”, 5% of which result in patient death. There are many indicators that suggest this is often due to poor information, and as summarised above, that we can make a significant difference by addressing the information needs of clinicians.

I’m sure I’m not alone in hoping I don’t have to be in hospital before they’ve got some of these information access and delivery problems sorted out!