Medical Approaches

Clinical decision making is the real challenge that meets the junior doctor on transition from medical student. Learning the contents of books outlining diseases is arduous but straightforward; moving from problem to diagnosis is tricky and there is less consensus on how to do it and much less on how to teach it. Not surprisingly then our medical student experience with its textbook knowledge and limited and disorganised clinical experience leaves us feeling rather passive and impotent when it comes to reasoning from problem to diagnosis to action.

Medical Approaches grew out of a sense of frustration with our own weakness reasoning at the bedside. We wished to formulate our medical knowledge as ‘bottom up’ strategies for clinical problem solving rather than as ‘top down’ encyclopaedias of diseases. Additionally, we came to recognise that moving from problem to diagnosis is a little like inverse problems in mathematics – there is no single solution only a set of solutions which work better than others. Progress with underdetermined problems like these benefits from drawing on a large pool of experience. What we needed then was a structure which housed approaches for clinical problem solving which could be freely accessed and exchanged amongst junior doctors. Its community of users would contribute their experiences and take away with them those from a broader medical community. The content would be open source.

We have now created and launched Medical Approaches along these lines. It comprises a resource, which is arranged in five main blocks. The structure is intended to simulate the way any professional would seek to have their knowledge organised: associative, and oriented towards aiding decision making. Block one is a collection of pivotal symptoms (subjective starting points) and block two is a collection of pivotal clinical states (objective starting points) be they examination findings or lab results. Each starting point comprises of an ‘approach’, which focuses on the practical diagnostic issues relevant to that particular problem. So in block one we have, for example, ‘headache’, ‘chest pain’, ‘abdominal pain’ and ‘transient alteration of consciousness’ while in block two we have ‘jaundice’, ‘lymphadenopathy, ‘hypernatraemia’. Each approach aims at a crisp and logical movement through the practical differential diagnosis, key discriminating features and potential pitfalls. With these two blocks we have attempted to provide a reasonably comprehensive coverage of the sorts of presenting problems the junior doctor faces in internal medicine. The remaining three blocks comprise: Basic Investigations, which provides a logical approach to the chest Xray, ECG and arterial blood gases; Common Medical Conditions, which provide practical points on most conditions which make up much of the daily workload; and Medical Emergencies, algorithms for action when the need to act outweighs the time needed to think. All the approaches cross-refer.

Navigating the Approach to Diarrhoea

The content isn’t very original (it has been synthesised from a scatter of well known sources) but the structure is novel because it orients everything back to initial patient encounters and therefore back to real life doctoring and its demands. The content will be modifiable in an on-going way to remain faithful to this structure. Critically, we want the resource to be modified by its community of users to ensure that it is always answerable to the realities of clinical practice. Our contribution facility allows users to submit their own recommendations on how specific clinical problems can be effectively tackled. Currently, the contributions are included in the next text revision, with acknowledgement, if the editors think they are useful. We are thinking of alternative ways of creating this open-source environment – possibly using an on-going log of comments on pages linked to a given approach as well as having designated topics for on-line discussion and invitations to send in case experiences which raise general points. Growing evidencebased medicine interest in applying clinical trials to diagnostic questions will additionally be an interesting source of content improvement.[1]

Of course interest in diagnosis and computers is not new. Previous efforts, for example, MYCIN[2], have worked on automating diagnosis so the clinician enters the basic data and the computer does the reasoning, generating the diagnostic likelihoods. For a variety of reasons these have not been taken up much by doctors. A key reason is perhaps the control they take away from the doctor – control over understanding and directing the interpretation of our patients and their problems. We wish to individualise each approach to each patient and therefore wish to keep the strategies we employ partially open, not entirely determined. Medical Approaches steers away from automated diagnostic systems in that it is all about augmenting the clinician’s thinking process about their patient’s problem and entering into exchange with a wider medical community about the strategies employed and their successes and failures.

The resource is housed at medicalapproaches.com and can be used on-line for free. For those with handhelds, the entire resource can be downloaded and used on the move, again without cost. Designed for the Mobipocket reader, it is thus readable on all handheld platforms, including Palm, Pocket PC and Psion. Since our launch in February the site is averaging 10,000 hits per week with visitors coming from the USA, Australia and across Europe and we have had over 1000 people registering to download the book to their handhelds. Handhelds make the resource even more interesting: approaches can be beamed to and from doctors and personalised changes made and shared all in reaction to the realities of the job. It can be brought close, literally, to the bedside.

We hope Medical Approaches can help junior doctors actively engage in problem solving and remind us all that thinking about your patient can be more reflective, more reasoned and more fun than at times we tend to suppose.

References

1 - J A Knottnerus (ed) ‘The Evidence Base of Clinical Diagnosis’ London: BMJ

Books 2002.

2 - P Szolovits et al. Artifical Intelligence in Medical Diagnosis Annals of Internal Medicine 1998;1:80-87; for a review of expert systems in medical diagnosis. Pubmed ID 3276267.

This article was written by Dr Gareth Owen and appeared in

issue 35 of Biomedical Informatics Today (BMiT) in the Spring of 2002.

BMiT is the official newsletter of the British Medical Informatics Society.