Richard Smith: Database of cases launched

Richard SmithEvery 36 hours the NHS treats a million people. Across the world there are billions of interactions between patients and health systems every year. Each of those patients is a “case,” and the potential learning from those cases is huge. Unfortunately most of the learning is lost, unrecorded and unshared. But now the launching of a database of cases by BioMed Central  opens up the potential of capturing much of that lost learning.
Case reports, as you must have noticed, have come back into fashion. Medical journals a 100 years ago were full of case reports, but after the appearance of randomised trials after the second world war and the subsequent birth of evidence based medicine in the 90s case reports came to be seen as embarrassing unscientific detritus. Editors (and I was one then) swept them out of their journals in pursuit of more rigour and higher impact factors.
This couldn’t last because cases are the very stuff of medicine. Doctors learn from individual cases, usually their own, remember them, and tell stories about them to other doctors. A chunk from a textbook or a study in a journal needs to be linked to a patient, a case, in order to be real. People, including doctors, relate well to individuals but not at all to thousands of people. Tell me that 20 million people died in the first world war, and I don’t manage an emotional response. Tell me the story, as vividly as possible, of somebody’s grandfather who died in the trenches, and I might well be in tears (especially if the story has a bagpipe accompaniment).
So it was inevitable that cases would be rediscovered, and one of the first signs was the launch of the Journal of Medical Case Reports  in 2007. BMJ Case Reports  followed in 2008 and has now published over 4000 cases. And now there is a project to produce a standardised way of reporting cases, just as there is for randomised trials, systematic reviews, and many other types of studies.
Journal of Medical Case Reports and BMJ Case Reports tend to prefer what is “original” and “important,” although BMJ Case Reports says that it is interested in “reminders” of an “important clinical lesson,” implying that originality may not be requisite. But if I submit a case of a patient with a cold failing to respond to antibiotics will the editors consider that an important reminder? Probably not. Perhaps, indeed, they’ll wonder about my competence.

I have an interest in this as I was the editor of Cases Journal,  which boldly said that it would publish any case that was ethically sound and “complete enough.” Our logic was that every patient is unique, that a reader could learn something from every case if the story was well told, and that we hoped to build a database of cases—and the more patients in the database, including “bog standard” cases, the more valuable it would be. We were a sister journal to the Journal of Medical Case Reports and were sold (or maybe given) to BioMed Central by the Science Navigation Group. BioMed Central couldn’t see a viable business model for Cases Journal, and so we were put into hibernation. We have hopes to be reborn in Nigeria, but so far progress is slow.

Despite our aspiration we never managed to launch a database, although we did publish an editorial showing how hard it would be.  But now BioMed Central has done it, and the database includes not only all the cases from the Journal of Medical Case Reports and Cases Journal but also those from BMJ Case Reports and other journals. There are over 11 000 cases from 100 journals. The database uses text mining, meaning that cases can be found using “standard controlled vocabularies,” identifying patients by diagnosis, intervention, medication, age, sex, ethnicity, symptoms, and other words.

The database will be free to all, and its usefulness will depend as much, probably more, on its users than its creators. It can be used for teaching, learning, researching, generating hypotheses, looking for associations, and much more.

I’m obsessed by the fact that most of health care in developing countries is now concerned not with patients with one condition but with patients with multiple conditions. Yet most of our evidence base comes from randomised trials that consciously excluded patients with multiple conditions. We can never do trials on every combination of age, gender, ethnicity, medical history, and combination of conditions but remembering the billions of patients treated every year we could build a database of cases in which we could find patients with any conceivable combination.

The value of the database will increase as the number of cases increases, and I can’t see why we shouldn’t be able to build a database with millions not just thousands of cases. This is a challenge to you. If every clinician were to write up just one case a year we’d soon have a database with millions of cases. And if you don’t see patients you could write up your own case: my report of my “Beijing cough” was one of the most accessed cases in Cases Journal.

Competing interest: RS was paid to be the editor of Cases Journal, but the payments stopped when the journal was sold. He is keen to resuscitate the journal in Nigeria, but 20 000 pounds is needed to do this. None of this money would go to RS, and if the journal is resuscitated and he is the editor he will not be paid. RS has been involved unpaid in selecting prizewinners for BioMed Central and is a fanatical supporter of open access. He was also involved in generating the guidelines on reporting case reports and had his expenses paid by the University of Michigan and was given two splendid dinners.

Richard Smith was the editor of the BMJ until 2004 and is director of the United Health Group’s chronic disease initiative.