The Equator network of guidelines on reporting health research, a hugely valuable resource, has 200 guidelines, but none of them are for case reports. So as case reports seem to be returning to respectability a group of us gathered last week at the University of Michigan to start the process of producing guidelines for publishing case reports.

I’m now so old that I see many things that have disappeared returning, some for the second or third time. Time is surely circular as well as linear. I remember when the fashion was to sweep case reports out of journals as horrible, unscientific detritus. And certainly we don’t want to see the return of one of the first papers I wrote—the 197th case of duodenal cancer badly reported with pages and pages of ill informed advice on how to manage future cases. But case reports are now back in fashion.

The Journal of Medical Case Reports appeared six years ago, and BMJ Case Reports started in 2008. Cases Journal, which I edited, promised to publish any case that had patient consent and was adequately reported, but the business model didn’t work well—and it’s currently hibernating in Nigeria waiting to rise again.

PubMed contains 1.6 million case reports (8% of all articles) compared with 400 000 randomised trials; and the number of case reports published each decade is increasing. So even if “top journals” eschewed case reports they have continued to find a home somewhere.

There are various guidelines available on reporting case reports, including on N of 1 trials, but there is no consensus. Reporting guidelines are needed because we have overwhelming evidence that everything—randomised trials, case control studies, systematic reviews—are badly reported. As one old hand at reporting guidelines said, “I’m sent a paper to review every week on bad reporting, but we don’t need any more. We know that everything is badly reported with vital information missing.” We also know that reporting guidelines can improve reporting, although, as with any guideline, it’s a long, slow process—and depends on much more than simply publication of the guideline.

There is now an agreed process on producing guidelines. It begins with a review of what is already available. A group of people (around 20) is assembled with diverse interests in the subject (clinicians, researchers, guideline junkies, editors, and others), and they are phoned before the meeting to gather views. The consensus meeting takes place over two days, and always, said one of the guideline junkies, it’s more complicated than you expect. In every case so far a consensus has been reached, and there is then feedback after the meeting on the final document together with review by  some who were not at the consensus meeting. Ideally the guidelines are also tested in the real world before wide dissemination.

We were a diverse group (I knew only three others before the meeting), and we did reach consensus. Our product from the two days was a checklist of perhaps 25 items. The hope is that the checklist will be useful to authors writing case reports, many of whom are inexperienced authors, and to editors reviewing case reports before publication.

One tension with case reports that doesn’t apply to, say, randomised trials is that the best case reports are stories and the more that they employ the techniques of the storyteller or novelist the more likely that they will  be read, remembered, and acted upon. We couldn’t write guidelines for such skilled story telling, and most authors of case reports don’t have such skills—but we could produce guidelines on what should be included in case reports, even those that might win literary prizes.

Was our aim more or better case reports, I wondered. We were certainly aiming at better case reports, and more good case reports but fewer overall would be a good outcome. But a better outcome would be better and more.

One point that we debated energetically was the involvement of patients. Hardliners (including me) thought that case reports should never be published without patients being given the report to read, told how it would be disseminated, and asked to give written informed consent. Others thought that to ask so much might be culturally insensitive in a world with different traditions. The hardliners prevailed.

More interesting is the move to include the patient’s story of the case. I’m very keen on this as I’ve read many case reports written by doctors that are accompanied by patients’ accounts of their case—and you sometimes wonder if the doctor and the patient actually met. I believe that the patient writing their story—and not just adding a few quotes to the doctor’s account—can be the source of rich learning, both for the doctors and the patient and readers. We agreed that we wanted to encourage such patient stories but that they shouldn’t be mandatory.

Our guidelines and checklist have much more process to navigate, but they should appear next year and should help improve the quality of case reports.

For more information on the meeting visit #casereports on Twitter. RS and others Tweeted about the meeting. The Twitter name of RS is @Richard56

Competing interest: RS had his expenses paid to attend the meeting, with much of the sponsorship coming from the University of Michigan. Despite prejudices about the Mid West being a “meat and potatoes kind of a place” the dinner we had on the middle night was excellent. RS was paid two days a month to edit Cases Journal, but that ended a few years ago. He does hope that the journal could reappear in Nigeria (without him being paid), and we’d be very grateful if anybody has £20 000 to get us started again.

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