Scientific journals trade in the currency of knowledge and ideas. JGIM’s commitment to readers (and therefore our expectation of authors) is that material published in the journal is original. But what does it mean for a piece of work to be “original?” Certainly all scientific writing has an intellectual heritage; even Newton stood on the shoulders of giants. Yet even in the best of circumstances, the provenance of an idea may not be readily discernable from the reference list. And circumstances are not always ideal.

Recently, it came to the attention of JGIM’s editors that a manuscript just published online in another journal (Article B) bore a clear resemblance to a manuscript published in JGIM approximately six months earlier (Article A). On closer inspection, there was reason for concern. Both papers reported on a quasi-experimental evaluation of a quality improvement intervention. The titles, by-lines, and abstracts were similar and the methods sections almost identical. Moreover, entire paragraphs of the introduction and discussion sections were almost the same.

Had an editorial crime been committed? And if so, was this a felony, a misdemeanor, or merely a technical breech akin to jaywalking? The first job of an editor in such circumstances is to establish the facts. Careful comparison of the two published papers showed that the same intervention was introduced (and evaluated) in two discrete inpatient settings: one involving housestaff, the other not. Furthermore, the patient populations, study designs, and analytic methods were distinct.

What then, to make of a case where the same authors publish two papers using many of the same words to report on two similar but distinct interventions? What is self-plagiarism and how is it defined (if it exists at all)? To answer this question, we turned to the Committee on Publication Ethics (http://publicationethics.org) and the World Association of Medical Editors (www.wame.org), among other resources. Within hours of submitting the redacted case to the WAME listserv, we received a number of helpful replies, including one from Miguel Roig of St. Johns University. In prior writings, Roig has argued that self-plagiarism exists in four forms. Duplicate publication is publishing the same results in more than one article. Salami science is slicing up one large study into multiple small articles. Copyright infringement consists of using previously published material without permission of the copyright holder (which in JGIM’s case is the Society of General Internal Medicine). Text recycling is reusing phrases, sentences, or paragraphs found in previous work without appropriate attribution, including quotation marks.

Each of these violations carries different moral weight. Duplicate publication is a clear ethical breech, in part because systematic reviews or meta-analyses may inadvertently count the same data twice, leading to mis-estimation of effects that could potentially harm patients. Salami science is undesirable but may be unavoidable without changes in the ways universities assess faculty research productivity (i.e., by counting papers). Copyright infringement is a legal issue, and interestingly, there have been few if any successful lawsuits involving this form of self-plagiarism. Finally, text recycling is a matter of degree. In our informal poll, many experts (including Roig) are fine with around 10% re-cycling of verbiage, some even arguing for the benefits of repeating complex methods verbatim. A few suggested limits of 15-20%, but none countenanced more than 30%.

JGIM has no established limit for acceptable text re-cycling, nor do we currently plan to impose one. However, the JGIM editors wish to be clear about two things. First, we expect that authors will disclose any substantial overlap between manuscripts submitted to JGIM and all other manuscripts published, submitted, or nearing submission. Often, the easiest way to disclose is simply to provide a copy of the other manuscript(s). Second, authors should be careful about the amount of text they recycle. While there are sometimes good reasons for re-using certain textual elements (particularly in the Methods and literature review), authors should be cautious and thoughtful in doing so. JGIM readers expect that work appearing in the journal is original. Small deviations can be tolerated. Bigger ones may give the appearance of deception and should be eschewed.