The numbers games.

Abstract

In 1979 Allan M. Cormack and Godfrey N. Hounsfield were awarded the Nobel Prize in medicine “for the development of computer assisted tomography”. By this time, Hounsfield had published eight papers inmedical journals, his seminal works in the British Journal of Radiology. The second and, thus far, last Nobel Prize awarded for medical imaging went to Paul C. Lauterbur and Peter Mansfield, belatedly, in 2003 “for their discoveries concerning magnetic resonance imaging”. Fourteen of Mansfield’s works, including his most important, had also appeared in the British Journal of Radiology. (He published twice in Pediatric Radiology [1, 2]). This year medical journals are expected to publish almost six times the number of radiologyrelated articles compared with the Nobel year 1979 and about twice as many as in 2003 (Fig. 1). At least part of this publishing explosionmust be explained by increased scientific activity. Witnessing such an unprecedented increase in productivity begs two questions: Where is the big discovery that will launch the next transformation of medical imaging? Or is there perhaps no direct link between article output and scientific impact? In the domain of journal publishing, the Impact Factor has been around for decades. Everyone agrees “it is just another number” and “we would never obey its dictate”. But try securing a research grant if you have not appeared in highimpact journals. Or try to publish something unconventional, controversial or counter-current (innovative?) in the most prestigious journals. Our money would be on rather disappointing outcomes. This is speculation and a virtual kicking of our own backsides, but it is at least an interesting paradox that the most important works on which two Nobel Prizes were founded were published in the fringes of the medical literature (the British Journal of Radiology, host to the laureates, had a 1992 impact factor of less than 0.7). Journal editors claim not to be blinded by numerics, but which editor is happy to see her journal slide down the ranking? A self-perpetuating relationship emerges: Authors need to publish high volumes and in high-ranking journals; journals favour works that are guaranteed citations. The mechanics of the phenomenon are fairly straight-forward: It is much easier to rank by numbers than by any qualitative dimension. The unfortunate consequence may be a profession that self-restricts its research to topics that are conventional, non-controversial and therefore guaranteed publication. In 2014 Pediatric Radiology received 850 new article submissions, about one-fifth the total number of radiology articles published in all medical journals in 1979. The drive to publish is clearly increasing, and there must be several underlying forces. One obviously is technology: It is easier than ever to collect clinical and imaging data, transform these into a manuscript and submit-to-journal by the tap of an index finger. That this is phenomenal can be boded for by anyone who published in science before the internet era, in times when a whole support apparatus had to be enrolled to clear all the required hurdles. Now, everyone can publish. With the internet came an astonishing democratisation of medical publishing (ease of article submission, and also the birth of the onlineonly journal), which is fundamentally sound. A second driving force takes the shape of employer and research-funding organisations, and in particular their strategies for selecting whom to employ, support and award. One way is by counting and weighting (by journal-ranking) publications; both fair and objective. With modern metrics it is no longer a question of * Oystein E. Olsen oeolsen.pedrad@me.com