BMJ 1995;310:1387-1390 (27 May)
Ronald E LaPorte, professor,a Eric Marler, independent consultant,b Shunichi Akazawa, information specialist,c Francois Sauer, managing partner,d Carlos Gamboa, regional adviser,e Chris Shenton, internet engineer,f Caryle Glosser, psychologist,g Anthony Villasenor, manager,h Malcolm Maclure, associate professor i
a Department of Epidemiology, University of Pittsburgh, 3460 Fifth Avenue, Pittsburgh, PA 15213, USA, b Tarrytown, NY, USA, c Geneva, Switzerland, d AT (GIS), Group of Enterprise, Design and Development, Boca Raton, FL 33434, USA, e Pan American Health Organisation, Washington, DC, USA, f Sterling Software, Washington, DC, USA, g Pittsburgh, PA, h NASA Scientific Internet, Washington, DC, i Department of Epidemiology, Harvard University, Cambridge, MA, USA
Correspondence to: Dr LaPorte (firstname.lastname@example.org).
The musky scent of aging paper in our medical libraries still evokes an atmosphere of scholarship. But the cloistered peace of the stacks is increasingly punctured by the faint sounds of the coming revolution: the clicks, beeps, and whirrs of computers linked to the internet. For whom do they toll? Are they the death knell of biomedical journals as we know them? Or are they the pealing spire of the global village summoning health scientists to the electronic commons to share the harvest of knowledge?
We are at a watershed in biomedical publishing. For some time the costs of paper journals have been mounting and the budgets of health science libraries contracting, while the number of have nots in poorer countries clamouring for access to medical literature has been growing.1 But now the information technology explosion that revolutionised banking and the airline industry is at the gateway of the biomedical community.
As the hard copy journal system has started to decay, there has been an information technology explosion that, some argue, will completely transform the exchange of information in the biomedical community. The current process of biomedical publication expanded in the late 1800s. The approaches towards delivery of information to the scientists have changed little during the 20th century: mailed journals, textbooks, and scientific meetings. Transmitting information through the journal system can be likened to the use of the Addressograph in the 1950s for producing mass mailings, or the vinyl records that we all remember. New technology came in to produce the mailings more effectively and to "deliver" music to the consumers. Within a short period the Addressograph and the record player became virtually extinct. We believe that biomedical journals as we know them will become extinct in the next few years as the result of the development and evolution of new, high powered electronic information delivery engines which will revolutionise information exchange among scientists and between scientists and the lay public.
The publication system
We publish to exchange information and to archive our work with some degree of permanence so as to leave a paper trail of evidence for future scientific work. We also publish for currency, to obtain promotion, to obtain grant support, and to obtain accolades from our peers. A new research communication system should be able to deal with these needs. It should be an efficient system that helps us to obtain the information that we need quickly and cheaply. It should also have as part of this a currency system whereby one can determine which communications are viewed by the people in the field as the "best," having the greatest "impact." We are in the process of developing such a system through the Global Health Network2 3 called the global health information server (GHIS). This has evolved as the result of the information superhighway2 and the internet's world wide web,4 which have transformed the ergonomics of access to electronic information. Before these technologies accessing the internet was (and still is for many users) "irkonomics." They had to remember addresses and search lists and know what to type and when. Now you just need to read, point, and click. How will the world of health information look when every original paper, letter of criticism, and review article, as well as every form, chart, and database in the computerised world, is accessible with a couple of dozen clicks of a mouse? We envision new forms of transmission of research communications.
The peer review system
The current peer review system involves prospective peer review: articles submitted to journals are subsequently sent to reviewers in the field, and these reviewers give comments and priority scores. The reviews are sent back to the editorial board for the final decision. If it is positive, the papers are posted back for changes, sent back to reviewers, and eventually published. Often there is a 12-16 month gap between submission and publication. The reviewers and the editors act as gatekeepers, and in some of the leading journals, such as the BMJ, Lancet, or New England Journal of Medicine, only about 10% of the articles are published. As scientists we want to publish in these journals because we reach a large number of people (the BMJ, for example, has a subscription of 110000); the articles in these journals are heavily cited, thus we have a greater impact on the field; and publishing in these journals is very prestigious when we look for tenure, grants, and new positions. After publication, journals are then sent to individual subscribers and libraries throughout the world. For an individual to subscribe to the Lancet, New England Journal of Medicine, and the BMJ it costs over $400, which is more than the per capita income in many developing countries. The journals accumulate quickly, occupying large amounts of space; many eventually find their way to the dustbin, an enormous waste of natural resources.
Global information servers
It is time to develop the architecture for a new global health information server, as all kinds of benefits could result. To do this, we can learn from what has been developed in physics, as this field is the first to have the courage to move beyond the print format and a print "mentality." It has developed the first generation of electronic research communications, and can serve as a starting point.
In physics Paul Ginsparg in 1991 started the first electronic research communication system with virtually no budget.5 It has grown rapidly and now has a user base of more than 40000, with the information retrieval on the system being the primary mode of information technology. The hep-th (high energy physics theory) is an automated system on the internet (see http://xxx.lanl.gov/). The system allows the community of research users to electronically submit and replace research communications and to search listings for subject and authors. It also allows ongoing corrections and addendums. When a researcher writes a communication, it is electronically submitted and archived. Users who subscribe receive a daily listing of titles and abstracts of new communications and can use various search engines to identify information of interest and obtain the communications electronically. This system is not an electronic journal, but rather an electronic archive of scientific research communications. There is no peer review in the system. The system runs smoothly, for virtually no cost. Why can't we do this in health, and in fact improve on this system?
Most people in health care have access to the internet. The storage of information is not a problem any more--the cost of memory has gone down very rapidly, such that collecting all the biomedical literature is indeed feasible.
A global health information server
The work in physics was developed with 1990 technology; if we begin there and move to 1995 technology there are truly exciting possibilities. A first step would be to archive existing and new papers electronically. Currently most papers worldwide are created on some form of word processor. It is not difficult to convert these to standardised coding of the text. The files can be transferred to the global health network information server by using the internet. The articles can be fully text searched with the wide area information system (WAIS), and the article retrieved. As with physics, daily listings of titles and abstracts would be distributed through the internet. The listings would be tailored to the interest of the readership. The search functions in the electronic document server will revolutionise publications, as they can be used to search electronically through text to find words and concepts, something that is not possible with paper based systems. This addresses one of the most important areas in research: that of information discovery. There is the concern, however, that the articles might have to be peer reviewed; thus we need to consider new technologies for peer review. This leads into the second step.
It is simple to develop a peer review system; before an article is put in the archive, it is reviewed and then adde to a moderated list. Opinion about whether this is needed is divided. Perhaps a better approach would be to make this a truly democratic system. When a title, abstract, or article is pulled up, a comment card can also be seen. This rates the article as being in the top, middle, or bottom third of articles in this subject. Also, comments can be attached to the abstract or article, or directed exclusively to the authors of the paper. In this system all the readers of the paper serve as the reviewers of the paper. Individuals from developed and developing countries alike could comment or ask for comments from the author or the other readers. When pulling up the abstract or the title so as to decide whether to read the paper, the reader could also pull up the average "priority score" as well as the comments. The selection could be based on reviewers who are researching that topic or are otherwise concerned with the content--for example, statisticians. If papers are poor, then the scientific community will most certainly indicate that they are poor; this is the nature of science, and this is the nature of the internet. This is analogous to choosing a film on the basis of film reviews in the newspaper or on the basis of surveys of filmgoers.
An important component of the current journal system is prestige and recognition. How should this be incorporated--if at all--into the proposed system? In the ideal world we as scientists would just communicate with one another, and prestige--"impact"--would not be relevant. We could design the system idealistically to develop the most efficient way to communicate. The paper journal system outlined above has within it measures of quality and impact. In the proposed electronic system, quality would be determined by the ratings and comments of the readers, perhaps the most democratic means possible, as everyone can "vote" with ratings and comments. Impact can be examined in several ways. The simplest is a count of how many times the article is retrieved. A second measure would be analogous to a citation. In a hypertext system it is easy to see how many times a research communication is referenced and in fact retrieved. The greater the number of times a communication is referenced and retrieved, the greater (presumably) the impact. This system would be in many ways fairer than that which exists now for judging articles; for someone coming up for tenure, the perceived quality of the papers, and impact by researchers in the same field and in other fields, can be measured directly.
Coping with information overload
The reader who wants to examine only "good" articles, or articles that are being read by many people, could easily have a filter on the articles to screen out, say, articles with a priority score of less than 2-3 or those read by fewer than 400 people a week. In this system the peer review is not left to a handful of people but involves the whole scientific community.
Coping with information overload
A major problem of the information superhighway is too much information. In the global health information server there can be software agents that learn what the reader selects to read. Thus if a reader typically chooses diabetes articles and medical technology articles, the agents will learn to select these, sifting out articles of no interest to that reader.6
Redesigning research communications
In general, electronic journals have failed as they have tried merely to take the print format and make it electronic, rather than creating a new, much more powerful and attractive format for biomedical research communications. The nature of the information superhighway also allows us to redesign the mode of information transfer--the journal article itself. We are not dealing with journal articles any more--we are dealing with research communications among scientists. Thus the terms "article," "paper," and "publication" should die.
The latest internet technology uses the web/hypertext language. An article might have certain terms highlighted in the text: if the Cox proportional hazard model were highlighted, a click on this would bring up information about the Cox model. Citations would not be needed: a click on "Jones 1994" would pull up the complete article. A click on the data in the article would allow direct access to the stored data. We need not be constrained by text alone; a click on "Ron LaPorte" could bring up a video with him saying how important the work is.
We need to start to consider a brand new format for research communications, as many of the constraints of paper journals are being lifted. As the terms "paper" and "article" convey a two dimensional print nature, it is best not to use these terms, but rather "research communication." In the new paradigm, research communications are n dimensional as one can "jump" from one idea to others, as in the example just given. The nature of the communications also differs in that there will be a change in philosophy in that a print journal article is permanent and almost impossible to change. With the new technology, as comments are received and new analyses done, the authors can make ongoing changes to the communication. Instead of scientists punctuating the corpus of literature with their papers, as happens now, they can construct ever changing stories as they do their research, and through hypertext these stories can be much more easily seen in the web of the total information in that discipline.
We have grown to accept the technological advances of adding colour to old black and white film; with technology we can change and update our research communications. Why do research communications have to be "permanent"? Why shouldn't they change as new information becomes available? We could also incorporate video and sound as part of research communications. Scientists will begin to demand information such as this. There could be a parallel archiving system for health information from the lay press, and lay people and scientists alike could go from the lay press to the scientific literature at will. Because these things are now possible, we believe that not only will the journals be replaced, but their nature and format will evolve away from conceptualisations of the paper or the article into something much more powerful.
Virtually all things are possible
Moving beyond this, we envision a virtual information facility with the establishment of a health "MOO" (multiple-use dimension object oriented).7 MOOs permit scientists the opportunity to "chat" in real time, share data and information, and present research. In the area of research communication, a user could "walk" down a virtual corridor and enter the public health room and browse through the articles there. These virtual systems are just being developed and hold considerable opportunity. Users could also strike up conversations with other people in the room. They can find the information they need as well as identify people with common interests, such as those who read the same articles. In the room they could also take classes and obtain a degree in the Global Health Network University8 with classmates from around the world. They could simultaneously show new research communications to their classmates. A global health MOO thus represents the integration of varied research communication systems with collegial interaction and training across the world.
An illustration of the power of the primitive electronic systems that already exist is the response to a letter we recently published in the BMJ requesting that people contact us concerning participating in the Global Health Network University.6 At about the same time as it was published in the journal it was sent to several lists on the internet. The publication in the BMJ brought six letters and two electronic mail messages. The response from the internet was overwhelming: 147 email letters. Few of us receive more than a handful of letters responding to our articles, yet a short electronic letter received an electronic response almost 20 times that of print format.
Compared with the paper journal system as currently constituted, the global information server most certainly is faster; research findings will be presented to scientists in days rather than years. It is also less expensive and potentially more equitable, at least for the type of information transfer entailed in a short research communication. The new technology will foster much more of an interchange among scientists through electronic mail. The internet is spreading rapidly to almost all countries; thus researchers in developing countries will have access to communication that they do not have through print. For these reasons and others presented earlier, we believe that the "printing" technology of the journals will rapidly be obsolete. If journals are to survive there will have to be a profound transformation to n dimensional knowledge structures. The people with the journals are skilled and can lead in part this revolution. However, they need to shake themselves away from the current mind set.
It is time that scientists begin to take control of their research communication; the global health information server allows this to occur. We are just touching the capabilities of the systems. A discussion concerning the development of the system will become available on the global health net home page (http://www.pitt.edu/HOME/GHNet/GHNet.html). We invite you to join the dialogue.
We should consider the development of a trial system for a specific discipline, such as epidemilogy or public health. This is similar to the development of the server in physics. We need to develop this not only with people in health but also with groups, such as the NASA Scientific Internet (NSI) and industry, who are more advanced than we are in telecommunications and who will allow us to take rapid advantage of the existing as well as future technologies. It is time to move beyond the vinyl records of journals articles to the CDs of research communication on the electronic information superhighway.