Epilepsia, 45(8):889–894, 2004 Blackwell Publishing, Inc.  C 2004 International League Against Epilepsy

Editorial

Peer-Reviewed Publication: A View from Inside The following letter commonly results in disappointment, frustration, puzzlement, and anger, especially in young academicians.

actions of the Royal Society (1). The first editor, Henry Oldenberg, wrote that careful review of a paper is required “before we give a publick testimony of it to ye world, as is desired of us.” In the same article, Tobin indicated that Sir Isaac Newton had such a rough time with the review of his submissions that it “sacrificed my piece [of mind], a matter of real substance.” Novice authors should take heart; even Newton had trouble with reviewers! In this article, we examine critically the review process and offer suggestions that may be useful to some authors and some reviewers. Our opinions result from years of participating in the process as author, reviewer, and editor, and recently as editor-in-chief (R.S.F.) and managing editor (L.E.P.) of a medical journal.

Dear Dr. Hopeful, We regret to inform you that we are unable to accept your paper for publication in our journal. Although the reviewers, whose comments are reproduced below, found some merit in your article, it did not achieve a priority sufficiently high to justify publication. With apologies, this decision is final. Publication matters for several reasons. First, it brings the ideas of scientists and clinicians into public awareness. Second, it stakes a claim of ownership of the ideas and data within the international medical and scientific community. Third, publication in a peer-reviewed journal demonstrates that the work has passed critical scrutiny. Most university promotion committees consider peer-reviewed publications seriously, even if an author’s primary responsibility is clinical. Other factors related to publication that influence promotion committees include the total number of publications, the citation indices of the various journals, whether the individual under consideration is first author, and if the work is independent of studies completed during training. Non–peer-reviewed publications are not ignored by any means, but chapters, books, and computer-based and audiovisual materials are not counted so highly as are critically reviewed original research articles. An excessively high ratio of abstracts to publications is often interpreted as indicating that the faculty member does not complete work. As of this writing, PubMed covers 3,799 journals; the Ingenta journal service reports on 27,915 journals and almost 15 million articles; and the ISI Journal Citation Reports covers 8,400 journals, of which 137 are related to clinical neurology. The acceptance rates for these journals vary markedly, such that Nature, Science, New England Journal of Medicine, Lancet, and other highly competitive journals accept between 1 and 15% of submissions. Other journals accept a large majority of articles submitted. By comparison, Epilepsia currently accepts 25% of submitted articles. Given the importance of the peer-review process to scientific publication, surprisingly little attention has been devoted to evaluating its use by journals. Peer review was in place as far back as 1665 for The Philosophical Trans-

THE REVIEW PROCESS Manuscripts submitted to Epilepsia are uploaded to a web server. The editor-in-chief, sometimes together with a small advisory group, makes a decision whether to proceed with external peer review. Through experience, the editor may recognize papers that have a very small or no likelihood of achieving a publishable priority score from external reviewers. In such cases, administrative rejection saves time and effort for everyone. In theory, an editor could accept a paper without external review, but such an occurrence is extremely uncommon. Typically, two or three expert reviewers are chosen to review papers electronically through the journal’s web site. Different editors use different methods for obtaining reviewers. Time-honored methods include skimming through the bibliography of the submitted manuscript, choosing professional colleagues who work in the same field, matching the subject to the expertise of the journal’s editorial board, and accessing internet databases to identify authors working in similar areas. Some editorial offices send papers to potential reviewers and hope to receive reviews, whereas others ask potential reviewers in advance if they are willing to review manuscripts. Epilepsia queries reviewers by e-mail first to gauge interest and response time. Editors ask that reviewers return their comments within time frames ranging from 1 week to several months. Delays in returning decisions to authors are usually due to tardy reviews. Reviewers contacted by fax, telephone, or e-mail a week after deadline will often take an additional 889

890

R. S. FISHER AND L. E. POWERS

7 days to complete their reviews; some reviewers may never respond (2). Performing anonymous reviews is a service with little reward or visibility, so editors can do correspondingly little to mandate a timely return. Reviewers are asked to make a recommendation as to immediate acceptability, acceptability after minor revision, reconsideration after major revision, or rejection. Different journals use different versions of this categorization. Some ask for stratification into priorities, such as top 10%, 10 to 30%, etc. In our view, such numeric categorization is questionable for reviewers who only see a small number of articles submitted to a given journal. Specific comments are solicited on the length of the article, number and suitability of tables and figures, references cited, and format. Reviewers also are given the opportunity to make confidential remarks only to the editor as well as specific comments that are sent to the authors. On some occasions, the confidential comments are negative, whereas the transmitted comments are benign or even modestly positive. The editor then must explain to the authors the seemingly mystifying reasons the manuscript was rejected. Reviews returned to the journal may be concordant or discordant. Few studies have evaluated the consistency of manuscript rankings among different reviewers. One analysis (3) examined the concordance between pairs of reviewers recommending acceptance, revision, or rejection. Agreement was nonsignificant at p = 0.08 for 179 consecutive papers submitted to one journal, and p = 0.28 for 116 papers submitted to a second journal. An editor has two choices in the face of discordant reviews: send them out for additional reviews, or adjudicate and decide. Which course is taken depends on the nature of the individual reviews and the expertise of the editor or associate editor in the particular subject area. A second round of reviews delays response time to the author. Furthermore, reviewers often diverge again. Despite this lack of concordance, peer review probably can distinguish very bad from very good papers. In the vast middle is a significant random element and “luck of the draw.” Authors may suggest a few reviewers. Some of these may be used, but the editor often assumes they are friendly reviewers. It is also reasonable for authors to ask that certain reviewers be excluded for conflicts of interest. Most editors act on the recommendations of their reviewers, but also reserve the right to override the opinions of the reviewers. If the reviews place the paper in a potentially publishable priority range for the journal, then the reviewers’ comments are sent back to the authors for incorporation into a revised manuscript. A cover letter is requested, documenting the changes and responses. The author has the option of documenting reasons not to follow every reviewer suggestion, but editors become quite good at detecting whether the key suggestions have been taken. If the paper is sent back for revision, the authors know that the editors hope the paper can be published. It is painful for Epilepsia, Vol. 45, No. 8, 2004

all to have a manuscript rejected after a round of revisions, but it happens if key changes are not made. Editors are regularly impressed with the insight and grasp of detail shown by most reviewers. However, many reviews focus only on certain features of the manuscript, and ignore others. In so doing, reviewers may miss key errors. One rather distressing work (4) circulated a fake study to 285 reviewers for the Annals of Emergency Medicine. The study claimed that intravenous propranolol helped migraines. Intentional errors included no definition of migraine, no inclusion/exclusion criteria, faulty randomization, use of an unvalidated pain scale for outcome, no data on safety and adverse events, impossibly favorable p value given the numbers, latest reference of 1989, and duplicate Table 1 and 2. Of the 185 reviewers, 108 appropriately rejected the manuscript, 62 requested revision, and 15 accepted this flawed piece of work. Such findings emphasize that the review process is valid only if reviewers are knowledgeable and careful. Unanimity is not required. The process catches some, but clearly not all serious errors. Oversight is required by an editor-in-chief. Aside from the difficulty of receiving a timely review at all, the greatest problem that an editor faces with reviews is bias. Every expert reviewer is biased to some extent. We are unaware of a study correlating favorability of reviews with concordance of the reviewers’ prior work, but it would not surprise us were there to be such a correlation. What reviewer can resist flipping to the bibliography to see if his or her work has been cited, and in what context? What reviewer is so emotionally centered that a positive or negative citation to their prior work does not have some influence on the review? It is the editor’s task, where possible, to perceive and compensate for such bias. Doing so is not always possible. WHAT CONSTITUTES A USEFUL REVIEW? Table 1 summarizes the elements of a useful review. TABLE 1 1. Is there a clear hypothesis? 2. Is the question important? 3. Is the subject of interest to the readership? 4. Is existing literature cited appropriately? 5. Are the methods (design, statistics, inference) sound? 6. Are the figures of high quality? 7. Is the length suitable for the message? 8. Is the writing unambiguous and properly formatted?

Few reviews encompass all of these elements. Reviewers tend to use one of several styles. We call the first “the judge,” who gives a brief go or no-go review with little detail. At the other end of the spectrum is “the silent coauthor” who provides a detailed revision of the paper in the review. The “strategic reviewer” focuses on the hypothesis and whether it is established. This is a useful feature of a review but does not always provide specific feedback

PEER-REVIEWED PUBLICATION to the authors for use in their revisions. A fourth type of reviewer is the “uncritical friend,” who generates inappropriate and nonspecific enthusiasm for the article. “The egotist” focuses on the paper’s treatment of the reviewer’s prior works. Elements of several of the styles combine to form the ideal reviewer, who would provide a strategic review, plus specific helpful comments directed toward the questions listed in Table 1. In science, hypothesis testing results in affirmation or rejection of the hypothesis. Either outcome can provide important information. Journals do publish negative results, provided that the method for obtaining them is good. Among manuscripts submitted to JAMA (5), 20.4% of those with positive results and 15.0% of those with negative results were published, a nonsignificant difference. However, authors may be less likely to submit studies with negative outcome for publication. Should reviewers be anonymous? This issue is debated and has been the subject of published literature (6). In the future, electronic publishing and preprint web servers may allow open commentary and real-time review. Current studies indicate that signed reviews are of high quality, but it is much harder to obtain reviews, takes longer for reviews to be returned, and generates a higher proportion of positive reviews. Candid negative assessments are of great value to the decision-making process of the editors. Such assessments are sacrificed with signed reviews. For that reason, most journals persist in the use of anonymous reviewers. The converse question of whether authors should be anonymous to reviewers is debated among journal editors. Some journals obscure the name and institution from the papers before sending them out for review. Studies suggest that about one third of “anonymous” papers are easily identified by reviewers because they are familiar with the work from abstract presentations or other sources (7). Effective blinding assumes that reviews will be less biased and free from revenge and reward motives, therefore leveling the playing field. The negative side of author blinding is loss of information about a coherent body of that author’s prior work, and sacrifice of the possibility that authorship experience and prior record should be considered in the ranking of the paper. Peer review is a flawed process. Like democracy, in the eyes of Winston Churchill, it is the worst system, except for all the rest. ADVICE FOR AUTHORS Authors should serve their manuscripts to reviewers like a well-presented meal on a tray. A good title is descriptive. The abstract should be specific. Tables and figures should illustrate the findings at a glance. Reviewers are distractible and busy. An effortful review is more likely to be a negative review. Keep the introduction brief. Papers

891

are generally too long. The discussion should be shorter than the results; otherwise, the author is not focusing on the data needed to support the theory. Table 2 lists the “seven deadly sins” of authors. TABLE 2 1. Failure to state a clear hypothesis 2. Excessive ambition and lack of focus 3. Expositing on a minor new wrinkle 4. Invoking the least publishable unit 5. Bad trial design 6. Overgeneralizing applicability of findings 7. Bad writing

To illustrate these shortcomings, we give examples, some of which are real examples, and others created to make a pedagogic point. 1. Failure to state a clear hypothesis We have examined 23 children with seizures after measles– mumps–rubella vaccinations. Seizures occurred within 2 weeks of the vaccinations in 18 of the children . . . When reporting new and important material, or for certain types of epidemiologic studies, it may be appropriate to issue a purely descriptive paper. But the vast majority of submitted manuscripts are not opening new frontiers. For these studies, a clear hypothesis—one that can be accepted or rejected by the end of the manuscript—is important. Present the research problem, and ground it in scientific theory. A common reason for a negative review is difficulty in discerning the hypothesis of the paper. 2. Excessive ambition and lack of focus In this study, we evaluate the effect of antiepileptic drug therapy on the quality of life for people with epilepsy. The study population consists of 100 consecutive patients in our clinic started on an antiepileptic medication. Quality of life is a big area, with many subdivisions. Antiepileptic drugs (AEDs) vary substantially in their side effects. This paper will not produce a clear answer pertaining to quality of life for only 100 patients in one clinic, with many seizure types, many lifestyles, and taking various AEDs. The subject of a submitted manuscript should be focused. Authors should limit the question and trim all material not directly pertinent to that question. 3. Expositing on a minor new wrinkle Postictal nose-wiping has been shown to have localizing value for an ipsilateral seizure focus (five citations). In this study, we report a patient who exhibits postictal lipwiping. Some studies present new material, but so little beyond existing boundaries of knowledge that the usual reviewer or reader reaction is a shrug of the shoulders. Production of Rocky I–V may be useful in Hollywood, but not in competitive science. A special example is that of the case Epilepsia, Vol. 45, No. 8, 2004

892

R. S. FISHER AND L. E. POWERS

report. A case report may be a junior author’s first foray into the field of publishing. Unfortunately, most case reports are not worth publishing and fall into the category of expositing on a minor new wrinkle. Journals are wary of case reports, because they generate very few citations. Possible exceptions include cases with unique findings, broad interest to readers, difficult-to-obtain data, and clear cause–effect relations. 4. Invoking the least publishable unit Zoxx (fake name) is a new drug with promise for the treatment of epilepsy. Herein we report two articles detailing findings in a pilot series of nine patients. The first summarizes the efficacy of Zoxx, and the second reports our experience with the side effects of Zoxx in our pilot trial. Experience and judgment are required to determine when to halt data acquisition and prepare for publication. Mistakes can be made in both directions. Attempting to publish too little too soon can result in rejection of the paper. The usually transparent method of breaking one paper’s worth of material into two or three papers (presumably to provide more lines in the authors curriculum vitae) usually results in negative review or a request to combine the works into one paper. In the other direction, some authors seem never to be quite ready to publish and accumulate several manuscripts’ worth of material before submission. Junior authors are encouraged to seek advice from more experienced colleagues on how much should be included in a manuscript. 5. Bad trial design The association between nonepileptic seizures and vocalizations was examined. In total, 30 seizures were examined in seven patients. Vocalizations occurred in 45% of the epileptic and 58% of the nonepileptic seizures (p = 0.20, not significant). Textbooks have been written on good and bad clinical trial design, and space is insufficient here to list all the ways to go wrong. This example illustrates a particularly common error, the underpowered negative study. Seven patients is not a large number. How likely is it that the two groups do differ, but sample size was too small to detect the difference? A sample-size calculation is necessary to know how the study should be powered. How was each group defined and entry validated? What statistical test was used to distinguish the outcome variables and why? Is this a retrospective or prospective study? If retrospective, what are the potential mismatches between the two groups? The example also misuses repeated-measures statistics. One of the seven patients may have generated many of the seizures and have been atypical for the other patients. Poor trial design is a common reason for negative reviews in clinical studies. International committees have derived uniform requirements for manuscripts (8) from a meeting in Vancouver, Epilepsia, Vol. 45, No. 8, 2004

British Columbia, in 1978, and for clinical trials (9). Uniform formatting requirements are listed in instructions to authors. They include requirements for permission to reproduce published material, protecting patient confidentiality, and meeting the ethical guidelines of the Helsinki Declaration of 1975, revised in 1983. Introduction, methods, results, and discussion should be kept separate. Consortia requirements give guidance as to the information that should be present in the title and suggestions for a structured abstract comprising background, methods, results, and conclusions. The methods should be sufficient to describe the study population, inclusion criteria, outcomes, statistical methods, randomization method, and blinding. The results section should indicate the flow of the patients through the trial, the nature of the follow-up, and the numbers studied according to the protocol and with deviation from the protocol. Discussion should comment on bias, precision, links to prior literature, and interpretation of results. References should be in order and correct. 6. Overgeneralizing applicability of findings We investigated the prevalence of nonepileptic seizures. Among 100 consecutive referrals to our University Epilepsy Center, 40% subsequently were shown by videoEEG monitoring to have nonepileptic seizures. Therefore we believe that nonepileptic seizures are more common than previously has been realized. Every important study is performed on a subgroup of the real population of interest. Clinical manuscripts tend to come from academic centers that see a highly selected referral population, not typical for the community at large. Nonepileptic seizures, for example, are probably uncommon in the general population, whereas they can be the majority of seizures in an academic epilepsy-monitoring unit. Many papers are rejected because their results do not appear likely to generalize to a larger population of interest. Even if results do so generalize, the discussion section should include caveats as to whether the study sample truly is representative. The author of a data-based publication should set results in the context of existing literature, but base conclusions on the work of the paper, not a synthesis of the literature. 7. Bad writing Seizures which result from a myriad variety of underlying factors that complicate management must be viewed in the context of the complete picture of the patient’s health and quality-of-life. This is a fabricated example, but many real paragraphs of impenetrable prose could have been used in its place. A manuscript is a medium for communication. It requires both a message worth communicating and a presentation that conveys a message understandable to the reader. Unnecessary complexity obscures the meaning. Begin each paragraph with a topic sentence and maintain a

PEER-REVIEWED PUBLICATION consistent, logical flow of scientific terms and ideas. Use a more conversational active voice. Think, write, and speak clearly. Read the words out loud. Good composition is like the precision of a scalpel in the hands of a skilled neurosurgeon. Science and writing are entwined. If neurologists find some peer-reviewed articles in Epilepsia tedious to read, imagine the frustration of researchers in other fields. Specialization and incomprehensible text have led to a disturbing trend in many basic science journals away from open scientific discourse and the free flow of ideas (10). Sometimes the problem is not just bad writing, but also a language barrier. If the language of the journal is not the native language of the author, then serious consideration should be given to having a native speaker (one who is a good scientific writer) rewrite the draft before submission. Editors and reviewers do not consider it their responsibility to polish the language in a submission, although they may do so to some extent. Poor writing and inaccurate data can result in rejection of a request for research funding as well as manuscript publication. Can the study be replicated? Accuracy of data and meticulous attention to detail are essential. Numbers in the text must match the data presented in the tables and figures. Statistics reflect how well or how poorly the research was conducted. SCIENTIFIC MISCONDUCT Albert Einstein is quoted as saying, “Many people say that it is the intellect which [sic, that] makes a great scientist. They are wrong: it is character” (11). Various types of scientific misconduct or fraud find their way into the literature. We do not know if this is a growing trend, but the ability to copy and paste electronic text may contribute to the problem. Scientific misconduct is a frequent topic of discussion among journal editors. The definition of scientific misconduct of the American Medical Association Manual of Style encompasses misappropriation, misrepresentation, and interference. Misappropriation is plagiarism. Recently an author submitted a manuscript to Epilepsia that was virtually identical to an article published in another journal. The major difference was the gender of the patient. A reviewer was the author of the original article and pointed out the similarities. Most authors recognize that extensive quotation without attribution is plagiarism. Fewer authors recognize that copying a complex sequence of headings, subheadings, and manuscript organization also can be considered misappropriation. Use of figures, tables, and charts from another publication is acceptable once permission is obtained, ideally from the author and the journal editor. Autoplagiarism, defined as copying from oneself, may be ethically superior to copying from somebody else, but it

893

still represents duplicate publication. Table 3 enumerates ways to avoid misappropriation. TABLE 3 1. Attribute sources 2. Put quotes in quotations 3. Do not extensively paraphrase 4. Use a unique sequence and organization 5. Obtain permission for figures, tables, and charts 6. Avoid autoplagiarism 7. Carefully consider coauthorship

Reviewers may react negatively to extensively reproduced material in an original article. Reproduction of previously published work tends to be more appropriate for review articles and editorial material. An author should carefully consider coauthors. Epilepsia published a paper in which one author belatedly discovered that the second author had misappropriated material for the paper. Unfortunately for the first author, Medline will list the retraction under both names. A separate question from one of misappropriation is that of who deserves coauthorship. Authors and editors will disagree on this issue in particular instances. Our view is that authorship should be awarded according to who performed the work. It requires an intellectual input, such as planning the study, participating in writing or rewriting, or making a key observation leading to the manuscript. Funding and running the laboratory in which work is done also constitutes grounds for authorship. Acknowledgments, rather than authorship, may be more suitable for a clinician delivering ordinary care for the patients in the study, clinical consultants, statisticians, and other associates of the lead author. Reviewers should excuse themselves from the review process if a potential conflict exists with their own unpublished work. Borrowing ideas is misappropriation. Other types of scientific misconduct are misrepresentation and interference. Misrepresentation is intentionally falsifying, fabricating, or omitting information. Interference is seriously compromising a researcher’s work by taking, hiding, or damaging property, including data and software. THE BUSINESS OF PUBLISHING Authors and editors are, for the most part, scientists, clinicians, or both. Editors serve as volunteers or sometimes receive token payments for services. Reviewers perform selflessly and almost entirely in the shadows. However, the journals themselves represent businesses, and successful journals generate profits. These profits may flow to publishing companies or to scientific societies. Many of the societies on which neurologists come to rely, Epilepsia, Vol. 45, No. 8, 2004

894

R. S. FISHER AND L. E. POWERS

in turn rely on income from subscriptions and advertising in journals owned by the societies. Success of a journal usually is judged by numbers of subscriptions and by its citation impact factor. The Institute for Scientific Information (ISI) in Philadelphia derives the impact factor and publishes a comprehensive database called Current Contents. The citation index is calculated as the number of citations in the past 2 years to articles in the journal, divided by the number of citable articles published by that journal (exclusive of comment letters and notices). Journal citation indices range from less than one to more than 50. The citation index is the “Nielsen ratings” of the journal business. It is, however, subject to some problems. For example, 2 years may not be the best time frame during which to determine the importance of an article, self-citations account for nearly 25% of some indices, review articles or method articles tend to be the most cited, and articles can be cited to correct errors in the original paper. Most important, no scientific study links the citation number to the importance of an article. Many editors keep one eye on the citation index, but also watch out for those submissions that seem to be important or likely to be popular. The success of the Internet and electronic publishing have enabled the international community to access scientific literature, most notably abstracts. The media have carried several stories recently on the open press movement (12). The open press movement argues for immediate public availability of all published works. Paid subscriptions and copyright restrictions would be eliminated. Minnesota Democratic Representative Martin Sabo has proposed in the U.S. Congress that government-funded research be available to all for free, and copyright not be allowed. In our opinion, the open press movement has made some justifiable points. Authors and readers want work to be available as widely and rapidly as is possible, but no tested financial model explains how journals could be funded in this system. Whether payment by the authors, universities, sponsoring societies, or the granting institutions would maintain the current diversity and quality of publication remains to be seen. Many submitted papers do not come from funded sources. In addition, almost all research already is immediately available, but at a price. Most researchers have free access through academic libraries. The lay public does not now have immediate free access. This debate will continue, but the increasing movement of publishing to electronic media likely will change the current financial models used for medical and scientific journals.

Epilepsia, Vol. 45, No. 8, 2004

CONCLUSIONS Publishing an article can be a trying experience. The peer-review process is flawed but well entrenched in academia. It is helpful for an author to start with something new and important to say and then to say it well. An author should be the most critical reviewer of the paper and should be scrupulous about the study’s limits and dependence on work of others. A publication confers a small piece of academic immortality but, in the process, also can preserve errors, omissions, and misinterpretations. If the manuscript represents good science and is well expressed, then it is likely to be published within the wide diversity of medical literature. Acknowledgment: Prof. Robert S. Fisher was supported by the Maslah Saul, M.D., Chair and the James and Carrie Anderson Epilepsy Research Fund. We thank Dr. Richard T. Johnson for critically reading the manuscript.

Robert S. Fisher Linda E. Powers Departments of Neurology and Neurological Sciences Stanford University School of Medicine Stanford, California, U.S.A.

REFERENCES 1. Tobin MJ. Rigor of peer review and the standing of a journal. Am J Respir Crit Care Med 2002;166:1013–4. 2. Pitkin RM, Burmeister LF. Prodding tardy reviewers: a randomized comparison of telephone, fax, and e-mail. JAMA 2002;287:2794–5. 3. Rothwell PM, Martyn CN. Reproducibility of peer review in clinical neuroscience: is agreement between reviewers any greater than would be expected by chance alone? Brain 2000;123:1964–9. 4. Baxt WG, Waeckerle JF, Berlin JA, et al. Who reviews the reviewers? Feasibility of using a fictitious manuscript to evaluate peer reviewer performance. Ann Emerg Med 1998;32:310–7. 5. Olson CM, Rennie D, Cook D, et al. Publication bias in editorial decision making. JAMA 2002;287:2825–8. 6. Godlee F. Making reviewers visible: openness, accountability, and credit. JAMA 2002;287:2762–5. 7. Katz DS, Proto AV, Olmsted WW. Incidence and nature of unblinding by authors: our experience at two radiology journals with doubleblinded peer review policies. AJR Am J Roentgenol 2002;179:1415–7. 8. International Committee of Medical Journal Editors (ICMJE). Uniform requirements for manuscripts submitted to biomedical journals. N Engl J Med 1997;336:309–15. 9. Schulz KF. The quest for unbiased research: randomized clinical trials and the CONSORT reporting guidelines. Ann Neurol 1997;41:569–73. 10. Hayes DP. The growing inaccessibility of science. Nature 1992;356:739–40. 11. Martyn C. Fabrication, falsification and plagiarism. Q J Med 2003;96:243–4. 12. Malakoff D, Bachtold D. Electronic publishing: who owns, who pays? U.K., U.S. offer answers for journals. Science 2003;301:29.