Prepublication Release of Medical Research

Exactly when and how findings from medical research studies should be disseminated to clinicians, other researchers, and the public is a critically important yet contentious issue.^{1 - 4} The tradeoffs between expediency and ensuring quality and accuracy have long been recognized and debated but have come under increasing attention recently, particularly with the use of electronic media for communication of medical research findings. Medical researchers, especially those involved in major studies, usually want to communicate the results of their research as widely and as quickly as possible. Journalists, well aware of the public's "right to know" and seemingly insatiable desire for information about health and medicine, are interested in promptly reporting medical news, especially about potential "breakthrough" research findings. However, journal editors, responsible for ensuring that the scientific information they publish is assessed for quality by peer review and editorial evaluation, generally discourage prepublication release so that the entire study, with full methods and results as well as proper caveats, can be appropriately evaluated and interpreted.

An important issue is whether the release of research findings prior to formal peer review and publication of the complete study leads to the appropriate or inappropriate adoption of new findings into clinical practice and whether such practice changes are supported by evidence from the study. In this issue of THE JOURNAL, Gross and colleagues⁵ report their evaluation of changes in the use of carotid endarterectomy (CEA) associated with various methods and times of disseminating the results of 2 pivotal clinical trials in stroke prevention: the North American Symptomatic Carotid Endarterectomy Trial (NASCET)⁶ and the Asymptomatic Carotid Atherosclerosis Study (ACAS).⁷

Both studies were stopped early because interim analyses showed evidence of efficacy of CEA in reducing the risk of ipsilateral stroke. In NASCET, the absolute risk reduction was 17% over 18 months of follow-up for symptomatic patients with 70% to 99% carotid artery stenosis; in ACAS, the absolute risk reduction was 5.8% over 2.7 years of follow-up for asymptomatic patients with more than 60% carotid artery stenosis. The National Institutes of Health (NIH), which funded both studies, issued prepublication clinical alerts for NASCET⁸ and for ACAS⁹ to notify clinicians of the study results. The clinical alerts preceded publication of the full articles in medical journals by 6 months and 7.5 months, respectively.

In their longitudinal analysis of hospital discharge data of 272,849 CEAs performed in 7 states from 1989 (2 years before the NASCET clinical alert was released) through 1996 (2 years after the ACAS clinical alert was released), Gross et al found that the CEA rate increased significantly following release of the NASCET clinical alert, but the rate of increase slowed after journal publication of the study. Likewise, the CEA rate increased significantly following the ACAS clinical alert, then declined after publication of the study. In addition, the authors found that the CEA rate increased more for patients older than 80 years (who were not included in either study) than for younger patients and that the proportion of CEAs performed in low-mortality hospitals (the setting recommended in both studies) did not change substantially, suggesting that the study results may have been extrapolated to patient groups and hospital settings not directly supported by evidence from the trials.

Several factors might help to explain the trends observed by Gross et al.⁵ First, both NASCET⁶ and ACAS⁷ addressed a relatively common clinical problem (carotid artery stenosis) for which effective treatment was limited and for which evidence-based interventions proven to prevent a potentially devastating outcome (subsequent stroke) were just becoming available. Moreover, the indications for CEA and the relative benefits of medical vs surgical therapy for stroke prevention for patients with asymptomatic carotid atherosclerosis were the subject of much debate and controversy with respect to efficacy, safety, and appropriateness. Accordingly, the medical community eagerly anticipated the results of these 2 studies.

Second, although the findings of Gross et al⁵ regarding the association between prepublication release of study results and changes in CEA rate appear valid and robust, the changes in rates of CEA use observed following full publication of the studies might not have been related to journal publication. As the authors acknowledge, their analysis cannot establish causality, and it is possible that the observed declines in CEA rate simply may represent a plateau in procedure use that roughly coincided with publication. On the other hand, even though the major findings reported in the clinical alerts and the published articles were essentially no different, the full articles, in particular ACAS, included additional subgroup analyses and emphasized several clinical caveats that were echoed by accompanying editorials.^{10 - 11} It appears that the CEA rate changed fairly abruptly after ACAS was published, with a greater decline in rate of CEA use among patients older than 80 years, suggesting possible uptake of additional information related to the published articles.

Third, the results of NASCET and ACAS were widely disseminated prior to journal publication, not only by the clinical alerts but also through presentation at several major scientific meetings. Thus, physicians and surgeons had ample additional opportunities to review and discuss the possible benefits of CEA. In fact, investigators from the European Carotid Surgery Trial,¹² which showed similar beneficial results of endarterectomy for patients with symptomatic carotid artery stenosis, had communicated preliminary results to the NASCET investigators just before the NASCET clinical alert was released.¹³ In addition, these studies generated extensive coverage in the news media each time the prepublication results were disseminated (ie, following the release of the clinical alerts and following presentation at scientific meetings).

Moreover, the possible influence of additional information about CEA available around the time of the release of the results of NASCET and ACAS should be considered. Between the time of the first release of the NASCET to several months following journal publication of the ACAS results, numerous articles and a multidisciplinary consensus statement on guidelines for CEA¹⁴ were published in medical journals. A PubMed search using the term carotid endarterectomy that covered most of the time period examined in the study by Gross et al⁵ (ie, from February 1991 through December 1996) showed 1463 citations. In addition, publication of NASCET and ACAS and other related articles in medical journals generated additional major press coverage. The continued appearance of articles on CEA in medical journals and continued coverage in the news media may have fostered ongoing discussion among physicians and surgeons regarding indications and patient selection and might have contributed to a more circumspect view of the anticipated benefit of CEA for reducing stroke risk in patients with asymptomatic carotid artery disease.

Clinical alerts are an effective mechanism for prepublication dissemination of research findings and, as Gross et al indicate,⁵ generate immediate attention. Clinical alerts are issued by the NIH through the National Library of Medicine to notify physicians and the public expeditiously about findings from NIH-funded clinical trials that are judged to have the potential to affect morbidity and mortality significantly.^{15 - 16} Clinical alerts enable urgent and widespread release of vitally important information but are used selectively. Since January 1991, only 22 clinical alerts and advisories have been issued.¹⁶ However, the wide distribution of clinical alerts to physicians and other health care professionals, medical specialty and professional societies, government agencies, and medical schools and libraries, combined with concurrent announcement to the news media through press releases and press conferences, undoubtedly generates much interest and consequently commands considerable attention.

In contrast to the infrequent nature of clinical alerts, presentation of research findings at scientific and clinical meetings is the most common form of prepublication dissemination of medical research. These presentations are an integral part of the orderly progression of scientific communication and scholarly discourse. However, they represent a preliminary exchange of information between researchers and other experts that generally is not designed for announcement to the public. Indeed, for many scientific meetings, acceptance of a paper for presentation is usually based on peer review of the work's abstract, not the full report. Research presented at scientific meetings, even though at times seemingly quite promising, should be considered preliminary until the full report of the study on which the findings are based is evaluated by rigorous peer review for scientific validity and clinical importance. However, a substantial proportion of abstracts presented at scientific meetings are never published as full articles.^{17 - 19} Nevertheless, despite the preliminary nature of the data and the potential lack of proper context and explanation of implications, information from presentations at scientific meetings has been touted by meeting organizers and investigators through press releases, press conferences, or exclusive interviews. These preliminary data have received substantial media coverage and dissemination, at times creating undue attention, confusion, and even untoward results.^{4 ,20 - 21}

Prepublication dissemination of research also occurs outside the realm of bona fide scientific meetings when researchers, their institutions, or study sponsors issue press releases, hold press conferences, or arrange interviews with the news media to unveil new study findings. Such "publication by press conference"²² of results that have not been peer reviewed formally are generally perceived as having limited scientific merit and might be viewed by some as an attempt to garner attention primarily to increase the chances for subsequent funding or to gain market share.²³

Medical journals have established policies regarding prepublication release of research findings^{24 - 25} to help ensure appropriate evaluation and dissemination of medical information. Some journalists (and some researchers) have criticized medical journals for these policies, suggesting that journals exert too much control over release of biomedical science. However, major studies often require several years for completion and an additional 1 to 2 years until the manuscript is submitted for publication.²⁶ Scientific journals go to great lengths to strive to ensure the accuracy and validity of information they publish, which also takes time.

Only complete publication of the research allows for full informed assessment and comment on the study findings. Accordingly, the best way to promote quality of scientific reporting, to increase the likelihood of proper application of study findings, and to help ensure patient safety for all medical research, is through rigorous peer review, careful editorial evaluation, and clear, objective presentation of study findings along with appropriate caveats in the published article and in accompanying editorials. Ultimately, and regardless of the inevitable professional tensions that may exist among researchers, journalists, and journal editors, they actually all share a common goal—ensuring accurate and timely publication of important medical research that will improve patient care.