Control of Data, Authorship, and Responsibility for Clinical Trials Publications

There is a growing sense of concern among the editors and editorial board members of several journals and among many ophthalmologists and vision scientists regarding publications of clinical trial data from industry-sponsored studies. These concerns relate to control of data and likely financial incentives to publish positive results on the part of drug company employees, people with financial interests in the company, or, possibly, the industry in general. Additional concerns relate to decisions not to publish certain analyses or entire clinical trials if they are negative. This editorial summarizes the results of a recent modified Delphi process survey of Ophthalmology editorial board members. It is hoped that by presenting to the readership some areas of consensus and lack of consensus, this editorial will serve as a format to express the concerns and offer possible solutions.1 In general, the editorial board endorses a call for greater transparency and disclosure in these areas, as is currently being done by, for example, the Journal of the American Medical Association, which requires reporting conflicts and relationships between investigators and research sponsors.2

Companies whose products are being evaluated have a vested financial interest in study outcomes. This truism at times conflicts with the values of our board members, who prioritize scientific knowledge and public health and well-being before commercial interests. What causes most discomfort is that decisions to submit trial results for publication and manuscript preparation, data, performance of and deciding upon statistical analytic strategy, and timing in general are typically under control of the sponsoring company, who has a major financial interest in the study results. It is not within the power of any journal to change the control over the study; however, the editorial board does feel that better disclosure of potential conflicts of interest may help our readers better interpret the articles that describe study results. In addition, our editorial board members are concerned that a pseudoindependent legitimacy to some clinical trials publications is implied or may be perceived when authors from academic or other nonaffiliated institutions are authors of such articles, particularly when no or few authors from the sponsoring company have names on the mastheads. This may lead the reader to conclude that these authors designed the study, had control over or access to the data, and guided or actually performed the data and statistical analysis. Often, this is not the case.

Almost always, ophthalmologists or vision scientist authors in clinical trials sponsored by industry do not have access to all of the data. With drug studies, for example, the data are voluminous, and a hardcopy version may fill a truckload of boxes. Seeing all the data is not feasible, and subsets of the data are typically shared with the investigators and authors. Similarly, authors may not guide statistical analyses. The statisticians are usually in-house employees of the sponsors. Requiring all authors to have access to data and the opportunity to be involved in analyses may be ideal, but at a minimum, particularly for multicentered clinical trials, at least one author, preferably the senior one, should provide a statement that he or she had full access to all of the data the authors may have requested and takes responsibility for data and analysis integrity. If this was not permitted by the sponsor or not feasible, it should be so indicated. In addition, the individuals in the sponsoring company who were responsible for data accuracy and analyses should be identified, should be authors, and should sign such a responsibility statement. If a clinical research organization performed the clinical study under a contract, it should be disclosed, as well as the funding sources for the study or project. All authors should disclose any conflict of interest but also should sign a statement indicating whether or not their potential conflicts of interest influenced the results being published.3 Although this is not directly related to the issue of data access, it underscores the importance of asking all authors to state whether they have had access to negative data or trial results with the drug or device described in a scientific publication. Having such information in an article would give the readership a much broader and perhaps more accurate perspective on the drug or device being studied.

Pharmaceutical or other corporate sponsors of clinical trials may be reluctant to submit negative results for publication; however, there may be much to learn from failures. In this regard, there is pending legislation in this country and internationally on fair access to clinical trials, which would require the registration of clinical trials before patient encounters. This might effectively permit editors, reviewers, and readers the opportunity to suspect the potential existence of negative results regarding a pharmaceutical or device. It also gives the public, as well as physicians and vision scientists, an idea of what clinical trials are active and when results may be forthcoming.4

It is not possible for any journal to assure that all results from clinical trials will be published and disseminated, although doing so in some venue would help advance our goal of understanding disease, new treatments, and improving patient care.5