Conflicts of interest may result from a large number of both financial and non-financial relationships in our private and professional lives1,2,3. Recognizing that it is unrealistic to eliminate all potential conflicts, medicine as a profession has attempted to develop mechanisms to manage these sources of bias2. Over the past decade, the concepts of transparency and full disclosure of potential conflicts of interest have rightfully become accepted in all areas of medicine, including research, education, and patient care. Indeed, measures such as the standardized format, developed by the International Committee of Medical Journal Editors, for the self reporting of potential conflicts of interest (adopted by JBJS for all manuscripts submitted after January 1, 2011) are an important first step in securing our professional credibility3. However, such efforts perhaps only alert the reader to the most overt potential sources of bias. Unfortunately, other important forms of bias may be pervasive but go unrecognized. One of these trends is the systematic bias of published reports favoring products produced by companies funding the research1,4. This type of bias may only be revealed through systematic reviews and meta-analyses of entire fields of research.
In this paper, Lee and colleagues present the results of a systematic review of previously published, prospective studies about venous thromboembolism prophylaxis after total joint replacement. The goals of the study were to define the proportion of industry-sponsored studies in this field over the past five years and to determine whether the authors’ conclusions about the thromboprophylaxis were associated with the source of the funding. Seventy-one articles were analyzed. The source of funding was identified for sixty-six of the studies but in five cases the sponsorship could not be determined, even after the authors and journal editors were contacted. Fifty-two (78.8%) of the sixty-six studies were funded by a pharmaceutical or medical device company, whereas fourteen were sponsored by non-industry sources. Furthermore, only two (3.8%) of the fifty-two industry-funded studies reported conclusions that were unfavorable to the funding source. In distinction, three of the fourteen non-industry-sponsored studies reported unfavorable results. These qualitative conclusions of the papers demonstrated a significant association with the sponsorship of the study (p = 0.033). The major weakness of this report was that the number of non-industry studies was small; consequently, if all five of the studies for which the source of funding could not be determined were actually non-industry-funded, the results would not have been significant.
Despite this limitation, the information is important and corroborates previous reports from other areas of medicine, including orthopaedics. Bekelman et al. performed a systematic review of biomedical research articles in MEDLINE between 1980 and October 2002 and reported a significant association between industry sponsorship and pro-industry conclusions (odds ratio, 3.6)1. In our own specialty, Khan et al. reviewed the articles published in five major orthopaedic journals over a two-year period, between 2002 and 2004, and noted a strong statistical association between industry funding and favorable outcomes4.
Potential explanations for this systematic bias include (1) publication bias, with positive studies being more likely to be published regardless of funding source and (2) suppression of publication by the funding source when unfavorable results are identified1,4. Additionally, Lee and colleagues note that bias may also result from the study design if the methods that are used are likely to confirm the superiority of the modality produced by the company supporting the study. For example, selection of an inappropriate comparator or use of a noninferiority test as the primary end point may favor the modality promoted by the funding source. Indeed, in the current study, Lee and colleagues note some evidence of this type of bias as thirteen of the industry-sponsored studies used a noninferiority test, compared with only one of the non-industry-funded studies. Importantly, this type of bias may be potentially dangerous as it may not be apparent when reviewing a single study but can taint the development of clinical guidelines when they are based on systematic reviews and meta-analyses. In the field of thromboprophylaxis after joint replacement surgery, Lee and colleagues note that the systematic bias of industry-funded studies reporting favorable outcomes may result in the development of guidelines that favor the use of aggressive thromboprophylaxis, which remains a particular concern to orthopaedic surgeons.
In order to address this type of bias, other mechanisms beyond full disclosure are likely required. In response to such concerns, in 2007 the American Academy of Orthopedic Surgeons adopted Standards of Professionalism on Orthopaedist-Industry Conflicts of Interest that hold us accountable to a mandatory minimum code of conduct5. Paraphrasing standard 17, an orthopaedic surgeon who is the principal investigator of a study shall make his or her best efforts to ensure that the results are reported in an unbiased manner regardless of the funding source5. In order to meet these standards, orthopaedic investigators should commit to only participate in studies with appropriate and balanced methods and should avoid participating in any study in which the funding source owns the data or has right of refusal to allow publication.
In summary, the value of the study by Lee and colleagues is that it highlights that orthopaedists need to continue to improve the way we manage potential conflicts of interest in research and other aspects of our professional lives; the development of consistent policies for the reporting of potential conflicts of interest and codes of conduct represent the first steps in this process.