Abstract
Background: Over the last few decades medical research and development has come to depend more heavily on the financial support of industry. However, there is concern that financial relations between the medical community and medical industry could unduly influence medical research and therefore patient care. Our objective was to determine whether conflict of interest owing to authors’/investigators’ financial affiliation with industry associated with their academic research has been identified in the surgical literature. In particular, we sought to answer the following questions: What is the extent of such conflict of interest? Does conflict of interest bias the results of academic surgical research in favour of industry? What are the potential causes of this proindustry bias?
Methods: We conducted a systematic review of the literature in May 2008 using the OVID SP search engine of MEDLINE, EMBASE, CINAHL, the Cochrane Database of Systematic Reviews, DARE and Health Technology Assessment. Quantitative studies that included a methods section and reported on conflict of interest as a result of industry funding in surgery-related research specifically were included in our analysis.
Results: The search identified 190 studies that met our criteria. Author/investigator conflict of interest owing to financial affiliation with industry associated with their academic research is well documented in the surgical literature. Six studies demonstrated that authors with such conflicts of interest were significantly more likely to report a positive outcome than authors without industry funding, which demonstrates a proindustry bias. Two studies found that the proindustry bias could not be explained by variations in study quality or sample size.
Conclusion: The conflict of interest that exists when surgical research is sponsored by industry is a genuine concern.
Evidence-based medicine has had increasing influence in day-to-day clinical decision-making within the practice of surgery. The best available evidence is founded on the comprehensive and systemic evaluation of the literature, ideally in the form of a meta-analysis or systematic review. The strength of a specific treatment recommendation is based on the quality (level) of evidence within the literature and on clinical experience and patient preference, the latter taking risk, onus and cost into consideration.1 Clinical experience and patient preference constitute subjective but necessary components of evidence-based medicine in patient care, whereas research is more objectively measured through methodologic evaluation. Good evidence is contingent on good research. The more rigorous the methodology of a study, the better the control of systematic bias and the higher the level of the evidence. However, high-quality research, such as randomized controlled trials, remain challenging, especially in surgery, owing to problems related to blinding, equipoise, patient preference, generalizability and, of particular relevance to this study, prohibitive costs.2
Funding available for medical research can be sought from numerous sources, including government agencies, foundations, academic institutions, medical or surgical societies and industry. Over the last few decades medical research and development has come to depend more heavily on the financial support of industry. In 1980, 32% of total biomedical research was funded by industry; this figure increased to 62% in 2000.3 In 1986, 46% of life science companies supported academic research; within 10 years this figure increased to 92%.3
Although industry funding can lead to medical and technologic advances and provide an opportunity to validate new technologies, the industrial support of biomedical research has come under increasing scrutiny within the scientific community, government, media and public. The major concern is that financial relations between the medical community and medical industry could unduly influence medical research and therefore patient care.4 It is well recognized that the collaboration of industry and academic researchers may create a conflict of interest.3,5–10
Conflict of interest, defined as a set of conditions in which professional judgment concerning a “primary interest” tends to be unduly influenced by “secondary interest,”11 introduces a common form of nonscientific bias into biomedical research. Primary interest to a physician is a patient’s welfare and to a scientist is the validity of the research. Secondary interest includes personal financial gain, research funding and academic pressure to publish. In 2003, Beckelman and colleagues3 illustrated the scope of this potential conflict: one-quarter of all academic biomedical researchers receive industry funding, one-third of all universities have financial ties with industry sponsors, and one-third of lead authors have personal financial interest in their study topic. Society’s perception of this potential conflict is still not clear, especially when the research may not be possible without industry support.12
It has been demonstrated in the medical literature that industry funding adversely affects the validity of the research by producing a proindustry result or conclusion.3,5–10,13,14 A systematic review of the medical literature reported that the odds of a study having a positive outcome is 3.6 times greater when the research is industry-sponsored.3 Whereas the potential influence of industry funding on the outcomes of medical research has been extensively discussed in the literature, the data related to surgical research are less clear. We undertook a systematic review of the literature to determine whether conflict of interest owing to authors’/investigators’ financial affiliation with industry associated with their academic research has been identified in the surgical literature. We sought to examine the extent of this conflict as well as the influence of industry funding on the research results.
Methods
We performed a systematic review to answer the following questions: What is the extent of author/investigator conflict of interest owing to industry funding of their surgical research or financial affiliation to industry associated with the research (e.g., royalties)? Does author conflict of interest bias the results of academic surgical research in favour of industry? What are the potential causes of the proindustry bias? We conducted a literature search in May 2008 using the OVID SP search engine of the databases MEDLINE (1966–2008), EMBASE (1980–2008), Cumulative Index to Nursing and Allied Health Literature (CINAHL; 1982–2008), the Cochrane Database of Systematic Reviews, the Database of Abstracts of Reviews of Effects (DARE) and Health Technology Assessment. We used the keywords “conflict of interest,” “publication bias,” “positive outcome bias,” “surgery,” “research support” and “biomedical research.” The search strategy is summarized in Box 1. The review was limited to studies published in English. We further reviewed the bibliographies of the selected studies meeting our inclusion criteria, outlined in the next paragraph, to complement our database search strategy.
Search strategy
Conflict of interest.mp or “conflict of interest”/
Publication bias/
Positive outcome bias.mp
Surgery/or surgery.mp
1 or 2 or 3
4 and 5
Research support as topic/or research support.mp
Biomedical research.mp or biomedical research/
7 and 8 and 4
6 or 9
We included quantitative studies that contained a methods section and reported specifically on conflict of interest as a result of industry funding in surgery-related research. Studies that included both medical and surgical research were included only if results pertaining to the surgical literature could be analyzed separately. Editorials, narrative reviews, commentaries and letters to the editor were excluded. The abstracts of all articles that matched the search terms and met our inclusion criteria were reviewed. We then obtained the full text version of abstracts that we felt were suitable. These articles were reviewed for data relevant to the research question. Any discrepancy on the selection of articles was resolved by discussion among 2 reviewers.
Results
The literature search identified 190 studies; 8 satisfied our inclusion parameters (Table 1). Six of 8 studied the orthopedic literature,4,15,16,18,19,21 1 the spine literature,20 and 1 the combined literature from the disciplines of orthopedic surgery, neurosurgery, general surgery and plastic surgery.17
Our systematic review demonstrated that authors’/investigators’ conflict of interest owing to financial affiliation with industry associated with their academic research was well documented in the surgical literature.4,12,15–21 Although specific to orthopedic research, Zuckerman and colleagues4 demonstrated that author self-reported conflict of interest involving research presented at a national annual meeting in the United States had substantially increased from 10% in 1985 to 32% in 2002 (p < 0.001).
Six studies demonstrated that authors with such conflicts of interest were significantly more likely to report a positive outcome than authors without industry funding, which demonstrates a proindustry bias.15–18,20,21 Bhandari and colleagues17 found that, although not statistically significant, the odds ratio of a proindustry conclusion was 5 times greater in a surgical trial than in an industry-sponsored drug trial.
One potential explanation for this proindustry bias is that industry-funded studies are of higher scientific quality than non–industry funded studies owing to disproportionate funding levels provided by industrial sources. However, 2 studies found that proindustry bias could not be explained by variations in study quality or sample size.17,21 Cunningham and colleagues21 found that prospective study design, use of control groups and larger sample size were not more prevalent in industry-sponsored than in non–industry sponsored research, yet the industry-sponsored studies were more likely to conclude with a proindustry outcome. Similarly, Bhandari and colleagues17 found that neither discrepancy in study quality (assessed using the Detsky Index) nor sample size could explain the proindustry bias associated with industry sponsorship.
A final study included in our review contradicted those mentioned previously. Lynch and colleagues19 analyzed manuscripts submitted to the Journal of Bone and Joint Surgery (the other studies analyzed only research published in journals or presented at meetings) for an association between both scientific and nonscientific factors (including funding source) and outcome bias.19 They found that industry-funded research was more likely to be published than non–industry funded research but that source of funding was not associated with a positive study outcome and that a positive study outcome was not itself a predictor of acceptance after peer review. However, a low-level form of bias against non–positive outcome studies during peer review was suspected because the quality of the non–positive outcome studies was superior to that of the positive-outcome studies, but the acceptance rate was similar.
Discussion
Our systematic review of the surgical literature suggests that authors’ financial conflicts of interest and industry sponsorship of studies examining medical devices or related products exerts a bias toward proindustry results. Furthermore, it appears that the outcome bias in the surgery-related research literature is at least equal to if not greater than that in the medical literature, specifically drug trials.17 Although the study by Lynch and colleagues19 used a more robust methodology than the others (examining all submitted manuscripts as opposed to only those published in journals or presented at peer-reviewed meetings), the discrepancy in findings does not negate the significance of the proindustry outcome bias demonstrated by the other studies. Lynch and colleagues19 themselves state that their methodology included only manuscripts submitted to 1 journal and was specific to hip and knee reconstruction and therefore not necessarily generalizable to other subspecialties in orthopedics, other specialties in surgery or other journals. The fact that submitted, rather than presented or published, manuscripts were analyzed cannot explain the difference among findings in the study by Lynch and colleagues and those of all other studies included in this review. It is unknown if the reason lies in the combination of the low-level bias against non–positive outcome studies in the peer review process and the greater likelihood of manuscript publication for industry-funded research.19
Limitations
There are limitations to our study. It is not generalizable beyond the English surgical literature. The true extent of conflict of interest and proindustry bias cannot be determined by the methodology of the included studies because conflict of interest was self-reported. Therefore, it is possible that the true incidence may be higher. Finally, as conflict of interest is self-reported, author compliance with reporting may be increasing and may account for some of the perceived increase in conflict of interest reported by Zuckerman and colleagues4 and others.3,5,6
We find it interesting that most of the inquiry into the extent and impact of industry-sponsored research in surgery was in the specialties of orthopedic and spine surgery. Okike and colleagues16 found that self-reported conflict of interest was common in orthopedic surgery, but mostly for hip and knee reconstruction and spine surgery, and represented about 50% of all research presented at the 2001 and 2002 annual meetings of the American Academy of Orthopaedic Surgeons. In fact, 3 of the 8 reports included in this study investigate research specific to these subspecialties.18–20 Considering that most orthopedic and spine surgeries require the implantation of a surgical device, it is not surprising that the industry-related conflict of interest literature focuses on those disciplines. Although we identified only 1 paper that satisfied our inclusion criteria and investigated conflict of interest in other surgical specialties,17 we found many editorials discussing concerns in other areas of surgery.22–28
The conflict of interest that exists when surgical research is sponsored by industry seems unavoidable but does not necessarily mean that the research is biased or that the researcher is corrupt.15,16,18 Nevertheless, a proindustry bias is a genuine concern in the surgical literature. The challenge is to identify and manage the conflict, hence the author’s obligation to disclose his or her conflicts of interest and allow the readers to decide for themselves if a bias exists.
One possible explanation for the apparent bias is a discrepancy in research quality. As industry funding typically provides superior financial support, it has been hypothesized that the result will be improved or higher-level study design with better study administration and an increased sample size, reducing the risk of type-β error. This concept, however, was challenged by the findings of 2 papers included in our review.17,21 Clearly, companies are going to be selective in their funding and are more likely to invest in research that will prove beneficial to their products. Biomedical companies conduct considerable in-house research to preferentially fund academic collaboration that will be successful and result in positive studies.6,29,30 We agree with Cunningham and colleagues21 that, unlike with drug trials, it is very difficult for a company to perform human in-vivo testing on surgical implants or devices without the involvement of an academic surgeon/researcher. Thus, this explanation is less likely to account for positive outcome bias in surgical trials and warrants further consideration.21
Another explanation for the apparent proindustry bias is documented in the medical literature. It has been suggested that trials may be stopped prematurely or a sponsoring company may attempt to obstruct publication of the results if the findings of the study are negative.31–34 Although these occurances have not been directly investigated in the surgical literature, 2 publications in our review demonstrated that studies involving researchers with consultant/employee status, royalties or stock options were associated with a higher likelihood of a positive outcome than those involving researchers who received only research support.16,18 This implies a more sinister possibility that some investigators may be swayed by secondary interests and, in doing so, disregard the primary interest of study validity.
Even if industry support does not influence investigator behaviour, a negative public perception could jeopardize trust in the scientific community. With this in mind, a recent survey of 245 patients in an orthopedic spine surgeon’s office found that “despite adverse publicity in the popular press in the past few years about physicians and conflict of interest issues, the public is overwhelmingly in favour of allowing surgeons to work with industry because of its potential beneficial effect for patients.”35
Distinct benefits exist in the collaboration between industry and academic research in surgery for the physician/scientist, industry sponsor and the public. The fact that proindustry bias is present in the surgical literature is worrisome, but the causes behind it are not clear and may be only partially related to true conflict. The findings of our study do, however, imply that the self-reporting of conflicts of interest is not sufficient to ensure integrity in the research and public trust and that further reflection and evaluation by all the key stakeholders is warranted. Further safeguards should be implemented, such as restriction of authorship for surgeons with significant financial conflicts, and partnership between major players, both academic and industry, to create independent clinical trial groups that minimize actual and perceived bias.
Footnotes
Competing interests: None declared for Drs. Bailey, Fehlings, Hall, Wai and Fisher. Dr. Rampersaud is a paid consultant for Medtronic – Spine.
Contributors: Drs. Bailey and Fisher designed the study and acquired the data, which all authors analysed. Drs. Bailey, Fehlings and Fisher wrote the article, which all authors reviewed and approved publication.
- Accepted February 22, 2011.