Secondary Logo

Journal Logo

The Spine Blog

Saturday, August 10, 2019

Does author conflict of interest affect study results?

There has long been concern that industry funded studies, especially those in which authors have a direct financial relationship with the device manufacturer, are prone to bias in study design, performance, interpretation, and publication. The main efforts to mitigate this effect have been around author conflict of interest (COI) disclosure—both voluntary and through the Open Payments website-- and listing of all studies prior to commencement on clincaltrials.gov. While this has led to an increase in transparency, it is not clear if these programs have had any effect on the level of bias in industry-funded studies. In order to assess the relationship between author conflict of interest and outcomes in the lumbar disk arthroplasty (LDA) literature, Jordan Guntin and colleagues from Rush University Medical Center performed a systematic review of the LDA literature from 2011-2018. They identified 57 eligible studies, about half of which were level 4 studies and one third of which were level 1 studies. Fifty-three percent of studies had at least one author with a COI, and the vast majority of these involved a direct financial relationship with the device manufacturer. Studies from the United States were significantly more likely to have an author with a COI (77% vs. 43%). Ninety percent of the studies listed the COI, while the authors identified the COI through other sources (i.e. Open Payments) in the remaining 10%. Studies including an author with a COI were 60% more likely to report a favorable outcome (83% vs. 52%, p=0.02).

The results of this study come as no surprise, and the authors have done a good job in quantifying the effect of author COI on the LDA literature. The main question left unanswered, and which may be unanswerable, is the mechanism through which author COI affects the outcomes of published studies. One would hope that it is not through direct fraud by the device manufacturer and authors, and this seems unlikely in an era of increasing transparency. More likely is some type or types of bias, which can involve study design, patient selection, defining "success", statistical analysis, data interpretation, and the decision to publish equivocal or unfavorable results. Given that these biases are seemingly not eliminated through COI disclosure, other mechanisms may need to be considered in the future. In an ideal world, spine investigators could receive non-industry funds that would allow them to perform high quality, unbiased research. However, government and foundation funding for studies of this kind is exceedingly hard to obtain. Another possibility would be to require industry to provide the funding to a neutral organization that then selects the investigators so that the device manufacturer is held at arm's length from the study. This would prevent any bias in the study design, performance, interpretation, or publication. This study and others like it have created an appropriately high level of skepticism among investigators, spine surgeons, and patients. While this skepticism helps to limit overly enthusiastic adoption of new technology that has not been vetted in studies without COI (LDA is a great case in point), it may also inappropriately slow adoption of effective new technology. While it may be non-intuitive, industry and patients may both be best served through the creation of neutral organizations tasked to perform studies evaluating new devices and technology.

Please read Mr. Guntin's article in the August 15 issue. Does this change how you view the association between author COI and published outcomes? Let us know by leaving a comment on The Spine Blog.

Adam Pearson, MD, MS
Associate Web Editor