Conflicts of interests have long been recognized as potential sources of influence in the conduct and reporting of clinical trials. This controversy was again rekindled after the publication of the latest statin guidelines and a series of studies regarding competing interests in leading medical journals. We investigate the association between declared author conflicts and the outcomes of large cardiovascular trials. We searched the Medline (PubMed) database to identify “phase 2” and “phase 3” clinical trials using the search term “cardiovascular” over the past decade using “10 years” as the filter. We perceived the competing interest as present regardless of the nature such as consulting fees, honoraria, travel imbursements, stock holding, and employment. Of the 699 titles retrieved, 114 studies met the inclusion criteria. Nearly 80% of studies had at least a single author with competing interests. The 114 studies had a total of 1,433 investigators, of which 725 had declared conflicts of interests (50.6%). A total of 66 studies (58%) had half or >50 percent of investigators who had some conflicts of interests. Of these studies, 54 studies had favorable outcomes and only 12 had unfavorable outcomes (p <0.001). Among the type of competing interests, consulting or personal fees was the most common present in 58 investigators (51%). This was followed by research grants present in 55 the researchers (48%). Among 25 (22%) studies, at least one investigator reported stakes in the industry, of which only 2 studies had unfavorable outcomes for the intervention being investigated. Just 1 of the 25 clinical trials with a sample size of >1,000 had no investigators with competing interests. In conclusion, authors conflicts are associated with favorable outcomes in cardiovascular outcome trials.
Cardiovascular disease continues to be the leading cause of mortality across the globe and therefore an unbiased dissemination of cardiovascular research data is critical to ensure the highest ethical standards, which may be compromised by existing competing interests among the researchers. The issue of the competing interests among the investigators was again highlighted by the latest statin guidelines and a series of studies in leading medical journals. Recently, we demonstrated that financial interests are associated with outcomes that are favorable to the industry and this relation holds true even in a prepublication database suggesting that the publication bias alone cannot account for such findings. The current trends in the frequency of competing interests among the leading cardiovascular researchers are unknown. Herein, we investigate the association between declared author conflicts and the outcomes of large cardiovascular trials.
Methods
We searched the Medline (PubMed) database to identify “phase 2” and “phase 3” clinical trials using the search term “cardiovascular” over the past decade using “10 years” as the filter. This generated 699 titles, from which only randomized controlled trials were selected. Subgroup analyses, interim analyses, and publications of study designs were excluded. The last date of the report search was end of July 2015. Two investigators performed the report search independently. A mutually agreed template was used for the extraction of data. We perceived the competing interest as present regardless of the nature such as consulting fees, honoraria, travel imbursements, stock holding, and employment. Results of the studies were labeled “favorable” if the results favored the intervention under investigation, or “unfavorable” if the intervention fared worse or similar to standard medical treatment. We examined the studies regardless of the funding sources, whether industry, federally (such as National Institute of Health) or independently sponsored. For the purpose of statistical analysis, the studies were categorized into 2 groups, based on whether the investigators reported financial interests or not, and the data were analyzed through the statistical package of social sciences (SPSS version 17.0). The 2 groups were compared for differences in outcomes using Fischer’s exact test. Results were expressed as 95% CIs and deemed significant if the p value <0.05.
Results
Using the search strategy described in the methods, a total of 699 titles were identified and 114 studies met inclusion criteria after full review. The reasons for exclusions of most studies were being noncardiovascular or study types other than the phase 2 and 3 clinical trials. Most of these studies (69%) were sponsored by industry and were from the United States (51%). Nearly 80% of all studies had at least a single investigator with competing interests. The 114 studies had a total of 1,433 investigators, of which 725 had declared conflicts of interests (50.6%). A total of 66 studies (58%) had half or >50 percent of investigators who had some conflicts of interests. Among these studies, 54 studies had favorable outcomes and only 12 had unfavorable outcomes (p <0.001). Among the type of competing interests, consulting or personal fees was the most common present in 58 investigators (51%). This was followed by research grants, present in 55 the researchers (48%). Among 25 (22%) studies, at least one investigator reported stakes in the industry, of which only 2 studies had unfavorable outcomes for the intervention being investigated. Five or more investigators with competing interests were present in 61% of the studies ( Table 1 ). Declaration of financial competing interests by at least one investigator was associated with a significantly greater likelihood of favorable results for the drug or intervention under investigation (p <0.005). Similarly, sponsorship from industry was associated with more favorable results (p = 0.03).Of the 8 (7%) studies which were based on devices, only 2 had unfavorable outcomes. Six of those 8 studies reported conflict of interests. Rest of the 106 studies were based on drug intervention (drug vs placebo = 62, drug vs standard treatment = 44). The mean sample size of the trials included was 1,342 ± 3,215 with 25 (21.9%) studies having a cohort of >1,000. Of these 25 studies, only one study had no competing interest.
