Bovine Thrombin Safety Reporting: An Example of Study Design and Publication Bias

• Published in: The Journal of surgical research 2010, Vol.158 (1), p.77-86
• Main Authors: Crean, Sheila, M.P.H, Michels, Shannon L., M.S.P.H, Moschella, Kevin, B.S, Reynolds, Matthew W., Ph.D
• Format: Article
• Language: English
• Subjects: adverse events; Animals; bias; biologic product; Biological and medical sciences; Cattle; General aspects; Hemostatics - adverse effects; Humans; Medical sciences; Publication Bias; Randomized Controlled Trials as Topic; Research Design; Safety; Surgery; thrombin; Thrombin - adverse effects
• ISSN: 0022-4804; 1095-8673
• DOI: 10.1016/j.jss.2008.09.006

Background Bovine thrombin, a popular hemostat and sealant since 1945, has recently been subjected to clinical trial testing due to reformulations in 1998. We sought to compare adverse event rates of early observational studies with those of later interventional trials. Methods A MEDLINE-based literature search in publications that report safety in bovine thrombin exposed surgical patients was extracted and reviewed. Results In 38 studies, about half were case reports and 31.5% were interventional trials. In case reports, 41% of authors reported severe coagulopathic adverse events. In contrast, whereas blood complications were common in large trials, no association of harm was established for bovine thrombin product exposure and/or immunization. Conclusions In this review, later clinical trials failed to reproduce the common and severe coagulopathy predicted by earlier observational studies in bovine exposed patients. This example illustrates that perceptions of safety can change as a function of study design, even for a widely adopted, well established biologic such as thrombin. Caution must be exercised in interpreting evidence from observational studies alone.