1409. Non-human Therapeutic Trials in Neuroscience: The Truth Behind the Headlines
Authors: Samer Zammar, MD; Ryan Jafrani, MD; Jennifer Dobson; Jessica Lane, MD; Neel Patel, MD; Robert Harbaugh, MD; Michael Glantz, MD, PhD (Hershey, PA)
In Neuroscience, marked therapeutic advancements have been published based on animal studies. Unfortunately, the “great” achievements in animal trials have not been significantly reproduced in human studies. The authors believe that inconsistent application of standards used in animal trials plays a role in this discrepancy and thus decided to grade the Class of Evidence (CE) of therapeutic animal trials in the neuroscience literature.
We reviewed 7 well respected journals in neuroscience over a 6-month period and identified the nonhuman-therapeutic trials. The authors used the American Academy of Neurology (AAN) evidence classification system to grade these trials. In addition to the AAN parameters, we examined the presence of inferential statistics, effect size, confidence intervals, multivariate analysis, power, and industrial funding and their correlation with CE.
35 articles were retrieved. 71% of the articles had a CE of 4. The median CE was 4 with IQR [2.5;4]. Significant factors that affected the low CE were: inability to establish similarity between treatment and control groups, lack of allocation concealment, and lack of masked assessment in 69%, 97%, and 54% of the articles respectively. Furthermore, effect size, confidence intervals, power, and multivariate analysis were not provided in 71%, 77%, 63%, and 86% of the articles. Outside the parameters included in the AAN grading system only the power calculation was significantly associated with higher class of evidence (p=0.018) on univariate analysis. None of these parameters had a significant correlation on multivariate analysis.
Despite the fascinating results reported in some non-human therapeutic trials in neuroscience, their low CE pose a significant threat to their reliability and reproducibility. Further attention to the design of the trials and statistics might help overcome these limitations.