It is tempting to try to predict what could have happened if the US Food and Drug Administration (FDA) had exercised greater oversight over the various production sites making losartan, a medication that was recently recalled across numerous distributors after the discovery that a potential carcinogen may have been introduced into the product.
But there is also the possibility that even in spite of increased oversight by the FDA, impurities could have still slipped into the supply chain. And, that despite FDA action, some issues with drugs or data quality that are revealed during inspections have the potential to be absent from the drug’s final label.
When trial data are problematic, argued New York University journalism professor Charles Seife in a 2015 paper, these data don’t always make it into an article transcript for consideration by a peer reviewer.1 Dr Seife wrote that despite red flags being raised about some falsified data on apixaban from the ARISTOTLE trial, the data quality issue was not reflected in the drug’s label, which he said still included a mortality benefit analysis that relied on data from 24 suspect Chinese sites that the FDA itself determined should be excluded. This research misconduct in ARISTOTLE, wherein researchers knowingly published falsified data, was not revealed in the peer-reviewed literature.
This presence of tainted data may not always be easy for readers to assess — even if an individual were to get access to full FDA reports that discuss departures from good procedures, the reports often contain large portions of information that has been redacted.
The problems with bad data can have lasting impacts, and can affect the conclusions of subsequent studies. It can sometimes be unclear to researchers there have been problems with the data included in a study, and if those are not resolved, the study is fair game for inclusion in future meta-analyses.
This is exactly what happened for the ARISTOTLE trial, said Craig A. Garmedia, PhD, and colleagues, who wrote about the topic in a research letter in JAMA Internal Medicine. Dr Garmendia conducted the work as a doctoral student with Florida International University’s Robert Stempel College of Public Health and Social Work’s Department of Epidemiology in Miami and, although he is an investigator with the Office of Bioresearch Monitoring Operations with FDA’s Office of Regulatory Affairs, he stated his comments reflect his own views and should not be construed to represent the FDA’s views or policies.
The research team looked at 22 meta-analyses that included the ARISTOTLE trial and determined that 10 overviews (46%) had conclusions that would have changed had the studies with falsified data been excluded. And, of the 99 original analyses that comprised the 22 meta-analyses, 32 (32%) had results that would have changed the original conclusions in the initial papers. Of these 32 affected estimates, 31 (97%) of the outcomes “no longer favored apixaban for the prevention of serious medical issues, and 1 (3%) favored the control.”
Cancer Therapy Advisor spoke to Dr Garmendia to get some insight into how these falsified data persist and how they have the potential to influence clinical decision making.