The initial Cooperative Research and Development Agreement (CRADA) between the US Food and Drug Administration (FDA) and CluePoints explored a data-driven approach to select sites exhibiting data anomalies, which may indicate fraud or misconduct.
The agreement – now extended to October 2021 – has been supplemented with additional testing ideas and will focus on moderators of treatment effect and real-world evidence (RWE), explained Patrick Hughes, co-founder and chief commercial officer, CluePoints.
Based on the moderators of treatment effect, further analyses will include geographic region, country, and supervising contract research organizations (CROs), according to the company. The regulator also plans to conduct statistical tests, such as key risk indicators (KRIs), which will augment the existing approach, Hughes said.
The FDA earlier this year expanded on its recommendations for taking risk-based approaches to clinical trial monitoring in a draft guidance. The document expands on the risk-based monitoring (RBM) guidance published in August 2013.
The FDA also has published a Q&A that reinforces the importance of performing risk assessments for all studies. Taking this risk-based approach to study execution is a requirement of ICH E6 (R2), which Hughes previously described as sparking “the biggest paradigm shift in the industry in decades.”
“FDA want to use these techniques to thoroughly interrogate submission data with a view to supporting the selection of sites for inspection based on the results,” he told us. “This unsupervised assessment of the data mirrors the insistence that sponsors carry out this type of risk-based approach in their own oversight of data quality and integrity.”
The data-driven approach is able to detect ‘anomalous’ sites previously undetected and determine the nature and extent of the data irregularities.
Said Hughes, “These benefits are expected to not only accrue to the site inspection process and improve data quality for all reviewers, but may also inform the efforts of clinical and statistical reviewers to conduct sensitivity analyses, subgroup analyses and site by treatment effect explorations.”
Sloppiness or an attempt to defraud?
Anomalies or items that could be interpreted as misconduct include intentional and unintentional errors, such as the propagation of vital signs data and invention of data, “usually for fiscal gain,” explained Hughes.
“Many sponsors would describe the former as sloppiness, but the latter is clearly a premeditated attempt to defraud the study,” he added, noting that CluePoints does not judge the signals, but ‘pinpoints’ them so sponsors can concentrate their efforts.
“The most common signals that we see are erroneous items that are usually to do with site staff not seeing value in recording the data accurately,” said Hughes, giving the example of recording a patient’s weight once and copying it for all subsequent visits, instead of reweighing.
“We see a lot of this type of sloppiness within vital signs,” he said. “The question is, would this be considered misconduct. Since the data is being recorded deliberately inaccurately then many sponsors see this as inventing data and, whilst not as serious as some discoveries, it can often be a clue that something more serious might also be going on at the site.”