Tufts Study: New Data Slows Trials Down
While the burgeoning amount of data collected in clinical trials promises to add more depth and breadth to study endpoints, leading to better studies and better drug therapies, for now it’s doing more harm than good by increasing data management burdens.
That’s according to a recent analysis by the Tufts Center for the Study of Drug Development (CSDD), which found that data volume coming from new technology like wearable devices, social media and real-world evidence is contributing to longer development times and posing technical and integration challenges to clinical data management staff.
“The change of endpoints is pushing us completely out of our comfort zone data-wise, and we don’t have the systems to easily support the demands of our new study designs,” said Ken Getz, associate professor and director of sponsored research at Tufts CSDD, who led the analysis.
During the past decade, the study found that data management cycle times have gotten longer, with the time from last patient, last visit to database lock increasing from an average of 33.4 days in 2007 to 36.1 days in 2018.
For Jill Johnston, president of WIRB-Copernicus Group’s Clinical Service Organization, the surprise was mutual: “This was very disappointing for me to see — really, no improvement in 10 years? With all of the advances in technology, the ability to transform and manipulate data, we are still no better than 10 years ago.”
Another shock? Getz said that research sponsors and CROs reported using their primary electronic data capture systems to collect and manage traditional data from case report forms and central and local labs — but not to collect and manage data from newer sources, including electronic health/medical records, mobile devices and social media communities.
This can create a silo effect that drags down the process.
“Historically, researchers have been reluctant to use primary health data because it has been too difficult to manage, and typically raised concerns with quality, completeness and usability,” said Wayne Kubick, chief technology officer of Health Level Seven International. “This has caused extra work for sites and sponsors, and often contributed to discrepancies between source data and study databases.”
However, things are changing for the better, Kubick said, thanks in large part to provisions in 2016’s 21st Century Cures Act that require providers to make data more accessible through APIs.
The Tufts CSDD study also found that 77 percent of sponsors and CROs have difficulty loading data into their primary EDC system due to compatibility, technical demands and integration challenges. Sponsors and CROs currently use an average of six applications to support clinical trial activities.
“The ultimate surprise was to see how troubling this complexity is to the development times,” said Getz.
“Companies are using everything from wearable devices and social media to mobile applications, collecting biomarker data, and it’s just remarkable that a high proportion of data collected is not captured in EDC but captured in other programs and only integrated much later in the process.”
The research also showed that protocol changes are the most common reason for delays in building study databases, accounting for 45 percent of database build delays reported by sponsors and CROs. Frequency of releasing the final study database after the first patient visit is associated with longer downstream delays and inefficiencies.
All 257 drug developers surveyed by Tufts CSDD reported using electronic data applications in clinical trials — but 26 percent of sponsors and 52 percent of CROs said they are still using paper case report forms in some investigations.
“I start to think about the root cause of that,” said Johnston, adding that it could be a lack of CRO involvement in early planning stages.
Companies may not have had enough time to deploy a full database, which would make getting started on paper easier for them. “A necessity due to their situation, rather than something they would prefer or plan,” she said.
In addition, there could have been a subset of studies more conducive to the use of paper forms, which could have been faster to change and implement. Those benefits could outweigh the risks for certain sponsors, Johnston said.
Are the troubles and delays just a natural part of the transition companies must make as technology advances, giving them more nifty forms of data with which to round out their studies? No, according to Getz, who says the problems run deeper than that.
“We have legacy systems that we rely on quite heavily, but all this new data — use of real-world evidence, patient-reported data — is part of an area that we know is going to increase, particularly with our focus on patient-centered medicine, and yet we really don’t have the flexibility and compatibility to easily integrate and use this data,” he said.
Getz said the purpose of the study was to stimulate discussion in the industry about solutions for more quickly integrating the influx of disparate data and speed studies up. For now, he said, the expertise is not there. Pharma companies do not have in-house IT teams positioned to handle this.
“It would be interesting to see this segmented further into traditional pharma and biotech companies,” said Johnston. “Is it a deep seated mind-set, a technology gap, regulatory gap? Are we too set in our ways to see if there is a better way of doing this?”
My other thought would be, maybe this is good enough for a majority of the cases? Disappointing, but good enough?” she said.
Today’s electronic data capture systems were never designed to handle the upcoming avalanches of data in future trials, including from case report forms and wearables, Johnston said.
“It is going to be an excessive amount of data and hidden within will be the proverbial pile of gold,” she said, adding that the traditional clinical operator will likely need to think about data in a different way, using purpose-built tools.
Getz predicts that the industry will begin outsourcing much of this work to niche CROs with expertise in data integration. Getz also expects that the industry may soon begin looking outside itself for experts in other industries — although sweeping, effective solutions may not come for a few years.
“I think companies are finding short-term approaches that are helping them meet the objectives of their protocol requirements, but as our study has shown, that really just adds more time and variability,” Getz said. “When we’ll really start to see acceleration is anybody’s guess.”