In brief
Jennifer Sheller of Merck argues that clinical trials have grown too heavy and need to be simplified without losing rigor. Her fixes are leaner protocols, less data collection, checking data in proportion to its risk, and more automation. She also says a hybrid staffing model and careful integration of acquired biotechs let a large company run trials quickly and reliably.
The central tension is between complexity and speed. Trials keep accumulating data points, checks and processes, often after a past inspection finding, while pricing and regulatory pressure demand faster delivery. Sheller says the updated international good clinical practice guideline (ICH E6 revision three) gives permission to do less, but only if teams define what is critical and document their choices. Technology helps, but people and partnerships come first.
Questions
What is a functional service provider model in clinical trials?
It is a hybrid staffing approach in which a company's own employees and staff from outside contract research organizations (CROs) work as one team under the same processes. Sheller says Merck runs about 60% in-house and 40% FSP, and trial sites cannot tell the two apart.
How did Merck avoid disrupting trials when it bought other biotech companies?
Sheller says Merck met staff face to face, set out clear retention plans and kept the in-flight trials running under the existing processes. It moved over only what was essential, such as the safety database, and it built a reusable integration playbook.
Why are clinical trials getting more complicated, and what can be done?
Sheller says protocols are often copied from earlier trials, so they pick up extra data points and eligibility rules that are not needed. Her fixes are starting from the research question, scoring how heavy a design is for patients and sites, and cutting extras like unneeded exploratory endpoints.
What is ICH good clinical practice revision three, and why does it matter?
It is the latest update to the international standard for conducting clinical research, which dates to 1996. Sheller says it makes risk proportionality a principle, meaning critical data gets the most rigorous checks and less important data gets lighter ones.
How does Merck want to use AI and automation in trials?
Sheller describes using AI to find trends in text reports of protocol deviations and to automate data cleaning and reconciliation across vendors. Merck is also building an automation hub linked to each function, with implementation of its new quality management model planned for 2026.
Facts
Stated by the speakers. Each quote matches the transcript, which shows what was said, not that it is true.
Merck invested nearly $18 billion in research and development in 2024 and conducts trials in over 60 countries at more than 21,000 sites.
“a company that invested nearly $18 billion in R &D in 2024 and conducts trials in over 60 countries at more than 21 ,000 trial sites”
0:00:06 · Ben Comer · R&D spending · 2024
Jennifer Sheller joined Merck in 2017 and initially worked on integrating acquired companies' R&D activities into Merck's operations.
“You joined Merck in 2017, Jen, and initially, I think, worked on integrating acquired companies' R &D activities into Merck's R &D operations.”
0:16:33 · Ben Comer · Career · 2017
Sheller says roughly half a million clinical trials are registered on ClinicalTrials.gov.
“I think there's a half a million clinical trials registered at clintrials.gov. And so it's just, it's exploded.”
0:04:41 · Jennifer Sheller · Trial volume
In the late 1990s, clinical research associates verified hospital records against handwritten triple-copy NCR paper forms, which later gave way to electronic data capture.
“when I first started in the late 90s, it was, you know, triple NCR paper. So we'd go out to the hospitals to make sure that the records in the hospital system matched what was handwritten on an NCR form.”
0:05:12 · Jennifer Sheller · Data capture · 1990s
Sheller spent about 13 years at Sanofi, then moved to the biotech Kyowa Kirin, and has now been at Merck for over eight years.
“I spent about 13 years at large pharma at Sanofi and went to a small biotech, Keowakirin, shortly thereafter, before coming back into large pharma here at Merck for over eight years.”
0:08:58 · Jennifer Sheller · Career
Andy Lee, a former Sanofi colleague of Sheller, moved Merck from a fully outsourced clinical operations model to an in-house one.
“So he came into Merck and really shifted the paradigm from outsourced, fully outsourced model to bringing everything in-house.”
0:10:06 · Jennifer Sheller · Operating model
Merck's clinical operations model uses about 60% of its own headcount and about 40% functional service provider staff.
“So he brought everything in-house and built an operating model where we have, you know, 60% headcount through our company and then 40% or so through functional service provider.”
0:10:35 · Jennifer Sheller · Staffing model
Sheller says unit-priced outsourcing contracts fit poorly because a safety signal can require extra monitoring, forcing change orders.
“if you have unitized pricing for a given study, X number of monitoring visits, X number of phone calls to a site, that's not the real world because you might have a safety signal where you need extra monitoring.”
0:15:14 · Jennifer Sheller · Outsourcing contracts
Merck received its first approval for a product from Peloton Therapeutics within two years of acquiring the company.
“I'll share the Peloton example that came forward with a product that we got our first approval within two years of acquiring Peloton.”
0:17:36 · Jennifer Sheller · Acquisition integration
Merck acquired a HIF2-alpha inhibitor from Peloton Therapeutics, with VHL-associated renal cell carcinoma as its first indication.
“we brought in the HIF2-alpha inhibitor from Peloton and renal cell carcinoma and VHL was our first indication.”
0:25:06 · Jennifer Sheller · Oncology drug
Sheller says the Acceleron acquisition, with a large phase three program, was probably Merck's most challenging integration.
“Accelerant had a large phase three program that was probably the most challenging integration.”
0:20:41 · Jennifer Sheller · Acquisition integration
During the Peloton integration, Sheller hired a summer intern to shadow the team and write the integration playbook that became Merck's operations standard.
“I actually hired one of my summer interns to extend her stay for a project to shadow us and build the playbook out.”
0:21:37 · Jennifer Sheller · Integration playbook
Sheller cites Merck's Prometheus acquisition as a return to immunology, an area Merck had not been in for a while.
“That was an area we haven't been in in a while with the Prometheus acquisition, for example.”
0:26:12 · Jennifer Sheller · Acquisitions
Sheller cites a study saying data collection in clinical trials has risen more than threefold over roughly the last decade.
“over the last decade or so, I think they quoted that we've increased data collection in trials more than threefold.”
0:29:00 · Jennifer Sheller · Data collection
The ICH good clinical practice guideline dates from 1996 and has now reached its third revision.
“this is, I think, from 1996, so revision three, and it's taken that long of time.”
0:34:29 · Jennifer Sheller · ICH GCP · 1996
Sheller says revision three of ICH good clinical practice makes risk proportionality a principle, applying more rigor to critical data and processes.
“Meaning, you don't treat all of your data the same. You focus on critical data, critical data and processes with a higher amount of rigor.”
0:35:05 · Jennifer Sheller · Risk proportionality
A TransCelerate study found only about 2-3% of critical data changed as a result of 100% source data verification.
“they showed only like two to 3% of critical data changed from all of that effort doing that cross check.”
0:35:39 · Jennifer Sheller · Source data verification
Sheller says Merck wants to use AI and natural language processing to trend non-important protocol deviations so teams can focus on serious ones.
“that's where AI can help me trend the non-important deviations for the oversight.”
0:38:30 · Jennifer Sheller · AI oversight
Some research sites told Merck they would not run a trial unless non-physician sub-investigators, such as nurse practitioners, were allowed under the R3 changes, and Merck changed its process.
“hey, look, we're not going to do this trial unless you do apply the R3 changes and having non-physician investigators or sub-investigators in the trial.”
0:44:11 · Jennifer Sheller · Site feedback
Sheller says about 5% to 8% of US cancer patients participate in a clinical trial.
“I think is 5 % to 8 % of cancer patients participate in a clinical trial, whereas many would be interested if they were asked.”
0:49:08 · Jennifer Sheller · Trial participation
Merck created the Let's Talk Trials education campaign for the whole clinical research industry rather than for Merck alone.
“Let's Talk Trials is something that we created for the industry, for clinical research as a whole. It's not specific to Merck.”
0:49:38 · Jennifer Sheller · Patient education
Sheller says Merck is resetting its operating model around risk-based quality management, with implementation planned for 2026.
“we're really focused, as I mentioned earlier, on resetting our operating model with risk-based quality management.”
0:52:52 · Jennifer Sheller · Quality management · 2026
Sheller says she has built an automation hub at Merck, linked to the organization's functions through matrix leads.
“So I build an automation hub and I'm pairing it up through the functions in our organization.”
0:53:28 · Jennifer Sheller · Automation