Sites Standing Strong: the 2025 Save Our Sites Conference

In addition to AI, patient recruitment, and SOPs, one conference session focused on the dreaded site audit. Panelists discussed ways sites can be prepared for an audit and what to do if they are faced with a Form FDA 483, the document inspectors issue when they observe violations of FDA requirements.

By Carolyn Gretton

The second annual Save Our Sites (SOS) conference, held in Oklahoma City March 13–15, offered attendees the opportunity to connect with each other and gain insights into the challenges and opportunities of operating a clinical research site in today’s environment.

SOS Co-founder Ashley Margo cited the conference’s goal as “clinical research professionals coming together to make sites stronger.” To that end, a couple of themes ran through the conference’s interactions and presentations. Here are a few highlights:

How AI fits into sites

Not surprisingly, given its growing prevalence in the life sciences, artificial intelligence (AI) was front and center at SOS. In “AI and the Future of Trials: Adapting, Innovating, Excelling,” panelists discussed the ways AI could be used in site operations and the benefits and risks inherent in the technology.

Mike Wenger, founder of FullStack Trials, said AI isn’t limited to one use case, and that its use cases inform its quality. “[AI] is not good or bad; it’s more nuanced than that,” he said. “Is it taking the wheel, or is there still a human in control ensuring quality?”

Wenger added that using AI creatively is key to getting the most out of it. “It’s a skill you need to develop.”

Andrew Mitchell, founder and CEO of YEZA.AI, called AI “a tool in the toolbox” and said like any other tool, “you use the best tool for the job.” He compared AI tools to a chef’s knives, which go with the chef from job to job.

“Make sure you’re using appropriate AI tools for appropriate uses,” said Denali Rose, VP of strategic partnerships for Haystack Health and producer of the Note to File Podcast. For example, she said she thinks SOPs are a great starting point for AI.

“Have AI do the things you don’t want to do,” she said.

Andrea Cordero, clinical project manager at Alto Neuroscience, said AI use needs to be carefully considered. “Take a step back and strategize its usage, rather than just throwing it at everything,” she said. “You need to have a workflow in place from the start, then incorporate AI.”

“Saying AI can do everything isn’t very helpful,” agreed Derk Arts, founder and CEO, Castor. “It’s good to consider if there will be a net benefit. Start with small problems you can wrap your head around. If it’s something you can train an intern to do, that’s a good use case for AI.” However, Cordero cautioned, “you need to understand what the [regulatory] risks are. Proprietary data and privacy are concerns.”

These concerns were echoed by many of the attendees. One said she was nervous about using AI, even though she felt she needed to learn it or be left behind. “How do you feed protocol info into AI and maintain privacy?” she asked the attendees.

Arts said there need to be data controls in place and an agreement with the AI supplier. “It’s no different or more dangerous than [using] any other cloud storage service,” he said.

Wenger said it also helps to have “rules of the road” or an SOP for how AI is used in the organization. “Otherwise you’re opening up your company to risk,” he said.

I was truly alarmed by one hypothesis offered when another attendee asked if AI could make it easier to fake research data. The panelists all agreed: Yes, it could. But Arts offered some reassurance by adding that experts are “looking into biometric ‘watermarking’ to help authenticate research data.”

Overall, the panelists were optimistic about the future of AI in pharma. “I think AI will reduce burnout in our industry,” Arts said.

“The genie is out of the bottle; the way we work has fundamentally changed,” Mitchell said.

“Change comes with risk, but once you get to the other side, it’s a better model,” Wenger said.

The discussion of AI continued in “Tech: Cutting Costs and Streamlining Site Success Through Efficiency.” Ali Imran, co-founder and chief intelligence officer at Plexus AI, said, “Like it or not, AI is coming to clinical research.” He noted that it’s “much bigger than ChatGPT” and that it can give clinicians more time with patients. “How we adapt it is based on the systems in our clinics,” he said.

“AI can eliminate redundancies,” said Deborah Tunick, head of clinical operations, Tilda Research. “We’re getting to the point where AI will be running in the background.”

When asked if tech should be symbiotic with sites, Imran said, “Tech should be ambient and learn your systems over time; AI enables that.”

When discussing technology use in general, the panelists agreed that tech platforms should be able to adapt to user needs as they grow and change over time. “Clinical research changes constantly, and if your platform doesn’t change with it, it can become quickly outdated,” Tunick said.

“Sites should have a lot of say in how tech operates,” said Courtney Alexander, site operations project manager, M3 Wake Research.

“You need to demand a comprehensive, easy-to-use solution that’s customer-centric,” said Ashish Sahni, founder and CEO of Clin Street. “The system needs to adapt to you, not vice versa.”

The skinny on SOPs

Standard operating procedures (SOPs) were the focus of “SOPs: Tailored or Universal? Crafting the Blueprint for Seamless Site Operations.” Eyde Edens, senior attorney at Kulkarni Law Firm, called SOPs “your opportunity to say this is how we interpret these regulations,” which at times can be vague. “SOPs are your armor,” she said.

Panelists agreed that SOPs should be as comprehensive and detailed as possible, but there was some debate as to how many SOPs are necessary. Scott Whitt, site owner and general manager, Triad Clinical Trials, said the most important aspect of SOPs is that they be “consistent and clear, reliable and defensible.” However, he noted that his organization looks to keep SOPs down, not scale them up, and that he believes in having a maximum of 10 to 12 SOPs. “We’re not going to create SOPs just to please one sponsor,” he said.

On the other hand, Jenn Pages, CEO of ClinMastery, indicated that new SOPs should be created regularly to make sure all aspects of business operations are covered. She also recommended staff be retrained on SOPs every two years to keep them up to date on any newly added SOPs.

Preparing for the dreaded audit

One compelling reason for strong SOPs was the subject of “483 Avoidance: Insider Tips for Legal and FDA Experts.” Panelists discussed ways sites can be prepared for an audit and what to do if they are faced with a Form FDA 483, the document inspectors issue when they observe violations of FDA requirements.

The panelists agreed the best way to handle a 483 is to avoid getting one in the first place. “Have your SOPs in place,” said Jessica Powell, chief development officer, Alto Neuroscience.

“Document everything,” said Suzanne Vyvoda, founder and principal consultant, Vyvoda Consulting, and fractional head of development operations for Thornhill Therapeutics. “Honest mistakes happen, and documentation can clear them up and explain what happened.” Ideally, the site has an escalation pathway in place for when errors are uncovered, she said.

“If you find a mistake and fix it, there won’t be a problem,” said Darshan Kulkarni, principal attorney, Kulkarni Law Firm. “It’s the hiding that’s the problem.” He recommends sites follow a “plan, do, check” cycle. “It’s not just about avoiding a 483, it’s about your whole operation,” he said.

“Make sure you have your PI sign off on everything they delegate,” said Robert Goldman, head of clinical operations, Contraline. “This will make you audit-ready.

“You need to have processes in place where fraud and falsified data are reported,” Goldman said.

Several panelists recommended sites bring in a third party to conduct a mock audit so that team members are prepared from the outset. “I think the idea of a mock audit is really powerful,” Vyvoda said. “It trains you to interact with an FDA inspector.”

“The biggest mistake I see is staff not knowing what to say and how to behave” with inspectors, Goldman noted.

Strategies for smart recruitment

In “Recruit Smart, Stay Strong: The Only Metric That Matters, Patient Recruitment,” panelists discussed different ways sites can up their recruitment game.

“It’s easy to say we need to follow up and speak the same language” as patients, said Ronak Ganatra, co-founder of PatienACE. “But it’s hard to do when a coordinator is wearing different hats.”

Ganatra recommended using tech like email whenever possible to streamline the recruitment process and automate follow up. He also said it was important to meet the patient on the same platform as where they find you.

“For instance, if you run a Facebook ad, don’t call the respondents because they might not pick up if they don’t recognize the number,” he said. “Try texting them instead.”

Alexis Reese, patient recruitment coordinator, Activian Clinical Research, said sites should not be afraid to go to the community. “The more we express and explain what we do, the more impact we can have,” she said.

Ganatra said when thinking about recruitment, sites should move beyond the study perspective and think about building a patient funnel. “It’s a less expensive way to get those leads,” he said. “Free [health] evaluations are a good way to build that database — it gets people familiar with the site.”

Having a clinic on site also helps with building a database, Reese said.

Audience members provided their own tips, such as offering an Uber account and other perks to enroll, and going to community events to get leads. One asked if it was more effective to conduct a light screening via phone just to get patients in the door, or to do a deeper screen to make sure the patients fit the trial.

The answer, according to Ganatra: It depends. “It’s site and indication-specific,” he said. “For example, a rare disease trial would need a deeper screen. Conducting a general screen is better for studies like obesity and NASH.”

“We tend toward a deeper phone screen,” Reese said. “If they don’t fit the trial, then you get their info anyway so you can contact them later for a study they may qualify for.”