Why Clinical Trial Coordinators Are Still The Human API

By John Oncea, Chief Editor, Clinical Tech Leader

Maggie Kelehan logs into her inbox, then starts seeing patients – seven to ten of them on a busy day. Between visits she’s in the EMR at whichever ophthalmology practice she’s embedded in that day, then her sponsor’s EDC, then the CTMS, then Google Drive and Sheets, then whatever regulatory paperwork one of her active studies needs. None of those systems reliably talk to the others. In practice, Kelehan becomes the connector.

Kelehan is a Clinical Research Coordinator II at an ophthalmology-focused research company, three years into the field. She works across multiple sites, multiple sponsors, and multiple EDC systems, which makes her a useful person to ask a blunt question: after a decade of the industry buying eSource, EDC, CTMS, and eConsent platforms, why does so much of the job still feel like manual data entry?

Her answer, in short: because nobody bought the connections between them.

“They don’t really communicate with each other,” she said of the systems she moves through every day.

By her estimate, roughly 30% of her time, more on a heavy regulatory day, goes to reconciling systems that should already be in sync. Asked what she’d get back if that friction disappeared, she didn’t hedge: “My job would probably have about 15 to 20 hours less a week.”

That’s not a complaint about any one platform. It’s the finding worth sitting with: individually, the systems mostly work. Collectively, they don’t cooperate. And the gap between “works” and “works together” gets filled, every day, by a coordinator.

The Myth Of Digital Transformation

Ask Kelehan whether her multi-site organization is standardized on tooling, and she’s blunt about it: “Each one is its own adventure,” she said of the sponsors she works with, right down to sponsors who, mid-trial, “decide to scrap their electronic data capture completely and just decide to do it all on paper.” Her best guess as to why: cost. Nobody had told her the real reason.

Every new study means learning a new system’s logic from scratch, “every single new study,” to figure out what a given EDC is going to ask of her and how its queries behave. That’s not a training problem. It’s the absence of any shared workflow standard across an industry that’s been digitizing for two decades.

Where The User Experience Actually Lives

She’s careful not to rank EDC platforms against each other; in her experience, no single system is inherently better at data entry than another. What separates a smooth study from a miserable one is how well the sponsor or CRO configured it.

“It almost all comes down to how well the sponsor or CRO has coded the data entry,” she said. Sponsors “have a lot of flexibility with all these systems in determining user experience,” and when they don’t invest in it, “the data entry tends to have a lot of logical errors,” auto-queries firing on questions that shouldn’t be answered yet or flagging required fields that are blank for a reason.

Her example: a question that generates a query saying it shouldn’t be answered until the study ends, but generates a different query, demanding data, if it’s left blank. “That’s illogical to me,” she said. “It should be a question that’s triggered after the patient has already completed the study.”

The fix she’d recommend to any sponsor: building or buying a new platform is almost embarrassingly simple; run one test patient through the entire study, start to finish, before it goes live. “Have you filled out an entire patient’s data history from A to Z in a test patient and seen if all of your queries make sense?” Nobody does this consistently, in her experience, and it shows up later as coordinator hours spent resolving queries that never needed to exist.

Automation On A Broken Foundation Doesn’t Help

Kelehan isn’t automation averse. She’s automation-skeptical about sequencing. Very little of her current workload is automated, and when asked whether that’s a problem she’d like solved, her answer complicates the usual “just automate it” pitch.

“If we do automate things, we’re going to have to get the basics down to be very smooth first,” she said. Layer automation on top of an EDC with inconsistent logic, and “I’m not sure automation would necessarily make it better,” she said. It might just create a bigger version of the same headache, faster. Her own estimate for when the basics will be there: roughly a decade out.

The fix she wants most isn’t AI. It’s her eSource data flowing directly into the EDC without a human retyping it. “If we had the capacity to have the source automatically upload into EDC, then there wouldn’t be as many transcription errors,” she said, and monitors could spend their time checking whether the data makes logical sense instead of whether it matches what’s already been keyed in twice.

Where She Actually Sees AI Helping

Given how much industry conversation orbits AI right now, it’s worth noting Kelehan doesn’t treat it as a silver bullet; she calls it “a little overhyped right now.” Her reasoning is grounded in daily use: “There’s still quite a bit of human error in AI since it does aggregate data from humans,” and she’s watched it fumble something as low-stakes as a recipe.

Where she does think it could help, concretely: patient pre-screening and outreach. “I think you could have an AI system pre-screen someone, find a patient in your CTMS or EMR system, call them and pre-screen them,” and scheduling logic across multi-site teams, flagging for a newer coordinator that she’s already committed to one site and can’t be booked at another, before the conflict happens instead of after.

Notice what both examples have in common: they’re not asking AI to replace judgment. They’re asking it to catch the kind of coordination failure that currently eats a coordinator’s morning.

What Vendors And Sponsors Rarely Ask

Asked what question sponsors should put to coordinators before rolling out new technology, Kelehan’s answer wasn’t about features; it was about testing for logic, the theme running through the whole conversation. She extended it to the patient-facing side, too: could an eighth grader complete this? Could an ordinary person, with a full life outside the trial, use this instrument or fill out this diary correctly every day for six months or a year? “Is that a reasonable ask of these people?”

That’s a question about respect for the end user, coordinator, and patient alike, and her interview suggests it gets skipped more often than it should.

The Payoff Is Patient Care

Kelehan doesn’t think any of this is abstract plumbing. Pressed on whether coordination-level fixes actually reach the patient, she was direct: “This is really going to improve patient care.” The more a cardiologist can see a patient’s full record – pulmonologist, PCP, ophthalmologist, all of it – instead of just looking at their own data, the better care that patient receives.

That’s the case for fixing interoperability that doesn’t require a coordinator’s math about hours saved, though the math helps. It’s the case that every system a coordinator manually reconciles today is a system a physician can’t yet see holistically tomorrow.

For sponsors and vendors reading this: the technology mostly works. The connections between them don’t. Fix that, and you’ve given coordinators back a fifth of their week, and given physicians the whole patient, not just their slice of one.