In the United States, only about 7% of adult cancer patients enroll in a clinical trial. Put another way, more than nine in ten people with cancer are treated entirely outside of any trial — the very same trials deciding the next generation of therapies for the next generation of patients.
Intuitively, one might assume that means patients simply don’t want to take part due to a lack of trust in experimental treatment; the data doesn’t support this, though. The largest meta-analysis to date led by Unger and colleagues that evaluated the motivation behind patient enrollment, found that the single biggest reason they steer clear of trials is structural; there’s simply no suitable trial to join at the place they receive their care. This alone is responsible for about 56% of non-participation with patients, who are genuinely asked but decline, representing a small minority. In other words, it’s not a problem of willingness. A patient and relevant trial both exist simultaneously yet never meet.
The problem, then, is the channel. For cancer patients in the United States, there’s pretty much one route and one route only to learn a given trial exists: the physician treating the patient. At any one moment, thousands of oncology trials are recruiting across the country — each with convoluted, multilayered eligibility criteria tweaked on a continuous basis. Expecting a physician, within the confines of a quick patient visit, to search this vast and ever-evolving space is unrealistic. So, physicians fall back on what they know: trials existing within their own institution and network.
From Blind Spots to Better Outcomes: How SNF Data Transparency Strengthens LTSS
What if health plans could identify member decline before an avoidable hospitalization occurs? Real-time clinical visibility into long-stay SNF members uncovers risk earlier and drives better outcomes.
A backup channel — the public clinical-trial registry — was supposed to help with the information right there and readily available. But it was never built for the patient, and it carries a structural flaw that the next paragraph makes concrete. The registry indexes trials mainly by one field (“tumor type”), so the patient can only search by diagnosis.
Modern oncology has been moving the other way for a while, though, and more and more trials are organized around a molecular target rather than an organ (e.g., a KRAS mutation trial recruiting patients with pancreatic, colorectal, and lung cancer since it’s after the mutation that is relevant rather than the organ the cancer happens to grow in.) In this case, pancreatic cancer patients with a KRAS mutation would type the only word they know into the search box —”pancreatic cancer” — only to have the trial they might qualify for perhaps not pop up at all. That’s because we are still organizing trials by the very classification oncology is abandoning– where in the body it originated–and thus systematically burying the most cutting-edge category: biomarker-driven, tumor-agnostic trials.
Beyond all of this is the element of time. Eligibility criteria are tweaked, trial slots open and close, and patients who’ve broken through the first two barriers to find a trial still face weeks of waiting and screening. So, their own disease and the trial’s status are moving targets during this stretch. By the time a patient acts on what they found, it may already be out of date.
The way the trials are classified and the fact that physicians are the sole conduit of information create problems not just for patients but also for trial sponsors.
Just as patients can’t find trials, trials can’t find patients; more than 20% of oncology trials fail to fill and are often terminated early as a result. Under-enrollment is a leading cause of early trial termination, and every day of delay is a steep cost to the sponsor — one ultimately priced into every drug reaching the market late. These are not two separate problems — the patient who can’t find a trial and the trial sponsor unable to recruit enough patients stand on opposite sides of the same exact wall, in fact.
Closing the Quality-Affordability Gap in Workplace Mental Health
In an interview, Kyan Health Co-Founder and Chief Commercial Officer Konstantin Struck discussed how Kyan gives mid-market and enterprise employers access to premium workforce mental healthcare, at a price point that is affordable.
Now, let’s be clear about which side of that wall the enormous investment here is aimed at — with AI impacts now in play as well. While AI has genuinely made parts of patient recruitment (a large market) quicker and more optimized, especially when it comes to identifying eligible populations, screening, and enrollment, the direction of said effort shows a pattern. Almost all of the investment points at the sponsor’s side of the wall — helping trial sponsors find patients faster — with almost none of it aimed at helping people find trials, understand whether they fit, and have a real say in decisions made about their own bodies. This is not an argument against AI, it’s instead one against asking the question from only one side. When every tool makes trials better at reaching patients, but none is helping patients see and seize their own options, the wall cannot be broken down even with a faster machine on the sponsor’s side.
This structural problem calls for a structural solution, requiring no new science, regulation, or data.
The first move? Re-indexing trials by what patients are actually looking for. A free-text trial description mentioning who the trial sponsor is seeking in a candidate can be more effective than simply mentioning the structured “tumor type.”
You can then use that same text to nail down the molecular target, any previous treatment prerequisites, and the eligible population, and thus point pancreatic cancer patients (for example, as mentioned earlier) to the KRAS trials they might never dig up otherwise.
In translating eligibility criteria into a language patients can understand, trial sponsors are not tossing out formal screening but allowing patients and physicians to walk into screening conversations already knowing the trial is worth talking about.
Given fresh regulatory tailwinds, these changes are not hard to achieve. In December 2025, the FDA rolled out its final guidance on boosting clinical trial participation before announcing two oncology “real-time clinical trial” pilots four months later: opening a request for comment on how AI and data methods can make trial efficiency better. The premise driving these moves is the very argument this piece makes from the patient’s side: data methods can help solve the problem of clinical research increasingly bottlenecked by information. Regulators are already using that same basis to rework how trials run — and the same logic applies just as well to how trials get found.
For too long, the problem of lackluster clinical trial recruitment has been portrayed as an awareness, education, or recruitment problem — names essentially blaming the patient or budget for the gap. The evidence points to something more uncomfortable, though, yet also more fixable: an information layer that indexes modern trials with an outdated classification, speaks to patients in technical jargon, and expires while the patients wait. For decades now, we’ve asked patients to compensate for the failures of this layer with a 7% enrollment rate coming out on the other side. It’s the infrastructure that’s failed them (not the other way around) calling on us to fix the index — not the patient.
Photo: FotografiaBasica, Getty Images, 503105612
Yuchen Gan holds a Master of Science from Columbia University and works on risk and data systems in regulated finance, where she builds explainable, rule-based systems for high-stakes decisions. She independently researches clinical trial discovery and matching.
This post appears through the MedCity Influencers program. Anyone can publish their perspective on business and innovation in healthcare on MedCity News through MedCity Influencers. Click here to find out how.