Most trials don't miss their enrollment targets because the science is wrong or the patients don't exist. They miss because eligible, findable patients hit a barrier somewhere between interested and enrolled, or between enrolled and finished, and nobody measured it in time to act. The scale isn't subtle: in Tufts Center for the Study of Drug Development benchmarks, roughly 11% of trial sites fail to enroll a single patient and another 37% under-enroll, leaving only about four sites in ten to hit their target.
The barrier is almost never on the eligibility checklist. It's a ninety-minute drive at 8am. A consent form written three grade levels above the patient's reading ability. A spouse who never agreed. A history of appointments the patient has quietly missed for years. These are predictable and, more usefully, addressable, if you can see them before the patient walks.
The appetite to participate is there. In CISCRP's 2023 global survey of more than 12,000 people, 87% said they would be willing to join a clinical trial, yet only about a third said they trust the pharmaceutical industry to communicate about trials openly and accurately. That gap between willingness and trust is where patients get lost, and it's only one of nineteen places it happens.
PRISM exists to make those places visible. It scores patient readiness across 62 models built from claims, patient-reported answers, geography, and behavior, then rolls those models up into 19 primary frictions: the actual reasons a specific patient declines or drops out. Each friction carries an intervention, because a signal you can't act on is just a more precise way to lose a patient.
The 19 frictions sort into four groups. This page is the map; each group has its own deep dive.
1. The barriers to saying yes
Six frictions decide whether a patient can consent, and mean it: whether they trust the process, understand it, can carry its mental load, feel settled in the decision, see a personal benefit, and aren't blocked by fear or low mood. These are the head-and-heart barriers, and they bite at consent. A patient can be eligible, willing, and ten minutes from the site, and still say no at the table because one of these went unaddressed.
2. The barriers outside the clinic
Some of the strongest predictors of who enrolls and stays have nothing to do with the protocol: the drive to the site, the cost of taking part, chronic instability in food and housing, no broadband or device, a study not offered in the patient's language. These social drivers of health decide who a trial is physically open to long before anyone reads an inclusion criterion. Congress took them seriously enough to legislate: the 2022 Food and Drug Omnibus Reform Act directed the FDA to require Diversity Action Plans aimed at enrolling underrepresented populations.
3. The people and support around the patient
A patient rarely decides about a trial alone, and rarely completes one alone. Caregivers, family, community, and a steady provider shape whether someone enrolls and whether they last. A patient can be fully able to participate and still lose the trial because the people around them don't back it, because they're the one doing the caregiving, or because a functional limitation the protocol never accommodated makes a procedure impossible.
4. The barriers to staying in
A signature is the start of the retention risk, not the end of it. Enrolled patients leave for reasons you can often see coming: a track record of inconsistent follow-through, a visit and dosing schedule their life can't sustain, fear of a specific procedure, or plain fatigue after too many prior lines of treatment.
The point isn't to screen patients out
It would be easy to read a list of frictions as a list of reasons to exclude people. That's the opposite of the point, and it's how trials end up enrolling only the patients who were always going to be easy, which is also how they end up unrepresentative. Among pivotal trials for drugs approved in 2023, only 23% had representative enrollment of Black participants and 30% of Hispanic participants, and those figures have been moving the wrong way.
Readiness scoring earns its keep by showing where to intervene. Low trust is a reason to invest in the relationship before consent, not a reason to skip the patient. A transport gap is a voucher, not a rejection. A comprehension mismatch is a plain-language consent summary, not a closed door. Seen early, most of these frictions are solvable, and solving them is how a trial reaches the patients it says it wants to reach.
Sources and methodology.
This article synthesizes published research and public benchmarks with Jumo Health's PRISM readiness framework. Statistics are presented with their publication context; trial conditions and patient populations vary.
- Tufts Center for the Study of Drug Development, site enrollment performance benchmarks. View source
- CISCRP, 2023 Perceptions and Insights Study. View source
- Communications Medicine (Nature), 2025, trial representativeness of FDA-approved drugs. View source
- National Assessment of Adult Literacy (NAAL), US Dept. of Education / NCES. View source
- Paasche-Orlow et al., New England Journal of Medicine, 2003, consent form readability. View source
- Johns Hopkins Medicine IRB, informed consent readability guidance. View source
- Journal of Clinical Oncology (ASCO), 2022, travel distance in early-phase trials. View source
- Geographic access to NCI-funded cancer research sites (PMC). View source
- Patient-reported out-of-pocket costs in early-phase oncology trials (PMC). View source
- Pew Research Center, Internet/Broadband Fact Sheet. View source
- FCC, Broadband Progress Report (100/20 Mbps standard). View source
- Kogan et al., Psycho-Oncology, 2022, caregiver role in phase 1 trial decisions. View source
- Cerutti et al., Cancer Medicine, 2025, family system and trial retention. View source
- Tufts Center for the Study of Drug Development, protocol amendment benchmarks. View source
- Trials (Springer), 2025, retention and missing primary-outcome data review. View source
- US FDA, Diversity Action Plans (FDORA 2022). View source
