This article was originally published in the International Clinical Trails Winter 2026 edition and is reprinted here. You can read the full original edition here. The article below was written by Rob Bedford.
Stop Adding More Sites: Why Recruitment Fails Before FPI in MedTech studies
Patient recruitment is routinely described as the single biggest risk to clinical trial timelines (it’s the #1 reason why trials fail1). In medtech, it is often treated as an unavoidable operational headache rather than a predictable outcome of earlier decisions. When recruitment stalls, the default response is all too familiar: more sites, more countries and consequently more budget.
That instinct is understandable. It is also frequently the wrong solution.
In many medtech studies, recruitment problems are not caused by a lack of patients. They are caused by feasibility assumptions that were never realistic, protocols that underestimate real-world care pathways and site strategies that prioritise coverage over accountability. By the time the first patient is screened, the conditions for delay have already been baked in.
The “more sites” reflex and where it comes from
The belief that more sites equal faster recruitment is deeply ingrained. It is reinforced by experience in large pharmaceutical programmes where recruitment volumes are vast, timelines are long and the cost of delay is absorbed across portfolios rather than individual studies.
In pharma, site proliferation can sometimes be justified. Large Phase III drug trials often require geographic diversity, statistical power across sub-populations and redundancy to hedge against slow-recruiting centres. The infrastructure, budgets and internal governance exist to support this complexity, even if inefficiently.
Medtech is different.
Many medtech trials are smaller, more targeted and more sensitive to operational variance. A device study with 150 to 400 patients does not benefit in the same way from hundreds of sites. In fact, spreading limited enrolment targets thinly across many centres often makes recruitment slower, not faster.
Yet the pharma-derived reflex persists, particularly when sponsors are under pressure to demonstrate momentum to boards, investors or notified bodies. Opening more sites feels like progress, even when it dilutes focus and accountability.
The “more is more” approach disproportionately benefits CRO delivery models. With an estimated 50–70% of industry-funded trial activities outsourced to third-party organisations 2,3,4, it is unsurprising that site expansion is often the default response to recruitment challenges.
Feasibility is where most recruitment failures begin
Feasibility is often framed as a box-ticking exercise: send a questionnaire, collect optimistic numbers, average the responses and move on. For medtech studies, this approach is especially risky.
Devices are embedded in care pathways. Eligibility is shaped not just by diagnosis, but by referral patterns, treatment sequencing, clinician behaviour and institutional incentives. A site may treat hundreds of patients with a condition each year, but only a small subset may ever be eligible for the device.
Common feasibility blind spots include:
- assuming prevalence equals availability
- ignoring where patients are lost between diagnosis and intervention
- underestimating competing procedures or alternative technologies
- failing to account for learning curves and workflow disruption at sites
When feasibility does not interrogate these realities, recruitment risk is built into the protocol. Adding sites later cannot fix a flawed assumption about how patients actually move through care.
Low recruitment is the #1 reason why trials fail
When trials are terminated early, it is often assumed that the underlying intervention simply did not work. The data tell a very different story.
A large retrospective analysis of terminated clinical trials published in Clinical Trials examined over 900 discontinued studies and found that only 21.3% of trials were stopped due to scientific data. Of those terminated due to non-scientific reasons, 56.5% were due to low enrollment and 9.4% were terminated due to trial administration or conduct issues, including problems with protocol execution, investigators or site performance1.
For medtech sponsors, this distinction matters enormously. Recruitment shortfalls and administrative failures are not acts of nature. They are predictable consequences of weak feasibility, unrealistic assumptions and insufficient early oversight. Once enrolment stalls or data quality deteriorates, adding more sites or increasing monitoring rarely fixes the underlying problem. It typically increases cost and complexity.
This evidence also challenges the industry’s instinct to treat recruitment as a downstream activity. If over half of trial failures are driven by insufficient accrual, then recruitment risk must be addressed at the design stage, not mitigated reactively once timelines are already under pressure.
Why site sprawl damages data quality as well as timelines
Recruitment challenges are usually discussed in terms of speed, but site sprawl has a second, less visible cost: data variability.
In medtech studies, outcomes are often influenced by operator technique, workflow integration and subtle differences in how procedures are performed. Increasing the number of sites increases the number of interpretations of the protocol, the number of training events and the number of opportunities for inconsistency.
This matters because medtech datasets are often smaller and statistically less forgiving. Losing data from a handful of patients due to protocol deviations, documentation gaps or inconsistent endpoint capture can materially affect interpretability. Unlike large pharma trials, there is rarely the option to simply enrol another cohort without significant delay or redesign.
A smaller number of high-performing sites, each enrolling a meaningful number of patients, typically produces cleaner, more interpretable data than a sprawling network of low-enrolling centres.
The patient burden problem is underestimated in medtech
Medtech trials often underestimate patient burden because the intervention itself may be less invasive than a drug regimen. That can be misleading.
Devices frequently require additional visits, imaging, procedural time or follow-up assessments that are not part of standard care. If these burdens are not carefully mapped and mitigated, participation drops, screen failures increase and retention suffers.
In contrast to pharma, where patients may be motivated by access to novel therapeutics, medtech participants are often choosing between an investigational device and an established procedure. Convenience, recovery time and disruption to daily life matter greatly.
Designing recruitment without a clear view of the patient journey is another way delays are engineered before the study begins.
Recruitment is a system problem, rather than a site problem
One of the most persistent misconceptions in clinical research is that recruitment is something sites “do for you”. In reality, sites operate within constraints: clinic capacity, staff availability, competing studies and local priorities.
In medtech, where patient identification may depend on procedure lists, imaging workflows or referral patterns, relying on sites alone to surface eligible patients is unreliable. This is not a criticism of sites, but a recognition of how healthcare systems function.
Effective recruitment strategies recognise this and design around it by:
- selecting sites based on demonstrated procedural volume and commitment
- aligning enrolment targets with realistic capacity, not theoretical reach
- supporting sites with clear screening logic and simple workflows
- reducing administrative friction that competes with clinical priorities
When recruitment is treated as a shared operational system rather than a site-level task, performance becomes more predictable.
Why adding countries often makes things worse
When recruitment stalls, expanding into additional countries is often presented as a solution. For medtech, this move frequently increases complexity faster than it increases enrolment.
Each new country introduces new ethics processes, regulatory expectations, contracting timelines and operational nuances. The marginal recruitment gain from an additional geography can be outweighed by MDR / IVDR delays elsewhere and increased variability in execution.
This is not an argument against international studies. A diverse patient population can strengthen generalisability and support multi-region regulatory strategies. A smaller number of countries, selected for both recruitment potential and operational coherence, often outperform a wide but shallow footprint.
Pharma programmes may absorb this complexity as a cost of doing business. Medtech programmes rarely can afford this luxury.
What good recruitment design looks like in practice
If recruitment success is designed upstream, what does good design actually involve?
First, feasibility must be treated as analytical work, not administrative work. That means pressure-testing assumptions about care pathways, not just collecting site enthusiasm. From a practical point of view, this means going beyond the responses in the site feasibility questionnaire and understanding in detail the site screening processes and the patient journey.
Second, site strategy should optimise for performance density, not geographic coverage. A site enrolling twenty patients is significantly more valuable than ten sites enrolling two each. Go for the sites with the track record and deep experience. This isn’t always the big teaching hospitals or prestigious institutions; in fact they can often have portfolios of studies that are either directly competing for patients or competing for the research coordinators’ time.
Third, protocols should be reviewed through the lens of participation burden. If an assessment does not materially support safety or performance claims, its cost to recruitment should be questioned. Put yourself in the shoes of the patient and ask yourself “would I come in for a 6-week visit and a 12-week visit”. Use electronic patient/clinician-reported outcomes (ePRO / eCOA) where possible.
Fourth, oversight must be early and targeted. Waiting months to discover that screen failures are high or visits are being missed guarantees delay. Early indicators are far cheaper to act on. Pick up the phone and speak with the coordinator; high screen failure rates can often be due to misunderstanding of overly-complex protocols, rather than genuine patient ineligibility. Embrace technology that improves the experience for both sites and subjects.
None of these principles require radical innovation or new technology. They require attention, experience and a willingness to challenge inherited assumptions.
Contrasting medtech with pharma and biotech
It is important to acknowledge that some of the structural issues discussed here manifest differently in pharma and biotech.
Large drug programmes often operate at a scale where inefficiency is tolerated. The cost of delay is real, but it is distributed. Medtech companies, particularly small and midsize manufacturers, experience delay as an existential risk.
Biotech startups face a different pressure: single-asset dependency. For them, recruitment failure can derail valuation, financing and strategic options. Yet they often inherit trial designs and operational models from pharma contexts that do not fit their reality.
Medtech sits somewhere between these worlds, borrowing practices from both, sometimes uncritically. The result is a tendency to overbuild trials that neither benefit from pharma-scale redundancy nor accommodate biotech-level fragility. Medtech studies also need to demonstrate health economics and commercial viability of products, not just clinical benefit.
Recognising this mismatch is the first step toward better design.
Closing thought
Medtech innovation often focuses on improving clinical outcomes through better technology. Clinical trials should be no different. Improving how trials are designed and run is not a peripheral concern. It is central to whether promising devices reach patients at all.
The evidence is clear: more than half of terminated trials fail because they cannot recruit enough patients, and nearly one in ten fail due to avoidable administrative or conduct issues — not because the intervention itself was ineffective. The uncomfortable truth is that many recruitment failures are predictable. The encouraging truth is that they are also preventable.
References
- https://pmc.ncbi.nlm.nih.gov/articles/PMC4444136
- Tufts Center for the Study of Drug Development (CSDD). Impact of CRO outsourcing on clinical development performance. Tufts CSDD Briefing, Tufts University School of Medicine, Boston, MA
- Drug Information Association (DIA). The next step toward a collaborative clinical ecosystem: CRO–sponsor partnerships. DIA Global Forum, October 2018.
- Arizton Advisory & Intelligence. Clinical Trials Outsourcing Market – Global Analysis and Forecast. Arizton Market Research, 2022
