Guide
How to Choose a CRO for an IVD Study
Most CRO selection advice is written for drug trials. Diagnostics run on different endpoints, different sites, and a different definition of done.
An in vitro diagnostic study and a therapeutic trial look similar on an org chart and almost nothing alike in practice. A drug trial asks whether an intervention changes outcomes, and it answers that with dosing, safety monitoring, and longitudinal follow-up. A diagnostic study asks a narrower question: does this test give the right answer, in the right hands, on the right patients?
That difference shapes everything downstream, and it is the reason CRO selection advice written for pharma translates poorly. Here is what actually matters.
The endpoints are different, so the expertise has to be
Diagnostic performance is measured in sensitivity, specificity, and agreement against a comparator or reference method, supported by precision, reproducibility, interference, and detection limit work. There is no dosing arm and usually no follow-up visit. Most subjects are enrolled, sampled, and finished in a single encounter.
A team that has spent its career on therapeutic protocols will not instinctively know that your reproducibility design is thin, or that your comparator method choice will draw a question from FDA, or that your intended use population does not match the sites you selected. Those are the errors that cost a submission cycle, and they are invisible until review.
Ask directly: how many IVD submissions has this specific team supported, and through which pathways? Not the company. The people who will be on your study.
Site type matters more than site count
This is the failure mode that catches sponsors most often, and it is particularly acute for anything headed toward the point of care.
A conventional research site is staffed by professional coordinators. That is exactly right for a 510(k) study running in a laboratory setting. It is exactly wrong for a CLIA waiver study, where FDA wants results generated by untrained intended use operators who resemble the people who will really run the test: a medical assistant, a front-desk staffer, someone with no laboratory training at all.
A CRO with a large network of traditional research sites cannot solve that by adding more of them. You need clinics that have the right operators and the right patient flow, plus someone to carry the research burden those clinics cannot absorb on their own.
Ask directly: for a waived-setting device, where do the intended use operators come from, and who handles the regulatory paperwork at those sites?
CRO, SMO, or both
Traditionally these are separate companies. The CRO designs and runs the study. The site management organization supplies and manages sites. Sponsors often end up contracting with both and absorbing the seam between them.
That seam is where schedule slips live. When enrollment stalls, the CRO points at the sites and the SMO points at the protocol, and the sponsor arbitrates. An integrated model removes the arbitration, though it is worth confirming the integration is real rather than a marketing label over a subcontract.
Ask directly: is the site network contracted directly by the CRO, or subcontracted? Who has authority to fix an underperforming site, and how fast?
What to actually ask before signing
- Which regulatory pathway do you recommend for this device, and why? A good answer engages with your predicate and intended use. A vague one is a signal.
- How long from executed contract to first patient enrolled, and what drives that number?
- Are the proposed sites already contracted and IRB-ready, or will they be qualified after we sign?
- Who is the day-to-day contact, and how many other studies do they carry?
- What happens if a site underperforms? What is the replacement process?
- Can you show the study design supporting both clearance and waiver, if the device is headed for point of care?
Red flags
A site count with no composition behind it. “Over 200 sites” means little if none of them match your intended use profile or can supply the operators your pathway requires.
A timeline that ignores startup. If the proposal is confident about enrollment but vague about activation, the schedule is optimistic. Startup is where IVD programs lose months.
Pathway advice that arrives after the protocol. Predicate strategy and intended use should drive the study design, not the other way around.
Reluctance to name the team. Diagnostics expertise lives in people, not capability decks.
When a generalist CRO is genuinely fine
Not every study needs a specialist, and it is worth saying so plainly.
If your device is destined only for moderate or high complexity laboratories, your predicate is well established, the comparator method is uncontroversial, and the evidence is largely analytical, a competent generalist CRO with laboratory experience can run that study perfectly well. The specialist premium buys you the most when the pathway is ambiguous, when waived settings are involved, or when site composition is doing real regulatory work.
Being honest about that distinction is, in itself, a reasonable thing to ask a prospective CRO to do.
Studybox works exclusively on in vitro diagnostics, as a combined CRO and site network. If you want a second opinion on a pathway or a study design, that is a conversation we are happy to have before anything is signed.