A physician runs a clinical trial. It goes reasonably well. She completes Good Clinical Practice training, gets through site qualification, learns the protocol, survives the monitoring visits, enrolls her patients, and closes the study out.
She would happily do another one.
Nobody ever calls.
Not because her performance was poor. Not because she withdrew. The clinical research associate who brought her the study changed jobs. The sponsor's medical director moved to a different therapeutic area. And with that, she vanished from every list she was ever on, silently, without anyone deciding anything.
Three years later, that same sponsor is complaining about a shortage of qualified sites and paying to onboard novices.
This is the one-and-done investigator, and it is one of the clearest examples in healthcare of an entire industry misdiagnosing its own problem.
The number, and the honest argument about it
Let us establish the scale, and then immediately complicate it, because the contested version is more interesting than the headline.
An analysis of 172,453 unique investigators filing FDA Form 1572 between 1999 and 2015 found:
- 49.5 percent were "one-and-done": exactly one trial, never seen again.
- 12.6 percent were "stop-and-go."
- 37.8 percent were "stayers."
A separate CTTI survey of 201 principal investigators found 54.2 percent in the one-and-done category.
Now the counterargument, which deserves airing. Analysts including Glass have argued that the true rate is likely below 20 percent, because the Form 1572 database is incomplete: investigators appear under different filings, some trials are not captured, and the denominator is unreliable. Their own analysis using Open Payments data produced roughly 20.3 percent.
That criticism is legitimate and should be taken seriously. Anyone quoting 50 percent without acknowledging it is being sloppy.
But here is why the argument about the denominator does not rescue the situation.
The most damning finding does not come from any database. It comes from asking investigators directly.
The finding that should have redirected twenty years of effort
In the CTTI survey, one-and-done investigators were asked why they had not conducted another trial.
The standard industry explanation is burden, and burden certainly featured: workload at 63.8 percent, time at 63.4 percent, reporting requirements at 56.5 percent. Those are real and the industry has spent two decades trying to reduce them.
And then:
44.4 percent said they wanted to run another trial and no opportunity was ever offered.
Not "it was too onerous." Not "I decided against it."
Nobody asked.
Now look at how these investigators entered research in the first place:
- 36.6 percent through a direct request from a sponsor.
- 29.0 percent through a colleague.
Two thirds of investigators entered clinical research through a single specific human relationship.
That is the whole mechanism, and it explains the disappearance precisely. When that one person moved on, the investigator did not decline anything. They were simply dropped by a system that had no memory of them beyond one person's contact list.
They did not leave clinical research. Clinical research lost their phone number.
Meanwhile the pool is shrinking
The context makes this worse rather than merely wasteful.
- The number of active US principal investigators declined from approximately 28,292 in 2014 to 26,115 in 2020, while trial volume increased.
- Roughly 40 percent of active principal investigators ran a single industry trial, and about 80 percent of some 37,600 investigators ran between one and five in the 2018 to 2020 period.
So a shrinking pool, concentrated activity among a small group of experienced sites, rising demand, and a large reservoir of trained-but-unused investigators that nobody is drawing on.
And every one of those idle investigators represents sunk training capital. Someone paid for their GCP certification, their site qualification visit, their protocol training, their regulatory familiarization. The industry paid for it, wrote it off, and is now paying again to onboard someone new.
This is not a talent shortage. It is an inventory management failure.
Willingness is a state nobody stores
Here is the structural diagnosis, and it is unusually crisp.
Sponsor and CRO databases store history: which trials an investigator ran, when, and how many patients they enrolled. That is a factual record of the past and it is reasonably well maintained.
What no database anywhere stores is intent.
Is this investigator available right now? Do they want another study? Do they have bandwidth this quarter? Has their patient population changed? Are they actively interested in this indication?
Those are the only questions that matter for site selection, and every one of them describes a live, private, changing state rather than a historical fact.
And there is no mechanism to express it. Consider the options available to a physician who wants more research work:
- They cannot advertise. No physician is going to post "available for clinical trials, please contact me" on a professional network. It reads as desperation and it is professionally awkward.
- They cannot register anywhere meaningful, because no neutral registry of willing investigators exists.
- They can contact sponsors directly, which almost nobody does, because it feels like cold-calling and they do not know who to contact.
- They can wait for someone to call, which is what they do.
Meanwhile the sponsor cannot see them. The sponsor sees a database of prior performance and a set of relationship lists, and has no way to distinguish "ran one trial and never wanted another" from "ran one trial, would love another, is waiting by the phone."
Those two investigators look identical in every existing system, and they are the opposite of each other.
CTTI already said this. In 2019.
The most striking fact in this whole area is that the diagnosis is not new and the recommendation has been sitting unimplemented for years.
CTTI, a public-private partnership specifically constituted to improve clinical trials, published recommendations on investigator retention. Recommendation number four was, in essence, to build mechanisms for "discovering opportunities for conducting additional trials."
That is exactly the right recommendation. It identifies the actual gap: not burden, but discovery.
As far as I can determine, no product has ever been built to do it.
The years since have produced an enormous amount of clinical trial technology: electronic data capture, decentralized trial platforms, patient recruitment marketplaces, e-consent systems, remote monitoring, site management systems. Almost all of it addresses trial conduct.
Almost nothing addresses the question of who should be running trials and how they find each other.
The reason follows the usual pattern. CROs will not publish their investigator lists, since those lists are a significant part of what sponsors pay them for. Sponsors will not maintain a shared registry with competitors. Professional networks that could theoretically host this monetize recruiter and advertiser access, which puts them on the wrong side of the market and makes physicians unwilling to disclose. And no neutral party has both verified physician identity and a reason to maintain investigator willingness data.
So the recommendation sits there, correct and unimplemented, while the industry keeps optimizing the parts that are easier to sell software for.
What would actually work
The design requirements are modest, which is what makes the absence conspicuous.
A private availability signal. An investigator states, visibly only to appropriate parties and on their own terms, that they are open to studies in specific indications. Private is essential: no physician will publicly advertise availability, and any design requiring them to do so has already failed.
Peer vouching in place of database history. The strongest predictor of a good investigator is not a row in a performance database; it is a colleague saying "she ran the study I was on, she enrolled well, she is meticulous." That is exactly the signal sponsors say they trust and cannot obtain systematically. It exists in abundance and circulates nowhere.
Member-controlled introductions. The investigator decides who can approach them and about what. This inverts the current arrangement, where the investigator is a record in someone else's database and the approach happens or does not happen without their knowledge.
Capacity as well as willingness. Phenotypes seen, approximate volumes, research staff available, competing studies. Sponsors have said explicitly in survey research that "patients ready" is their top selection criterion at 42 percent, and that 75 percent would use an inexperienced site with a large eligible population. They want this information. Nobody collects it in a form they can trust.
A peer question channel alongside it. Perhaps the most immediately useful piece: "who has run a trial in this indication and would talk to me for thirty minutes?" That question currently gets answered only by luck, and it is the difference between a nervous first-time investigator and a competent one.
And an obvious untapped supply. Retired and semi-retired physicians are a substantial potential sub-investigator pool for colleagues' sites. They have deep clinical experience, they have time, they frequently want to remain professionally engaged, and essentially nobody is recruiting them for this.
What you can do
If you have been an investigator
Tell someone you are available. The evidence says you are probably one of a very large group whose only barrier is that nobody knows. Contact the medical monitor or CRA from your prior study, and the research office at the nearest academic center. Both take twenty minutes and almost nobody does either.
Keep your own record. Trials run, indication, enrollment achieved, dates, contacts. Sponsor databases lose you when their staff turn over. Your record will not.
Notice your single point of failure. If your research career came through one person, as it did for roughly two thirds of investigators, deliberately build a second connection. This is the highest-value thing you can do for your own research continuity.
Say yes to being a sub-investigator. It is a lower-burden route back in and frequently leads to principal investigator opportunities.
If you select sites
Call your one-and-done list. You have a list of investigators who completed a study and were never approached again. The survey evidence says roughly 44 percent of them wanted more work. This is the cheapest site identification exercise available to any sponsor, it requires no new technology, and I am unaware of any organization that does it systematically.
Ask for willingness, not just history. Your database tells you what happened. It does not tell you who is available now, which is the only thing you need to know.
Track the relationship, not just the site. When your CRA changes jobs, their relationships walk out with them. Institutional memory of investigator relationships is an asset almost no sponsor manages deliberately.
Take peer vouching seriously as a signal. Investigators know which of their colleagues are good. That judgment is more informative than an enrollment number from a trial with a different competitive landscape three years ago.
If you run a research site or academic center
Map your idle capacity. How many physicians at your institution have run exactly one trial? Most research offices have never asked, and the answer is usually surprising.
Pair first-timers with someone who has done it. The single biggest determinant of whether a new investigator becomes a repeat investigator is whether the first experience was well supported.
Frequently asked questions
What is a one-and-done investigator? A physician who conducts exactly one clinical trial and never conducts another. An analysis of 172,453 unique FDA Form 1572 investigators from 1999 to 2015 found 49.5 percent in this category, and a CTTI survey of 201 investigators found 54.2 percent, though analysts have argued the true rate may be below 20 percent because the underlying database is incomplete.
Why do investigators stop running clinical trials? Burden is part of it, with CTTI survey respondents citing workload at 63.8 percent, time at 63.4 percent, and reporting at 56.5 percent. But the largest single finding was different: 44.4 percent of one-and-done investigators said they wanted to run another trial and no opportunity was ever offered.
How do physicians become clinical trial investigators? Overwhelmingly through personal relationships. CTTI survey data found 36.6 percent entered through a direct sponsor request and 29.0 percent through a colleague, meaning roughly two thirds entered through a single specific human connection, which is also why they disappear when that person moves on.
Is the number of clinical trial investigators shrinking? The active US principal investigator pool declined from approximately 28,292 in 2014 to 26,115 in 2020 while trial volume increased. Roughly 40 percent of active investigators ran a single industry trial, and about 80 percent ran between one and five over a recent three-year period.
How can a physician get more clinical trial opportunities? Practically: contact the medical monitor or clinical research associate from any prior study, approach the research office at the nearest academic medical center, tell colleagues who run trials that you are interested, and consider sub-investigator roles as a lower-burden route. There is no neutral registry of willing investigators, which is precisely the gap.
What did CTTI recommend about investigator retention? Among its recommendations was building mechanisms for discovering opportunities to conduct additional trials. The recommendation correctly identifies discovery rather than burden as the binding constraint, and no product appears to have been built to implement it.
The bottom line
An industry that spends enormous sums on trial technology, complains continuously about a shortage of qualified sites, and pays repeatedly to onboard novice investigators, has a large population of fully trained, GCP-certified, site-qualified physicians who wanted more work and were never contacted again.
Almost half of them, by their own account.
They did not quit. They were not dropped for cause. A clinical research associate changed jobs, and a career quietly ended without anyone making a decision.
The industry calls this attrition, which implies people leaving. The evidence says it is closer to neglect, which implies nobody looking.
Willingness is a live, private state that no database stores, and the only mechanism for expressing it is waiting for the phone to ring. For roughly 44 percent of them, it never does.
Part of a series on the missing professional infrastructure of healthcare. Previously: The Reference Calls the Vendor Chose For You
Evidence note: sources include CTTI research published in Contemporary Clinical Trials Communications (2017 and 2019); an analysis of 172,453 unique FDA Form 1572 investigators published in Contemporary Clinical Trials Communications (2019); Glass and Guy's analysis in ACRP Clinical Researcher (2022), which argues the one-and-done rate is substantially overstated by incomplete databases; and site selection preference data from Trials (2019). The contested nature of the headline one-and-done figure is discussed explicitly in the text, and the survey finding on unoffered opportunities does not depend on it.
