Contributed Commentary by Maria Perry Ladd, Co-Founder, Clinical Research Site Collective 

September 16, 2026 | Clinical research has never been only about generating evidence that a therapy is safe or effective. It is equally about generating confidence in the evidence itself. Every observation, every assessment, every protocol deviation, every adverse event, every laboratory result, and every conclusion must be supported by documentation that can withstand scrutiny. We work in a profession built on verification. “Trust but verify” is woven into our quality systems. “If it wasn't documented, it wasn't done” is probably the most recognized sentence in our field.

Documentation is not an administrative requirement. It is one of the primary ways our industry demonstrates integrity. It allows another person—whether a monitor, auditor, inspector, sponsor, investigator, or regulator—to reconstruct what occurred and reach the same conclusion. That expectation has shaped nearly every operational process we use.

There is an interesting consequence of working in an industry where processes are so carefully designed, regulated, and validated.

Every mature profession develops practices so embedded in daily work that they stop being viewed as decisions. They become assumptions. The people who designed them retire. New professionals inherit them through training. Eventually entire generations know the workflow without knowing why the workflow exists in the form it does.

That is not unique to clinical research. It is simply what mature industries do.

The difficulty is that environments change. Technology evolves, scientific understanding advances, patient expectations shift, operational complexity increases. Yet some of the workflows supporting those changes remain remarkably familiar. Perhaps they should. Or perhaps, from time to time, they deserve the same thoughtful reassessment we routinely apply to protocols, endpoints, statistical methods, and quality systems. That is not an argument for change. It is an argument for curiosity.

One of the great strengths of scientific inquiry is the willingness to revisit assumptions when new information becomes available. Clinical research has done this repeatedly in science, medicine, and regulation. Operationally we have been far more cautious, and with good reason. Patient safety, data integrity, regulatory compliance, inspection readiness, and business risk all reward consistency. Operational change carries uncertainty, and uncertainty is often interpreted as risk.

The result is an industry that has become exceptionally good at managing operational risk. Which raises a question worth sitting with: at what point does appropriate risk management become resistance to operational evolution?

That is not a criticism. The caution characterizing clinical research has protected patients, safeguarded scientific integrity, and contributed to the credibility of medicines that have improved millions of lives. But systems that deserve our respect also deserve our willingness to periodically examine whether they remain fit for purpose.

This particular question has been sitting in the back of my mind for years. Recent work surrounding Real-Time Clinical Trials gave these thoughts a solidity in my mind; the technology exposes something that is so similar as to not be ignored.

The most interesting questions in mature industries are not about the technologies we build. They are about the assumptions those technologies quietly reveal.

How We Got Here

It is easy to look at current technology and wonder why some of our workflows seem unnecessarily complex. It is harder, and considerably more important, to understand why they exist.

A recurring mistake in conversations about innovation is evaluating yesterday's decisions using today's capabilities. That almost always produces the wrong conclusion. The separation between source documentation and protocol data capture was not created because the industry lacked imagination, and it was not created merely because paper existed. It evolved because it solved real operational, scientific, and regulatory problems at a time when those problems demanded exactly that solution.

Investigators needed a reliable way to document the care and observations made during a participant's visit. Sponsors needed standardized information that could be aggregated across hundreds, and eventually thousands, of participants enrolled around the world. Regulators needed confidence that submitted data accurately reflected what occurred during trial conduct. Inspectors needed to reconstruct events months or years later.

Each stakeholder had a different need, and the workflow that emerged served them all. Source documentation captured what happened. Case Report Forms captured what needed to be analyzed. Those purposes overlapped but were never identical, and that distinction mattered. It still does.

Think about how many decisions have been built on top of that distinction since. Technology platforms, site workflows, training programs, monitoring strategies, quality systems, staffing models, vendor solutions. Even conversations about artificial intelligence tend to begin by assuming information will continue moving along the same pathway. Very little attention goes to whether the pathway itself deserves reconsideration.

Why? Because it has worked. But history should inform operational decisions rather than automatically determine them.

Source Is a Principle, Not a Place

When we discuss source documentation, we slip easily into language that treats source as a location: The source is in the chart. The source is in the EMR. The source is in the worksheet. The source is in the eSource system.

Operationally, we all understand what those statements mean. Conceptually, they may be doing us a disservice.

Imagine explaining source documentation to someone entirely unfamiliar with clinical research. Would you begin with binders, or with systems? Would you describe monitors comparing one document against another?

Probably not. You would describe something much simpler. Clinical research requires a reliable record of what occurred during a participant's involvement in a study. That record must accurately reflect observations, decisions, procedures, assessments, and clinical judgment. It must identify who documented the information, when it was documented, and what changed afterward. It must allow another qualified individual to reconstruct the events of the study.

Only after describing those principles would you get to where the information lives.

For decades our conversations about source have centered on the record itself. Paper source became electronic source. Worksheets became eSource. Files became systems. Somewhere along the way, we allowed the medium to become the discussion rather than the purpose.

The purpose was never the paper, the binder, the worksheet, or the application. The purpose was a contemporaneous, attributable, accurate, complete, and reconstructable account of what occurred. The purpose was confidence. Not confidence because information lived in a particular place. Confidence because another qualified person could examine the record and understand exactly what happened, when it happened, who performed it, what changed, and why.

Those principles are more important now than they were decades ago, not less.

Which is where many of these conversations go off track. They quickly become a debate about whether we are proposing to eliminate source documentation, or whether today's Electronic Data Capture system should become the medical record. Those questions are easier to argue because the positions are familiar. The harder and more useful question is whether we have confused the principles of source documentation with the workflow we currently use to achieve them.

One response I hear nearly every time this subject comes up is that sponsors cannot create source documentation, because Good Clinical Practice places responsibility for source records with the investigator and the site. That is precisely why sponsors have historically not developed source documentation for sites. It is a sound regulatory principle.

But notice what that statement does and does not do. It explains why the historical workflow developed. It does not establish that every implementation of that workflow must remain unchanged. Those are different questions. The first explains the past. The second explores the future.

Clinical research has repeatedly shown that foundational principles can hold constant while implementation changes dramatically. Electronic signatures did not eliminate accountability. Risk-based monitoring did not eliminate oversight. Electronic Trial Master Files did not eliminate essential documentation.

Each of those asked a version of the same question. Can we preserve the principle while improving the process?

Duplication Is Not Automatically Quality

Consider a routine study visit. A coordinator measures protocol-required vital signs, reviews concomitant medications, documents protocol-required assessments, discusses adverse events, and answers participant questions. She coordinates investigator review and communicates with ancillary departments. She documents the encounter, transcribing this study-generated data from either paper or an electronic source. She enters the data, responds to queries, and prepares for monitoring.

No experienced research professional would describe that as a sequence of isolated tasks. It is a continuous clinical interaction requiring judgment, prioritization, communication, and problem-solving.

Now remove this simple act of transcription. Not the review, not the oversight, not the investigator's responsibility. Only the transcription and the time it took. What becomes possible?

Maybe nothing meaningful. Or maybe five additional minutes with a participant. Time to notice an emerging concern. Time to reinforce study compliance. Time to answer the question that otherwise becomes an email three days later. Time for an investigator to have a real conversation instead of reviewing duplicate documentation.

None of those is easily measurable. Neither is the opportunity cost of never having them. We are good at measuring the efficiency gained by removing a task; we rarely measure what the reclaimed minutes make possible. This is where I expect the most resistance, and I understand it. One assumption I’ve encountered is that duplicate documentation inherently improves quality. Sometimes it does. Independent verification has enormous value, and replicate trials exist precisely because independent confirmation matters. Redundant safety systems save lives in aviation, nuclear power, and medicine. Clinical research should never remove a safeguard simply because technology makes removal possible.

But duplication and verification are not synonyms. Entering the same information twice is not automatically validation. Sometimes it is transcription, and transcription is itself a recognized source of error. A workflow that introduces a known error mode in the name of quality deserves at least to be examined on those terms.

There is an asymmetry here that we rarely name. We require evidence to add a procedure to a protocol. Every blood draw, every scan, every questionnaire has to justify its scientific contribution before it earns a place in the schedule of assessments. Operational steps face no equivalent test. They are inherited rather than justified. The burden of proof runs in one direction only, and it runs away from the things we do most often.

If we are willing to remove an unnecessary blood draw because it no longer contributes meaningful scientific value, it seems fair to ask whether an unnecessary operational step continues to contribute meaningful quality.

What Happens to Verification

This is the part of the argument most people arrive at on their own, usually about a minute after the rest of it, and I would rather name it than leave it sitting unspoken.

Source data verification exists because two records exist. A monitor compares the case report form against the source to confirm the information was carried across correctly. It is a control designed around one specific failure mode: transcription. If, for a given data element, there is one record instead of two, that failure mode is gone, and there is nothing left to compare.

That is a real consequence and I don't want to minimize it.

But notice what it does not touch. Consent. Eligibility. Safety reporting. Investigator oversight. Protocol compliance. Data plausibility. Site qualification. Everything monitoring exists to protect remains exactly where it was. What changes is one activity within monitoring, and it happens to be the activity we have already spent a decade steadily reducing.

It is also worth acknowledging plainly that verification is not only a quality practice. It is an operational and commercial structure, with budgets, headcount, and business models attached to it. That does not make it wrong. It does mean some resistance to reconsidering it will be about something other than science, and we will have a far more productive conversation if we are able to say so out loud.

Real-Time Clinical Trials as Catalyst, Not Destination

RTCT has generated interest because it challenges our assumptions about when meaningful information becomes available. Rather than waiting for information to move slowly through traditional operational pathways, it argues for earlier visibility, earlier decision-making, and earlier intervention. Those are worthwhile goals on their own terms.

But its most valuable contribution may not be technological at all. It may simply be that it forces questions we have not asked often enough.

If we want information available closer to the moment it is generated, shouldn't we also examine the number of operational steps standing between generation and availability? If the goal is to reduce delay in decision-making, shouldn't we examine the workflows producing the delay? If technology increasingly allows protocol-generated information to be captured accurately, contemporaneously, and transparently, should we keep assuming that multiple documentation steps always represent the highest-quality approach?

I am not suggesting that today's EDC becomes tomorrow's source documentation. I suspect that would be the wrong conclusion entirely. Today's EDC systems were designed around today's workflows. If the workflows change, the platform that serves them may not resemble EDC at all. It may be something built from the beginning around modern clinical operations rather than around historical technical constraints.

If that happens, the interesting question stops being whether EDC becomes source. It becomes whether the distinction between source documentation and protocol data capture remains operationally necessary for every data element.

Every data element is the part that matters. Not every piece of information generated during a clinical trial belongs in a data capture platform. Hospital records, consultation notes, radiology and pathology reports, external laboratory results, emergency department documentation, and specialist correspondence all have clinical purpose well beyond the trial, and they should continue to live where they belong. Clinical judgment should never be constrained by software design. An investigator's assessment is more than structured data, and a participant's clinical story is more than a set of fields waiting to be completed.

Acknowledging that strengthens the argument rather than weakening it, because it demonstrates this is not an all-or-nothing proposition. The question is not whether source documentation disappears but which data elements genuinely require duplicate operational workflows and which no longer do.

We have become remarkably sophisticated at applying risk-based thinking to monitoring, quality management, and oversight. Are prepared to apply the same thinking to workflow design itself?

The Hardest Part Isn't Technology

If these ideas have merit, technology will not be the limiting factor. Technology will continue to evolve, as it always does. The harder problem is human, and it is worth being specific about why.

Every operational model creates familiarity. Familiarity creates confidence. Confidence creates standardization. Eventually an entire professional community becomes highly skilled at executing a particular workflow, and that skill is not a small thing to have built. Questioning a workflow can feel like questioning the professionalism of the people who spent careers mastering it. That reaction is not irrational. It is predictable, and any conversation that ignores it will fail regardless of its merits.

There is also the shadow of the devil we know. A familiar inefficiency is a known quantity. Its costs are absorbed, distributed, and largely invisible. An unfamiliar model, even a better one, carries uncertainty that is concentrated and highly visible. Organizations are structurally inclined to prefer the first, and the individuals inside them absorb that inclination whether they intend to or not.

It would be easy to stop there, except that this industry has already done the harder version of this at least twice.

There was a time when paper Trial Master Files were simply how research was conducted. Every organization had its own filing conventions, its own naming structures, its own methods for organizing essential documents. Those differences were accepted as part of doing business. Few people seriously asked whether an entire industry could align around a common framework, because the question sounded naive. Then the TMF Reference Model emerged.

Looking back it reads as an obvious progression. It wasn't. It required organizations that compete with one another to agree that consistency across the industry created more value than preserving individual preference. It required technology providers to adapt. It required sponsors, CROs, and regulators to get comfortable with a different way of thinking about documentation. Most of all, it required people to question assumptions they had held for entire careers. The achievement was not technological but cultural, and it changed how studies are run.

Risk-based monitoring required something similar, in the opposite direction. TMF asked competing organizations to adopt a shared standard. Risk-based monitoring asked the industry to release a practice it had long treated as essential. One was addition and one was subtraction, but both depended on the same willingness to examine an inherited assumption and conclude it was no longer the best available answer.

These are precedents worth holding onto, and they’re recent. Clinical research does change when there is sufficient reason to believe the change advances the industry without compromising its principles. It has done it with documentation specifically, within the working memory of most people reading this.

Identifying the Right Stakeholders

Which brings me to something I think is underappreciated in these conversations.

The people best positioned to question an operational model are usually the people who understand it most deeply, and in this case the view is not evenly distributed. Sponsors see the data that arrives. Patients experience the visit. Only the site holds both records, and so only the site can see the distance between them.

That is a structural observation rather than a complaint. It also means the party with the clearest view of where duplication actually lives is generally the party with the least authority to change it. Sites know which fields are transcribed rather than generated. They know which documentation exists mostly for its own sake. They know which steps consume attention that could have gone to a participant, because they are the ones who watched it happen.

Sponsors understand where operational complexity affects execution across a program. CROs understand where processes create friction across dozens of studies. Each perspective is incomplete on its own. But a reassessment of these workflows that does not begin where the duplication is visible is not really a reassessment. It is a redesign conducted at a distance from the thing being redesigned.

Where Do We Go From Here?

Perhaps the next step is not a technology roadmap. Perhaps it is simply a conversation about asking better questions to assess and reassess what we do. Which operational steps continue to create measurable value? Which exist because they always have? Where does duplication genuinely strengthen quality, and where does it simply consume attention that could be directed elsewhere? At what point does appropriate risk management become resistance to operational evolution?

These do not have simple answers. The purpose of asking is not to reach immediate consensus. It is to create room for exploration.

Clinical research has repeatedly demonstrated that it can evolve without compromising the principles that define it. Paper became electronic. Monitoring evolved. Documentation frameworks matured. Global collaboration became possible in ways few would have imagined a generation ago. Each of those began the same way, with someone asking whether the existing approach was still the best approach.

“If it wasn't documented, it wasn't done.” That sentence has followed most of us through our entire careers. It never said, “documented twice.”

If we were designing clinical research today, knowing everything we now know about technology, would we intentionally separate source documentation from the system built to capture protocol-generated study data?

Maria Ladd has been in the clinical research industry for 20+ years. After years in the CRO landscape, she turned to Site Advocacy and seeking ways we can improve in daily work. Maria is co-founder of the Clinical Research Site Collective, speaker and status quo questioner. She can be reached at maria@clinicalresearchsitecollective.com.