You are not the exception
If your site still captures much of its source on paper, that is the majority position rather than a lag. A 2026 Tufts study found paper still ahead of both the EHR and research eSource for every kind of information it looked at (see Sources). And it is not only sites that never went electronic: a site director recently posted that his coordinators had voted unanimously to go back to paper source, having tried the alternative, and the thread filled with others saying the same.
The reasons are ones you know better than we do. Paper keeps the coordinator’s attention on the patient rather than on a screen. And the alternative is a purchase: an eSource system has to be chosen, bought and implemented. That is a budget and priority decision, not inertia.
Nothing here argues for moving off paper. The question is what is still missing for a site that captures its source on paper. The Layer Nobody Owns makes that case for any site. This article is about the part that is specific to paper.
What is missing when the source is on paper
A coordinator’s day contains two kinds of work. One is capturing the source, and eSource, the EDC and the EHR are all good answers to it. The other is running the visit: booking a room and a certified rater for the same hour, keeping the assessments in the order the protocol requires, and finishing the work that outlasts the visit itself. None of that produces a data point. Systems hold pieces of it — a CTMS will schedule the visit and record what was done — but none of them owns the running of the visit itself, which is why coordinators keep their own visit worksheet. Asked to name their biggest barriers, sites in the Tufts study put these coordination problems ahead of capture.
Every site has that worksheet, whatever it captures into, and the companion paper makes that case at length. What it does not cover is what paper changes. The work is the same either way, before the visit, during it and after. What differs is how much of it a system can keep track of, and how much a person has to. One visit touches several systems — the clinical care goes into the EHR, any dispense is transacted in the IRT, the visit is recorded in the CTMS — and a good deal of it touches no system at all. Each system knows about its own part and nothing beyond it. A paper form knows nothing at all: it holds the observation without knowing a visit happened. So joining it up falls to a person: what the visit needs booked against, what order the assessments run in, where each result goes afterwards, and what is still open three days later.
All of that is what the coordinator’s worksheet is holding, and it has a name: the work layer. Somebody builds it by hand for every study, and it lands in a Word document in a shared folder, which is why it goes stale, why nobody knows which version ran, and why none of it adds up across a week.
The Layer Nobody Owns: what the layer is, what being without one costs, and the five tests it has to pass.
What a work layer has to do
Paper is a choice about the patient encounter, not about the whole operation. Where the layer meets the encounter is a sheet of paper: the printed checklist, alongside the forms the site already uses. Everything around it — scheduling, versioning, tracking what is outstanding — happens at a desk or on a phone between visits, which is where the layer is of most use. The checklists themselves are built the way they would be for any site, from the protocol, the manuals and the site’s own material. What changes once they are printed is seven things.
Schedule properly, whatever the site uses at the bedside. Choosing paper there says nothing about how the week should be booked. A visit competes for a room, equipment and staff certified on the right instrument, for a length of time, and scheduling against all of it is what surfaces a clash when the visit is booked rather than on the morning. That is desk work, done days ahead, and no paper diary resolves it however carefully it is kept.
Print the checklist, and use it to drive the site’s own forms. One line per step the visit takes, in order, each carrying its timing and ordering constraints, whatever the site knows about it, and what is specific to this subject today: that they are still on the old consent version, that the kit assigned to them expires before the visit date. Steps that get recorded name where they go, on the printout or in a named system; steps that don’t are just checked off. The coordinator is not working it out mid-visit.
Put a QR code on the printout, for what happens away from the patient. Scanning it opens the visit on a phone. Not to work from while someone is talking, since the printout is for that, but to release a room, mark the visit complete once they have gone, and pick up the steps it leaves behind. Those are the ones that go unfinished when nothing is holding them.
Guide the work that happens in other systems. Most of what surrounds a visit is done at a screen anyway: booking it, ordering the kit, entering it into the EDC, reconciling the dispense in the IRT. On the online checklist each of those steps carries the way into the system that owns it and whatever the site has worked out about doing it — which screen, which fields this sponsor wants, the quirk that catches people out. A coordinator who has not done that step before is not asking someone, and one who has is not remembering.
Version the printout, and record which version ran. This is the one thing paper can’t do for itself. The checklist is republished when any of the documents behind it change, and every printed copy carries its version. “Is this the current one?” stops being a question answered from memory.
Make transcription a tracked step rather than a memory. Entering the visit into the EDC sits on the list like anything else and stays outstanding until it is done. That doesn’t remove the hour; it stops the hour going missing, which is the difference between a visit that is late into the EDC and one that quietly never gets there.
Leave the source where it is. The layer holds how the visit runs and what is outstanding, not what was found. Your forms stay your source, nothing is copied out of them, and adding the layer does not pull them into a new system’s validation scope.
One question worth asking your coordinators
What would a new coordinator need on their first dosing day that isn’t in any system you own?
If the answer is a document somebody built, the one with the courier cutoff and the order the assessments have to happen in, then that is your operation, held outside your systems and maintained by hand. It is also what a departure costs you, and what a new hire spends months rebuilding.
Send us that document and we will show you what holding it properly looks like: CRCHub@triradial.com.
Sources
- Harper B, Ford RM, Krayem R, Nomizu R, Andrus J, Getz K, “Characterizing the Protocol-Guided Source Preparation Process at Investigative Sites”, Applied Clinical Trials, April 2026. Tufts CSDD with CRIO; 209 investigative sites surveyed Sept–Dec 2025. Capture location by information type: paper leads at 67% for signed consent, 54% for protocol-specific visit information, 51% for medications and for adverse events, and 48% for laboratory results, against 19–30% for research eSource and 16–39% for the EMR/EHR. Also the source of the 20–40 hours of source preparation per complex protocol, and the finding that transcription ranks below coordination problems among sites’ reported barriers.
- Brad Hightower (Hightower Clinical), “My CRCs unanimously voted to go back to paper”. The post does not name the system they left, asking only “what’re we doing wrong that paper is still preferred over an electronic system in 2026?” Worth reading for the comments, where coordinators and CRAs describe what the electronic alternatives cost them in practice.