The behind-the-curtain look at how a clinical research site actually runs. Written for people looking to enter the clinical trial world, for new coordinators learning the ropes, and open to anyone curious what happens at clinical research sites.

In this file:

·        Who the CRA is and who actually employs them

·        Source data verification, and why the paper beats the database

·        What a monitor looks at first, and what that choice tells you

·        The follow-up letter, and why it is not a punishment

Mira arrived at eight, with a laptop, a badge on a lanyard and a printed list. She had initiated this site herself, months earlier, in a conference room with cold coffee and a projector that would not focus. Nani had spent two days pulling charts. Marcus, the research assistant who had joined in the spring with a biology degree and no idea what a monitoring visit was, had asked her that morning whether they were in trouble.

Mira is a clinical research associate, a CRA, and the answer to who she works for surprises almost everyone new.

She does not work for the site. She also does not, in most studies, work for the sponsor. She works for the contract research organization the sponsor hired to run monitoring on its behalf. The sponsor carries the legal obligation to monitor the conduct and progress of the investigation, and it is allowed to transfer that obligation to a CRO in writing. Mira is how a sponsor discharges a duty it cannot discharge from a desk two time zones away.

That matters for how you read her. A monitor is not an auditor and not an inspector, and the three get confused constantly.

·        A monitor visits during the study, repeatedly, to catch problems while they can still be fixed.

·        An auditor comes from quality assurance, independently of the study team, to check whether the systems themselves are working.

·        An inspector is a regulator. They arrive after the fact, and by then nothing can be fixed, only explained.

So when Marcus asked whether they were in trouble, Nani gave him the version that took her a year to arrive at herself. Mira is not here to find out whether we are good people. She is here to confirm that what we wrote down matches what actually happened, because in six months nobody in this building will remember, and the record will be all that is left.

The core activity of the visit has an unglamorous name. Source data verification, usually shortened to SDV, means comparing what sits in the sponsor's database against the original records here at the site.

Mira opens Walter's chart, the lab report, the ECG tracing, the signed consent form and the diary he filled in at home. Then she opens the same data points in the electronic case report form, the eCRF, which is the sponsor's structured record of every patient in the study. Then she reads them against each other, line by line, looking for the two to agree.

What surprises people is that she does not check everything. For most of the history of this industry she would have, and 100 percent SDV was the default. That is no longer the expectation. FDA's guidance on a risk-based approach to monitoring pushed sponsors toward targeting monitoring at the data and the processes that carry the most risk to patient safety and data integrity, rather than verifying every field on every patient with equal effort. In practice that means Mira verifies all of the primary endpoint data, all of the safety data, and consent for every single patient, then samples the rest according to a monitoring plan written before she ever arrived.

Consent is the exception that stays at 100 percent almost everywhere, and the reason is in the next section.

Mira checked consent first. She checks consent first at every site she visits, and the order is not arbitrary.

She confirmed the version of the consent form matched the version the IRB had approved at the time Walter signed. She confirmed the date on his signature came before the date of the first study procedure, including the screening labs. She confirmed the form was signed by Walter himself, and separately signed and dated by the person who conducted the conversation. Then she moved on.

Consent goes first because it is the only error on the list that cannot be repaired afterwards. A wrong blood pressure can be corrected against source. A missing lab can be retrieved. A patient who underwent a study procedure before consenting to it underwent a procedure they had not agreed to, and no amount of documentation fixes what already happened to a person.

After consent comes eligibility: did this patient actually meet the criteria the protocol set, and can the site show it. Then safety reporting: were events captured, assessed and reported on time. Then, last, the endpoint data the whole study exists to produce.

Marcus noticed the order before Nani pointed it out, which is when she decided he was going to be good at this. Rights first. Safety second. Data third. That is not a monitoring convention. It is the priority order of the entire enterprise, and a monitoring visit is just the one day a quarter when you can watch it operate.

Note to File: What makes a monitoring visit go badly is almost never the errors. It is not being able to find things. Mira has a fixed number of hours on site and a list she has to get through, and every minute she spends hunting for a lab report is a minute she does not spend on the data that matters. Prepare so the record can be walked, not so it looks clean. Charts in visit order, source clearly labelled, the regulatory binder current, and a written answer ready for anything you already know is imperfect. Telling a monitor about a problem before she finds it costs you nothing and buys you an enormous amount.

When Mira leaves, two documents get written, and the site only ever sees one of them.

The monitoring visit report goes to the sponsor. It records what she reviewed, what she found, and what she escalated. It is an internal sponsor document, and asking to see it is not how any of this works.

The follow-up letter comes to the site. It lists action items, usually with owners and due dates: a missing training record, three source pages needing a correction, a consent form where the person conducting the conversation signed but forgot to date.

New sites and new coordinators read that letter as a report card and take it personally. It is not one. Findings are the product of the visit, not evidence the visit went badly. A follow-up letter with no findings at all is unusual, and experienced staff find it slightly worrying rather than reassuring, because a visit that found nothing usually means nothing was looked at hard.

What actually matters is closure. Every item answered, in writing, with the correction documented and dated, before the next visit. An open action item from two visits ago is a real finding. A dozen fresh ones from yesterday are just Tuesday.

A monitoring visit is the sponsor checking, through a CRA who usually works for a CRO, that what the site recorded matches what actually happened. The main activity is source data verification, comparing the original site records against the sponsor's database, and under a risk-based approach it is targeted rather than exhaustive, concentrating on safety, consent and the primary endpoint. Consent is checked first because it is the only error that cannot be fixed after the fact, and the order of everything after it, eligibility then safety then data, is the priority order of clinical research itself. The visit ends with a follow-up letter of action items, which is the normal product of a normal visit, not a punishment. Mira closed her laptop at four, told Marcus he had asked better questions than most people do at their third visit, and booked the next one.

·        Previous issue: Issue 7, how causality is decided

·        Next issue: Issue 9, source documents

·        The whole file: roottorx.com/t/open-site-file

·        Root to Rx: the newsletter and The Root Room, the patient-facing side of the same story, at roottorx.com

·        21 CFR 312.50 — General responsibilities of sponsors (eCFR)

·        21 CFR 312.52 — Transfer of obligations to a contract research organization (eCFR)

·        21 CFR 312.56 — Review of ongoing investigations (eCFR)

·        FDA — Oversight of Clinical Investigations: A Risk-Based Approach to Monitoring

·        FDA — E6(R3) Good Clinical Practice, Guidance for Industry

•  •  •

The Open Site File is general education about how clinical research sites work. It is not regulatory, legal, or medical advice, and it is not tied to any specific study, sponsor, or employer. Study OLS-114 and all names are invented. Always follow your own site's SOPs, your protocol, and your PI and QA team. Characters are composites.

Filed by Justin Yamashita, MSc., Open Label Media

Next in the file: Issue 9, the documents Mira was actually reading, and why paper still outranks the database.