Medical monitor queries, explained.

What a medical monitor query is, and why one short email can matter for years

One washout question, from a site's screening room to the trial master file

1. A site screens a patientParticipant 226-031 stopped an IL-17 biologic on August 4. The site hopes to start treatment on October 15.
2. The site asksDr. Lindqvist writes through the site portal. Another site told him 8 weeks is enough.
3. The medical monitor finds the rule and answersWhich version applies, which section and page, what earlier answers said, and what date follows. This is where the 12 weeks and the Boston answer are caught, or missed.
4. Anyone told otherwise is correctedSite 114 gets a correction. The participant it randomized on the old figure is raised with the site's monitor.
5. The answer is filed and sharedThe decision goes into the trial master file (TMF), the study's official record, and into a shared FAQ so the next site need not ask.

Step 3 is the subject of this guide. The people, sites, drugs and figures in this example are fictional.

Why a short reply is hard to get right: four findings that surprise newcomers

Washout, deviation, amendment, clarification letter: the words to know

Fourteen terms are defined in the one-page summary.

Eligibility, washouts, missed doses, prohibited medicines: what investigators ask a medical monitor

In shortMost site questions fall into a handful of types. Eligibility questions press hardest, because a patient's screening window is running, and they lead in the guide's fictional sample. Each type has its own place in the protocol and its own urgency. A reply is useful when it states the facts it relied on, quotes the rule with its version and page, and says what the site may do next.

The questions are concrete. They arrive with a participant number, a date and a lab value, rarely with the rule attached. A public protocol from one US research network shows how such questions get routed: "For questions concerning entry criteria", medicines taken alongside the study drug and similar matters, sites write to a clinical management committee rather than to the general team address.5

A sponsor may name one physician and a backup. ICH E6(R3), the current edition of the good clinical practice standard from the International Council for Harmonisation (ICH), adds that each site "should have an assigned monitor as their contact point", so many questions reach a clinical research associate (CRA), the sponsor's visiting monitor, first.1

Six kinds of protocol question, and where each answer lives

The questions are written for this guide, drawn from the fictional study in the example.

Six things that make a site reply safe to act on

ICH E6(R3), FDA's deviations draft, EMA's waiver answer and 21 CFR 312: what governs a site reply

In shortNo regulation we read mentions a "medical monitor query". The duty comes from good clinical practice, which asks a sponsor to keep medical advice available, and from the rules that bind investigators to the protocol, such as 21 CFR 312, FDA's regulation on investigational drugs. E6(R3) became one consolidated text in June 2026. FDA's guidance on deviations is still a December 2024 draft. EMA's answer on waivers dates from March 2022 and is the plainest statement we found.

Protocol versions after an amendment: why the newest page may not be the page that applies at this site

In shortA protocol changes while a trial runs, three or four times on average. Each change reaches each country and site at a different time, and the gap averages 215 days. So every answer has two questions inside it: what does the newest version say, and which version is this site working under? A reply that never asks the second question can be wrong in either direction.

Those figures come from a Tufts Center for the Study of Drug Development study of 950 protocols and 2,188 amendments, from 16 companies and CROs.2 Oncology runs higher: 91.1% of protocols amended and 4.0 amendments on average, against 72.1% and 3.0 elsewhere.21 The same study says 77% of amendments were "deemed unavoidable", with regulators' requests and changes in strategy the top reasons, so better drafting will not make the problem go away. Answers go stale as a normal result of how trials run.

The documents a site reply can rest on, and how far each one reaches

Washout dates, screening windows and times-ULN: working a site's eligibility question by hand

In shortMost eligibility replies come down to small arithmetic: count the days from a last dose, compare them with a washout, check the result still fits inside screening, or divide a lab value by the upper limit of normal. It is simple, and it goes wrong often enough to appear in FDA letters. Here are the three calculations from the guide's fictional study, worked through.

The washout: from August 4 to the earliest Day 1

The lab value: is 107 over the limit?

A coordinator reports ALT, a liver enzyme, at 107 units per liter. The central lab's upper limit of normal (ULN) is 41. So the value is 107 ÷ 41 = 2.6 × ULN. The sample criterion excludes anything above 2 × ULN at screening, which here means above 82 units per liter. The participant is excluded at this screening, whatever the suspected cause.

Two traps sit in this sum. The ULN belongs to one lab, can differ by age or sex, and is not interchangeable with another lab's. And values get converted: in one FDA warning letter, an investigator explains an enrollment error by saying a study coordinator "made an error in converting" lab values. FDA redacts which values and does not say how.3

Re-screening: when may the site try again?

If the protocol allows one re-screening no sooner than 4 weeks after a failed screening, and it failed on October 7, the earliest date is October 7 + 28 days = November 4. That date belongs in the reply. EMA's inspectors list "Protocol deviations related to the re-screening for eligibility of study participants" among their findings, so a re-screen rule deserves the same care as the first screen.22

Can a medical monitor approve an exception to eligibility criteria? What EMA and FDA say about protocol waivers

In shortSponsor approval alone is not enough under any of the texts here. In the EU, EMA says sponsors should not run systems of prospective waivers and should amend the protocol instead. In the US, FDA's draft accepts that a rare one-off departure can happen, with sponsor approval and prior IRB approval, and says it should be extremely rare. So the word "exception" in a site's email needs a careful reply, whichever country the site is in.

The request usually sounds reasonable. The patient is motivated, the lab value is just over, the investigator believes it is harmless. EMA's inspectors have seen sponsors build systems around such requests. Two regulators address them directly, in different terms.

EMA's waiver answer and FDA's draft, side by side

What a good reply does with the word "exception"

From a wrong eligibility answer to an important protocol deviation: who classifies it, who reports it and how often

In shortEnrolling someone against a key eligibility criterion is on FDA's own list of important deviations. The sponsor sets the trial's criteria for what is important, so not every slip qualifies. Phase 3 protocols average 119 deviations of all kinds, involving nearly a third of participants. We found no published count of how many trace back to a wrong answer. In the EU a serious breach goes to regulators within seven days.

The direction is upward: in a January 2026 trade column, the editor says Tufts found that mean deviations in the late-stage trials supporting approval "have more than doubled", and gives no figure or period.25 How much of that is eligibility is unknown.

A 2026 audit at one research department in India found 526 deviations across 21 regulatory trials, with 42% about visits outside the allowed window and 37% about sample collection, processing and storage, and eligibility barely registers in its table.26 Its setting is narrow. What makes an eligibility deviation matter is what one case can cost a trial, and no audit we found prices that.

Which eligibility mistakes count as important

When a deviation becomes a report to a regulator

One participant on a wrong washout may not reach that level alone. EMA's version example above turned on 6 of 10 patients at a site, so the same wrong answer given to 30 sites might.12 The study team decides, and it needs to know how far the answer travelled to do so.

Seven ways site answers go wrong after a protocol amendment, and why nobody sees them

In shortThe physician usually knows the medicine. Drift comes from the system around the physician: several people answering, mailboxes that cannot be searched by criterion, an old version still on someone's desktop, and no step that compares a new answer with old ones. Each failure below looks fine in the email that caused it.

What FDA warning letters and EU inspectors find on eligibility and protocol compliance

In shortInspectors look for ineligible participants and for proof of how eligibility was decided. Eligibility is a small share of counted findings. It does appear in FDA's recent warning letters, and in those letters the investigator answers for it. Sponsors and CROs are cited for failing to ensure a study follows its protocol.

What three FDA warning letters say about eligibility

None of the three letters discusses a medical monitor's advice. One trade article says about 30% of warning letters are due to failure to follow the investigational plan. It gives no year or sample, so treat it as a rough guide.35

What EU inspectors wrote under protocol compliance on selection criteria

On FDA's list of common sponsor and CRO observations: failure to ensure proper monitoring and that the study "is conducted in accordance with the protocol and/or investigational plan".30 On its clinical investigator list: "Failure to follow the investigational plan; protocol deviations".30 FDA's slide does not rank them by count.

Clarification log, clarification letter, TMF and study FAQ: where a medical monitor's answers belong

In shortAn answer given to one site is correspondence, and it belongs in the trial master file. If many sites need the same explanation, a clarification letter tells them all at once, as long as it changes no rule. If the rule has to change, only an amendment will do. A clarification log and an FAQ help daily work, and each needs an owner who retires what an amendment makes wrong.

A letter explains, and an amendment changes: telling them apart

Four duties that apply to every answer on file

What AI can do with protocol questions, what FDA and EMA expect, and where the evidence stops

In shortSoftware can read a question, find the passage, count the days and compare earlier answers. We found no independent published study of AI answering site questions about a protocol, only vendor case studies. The nearest evidence comes from screening charts against criteria and from legal research tools. FDA's AI guidance is still a draft. It leaves out operational uses only when they do not affect patient safety, drug quality or the reliability of results, and an eligibility answer can affect that reliability. EU GMP Annex 22, also a draft, covers the making of medicines.

Having a model answer from retrieved text helps, and still leaves errors. In legal research, a field with similar habits of citation, one study measured made-up content "between 17% and 33%" of the time in commercial tools built on retrieval.40 That is a different field and a dated measure, and it is the best warning we found: a reply can carry a citation and still be wrong. Whichever tool you use, the test is whether it can cite the wrong version with confidence.

FDA's AI draft, EMA's reflection paper and the EU AI Act: what each one reaches

What software can take on, and what stays with the physician

Who answers a site's protocol question today, and where the two days go

In shortA question passes through a coordinator, an investigator, often a CRA, and a medical monitor or an outside medical desk. Then a clinical scientist and a study manager deal with what follows. On the one estimate we have, most of the elapsed time is waiting in a mailbox and the active work takes a few hours. Nothing in that process checks a new answer against an old one.

Investigator, CRA, medical monitor, CRO desk: who touches one site question

In house, through a CRO, on a portal or with drafting software: four ways to run it

No public figure gives the market size of medical monitoring services or query software from an independent dataset. Vendor brochures offer numbers without a method, so we leave them out.

From Dr. Lindqvist's washout question to a signed reply and a correction, in one hour

In shortThis is the example from the top of the page, inside the product. The question arrives at 8:46 am with a reply already drafted. A banner shows that Site 114 was told something different. Dr. Maya Chen, the lead medical monitor, drafts a correction, flags the participant to the site's monitor, and signs once. By 9:45 am the reply and the correction are out, and the decision is filed.

1

The inbox opens on the question, with the conflict already found

Maya's inbox holds seven open questions across two studies, each with its site, its criterion and its clock. Those that differ from an earlier answer carry a badge and have their own tab. MI-0422 from Site 226 has 23 hours and 6 minutes left on a 24-hour eligibility clock.

The site's message is highlighted: the participant, the last dose, the planned date, and in red the line that matters, "A colleague at another site understood that 8 weeks is enough". Below it sits the banner: "Site 114 was told 8 weeks on Sep 15. Protocol v3.0 says 12 weeks." Both answers appear side by side, each with the version it rests on. On the right are the passages the draft quotes, with the replaced text marked in red.

2

A reply that opens with the protocol text

The draft took 24 seconds on this sample question. It opens with the rule: VLT-302 Protocol v3.0, section 5.2, exclusion criterion 7, page 46, Amendment 2. "Day 1 cannot be on Oct 15." It works out October 27, says that October 27 or 28 stays inside the 28-day screening period so no re-screening is needed, and says the 8-week figure comes from version 2.0 and no longer applies.

Every sentence that states a rule carries a numbered citation. The reply never names the other site. Part A worked the same dates by hand, and they match.

3

One click drafts the correction to Site 114

Maya chooses "Draft a correction to Site 114". It is addressed to Dr. James Whitaker, copied to the site's CRA and the study manager, Sam Patel. It says where the 8-week figure came from, gives the 12-week rule, and asks the site to record participant 114-019 with its CRA as a possible eligibility deviation. "The study team will contact you about continued treatment."

It goes out with the reply, once she signs. Nothing has left the building yet.

4

The participant goes to the people who assess it

Participant 114-019 was randomized on September 18, 8 weeks and 3 days after the last dose. The screen says "possible", because whether this is an important deviation is for the study team to decide against the trial's own criteria, as E6(R3) section 3.9.3 expects.1 Maya flags the participant to the Site 114 CRA and Sam Patel. Whether it needs reporting further, and what happens to the participant, stays with those people and the investigator.

The banner turns green and shows what has been done.

5

One signature, with its meaning

The signing screen states the decision, "Not on Oct 15, earliest Day 1 is Oct 27", with the question, the study, the site, the criterion and the protocol version it rests on. The signer is locked to Dr. Maya Chen. She chooses a meaning, "Approved as medical monitor response", and enters her password.

Three boxes show what goes with it. The correction and the filing in the TMF are ticked and fixed. Publishing to the study FAQ, without participant numbers, is hers to choose. Only a physician with the medical monitor role can sign a medical decision: that is the solution's rule, stricter than E6(R3)'s "medical personnel".1

6

Sent, corrected, filed and published

At 9:45 am the signed reply goes to Site 226, the correction to Site 114, and the decision into the TMF with its signer, time, meaning and protocol version. The FAQ entry appears without participant numbers. The question moves off the open list, 59 minutes after it arrived.

That hour is what this sample shows. The solution's design target is a median of about 3 hours across all questions, to be measured in your pilot.

Select any screen to open it larger. The sample treats version 3.0 as approved at both sites. Part A explains why a real answer has to check that for each site, and Part C asks every vendor about it, us included. The investigator still decides for the participant: the reply says what the protocol allows.

A hepatitis C question drafted in 19 seconds, and a request for an eligibility exception answered with the re-screening route

In shortTwo more questions show the rest of the range. One arrives while you watch and needs no correction, because nobody has answered it before. The other asks for an exception on a liver test. The full review breaks it down criterion by criterion, says plainly that the protocol allows no exception, and offers the date the site can try again.

1

A new question, each step in the open

Dr. Henrik Lund at Site 207 Aarhus asks about participant 207-019: hepatitis C antibody positive, RNA undetectable, curative treatment finished in March 2023. The screen shows who is working on it, one step at a time. The Inquiry Reader finds four facts. The Topic Classifier files it under E-12 and starts the 24-hour clock. The Protocol Navigator finds section 5.2, criterion 12(c).

The last line never completes on its own: "Waiting for Dr. Maya Chen to sign."

2

No earlier answer, so this one becomes the reference

Nineteen seconds later the reply is ready. It quotes criterion 12(c), page 48, and says the criterion "does not exclude participant 207-019", and that "All other criteria still apply." The check finds nothing: "No earlier answer on E-12 — once signed, this one becomes the reference."

Maya can edit, shorten, or open the full review. The clock keeps running until she signs.

3

The full review, decision by criterion

Site 114's coordinator writes about participant 114-022: ALT 107 units per liter, 2.6 times the upper limit, and two hepatitis B results: anti-HBc positive, HBV DNA not detected. Then: "Given the patient's strong motivation, we would appreciate an exception to enroll him."

The review takes it apart. Liver tests, E-11: excludes. Hepatitis B, E-12: allowed with conditions, with HBV DNA tested every 12 weeks. Re-screening, 8.1.3: allowed once, no sooner than 4 weeks after this failure, so from November 4. A band notes that an exception was requested and that the protocol allows none, "so the reply says so plainly and offers the re-screening route".

4

How the draft was built, step by step

"How it was built" replays the work with times: 7 facts from the portal form and 2 attachments, 6 passages from version 3.0, 3 prior answers with 1 out of date, 7 sentences with every rule cited, and an exception request found. This answers the question a reviewer asks first: where did this come from?

The Activity tab lists the same steps with their times. Every agent step, signature and settings change goes on the audit trail.

5

An activity trail for every question

Who did what, in order, from the moment the site portal took the question. An inspector who asks how an answer was reached gets a timed list of steps. The site's message is kept beside it, with its attachments.

All questions, protocol hot spots, the study FAQ, the decision log, the dashboard and your settings

In shortAround one question sit the screens a medical monitoring team uses all study long: every question with its clock, a map of where the protocol confuses sites, an FAQ sites can search, a log an inspector can read, a dashboard and the settings that hold your own rules.

Seven site-question agents and the physician who decides and signs

In shortSeven specialist agents read, classify, find, compare, draft, check and file. None of them can send a reply. Only a physician with the medical monitor role signs a medical decision, and nothing reaches a site before that signature. The site's investigator keeps the decision about the participant.

What each site-question agent does, and where it stops

Who signs a reply, who confirms an out-of-date answer and who owns a deviation

From a site's question and your protocol to a signed reply and a filed record

Site portal, mailbox, eTMF: the systems a query desk already runs on

The solution works with the channels and documents a study already uses: your site portal and medical monitor mailbox for questions, your document system for protocols and letters, your central laboratory's reference ranges for reading a value as × ULN, and your electronic trial master file (eTMF) for filing.

It is built for enterprise use: roles and maker-checker separation between drafting and signing, a full audit trail with export, and integrations with the systems you name. Botminds is SOC 2 Type II audited, and its information security is managed to ISO 27001. GxP and Part 11 are duties on the sponsor, so no product can be certified against them. These are controls that support your compliance.

Design targets for reply time, consistency and eligibility deviations

These are goals to test on your own studies, and no customer result stands behind them. One cancer center's second eligibility review came with a fall from 7.37% to 2.56% of accruals, covering consent and eligibility together, through a different method.24

Five people in the two sample studies, and what each needs from a site reply

What late replies, conflicting answers and eligibility deviations cost a sponsor

In shortStart from your own records. Published figures frame the scale: what a day of a running trial costs, how long sites work from mixed versions and how many deviations a Phase 3 protocol carries. We found no published figure for what one wrong answer or one late reply costs. You can find most of your numbers in a week, from your mailbox, your deviation log and your screening data.

Seven query costs you can count in a week, from physician hours to inspection preparation

A patient kept waiting, a wary site, a physician's lost hours: what the totals leave out

What faster, consistent site answers return: physician hours, fewer eligibility deviations and fewer re-screens

In shortPut in your own numbers. The calculator values physician time released, eligibility deviations avoided and re-screens avoided, against what the tool costs you. Almost every default is illustrative, because public figures do not exist for this work. The hourly cost is an illustrative loaded cost, meaning pay plus overhead: replace it with your own.

From physician time to inspection records: six places query automation saves

Site questions a year, deviations and re-screens on your own figures

How ready your medical monitoring team is: log, versions, cover, waiver stance and sign-off

In shortPick the answer closest to your team today. Eight questions score 0 to 2, and the last is for reflection. The meter adds up your score as you go, and each question says why it matters and what a strong answer sounds like.

Twelve questions that test a medical monitor query tool, ours included

In shortAlmost every tool demonstrates well on a clean question. These questions test what a demonstration hides: whether it knows which version applies at a site, whether it finds a conflict you planted, whether a reply can read as a waiver, and who can sign what. Ask us too.

Testing query software in ten weeks: past answers, planted conflicts, one live study in parallel

In shortStart with questions you already answered, because you know how they turned out. Read every answer on your three most-asked criteria, plant conflicts to measure the check, then run one live study in parallel with your usual process. Decide on three numbers: reply time, conflicts caught and how much of each draft the physician keeps.

Reply time, planted conflicts and the share of each draft a physician keeps

Medical monitor questions answered: queries, eligibility, waivers, washouts, clarification letters and AI

In shortThe questions medical monitors, investigators and buyers search for, each answered on its own.