Health Authority Questions Across Markets: Tracking Queries and Responses in One Place

Why Health Authority Questions Need Their Own Record

Most regulatory teams already track submissions well. They know which sequence went to which agency on which date. What they track badly is the conversation that happens between submissions: the information requests, lists of questions, clarification requests, and inquiries that arrive during review and after approval. These questions usually live in email threads, in a shared folder of letters, or in a spreadsheet that one person updates when they remember. The answers live in the eCTD, but the link between a specific question and the specific paragraph that answered it is often lost once the response sequence is published.

That gap matters more than it used to, for three reasons.

The Same Question, Asked Five Ways

Agencies review the same science. When the dossier has a weak point, more than one reviewer tends to find it. FDA may ask for a justification of a dissolution acceptance criterion through an information request. EMA may raise the same issue as a major objection in its Day 120 list of questions. Health Canada may send a clarification request with a 15-day response window. PMDA may send an inquiry that frames the question around Japanese patients. Each question is worded differently, so a team that tracks questions by letter rather than by topic will treat them as four separate problems and write four separate answers.

Scale makes this worse. An independent assessment commissioned by FDA looked at product quality information requests alone (chemistry, manufacturing, and controls questions) for 40 applications reviewed between October 2023 and January 2025. Those 40 applications drew 410 information request letters containing 2,830 individual requested items, and applicants answered with 588 amendments containing 3,332 amendment items.¹ That works out to roughly 71 product quality questions per application, from one agency, before counting clinical, nonclinical, labeling, or statistical questions.

Where Answers Drift

Drift rarely comes from anyone trying to tell agencies different stories. It comes from timing and from people. The EU response is written in month seven by the CMC lead. The Canadian response is written in month nine by a contractor, after a new stability time point has come in. The Japanese response is written in month eleven by the local affiliate, working from a translated summary rather than the original data. Each answer is defensible on its own. Read together, they describe three slightly different control strategies.

Some differences between markets are legitimate. A joint analysis by EMA and FDA staff of 107 applications assessed by both agencies found that where outcomes diverged, the reasons were mainly different conclusions drawn from the same data, or different clinical data submitted to each agency. Part of the data difference came from timing: when an application went to FDA first, the later EMA submission often carried more mature data.³ The point is not that every answer must be identical. The point is that every difference should be known, explained, and chosen on purpose.

Agencies Now Read Each Other’s Files

The third reason is visibility. Answers given to one agency are increasingly seen by others, and by the public.

• Reliance pathways ask for the questions and answers. Swissmedic’s Article 13 procedure is built on a product’s prior approval by a reference authority. Its guidance says that an eCTD reference dossier must contain all sequences submitted to the reference authority, and that a paper or NeeS reference dossier must include the answers to questions. It also requires a tracking table of the reference-country lifecycle in Module 1.7.6.⁴ The answers you gave FDA or EMA are part of what Swissmedic reads.
• Collaborative reviews share the workload. FDA’s Project Orbis is a framework for concurrent submission and review of oncology products with partners that include Health Canada, PMDA, Swissmedic, the MHRA, and others.⁵ WHO’s good reliance practices note that agencies in a joint assessment can combine their lists of questions, and that industry can benefit from a single set of questions.⁶
• FDA now publishes complete response letters. On July 10, 2025, FDA published more than 200 complete response letters for applications submitted between 2020 and 2024, on openFDA. In announcing it, FDA said sponsors often misrepresent the reasons for its decisions, and cited its own 2015 analysis finding that sponsors avoided mentioning 85% of FDA’s safety and efficacy concerns when announcing non-approvals.⁷

Put together, these changes mean the full set of questions a sponsor received, and the answers it gave, is more and more likely to be read side by side by someone outside the company. A query log is how the company reads them side by side first.

How Each Agency Asks, and How the Clock Behaves

A query log has to handle four different question formats and four different clock rules. Getting these right matters, because the timing rules decide what the internal deadline should be, and the format rules decide how a letter should be broken into individual questions.

FDA: Information Requests on a Running Clock

During an NDA or BLA review, FDA asks most of its questions through information requests (IRs). FDA’s guidance on IR and discipline review (DR) letters describes an IR as a letter sent during review “to request further information or clarification that is needed or would be helpful to allow completion of the discipline review.” It is explicit that IRs and DR letters are not action letters and therefore do not stop the user fee review clock.⁸ That is the single most important difference from the EU. FDA’s goal date keeps running while the sponsor prepares an answer. A slow response does not pause the review; it shortens the time FDA has to read the answer.

The same guidance says FDA normally expects applicants to respond to IRs as quickly as possible, and that a response of a significant nature could count as a major amendment.⁸ Under the current regulation, a major amendment to an NDA submitted within three months of the end of the review cycle extends that cycle by three months, and the cycle can be extended this way only once. FDA may also defer review of an amendment to the next cycle.⁹ The PDUFA VII goals letter, which covers fiscal years 2023 through 2027, states that a major amendment submitted at any time during the review cycle may extend the goal date by three months.¹⁰ So the content of an IR response affects the review timeline, which is one reason a query log should record whether a response contains new data.

The PDUFA VII goals letter also committed FDA to a format for CMC information requests called Four-Part Harmony: what was provided, what the issue or deficiency is, what is needed, and why it is needed.¹⁰ The independent assessment of that commitment found that use of the four elements rose between the baseline and current samples, though the “why it is needed” element was still often missing.¹ For a query log, Four-Part Harmony is useful because it already splits each question into the parts a tracker needs: the dossier location FDA looked at, the gap it found, and the specific deliverable it wants.

When a review cycle ends without approval, FDA issues a complete response letter (CRL). Under 21 CFR 314.110, the letter describes all the specific deficiencies FDA has identified, and the applicant must resubmit, withdraw, or request a hearing. A Class 1 resubmission starts a two-month review cycle and a Class 2 resubmission starts a six-month cycle. If the applicant takes none of these actions within one year, FDA may treat that as a request to withdraw, unless the applicant asked for an extension.¹¹ Each CRL deficiency should become its own row in the query log, because each one must be addressed in the resubmission.

EMA: Lists of Questions and Clock Stops

The EU centralized procedure works on the opposite principle: the clock stops while the applicant answers. EMA describes an assessment of up to 210 active days. At Day 120 the CHMP adopts a list of questions, and the evaluation pauses (the first clock stop) while the applicant prepares responses. At Day 180 the CHMP may adopt a list of outstanding issues, followed by a second clock stop. EMA says the developer generally has up to three months to answer the Day 120 questions and up to one month for the outstanding issues, and that the clock-stop length must be agreed by the CHMP.¹²

The CHMP’s policy on response times adds detail that matters for planning. Applicants may ask for up to three additional months for the Day 120 responses, with a scientific justification, and extensions beyond six months from the date the list was issued would not normally be accepted. For the Day 180 list of outstanding issues, a one-month extension may be granted only in exceptional circumstances. The policy also reminds applicants that responses should not, in general, introduce substantial data from new studies that the CHMP did not request, though new analyses such as follow-up data from studies already submitted may be included. It also says the rapporteurs and EMA should be available to meet applicants after the Day 120 list to clarify the intent behind the questions.² That document dates from 2009, so teams should confirm current practice against EMA’s live pre-authorization guidance, but its structure still matches how the procedure runs.

EMA’s questions are not all equal. EMA notes that an oral explanation before the committee is usually organized when the CHMP still has major objections with the application.¹² The log should record which questions are major objections, because those are the ones that can decide the outcome, and they deserve the most senior review.

Health Canada: Clarification Requests, NODs, and NONs

Health Canada uses several types of query, each with its own response time, and its clock rules differ by type.

• Screening clarification requests. During screening, the solicited information is due within 15 calendar days of the request, or as stated in the request. A screening deficiency notice (SDN) must be answered within 45 calendar days, and a screening rejection letter is issued if there is no response within that time or the response is incomplete or contains unsolicited information.¹³
• Review clarification requests. During review, Health Canada asks the sponsor to expand on, clarify, or re-analyze information. Response times range from 2 to 15 calendar days depending on the performance standard of the submission, with 15 days for submissions on 180- to 300-day standards. The review does not stop if a complete response arrives on time. There is no limit on how many clarification requests one submission can receive, but the guidance says a particular issue will only be addressed once.¹⁴
• Notice of deficiency (NOD) and notice of non-compliance (NON). For an NDS, SNDS, or ANDS, the response to an NOD or NON is due within 90 calendar days. When an NOD is issued, all scientific review streams stop.¹⁴

The short clarification windows are where cross-market consistency is most at risk. Fifteen days is not enough time to rebuild an answer from scratch, so the team either reuses an answer it already gave elsewhere or writes something quickly. A query log with approved, topic-coded answers makes the first option safe.

PMDA: Inquiries and Replies

In Japan, review questions arrive as inquiries from the PMDA review team, and the sponsor replies in writing. PMDA’s public material in English on the exact inquiry process is limited, but a joint question-and-answer document from Taiwan’s TFDA and CDE and Japan’s MHLW and PMDA, written for their bilateral new drug review scheme, gives practical guidance that applies more broadly. It advises that submitting accurate responses in batches, after preparing the necessary information, leads to efficient review. It tells applicants to avoid submitting responses intermittently or incompletely, since that may lead to further inquiries, and to consult the reviewing authority in advance if an inquiry will take time to answer. It also says that if the applicant can provide its responses to inquiries from the authority that already approved the product, the reviewing authority might not raise the same question again.¹⁵

That last point is the case for a query log in one sentence. A clean record of how you answered one regulator can reduce the number of questions the next one asks.

After Approval, the Questions Keep Coming

Questions do not stop at approval. In the EU, a Type II variation can draw a request for supplementary information (RSI), and EMA sets the timetable for assessing the responses when it adopts the RSI.¹⁶ In the US, prior approval supplements draw IRs and, if needed, CRLs. Post-approval questions are where drift is most likely, because they often arrive years after the original answers were written, by people who were not there. A post-approval answer about a specification should be checked against what the company told every agency during the original review.

Designing the Query Log

A query log is a controlled register of every question a health authority has asked and every answer the company has given. It can start as a well-controlled spreadsheet and grow into a module of a RIM system. Either way, the design decisions are the same, and most of them are about granularity.

One Row Per Question, Not Per Letter

The most common design mistake is to log letters. A Day 120 list of questions with 85 items becomes one row that says “D120 LoQ received, responses due in three months.” That row tells you nothing about which questions are hard, which ones overlap with FDA’s, or which ones have already been answered elsewhere.

Log each question as its own record, and keep the letter as a parent. An FDA IR letter with six numbered items becomes one letter record and six question records. If a single agency item contains two distinct requests (for example, “provide the batch data and justify the proposed limit”), split it into two, because they may have different owners and different evidence.

Give each question an internal identifier that never changes, and record the agency’s own numbering next to it. The FDA-commissioned assessment of product quality IRs reported, among its interview themes, a suggestion that FDA expand the practice of numbering IR letters to help tracking.¹ Until every agency numbers every question, the sponsor’s own identifier is the only reliable key.

The Core Fields

The fields below are the minimum that make a log useful across markets. Teams add more, but removing any of these breaks one of the later steps.

Topic Codes That Work Across Agencies

Topic codes are what turn a list of questions into something you can compare across markets. They need to be specific enough to group the same issue and broad enough that people apply them the same way.

A workable approach is two levels. The first level follows the CTD structure, since every agency organizes its review around it: drug substance, drug product, nonclinical, clinical pharmacology, clinical efficacy, clinical safety, labeling, risk management, administrative. The second level names the specific issue within the product: “DP dissolution specification,” “DS genotoxic impurity limit,” “efficacy in patients over 75,” “hepatic safety signal,” “device component shelf life.” Second-level codes are product-specific and should be set by the regulatory lead early in the review, then extended as new issues appear.

Avoid free-text tags. If one person writes “dissolution,” another writes “diss spec,” and a third writes “release testing,” the log cannot find overlaps, and the consistency check fails without anyone knowing.

Status Values That Mean One Thing

Status fields cause more confusion than any other field. Use a short, fixed list and define each value in writing. A sequence that works:

• Received: logged, not yet assigned.
• In drafting: owner assigned, answer being written.
• In review: draft complete, under functional and regulatory review.
• Approved for submission: signed off by the named approvers.
• Submitted: included in a sequence sent to the agency, with the sequence number recorded.
• Resolved: the agency has accepted the answer, either explicitly or by not raising the issue again at the next stage.
• Reopened: the agency has asked a follow-up on the same question.

The difference between “submitted” and “resolved” is the one most logs miss. An answer is not finished when it leaves the building. It is finished when the agency stops asking.

Linking Every Answer to Its Evidence

An answer to a health authority is a statement in a regulatory application. It has to be true, and the company has to be able to show why it is true. That is why every answer in the log should carry links to the evidence behind it, not just the text that was submitted.

Link to the Source Data

Most answers rely on something: a batch analysis table, a stability report, a statistical output, a clinical study report section, a validation report, a deviation investigation. The log should record the exact document and version, not just a folder. If the answer quotes a number, the link should point to the controlled record where that number comes from, such as the approved analysis output or the signed report, rather than a slide or an email.

This does three things. It lets a reviewer check the answer before sign-off without hunting for the data. It lets a later author see what the earlier answer was based on, which is the only safe way to decide whether it can be reused. And if an inspector or a later reviewer asks where a figure came from, the company can show the chain from the answer back to the record in minutes.

Link to the Dossier Section and Sequence

Each answer should also record where it went in the eCTD. The EU Module 1 specification gives a useful model. Responses to questions are submitted as a sequence with the submission unit type “response,” and the related-sequence field points back to the sequence that started the regulatory activity, so the response is tied to the procedure it belongs to. Module 1 also has a dedicated responses section, and the specification’s file-naming examples include leaf titles such as “Day 106 Clinical Responses, MAA.”¹⁷ The query log should record the sequence number, the section, and the leaf title for every answer, in every market.

The FDA-commissioned assessment listed, among its interview themes, the practice of applicants sending IR responses by email first, to get the information to FDA quickly, before submitting the formal amendment and updating the eCTD.¹ That is a reasonable practice, but it creates two versions of the same answer. The log should record both: the date and content of the informal response, and the sequence number of the formal one. If they differ, the formal one wins, and the difference should be noted.

When the Answer Changes the Dossier

Some answers do more than explain. They change a specification, update a section of Module 3, revise the label, or add a commitment. These answers need a second link: to the dossier section that changed, and to the change record behind it. If an answer to EMA tightens a specification, the same change has to reach every other market’s dossier, either through the same review cycle or through a planned variation or supplement. A query log that flags “dossier changed: yes” and names the affected sections is how the regulatory lead sees that the change is not yet reflected in the US or Canadian filing.

Reusing Approved Answers With Control

Reuse is the biggest efficiency gain a query log offers, and the biggest risk if it is done carelessly. A good answer to EMA about a genotoxic impurity limit is probably a good starting point for Health Canada’s clarification request on the same topic. Pasting it in without checking is how a company tells Canada something that stopped being true four months ago.

What Counts as an Approved Answer

An answer becomes reusable only when three conditions are met: it was submitted, the agency accepted it (the question is resolved, not just answered), and the facts it relies on have not changed since. The log should mark these answers clearly, for example with a “reusable” flag set by the regulatory lead, along with the date of the facts it relies on (the data cutoff, the stability time point, the report version).

Answers that were submitted but drew a follow-up are not reusable as written. The follow-up shows the agency was not satisfied, and the next agency may not be either. These answers are still useful as history, but the reuse flag should stay off until the follow-up is resolved.

Rules for Reuse

When Not to Reuse

Some answers should never be reused without a fresh look by the functional lead. These include anything involving safety data with a cutoff date, any statement about ongoing studies, anything that commits the company to a future action, and any answer the first agency did not accept. EMA’s response-time policy is also a reminder that agencies limit what can be introduced at each stage: it says substantial new study data not requested by the CHMP should not, in general, be introduced in responses.² An answer that was fine at FDA in a resubmission may introduce new data that is not appropriate at Day 180 in the EU.

Tracking the Clock on Every Question

Time is the constraint that drives most bad answers. Teams that track only the agency’s due date discover in the final week that the stability report is not signed, the statistician is on leave, or the affiliate needs a week to translate. The log should track more than one date for every question.

Two Clocks Per Question

Every question has an external clock and an internal one. The external clock is the agency’s deadline, set by the rules above or by the agency’s letter. The internal clock is the set of dates the team must hit to meet it: first draft, functional review, regulatory review, sign-off, publishing, and dispatch. The internal clock should be built backward from the external one, and the log should show both.

The agencies’ rules affect how the clocks interact. At FDA, the review clock keeps running during IRs, so every day the sponsor spends on an answer is a day FDA does not have to read it.⁸ Interview themes in the FDA-commissioned assessment included the unpredictability of IR timing, IR due dates near the end of reviews that become shorter and less realistic, and the challenge of publishing responses and submitting them through the electronic gateway within FDA’s timeline.¹ At EMA, the clock stops, but the length of the stop is agreed in advance and extensions need a justification.² At Health Canada, a 15-day clarification window leaves almost no room for internal review.¹⁴

Building the Internal Timeline Backward

A practical internal timeline for a question with a 15-day window might allow five days to draft, three for functional review, two for regulatory review and sign-off, two for publishing and quality checks, and a margin of three days. For a three-month EMA clock stop, the proportions change, but the principle holds: publishing and dispatch always take longer than people expect, and the margin is what absorbs a late data point.

The log should show, for each open question, how many days remain on the external clock and whether each internal milestone has been met. A weekly view sorted by days remaining, filtered to questions that are behind on any internal milestone, is the single most useful report a query log produces. It is what lets the regulatory lead ask for help on Tuesday instead of discovering the problem on Friday.

Extensions and What They Signal

Every agency allows some form of extension. EMA allows a justified request for more time on the Day 120 list.² Interview themes in the FDA-commissioned assessment noted that FDA is generally open to extending IR due dates when asked and when possible.¹ FDA will grant any reasonable request to extend the one-year window to respond to a complete response letter.¹¹ PMDA’s joint guidance with Taiwan advises consulting the reviewing authority in advance when an inquiry will take time.¹⁵

The log should record every extension requested, granted, or refused, with the reason. Two patterns are worth watching. A cluster of extension requests on the same topic across agencies usually means a gap in the data that the company has not admitted to itself. And a habit of extending every clock stop to the maximum usually means the internal timeline is not built backward from the deadline. Swissmedic’s summary of the 2026 CIRS benchmark of six major authorities describes scientific assessment time as including company time for responding to questions, so the sponsor’s own response speed is part of the published comparisons.¹⁸

Keeping Answers Consistent Across Markets

Consistency is the main reason to keep one log rather than one per agency. It is also the step teams skip when they are short of time, which is exactly when inconsistencies happen.

Legitimate Differences Versus Drift

Not every difference is a problem. Answers can properly differ because the data has matured between submissions, because the regional requirements differ, because the agencies asked for different things, or because the company has made a deliberate regional decision such as a different presentation or a different labeled population. The EMA and FDA analysis of 107 applications showed that differing clinical data, often reflecting different submission timing, was a real source of different outcomes.³

Drift is different. Drift is when two answers on the same topic describe the facts differently and nobody decided they should. The test is simple: could the company explain the difference to both agencies, in writing, without embarrassment? If yes, it is a legitimate difference and the log should record the explanation. If no, it is drift and one of the answers needs to be corrected, either in the pending response or through a follow-up to the agency that received the outdated version.

A Cross-Market Consistency Check

Build the check into the workflow, not into a separate review. Before any answer moves to “approved for submission,” the log should show the reviewer every other answer with the same topic code, across all agencies and all products that share the same drug substance or drug product. The reviewer confirms, with a recorded sign-off, that the new answer is consistent with the earlier ones or that the differences are explained.

For high-risk topics, add a periodic view. Once a month during active reviews, the regulatory lead pulls every answer on each major topic code and reads them together. This catches drift that crept in through answers that were each consistent with the one before but, after four rounds, no longer match the first.

Pay particular attention to reliance markets. Because Swissmedic’s Article 13 procedure expects the reference dossier, including all sequences or the answers to questions, an answer given to FDA or EMA may be read by a reviewer in a market the company has not yet filed in.⁴ WHO’s guidance stresses that reliance depends on sameness of product, meaning the product submitted to the relying authority should be essentially the same as the one approved by the reference authority.⁶ Answers that describe the product differently across markets put that sameness in doubt.

Who Decides

Someone has to own the cross-market view. In most companies this is the global regulatory lead for the product, supported by the regional leads. The regional leads know the local requirements; the global lead knows what every agency has been told. When a difference is found, the global lead decides whether it is legitimate, records the reason, and, if it is drift, decides which answer is right and how to correct the other. The log should record that decision, because it is the evidence that the company manages consistency on purpose.

How the Query Log Connects to RIM

A regulatory information management (RIM) system holds the company’s record of products, registrations, submissions, commitments, and variations. The query log holds the questions and answers that happen within those submissions. The two must be connected, but they do different jobs.

What RIM Should Hold and What the Log Should Hold

RIM is the system of record for the structure: which product, which registration, which procedure, which sequence, which commitment. The query log is the system of record for the content of the conversation: the verbatim question, the answer, the evidence, the reuse status, and the consistency decisions. Many RIM platforms include a health authority correspondence or query module, and if yours does, the log can live inside it. If it does not, or if the module only stores letters as attachments, the log can live alongside RIM, provided it uses the same identifiers.

Integration Points

Five links between the log and RIM matter most:

• Registration and procedure. Every question links to the RIM registration and procedure record, using RIM’s identifiers. This is what lets anyone see all questions for a product or a market.
• Submission sequence. Every submitted answer links to the RIM submission record for the sequence that carried it.
• Commitments. Any answer that makes a promise creates or updates a commitment record in RIM, with a link back to the question that produced it. Post-approval commitments often start as an answer to a question late in review, and that origin should never be lost.
• Variations and supplements. Any answer that changes the dossier in one market triggers a check in RIM for the other markets, so the change can be planned as a variation or supplement where needed.
• Document management. Evidence links point to controlled documents in the document management system, by document ID and version, so that a superseded version is visible as superseded.

Start Small: The Spreadsheet Phase and When to Move

A company with one product under review in two markets does not need a new system. A controlled spreadsheet with the fields above, a fixed topic code list, restricted edit rights, and a regular backup will work for the first review cycle. What makes it work is discipline: one row per question, verbatim text, evidence links, and the consistency check before sign-off.

The signs that a spreadsheet has been outgrown are familiar. More than one person needs to edit it at the same time. The team cannot reliably find every answer on a topic. Reuse links break when rows are moved. The log cannot be joined to RIM records without manual matching. Audit questions about who changed an answer and when cannot be answered. When two or more of these appear, it is time to move the log into the RIM platform’s correspondence module or into a purpose-built tool connected to it.

Where AI Helps

Language models are good at some parts of this work and should be kept away from others. They are useful for splitting a long letter into individual questions, suggesting topic codes, finding earlier answers on the same topic even when the wording differs, and drafting a first comparison of two answers for a reviewer to check. They are also useful for spotting answers that cite a figure different from the one in the linked evidence.

They should not be the source of facts in an answer, and they should not decide whether a difference between markets is legitimate. Every answer must still trace to a controlled record and be signed by a named person who understands the data. Interview themes in the FDA-commissioned assessment included a suggestion that FDA build an office-wide repository of past IRs to improve consistency in how questions are written.¹ Sponsors need the same thing on their side, and the repository has to come before any AI tool that searches it.

Conclusion

Health authority questions are where the details of a dossier are tested, one agency at a time, on clocks that do not wait. The answers a company gives become part of its regulatory record in every market, and with reliance pathways, collaborative reviews, and published complete response letters, those answers are increasingly read side by side. Treating questions as a stream of separate letters, handled by whoever is free, is how answers drift. Treating them as a single, controlled record, broken down by question and topic, linked to evidence and dossier location, reused under clear rules, tracked on two clocks, and checked for consistency before sign-off, is how a company keeps telling every agency the same true story, with every difference explained.

Sakara Digital works with pharma and biotech organizations on regulatory operations, RIM, and the data and process design behind them. If you are planning a multi-market filing, or your team is already managing questions from more than one agency and wants an independent view of how to track them in one place, we are happy to have that conversation.

For Further Reading