Health authority questions — HAQs, information requests, deficiency letters, lists of questions, depending on the agency and procedure — are the most predictable surprise in regulatory affairs. Every team knows questions are coming after a major submission. Most teams still experience them as a fire drill: an agency letter arrives with a response clock running, and the organization scrambles to reassemble the people, data, and rationale needed to answer. This guide covers how question cycles actually work across agencies, why the questions are far more predictable than they feel, and what preparing before submission looks like in practice.

The Question Cycle, Agency by Agency

The mechanics differ, but the shape is universal:

  • US FDA issues Information Requests during review and, for unresolved deficiencies, a Complete Response Letter. IR response windows can be brutally short — days to a few weeks — and the review clock context makes speed strategic.
  • EMA formalizes the cycle: a Day 120 List of Questions stops the clock, followed by a Day 180 List of Outstanding Issues. Responses are substantial submissions in their own right.
  • PMDA conducts iterative inquiry rounds during review, often with rapid-turnaround written questions.
  • Other authorities — Health Canada, MHRA, TGA and beyond — run recognizable variants of the same pattern: submission, question wave, response, sometimes a second wave.

Two facts follow. First, response capacity is a core regulatory competency, not an exception process. Second, because every cycle adds months, each question wave you can shrink or pre-empt is calendar time recovered on approval.

Questions Cluster — Predictably

Here is the empirical part. In an analysis of more than 1,000 real health-authority questions across programs, the clustering was striking: questions map to discipline — quality/CMC, nonclinical, clinical, biopharmaceutics, labeling — with near-deterministic regularity based on which CTD sections they target, and within each discipline the same themes recur. Stability data sufficiency, specification justifications, impurity qualification, dissolution method robustness on the quality side; exposure margins and study design choices on the nonclinical side; endpoint and subgroup questions on the clinical side. The mapping from CTD module to question discipline is essentially fixed, and the base rates of question types within a discipline are stable enough to plan against.

What that means practically: given a submission’s content — which modules, which submission type, which known weak spots — the likely question set is estimable before the agency sees the dossier. Experienced regulatory leads do a version of this in their heads. The improvement is doing it systematically.

Preparing Before You Submit

1. Run a question-readiness review per discipline

For each discipline, list the questions similar submissions actually received. Rate your dossier’s exposure honestly: where is the data thin, the justification implicit, the precedent unfavorable? This is different from QC review — you’re not checking the dossier is correct, you’re anticipating where a reviewer will push.

2. Draft response skeletons for high-likelihood questions

For questions rated likely, draft the response backbone now: the argument, the data sources, the studies you’d cite, the SMEs who own it. A skeleton drafted calmly before submission beats a full response drafted in a two-week IR window.

3. Build your answer memory

Your best predictor of a good response is your own past responses. Teams that keep approved historical answers in a searchable, structured knowledge base respond faster and more consistently than teams that hunt through old PDF letters. Every response cycle should make the next one easier.

4. Track the questions as structured work

When questions arrive, decompose the letter into atomic questions, route each by discipline to a named owner with a deadline, and track them like the mission-critical work items they are — connected to the correspondence record, not in a fresh spreadsheet per letter.

Where Software Helps

This preparation loop is exactly what DnXT built Predictive HAQ Readiness for: it reads a submission’s profile and CTD content, applies discipline base-rates from that 1,000+ question corpus, and produces a likelihood-ranked, fully explainable list of probable questions with recommended preparation actions — before submission. On the reactive side, an RFI workbench decomposes incoming agency letters into routed atomic questions and retrieves grounded draft responses from your own approved answer history, with every AI draft held for human approval. Teams using this pattern target cutting a review round entirely — the difference between three cycles and two is often 500–900 hours of response work and months of calendar time. We’ll cover the predictive engine in depth in a dedicated post.

Frequently Asked Questions

What is a health authority question (HAQ)?

Any formal question a regulator sends about a submission under review — FDA information requests and complete response deficiencies, EMA Day 120/180 questions, PMDA inquiries, and equivalents. Companies typically manage them under one umbrella process because the response mechanics are similar.

How fast do we have to respond?

It varies from days (some FDA IRs and PMDA inquiries) to months (EMA clock-stops, CRL resubmissions). The operational point: assume short windows and pre-position answers, because the timeline is the agency’s choice, not yours.

Can you really predict agency questions?

Not the exact wording — but the discipline, topic, and likelihood, yes, to a useful degree. Question themes are driven by CTD content and well-known review priorities, both of which are visible before submission. Prediction produces a preparation list, not a crystal ball.

What’s the single highest-leverage change?

Keep structured, searchable records of every question you’ve received and every approved answer you’ve given. It’s the foundation both for faster responses and for meaningful prediction.

If your last question cycle was a fire drill, talk to a regulatory expert — we’ll show you what a predicted-question readiness report looks like for a submission like yours.