The Workbench · Craft

Section 314.70 sorts a drug change by risk, not size

This blog has already drawn the device-side version of this line, in not every PMA change earns a full supplement, and the production-floor version, in this week's note on a change control that has to clear review before it runs. The drug-approval version of the same question lives in 21 CFR 314.70, and it answers it with a three-way split that a change-control record can get wrong in either direction — treating a quality-relevant change as routine, or over-filing a routine one as if FDA had to bless it first.

Three categories, one regulation

21 CFR 314.70 sorts every post-approval change to an approved NDA or ANDA into one of three categories, and the category, not the applicant's own sense of how disruptive the change feels, decides what has to happen before the changed product ships. A major change needs a prior approval supplement (PAS), reviewed and approved by FDA before distribution. A moderate change needs a Changes Being Effected supplement, filed either thirty days ahead (CBE-30) or at the moment of filing (CBE-0). A minor change doesn't need a supplement at all — it gets implemented and then listed in the product's next annual report. Three tiers, three different clocks, and nothing in the regulation lets an applicant pick the convenient one by feel.

The sorting test is effect on quality, not size of effort

21 CFR 314.70(b)(1) sets the major-change threshold: a supplement is required, and must be approved before distribution, for any change to the drug substance, drug product, production process, quality controls, equipment, or facilities that has a substantial potential to have an adverse effect on the identity, strength, quality, purity, or potency of the drug product. Nothing in that sentence measures the change by capital cost, engineering hours, or how many departments signed off on it. A supplier swap that looks trivial on an org chart can carry a substantial potential to affect purity; a facility expansion that looks enormous on a capital budget can leave every quality attribute untouched. The regulation sorts by what the change can do to the product, and a change-control record that sorts by what the change cost to implement is answering a different question than the one 314.70 actually asks.

CBE-30 and CBE-0 aren't the same clock either

Moderate changes split again under 314.70(c), into two supplement types that look similar on the page and behave differently in practice. A CBE-30 supplement lets distribution begin thirty days after FDA receives it, unless FDA objects in that window and tells the applicant a prior approval is actually required. A CBE-0 supplement lets distribution begin on receipt, with no waiting period at all — reserved for a narrower, FDA-designated set of changes, including certain safety-driven labeling updates such as adding or strengthening a warning. Filing a moderate change as a CBE-0 when it only qualifies for CBE-30 doesn't speed up compliance; it starts distribution before the applicant has actually earned the right to skip the wait.

A category chosen by habit is a category chosen wrong

The practical failure isn't usually a firm claiming a major change is minor outright — it's a change-control system that defaults to whatever category the last similar-sounding change used, without re-running the actual 314.70(b) and (c) tests against the specific change in front of it. A raw-material supplier change gets filed CBE-30 because the last supplier change was CBE-30, not because this particular supplier change was separately evaluated for its own potential to affect identity, strength, quality, purity, or potency. FDA's own guidance on this framework exists because the category depends on the change's substance, not its resemblance to a prior entry in the same log, and a system that stops checking the substance has stopped actually classifying anything.

Where this meets the file

A change-control record that keeps its 314.70 category determination as its own reasoned field — citing which quality attribute the change could affect and why, not just which supplement type the last similar change used — is previewed in the launch catalog alongside this week's other change-management templates. If your program runs that determination differently, the shelf takes that correction directly.

The Regulatory Toolkit launches soon — a free shelf of source-mapped templates, checklists and browser-only tools for regulatory teams. Get one email when it opens, or contribute a template.

All Workbench notes