EDDO Is Now EDDC: Sorting the Combination Product Alphabet Soup

In under three years FDA has given the same concept three different names. It was an essential performance requirement, then an essential drug delivery output, and since October 2026 it is an essential drug delivery device performance characteristic. Meanwhile two standards use terms that sound almost identical to the first of those and mean something else entirely. A team building an electromechanical autoinjector is subject to all of it at once — which is why two competent engineers can discuss "the essential requirements" for twenty minutes before discovering they were never talking about the same list.

The confusion is structural, not careless

It is tempting to treat this as sloppy terminology that a good glossary would fix. It is not. Each term entered the vocabulary from a different direction, for a different purpose, and none of the source documents cross-references the others.

One arrived from the electrical safety world. One from the needle-based injection systems standard. One from FDA's combination product review practice — and that one has now changed name twice. Two more, EPR and EDO, are industry conventions that grew up inside company procedures without any external definition at all. They were never designed to fit together, so they do not.

Put them on the design control chain

The single most useful move is to stop treating these as competing synonyms and start asking where each one sits in the design control sequence. Almost all of the confusion dissolves at that point.

This also explains a common failure mode. A team maintains a legacy EPR list, is asked for its EDDCs, and hands over the EPR list. The content overlaps, but the list is at the wrong level and the wrong scope: it contains requirements rather than outputs, and it includes safety characteristics that have nothing to do with drug delivery while potentially omitting delivery characteristics that were never classified as safety-critical.

The two that sound alike and are not

The sharpest trap in this vocabulary is that essential performance and essential performance requirement are not the same idea, despite the near-identical name.

Essential performance is a defined concept in IEC 60601-1, the medical electrical equipment standard. It applies to electromechanical devices, and it is framed in terms of basic safety and unacceptable risk — performance whose loss or degradation beyond specified limits produces an unacceptable risk. If your delivery system contains electronics, it is in scope.

Primary functions is the parallel concept in ISO 11608-1:2022 for needle-based injection systems, defined at 3.18 with its requirements set out in 5.7.2 and risk-based acceptance criteria in 7.4.5.

This is not a distinction anyone invented after the fact. ISO drew it deliberately: among the listed changes in the 2022 fourth edition is the explicit aim of resolving confusion with the use of the term essential performance in IEC 60601-1. The committee saw teams conflating the two and introduced separate language to stop it.

The practical consequence for an electromechanical injector is that you are legitimately subject to both schemes, and they do not produce the same list. One is anchored in electrical safety and unacceptable risk; the other in delivery function and patient harm. A characteristic can be a primary function and not essential performance, or the reverse.

What FDA actually said — twice

In June 2024 FDA issued a draft guidance on essential drug delivery outputs, and it settled one question directly. In FDA's own words:

"Prior to this guidance, the term essential performance requirements (EPR) was generally used in communications between FDA and applicants for the EDDOs described herein. FDA is now using the term EDDO as we believe it is more descriptive."

So in the specific context of FDA correspondence about drug delivery, EPR was the previous name for the concept. A rename, not a new idea — and narrower than the general-purpose EPR many companies maintain internally.

Then, in October 2026, FDA did it again. The revised draft states that it "revises the draft guidance titled Essential Drug Delivery Outputs for Devices Intended to Deliver Drugs and Biological Products that issued in June 2024," updated to clarify testing and control of drug-delivery performance and to align with the revised 21 CFR part 820 Quality Management System Regulation. The term becomes essential drug delivery device performance characteristic — EDDC.

Note the lineage. The same concept has carried three FDA names inside three years: EPR, then EDDO, now EDDC. If you are keeping a terminology map, that column needs a date against every entry.

Both versions anchor the term in existing design control law rather than inventing a parallel requirement. If you already run design controls properly, EDDCs are not a new artifact. They are a classification applied to artifacts you already have.

The citations moved too, and this is the substantive half of the revision. The June 2024 draft cited the design output provision at 21 CFR 820.30(d) — the pre-QMSR clause structure. The October 2026 draft drops that framing: it points instead at ISO 13485 clause 7.3, incorporated by reference through 21 CFR 820.7(b), with design and development at 21 CFR 820.10(c), and notes the conforming updates to 21 CFR part 4 subpart A. The phrase "design output" does not appear in the revised draft at all.

The obligation did not change; the citation did. But expect the older reference to persist in templates, supplier procedures and third-party guidance for years — and when you find 820.30(d) in your own documents, map it forward to the QMSR structure rather than assuming it still resolves.

Note also what the guidance deliberately leaves out — unchanged between versions. It does not address drug-device compatibility, biocompatibility, sterility, human factors, electrical safety, electromagnetic compatibility, radio frequency wireless technology or cybersecurity. EDDCs are a drug-delivery lens, not a general safety lens.

Two additions in the revised draft are worth having. It says plainly that the "terminology used in this guidance is not intended to constrain how an applicant establishes their quality management system," and that where other terms cover the same ground a sponsor may leverage that existing work — which is as close as you will get to permission not to rename everything. And it adds worked appendices: example characteristics for a prefilled syringe and an autoinjector, and a set organized by product type. Those examples do more to settle an argument than any definition.

The crosswalk, now half-built

When this article first published, the honest summary was that nothing connected the standards vocabulary to the FDA vocabulary. The October 2026 revision closes part of that gap, and the way it closes it is instructive.

FDA now names ISO 11608-1:2022 directly, recognising it as a consensus standard that supplies test methods, and then says this:

"Although the standard's primary function definition identifies some of the functions identified as EDDCs in this guidance, primary functions and EDDCs do not necessarily align."

That is the agency confirming, in writing, the exact problem this article was written about. It goes on to say applicants may leverage the ISO 11608-1 information for an EDDC when the test method is appropriate for that EDDC — permission to reuse the test method, explicitly not permission to reuse the list.

The revised draft also now cites ISO 14971, alongside ICH Q9(R1), in the context of risk management. IEC 60601-1 is still absent. So the map now reads: ISO relates primary functions to essential performance in its own Annex H; FDA relates primary functions to EDDCs, loosely, and says they do not necessarily align; and nothing at all relates IEC 60601-1 essential performance to anything FDA writes about drug delivery.

Which means the mapping is still a deliverable you own. It is simply smaller than it was, and you now have an authoritative sentence to cite when someone insists the primary function list will do.

What the classification actually obligates you to do

This is where the terminology stops being an academic exercise. Classifying something as an EDDC is not a label — it triggers obligations, and most of them are quantitative.

Verification, and the sampling gap

The revised draft sharpens this, and not in your favour. A design verification testing protocol should include a statistical sampling plan; the plan "should include justification for the number of lots to be tested, taking into consideration the potential impact of lot-to-lot variability, and acceptance criteria." The tested lots should be the final finished product.

So FDA now asks explicitly for a justification, and names the source of variation it expects you to have reasoned about. What it still does not supply is any method for producing one. The word "confidence" does not appear anywhere in the document. There is no reliability target, no sampling methodology, no worked statistical example — the appendices give example characteristics, not example sample sizes.

That is not an oversight to complain about. It is how statistical expectations usually arrive: the agency names the obligation and leaves the justification to you, because the right answer depends on your product, your risk and your claim. But it does mean "we tested three lots of thirty" is a description, not a justification. Our framework for sample size justification applies directly: decide the claim first, and the number follows.

Preconditioning before you test

EDDC verification is not bench testing on pristine units. The guidance treats preconditioning as part of design verification — simulating storage conditions such as temperature, temperature fluctuation, pressure change and humidity, as well as shipping, before the functional testing happens. A dose accuracy result on a unit that never left the lab is not evidence about the product a patient receives.

Shelf life and expiry

EDDC performance has to be maintained across the proposed shelf life. The guidance expects design and development verification testing to evaluate EDDCs that may change over time or have age-related failure modes, and the final expiration date is informed by both the maintenance of that performance and drug stability. Usefully, it also says the converse: a characteristic that cannot change over time — a physical dimension such as needle length — does not warrant that evaluation. Classification therefore drives your stability program, in both directions.

The link to the risk file

Both of the standards concepts are defined against harm, so your primary function and essential performance determinations are risk determinations and should be traceable into the risk management file. The June 2024 draft never mentioned ISO 14971; the October 2026 revision now cites it, together with ICH Q9(R1). That is a prompt, not a procedure — the traceability between a delivery characteristic and the hazard it controls is still a link you are expected to build and maintain yourself.

A practical way through

Given that the FDA guidance is still a draft and the vocabularies do not align, a workable position looks like this.

1. Pick one controlled vocabulary internally and define it in a procedure. The failure mode is not using the "wrong" term; it is using several, undefined, inconsistently. Whatever you choose, write the definitions down.
2. Keep a mapping table as a living document, and date it. One row per characteristic, with columns for EPR, EDO, EDDC, CQA, primary function and essential performance as applicable. Record which FDA term vintage each entry was written against — a column that would already have saved you one rename, and will save you the next.
3. Use FDA's current vocabulary in FDA-facing documents. A submission that speaks of EDDCs is answering the question that will be asked. The revised draft also says in terms that its terminology "is not intended to constrain how an applicant establishes their quality management system," and that work done under other terms can be leveraged — so the translation belongs at the submission boundary, not in a rewrite of every internal procedure.
4. Never let a legacy EPR list stand in as the EDDC list. Derive them deliberately, from the delivery function, and record the reasoning for both inclusions and exclusions. The revised draft's appendices — worked examples for a prefilled syringe and an autoinjector — are a useful calibration for what belongs and what sits a level below.
5. Treat the draft status honestly. Non-binding does not mean ignorable; it means you may justify a different approach. If you depart from the guidance, say so and explain why, rather than quietly not doing it.

What to document

A requirement architecture that survives review states: the definitions your organization uses for each term, in a controlled procedure; which external schemes apply to this product and why, driven by the product architecture rather than habit; the mapping between those schemes; the derivation of each EDDC from the delivery function, including what was considered and excluded; the verification approach for each, with a sampling plan, preconditioning regime and acceptance criteria justified in advance; the shelf-life evaluation for every EDDC capable of changing over time, and the rationale for those excluded; and the traceability from each classification into the risk file.

None of this is exotic. It is ordinary design control discipline applied to a vocabulary that happens to be unusually confusing — and the confusion is precisely why writing it down is worth the afternoon it costs.

This article is general information, not regulatory or legal advice. The FDA guidance referenced is the revised draft of October 2026, which revises the June 2024 draft; both contain non-binding recommendations and are not for implementation. Confirm the current status and text before relying on either. Clause references to ISO and IEC standards are provided for navigation only — consult the standards themselves, which are copyrighted and available from ISO and IEC, for their definitions and requirements.