25.13 — The FDA: History, Structure and How a Decision Is Made

In 1902, a government chemist named Harvey Wiley recruited twelve healthy young men to eat meals dosed with the preservatives then routinely added to American food — borax, formaldehyde, salicylic acid, copper sulphate. He fed them, weighed them, and recorded what happened as they got sick. The newspapers called them the Poison Squad.

Wiley was trying to prove a point he could not otherwise make: nobody knew what was in food or medicine, and nobody had to say. Patent medicines contained morphine, cocaine and alcohol, sold for infants, with no labelling requirement at all.

Four years later, helped along by the public reaction to Upton Sinclair's descriptions of the meatpacking industry, the Pure Food and Drugs Act of 1906 passed. Wiley's small chemistry division became, in 1930, the Food and Drug Administration.

Everything the agency now does grew from that one idea — that a person buying a product has a right to know what is in it and whether it is safe — extended, disaster by disaster, over the following century.

The laws, and what each one added

The agency's power is entirely statutory: it can do what Congress has empowered it to do, no more. So its history is a list of laws, and each one is a permanent layer that clients still plan around.

Year	Law	What it added
1906	Pure Food and Drugs Act	Truthful labelling
1938	Food, Drug, and Cosmetic Act	Proof of safety before sale
1962	Kefauver–Harris Amendments	Proof of effectiveness
1976	Medical Device Amendments	Device regulation, device classes
1983	Orphan Drug Act	Rare disease incentives
1984	Hatch–Waxman Act	Generic approval pathway
1992	Prescription Drug User Fee Act	Fees, review deadlines
1997	FDA Modernization Act	Fast track, one-trial flexibility
2007	FDA Amendments Act	Post-market study powers, REMS
2012	FDASIA	Breakthrough therapy designation
2016	21st Century Cures Act	Real-world evidence, patient input

Three of those need a sentence beyond the table because they explain the agency's daily behaviour.

The 1938 Act came from the Elixir Sulfanilamide deaths of Chapter 25.2 and created the requirement to submit a new drug application before marketing. The 1962 Amendments came from thalidomide and added effectiveness, controlled clinical investigation, informed consent, and adverse event reporting. And the 1992 user fee act changed the agency's character more than any of the others: companies pay substantial fees, and in exchange the agency commits to review performance goals. Before it, applications could sit for years. After it, review times became predictable — and the arrangement has been criticised ever since for making the regulator partly funded by the industry it regulates, which is a fair argument worth being able to state on both sides.

How the agency is organised

The FDA sits inside the Department of Health and Human Services. It is organised into centres by product type, and knowing which centre owns a product tells you almost everything about which rules apply.

CDER — the Center for Drug Evaluation and Research. Small-molecule drugs, generics, and most therapeutic biologics including monoclonal antibodies. The largest centre and the one most clients mean when they say "the FDA".

CBER — the Center for Biologics Evaluation and Research. Vaccines, blood and blood products, tissues, and cell and gene therapies. If your client works in cell or gene therapy, their regulator is CBER, and the review culture there is noticeably different because the products are individualised and the science is newer.

CDRH — the Center for Devices and Radiological Health. Devices, diagnostics, and software as a medical device. Chapter 25.16 is entirely about this centre's rulebook.

And then CFSAN for food, CVM for veterinary medicine, CTP for tobacco, and NCTR, the agency's own research arm.

Separately from the centres sits the field force that actually inspects factories, laboratories and trial sites. This was the Office of Regulatory Affairs for decades; in the agency's largest reorganisation in recent history, effective 1 October 2024, it became the Office of Inspections and Investigations. You will see both names in client documents for years, and knowing they are the same organisation prevents confusion.

What happens inside a review

A company submits a new drug application or a biologics licence application. What follows is far more structured than most people imagine, and the milestones drive every client timeline.

Long before submission there are meetings. A pre-IND meeting before the first human study. An end-of-Phase-2 meeting to agree the design of the Phase III programme — the single most valuable meeting in development, because agreeing the primary endpoint and the analysis with the agency in advance removes the largest risk in the entire programme. A pre-submission meeting to agree the contents and format of the application.

Then the application arrives. It is enormous — the full nonclinical package, every clinical study report, the manufacturing section, the proposed labelling, and the datasets from Chapter 25.10.

Within 60 days the agency decides whether to accept it for review, which is called filing. A refuse-to-file decision is a serious event: the application was incomplete enough that review cannot start.

From the filing date the review clock runs — a goal of about ten months for a standard review and about six for a priority review. This is why applications are commonly described as having a "PDUFA date": it is the date by which the agency has committed to act.

The review itself is done by a team of specialists working in parallel, and each writes an independent assessment. A medical officer reviews the clinical evidence. A statistician re-analyses the data independently from the submitted datasets. A pharmacology and toxicology reviewer assesses the animal package. A clinical pharmacology reviewer assesses dosing and interactions. A chemistry reviewer assesses manufacturing and specifications. A labelling team works on the wording. And a project manager holds it together and is the company's day-to-day contact.

During review, the agency sends information requests — sometimes hundreds — and the company answers them under short deadlines. This is why a regulatory affairs team's workload peaks after submission rather than before it, which surprises people planning resources.

Inspections happen in parallel. A pre-approval inspection of the manufacturing site checks that the plant can actually make what the application describes. Bioresearch monitoring inspections check clinical trial sites and the sponsor's own trial oversight. An approval can be delayed by a manufacturing inspection finding even when the clinical data is impeccable, and this happens regularly.

For novel or contentious products, the agency may convene an advisory committee — an open public meeting of outside experts who hear the company and the agency present, and vote on questions put to them. The vote is advice, not a decision. The agency usually follows it and is not obliged to, and the cases where it did not are studied for years afterwards.

Finally the agency issues one of two letters. An approval letter, with the agreed labelling and any required post-marketing studies and risk management programme. Or a complete response letter, which sets out everything that must be resolved before approval is possible — more data, a new trial, or fixing a manufacturing problem. A complete response letter is not a refusal forever; many products are approved after resolving one.

What the label actually is

The prescribing information — the label — is not marketing copy. It is the legally approved statement of what the drug is for, who may take it, at what dose, and what is known about its risks. It is negotiated word by word during review.

Its structure is fixed, and knowing the order helps you read one quickly: the highlights section and any boxed warning, then indications, dosage and administration, dosage forms, contraindications, warnings and precautions, adverse reactions, drug interactions, use in specific populations, description, clinical pharmacology, nonclinical toxicology, clinical studies, and how it is supplied.

The boxed warning is the strongest warning available and appears at the top in a box — the industry calls it a black box warning. A drug may be approved with one; it means the risk is serious enough that every prescriber must see it before anything else.

And the label matters commercially as much as clinically, because a company may only promote a drug for what the label says. Promoting a use outside the label is a violation that has produced some of the largest corporate settlements in American legal history, and it is the reason for the strict separation of medical affairs from commercial functions described in Chapter 25.23.

When approval comes with strings

Two mechanisms attach obligations to an approval and both generate long-running work.

Post-marketing requirements and commitments. The agency can require studies after approval — a specific safety study, a trial in children, or a confirmatory trial for an accelerated approval. These are tracked publicly, and failure to complete them has consequences.

REMS — a Risk Evaluation and Mitigation Strategy. For drugs whose risks require more than a label, the agency can require a formal programme: prescriber training and certification, patient enrolment in a registry, mandatory laboratory monitoring before each dispense, or restricted distribution through certified pharmacies only.

A REMS is a systems problem as much as a clinical one. Somebody must verify a prescriber's certification, record that a patient's test was done before a pharmacy may dispense, and report compliance to the agency. These programmes are frequently outsourced, and they are a substantial and stable line of business for service companies. Chapter 25.25 covers the closely related patient support programmes they often sit inside.

How the agency enforces

Inspection findings and enforcement escalate in a defined ladder, covered fully in Chapter 25.18. In summary: observations at the end of an inspection on Form 483; then a warning letter if the response is inadequate; then, in serious cases, seizure, injunction, or a consent decree placing a company's operations under court supervision.

One point about tone is worth making, because it is misunderstood by people new to the industry. The agency's normal working relationship with companies is collaborative, meeting-heavy and technical. Enforcement is what happens when that fails. A regulatory affairs professional's central skill is keeping problems inside the collaborative channel, and a services engineer helps by making evidence easy to produce rather than by hiding problems.

What this means when you build systems

Three consequences follow from everything above, and they should shape any client conversation.

Dates are real and immovable. A PDUFA date, a submission date, a study start — these are commitments with commercial and sometimes clinical consequences. A system delivery that slips a submission is a serious failure, not an inconvenience, and it will be remembered.

Everything is inspectable. Your system, its validation, its audit trails and its change history may be examined by an inspector years after you have left the engagement. Build for the reader who arrives in 2033 with a badge.

And the agency publishes almost everything. Approval packages, inspection findings, warning letters, advisory committee materials and adverse event data are all public. A few hours reading a client's own published inspection history before a proposal meeting will tell you exactly what they are worried about, and almost nobody bothers to do it.

Next: Chapter 25.14, the rest of the world's regulators — how Europe, India and Japan differ, what ICH actually harmonised, and why a global submission is not one submission.