openFDA Drugs@FDA API

Every product on an American drugstore shelf has a paper trail that starts before the product exists, and this is where that trail lives. The openFDA Drugs@FDA API is the agency's official approval record made addressable: 29,273 applications at the August 2026 count, covering most human drug products approved since 1939 - prescription brand names, the generic challengers that followed them, many therapeutic biologicals, and over-the-counter products alike. One record sits over one application, with product rows and submission rows nested inside it.

Its position in the openFDA drug family decides what it is for. The openFDA NDC Directory API lists what is marketed (137,206 listings); the Drug Label corpus carries what packages claim about themselves (261,996 SPL documents); FAERS logs what went wrong afterward (20,692,690 reports). This is the only feed among the five that holds the approval event itself - who applied, what they filed, when the agency said yes, and whether a copy may substitute at the counter. Source: U.S. Food and Drug Administration (openFDA). Get a sample of this dataset and pull the approval biographies behind your own categories.

One real application, flattened for reading - an original approval followed by a labeling supplement, both captured during the August 2026 research pass:

application_number       N076422
sponsor_name             (applicant firm)

submissions[0]           ORIG · submission_number 1
                         submission_status        AP (approved)
                         submission_status_date   20040806
application_docs[0]      id 8910 · type Letter · date 20040806

submissions[1]           SUPPL · submission_number 4
                         submission_class_code_description   Labeling
                         submission_status        AP (approved)
                         submission_status_date   20060111

Read it as a clock instead of paperwork. The original reached approved status on 20040806 with its approval letter riding along as document 8910, typed Letter; a fourth-round labeling supplement cleared on 20060111 - roughly seventeen months between the first yes and the labeling revision. That interval is the unit of analysis here: subtract status dates inside one application and you have approval lag; stack the applications sharing a molecule and you have a generic-entry curve. Two honesty notes before anyone builds on it - firm-name strings were truncated in the captured payload, so confirm spellings before exact-match joins, and linked documents attach mostly for products approved since 1998. Your sample confirms current totals and exact header spellings before any commitment.

Fifteen columns carry the working payload, all verified against live records - the table below gives each with its type, definition and a genuine example:

Additional fields on request. Product rows repeat once per marketed presentation inside an application, and submission rows repeat once per regulatory filing, so a single application fans out into a small family of child tables - the nesting is the design decision to plan around, and deliveries arrive pre-flattened if you would rather see one wide row per product or per submission. Because the therapeutic equivalence column is the one everyone asks about, ask the sample to pin down how completely it populates across your molecule set rather than assuming it fills evenly. Get a sample.

• Geography: the United States regulatory perimeter. These are the approvals that gate American marketing, keyed to applicant firms rather than storefronts - there is no state or store dimension beneath the national level.
• Temporal: approvals reaching back to 1939, with labels, approval letters, reviews and patient information mostly attached for products approved since 1998. Each submission row states its own status date, so any delivery window you keep becomes a longitudinal archive rather than a snapshot.
• Granularity: one record per application, exploding downward - product rows repeat per marketed presentation, submission rows repeat per filing, and documents hang off individual submissions.

That combination makes this the regulatory-biography spine of the drug retail data hub: catalogs describe assortment, prices describe demand, recalls describe exits - and this feed describes the permission each product needed in the first place.

API, files, or your warehouse. Daily, weekly, or hourly.

Pick the slice, pick the cadence, pick the landing zone. Because an application keeps accumulating submissions across a product's whole life, keeping consecutive deliveries is what converts a reference lookup into a research corpus: new supplements arriving, tentative approvals converting, marketing statuses flipping. Diff two deliveries and the week's new filings fall out on their own - no re-pulls, no missed milestones. The sample comes first, so record shapes and totals are settled facts before any commitment. Get a sample of this dataset and wire it into your own stack.

• Generic-entry timing and pricing-pressure modeling - therapeutic equivalence ratings plus submission status dates turn the corpus into entry windows per molecule, ready to lay under a price panel; methods continue on our ML model training page.
• Event studies around approvals - reconstruct the filing chain from the original application to the latest supplement and align it against sales or volume series; playbooks sit on the data scientists use cases page.
• Competitive watch on challenger entrants - a new ANDA application naming an incumbent's molecule shows up as rows long before it shows up as headlines; screens live on the competitive intel use cases page.
• Regulatory-standing checks before listing - marketing_status separates Prescription, OTC, Discontinued and None (tentative), which is the difference between stocking a product and stocking a placeholder.
• Citation-grade verification - writers and researchers check approval dates and sponsors against the official record with the agency's own letters attached; reporting frames sit on the journalists academics use cases page.
• Category milestone marking - analysts stamp approval milestones into category reviews so competitive resets are dated rather than remembered.

Data scientists and quant researchers get the rare regulator-grade corpus whose nesting is a feature rather than a chore - flatten submissions into one timeline per application and the event study falls out; catalyst-timing plays collect on the investors quants use cases page. Developers and builders get flat JSON records that load without parsing gymnastics, with coded values - AP, TA, ORIG, SUPPL - that resolve through short dictionaries instead of bespoke mappings; integration notes sit on the developers builders use cases page. Market researchers and consultants get dated category resets for briefings that survive scrutiny; patterns sit on the market researchers use cases page.

• openFDA NDC Directory API - the regulated identity spine; resolving brand and ingredient rows to harmonized codes turns approval histories into panels you can roll up by labeler.
• openFDA Drug Label SPL API - the labeling documents behind each approved product, adding claims and warnings text to the regulatory biography.
• openFDA FAERS Adverse Event API - the post-market harm ledger; setting regulatory age beside report counts is exactly what the vs openFDA FAERS Adverse Event API comparison works through.
• openFDA Drug Enforcement / Recall API - what left shelves; pair it with what entered them and the product lifecycle closes at both ends.
• GoodRx Drug Prices & Pharmacy Finder - cash prices keyed to overlapping molecules, so approval age sits next to price movement.
• Walgreens Pharmacy & Drug Catalog - one chain's assortment, a direct test of how much of the approved universe actually reaches a shelf.

Patent-and-exclusivity boundary worth internalizing early - patent and exclusivity details live in the agency's separate Orange Book data rather than in Drugs@FDA, so full patent-cliff work pairs the two instead of expecting columns this feed never carried. The equivalence code speaks to substitutability at the counter, not to patent expiry.

Where to go next - the rail below collects the identity spine, the label and safety counterparts, the removal ledger, the consumer-price layer, the comparison against alternatives, and the glossary entries decoding application vocabulary, starting with Drugs@FDA approval application and the National Drug Code explained.