openFDA Drug & Device Data Crawler

Query the FDA's public API for drug adverse events, recalls, drug labels, NDC codes, device events, and 510(k) clearances. Eight datasets, one actor, no authentication required.

Features

• Queries all eight openFDA endpoints — drug adverse events (FAERS), drug recalls, drug labels (SPL), drug approvals (Drugs@FDA), NDC directory, device adverse events (MAUDE), 510(k) clearances, and device recalls
• Filters by drug name, manufacturer, and date range — field mapping is handled per-endpoint automatically
• Limits adverse event results to serious reports — death, hospitalization, life-threatening outcomes, and permanent disability
• Filters recalls by severity class — Class I (risk of death), Class II (moderate), or Class III (least severe)
• Paginates past the 25,000-record API cap — uses date-range windowing for large result sets
• Decodes numeric codes into readable values — drug characterization, sex, reaction outcome, and reporter qualification are translated from FDA codes to plain text
• Returns a unified schema across all endpoints — adverse event fields and recall fields coexist in each record; unused fields are empty rather than absent

Use Cases

Who is pulling this data and why?

• Pharmacovigilance analysts monitoring adverse event signals for a specific drug or drug class
• Compliance teams tracking open recalls by class and distribution region before a product launch
• Healthcare data engineers building pipelines that enrich internal drug databases with FAERS data
• Medical device researchers screening 510(k) clearance history for a competitor's device portfolio
• Insurance underwriters assessing liability exposure tied to Class I recall history for a manufacturer
• Academic researchers studying patient demographics and reaction outcomes across a drug category

How It Works

1. Select an FDA dataset endpoint and apply filters (drug name, manufacturer, date range, severity).
2. The actor constructs the openFDA search= query, paginates through results in batches of 100, and respects the 240 req/min rate limit with a 250ms delay between requests.
3. Raw API records are decoded — numeric sex/outcome/characterization codes become human-readable strings, dates are normalized to YYYY-MM-DD.
4. Each record is written to the Apify dataset in the unified schema. Adverse event fields are populated for drug/event and device/event runs; recall fields are populated for enforcement and recall runs.

Input

{
  "endpoint": "drug/event",
  "drugName": "metformin",
  "manufacturer": "",
  "dateFrom": "2023-01-01",
  "dateTo": "2023-12-31",
  "seriousOnly": true,
  "recallClass": "",
  "maxItems": 500
}

Field	Type	Default	Description
endpoint	string	drug/event	FDA dataset to query. See endpoint table below.
drugName	string	—	Drug brand or generic name. Partial match supported.
manufacturer	string	—	Manufacturer or recalling firm name. Partial match supported.
dateFrom	string	—	Start date (YYYY-MM-DD). For adverse events, filters by FDA receive date.
dateTo	string	—	End date (YYYY-MM-DD). Defaults to today when dateFrom is set.
seriousOnly	boolean	false	Return only serious reports (adverse event endpoints only).
recallClass	string	—	Class I, Class II, or Class III. Recall endpoints only.
maxItems	integer	100	Maximum records to return.

Available Endpoints

Value	Dataset	Records Available
drug/event	Drug Adverse Events (FAERS)	20M+
drug/enforcement	Drug Recalls	18K+
drug/label	Drug Labels (SPL)	257K+
drug/drugsfda	Drug Approvals (Drugs@FDA)	29K+
drug/ndc	NDC Directory	134K+
device/event	Device Adverse Events (MAUDE)	24M+
device/510k	Device 510(k) Clearances	174K+
device/recall	Device Recalls	—

Output Fields

{
  "report_id": "12345678",
  "endpoint": "drug/event",
  "drug_name": "METFORMIN HYDROCHLORIDE",
  "drug_active_ingredients": ["METFORMIN HYDROCHLORIDE"],
  "drug_manufacturer": "Amneal Pharmaceuticals",
  "drug_characterization": "Suspect",
  "drug_indication": "TYPE 2 DIABETES MELLITUS",
  "drug_route": "Oral",
  "reactions": ["Nausea", "Vomiting", "Lactic acidosis"],
  "reaction_outcomes": ["Recovered", "Recovered", "Hospitalized"],
  "serious": true,
  "seriousness_death": false,
  "seriousness_hospitalization": true,
  "patient_age": "67",
  "patient_sex": "Female",
  "report_country": "US",
  "report_source": "Physician",
  "receive_date": "2023-04-18",
  "recall_number": "",
  "recall_class": "",
  "recall_status": "",
  "recall_reason": "",
  "recall_product": "",
  "recall_firm": "",
  "recall_state": "",
  "recall_distribution": ""
}

Field	Description
report_id	Unique report or record identifier (FAERS safety report ID, recall number, NDC code, etc.)
endpoint	FDA dataset this record came from
drug_name	Drug brand name or generic name
drug_active_ingredients	Active ingredient(s) in the drug
drug_manufacturer	Drug manufacturer or reporting company
drug_characterization	Role of drug in the report: Suspect, Concomitant, or Interacting
drug_indication	Condition the drug was prescribed for
drug_route	Route of administration (Oral, Intravenous, Topical, etc.)
reactions	Adverse reaction terms (MedDRA preferred terms)
reaction_outcomes	Outcome per reaction: Recovered, Recovering, Not Recovered, Recovered with Sequelae, Fatal, Unknown
serious	Whether the report is classified as serious
seriousness_death	Whether the event resulted in death
seriousness_hospitalization	Whether the event resulted in hospitalization
patient_age	Patient age at time of event
patient_sex	Patient sex (Male, Female, Unknown)
report_country	Country where the report originated
report_source	Reporter type (Physician, Pharmacist, Consumer, etc.)
receive_date	Date the FDA received the report or recall initiation date (YYYY-MM-DD)
recall_number	FDA recall number (recall and enforcement records)
recall_class	Class I, II, or III
recall_status	Ongoing, Completed, or Terminated
recall_reason	Reason for the recall
recall_product	Description of the recalled product
recall_firm	Firm initiating or subject to the recall
recall_state	State of the recalling firm
recall_distribution	Distribution pattern of the recalled product

FAQ

How much data is available for drug adverse events? The openFDA Drug Adverse Events (FAERS) dataset contains over 20 million reports accumulated since the 1960s. The Device Adverse Events (MAUDE) dataset has over 24 million. This actor paginates through results up to the maxItems limit you set.

Does this actor require an API key? No. The openFDA API is a free government service that requires no authentication. Without a key, the rate limit is 240 requests per minute. This actor stays within that limit and does not currently support API key injection.

What happens when results exceed 25,000 records? The openFDA API has a hard cap of 25,000 on the skip offset (skip + limit must not exceed 25,000). This actor detects that boundary and stops pagination at the cap. For datasets larger than 25,000, use a narrow date range to retrieve specific windows of data across multiple runs.

Can I get only deaths or hospitalizations? Set seriousOnly: true to restrict drug/event and device/event results to serious reports. Within those results, seriousness_death and seriousness_hospitalization are boolean fields you can filter on in your downstream pipeline. The actor does not currently support single-outcome filtering as a direct input parameter.

Are all fields present in every record? The actor uses a unified schema across all eight endpoints. Fields that do not apply to a given endpoint (e.g., recall fields on an adverse event record) are returned as empty strings or empty arrays rather than omitted. This keeps downstream schema handling consistent.

Need More Features?

Open an issue or contact us at apify.com/jungle_synthesizer/openfda-drug-crawler to request additional endpoints, new filter parameters, or a custom build.

Why Use This Actor

• No scraping, no rate-limit workarounds — the openFDA API returns clean JSON; this actor handles pagination, decoding, and schema normalization so you do not have to
• Eight datasets in one interface — adverse events, recalls, labels, approvals, NDC codes, and device data are all accessible through the same input format
• Numeric codes are decoded — FDA uses integer codes for sex, reaction outcome, drug characterization, and reporter type; this actor converts them to readable strings before writing to your dataset