openFDA Scraper: Drug Events, Recalls & Device Reports

Pulls FDA safety surveillance data into one flat, consistent dataset: adverse event reports, enforcement actions, product recalls and drug labels across drugs, medical devices, food and tobacco.

openFDA is the FDA's own public API, so there is no login and nothing to solve with a proxy. Nine dataset combinations are supported and every one normalises into the same record shape, so a drug adverse event and a food recall land in the same columns.

Agent skill: SKILL.md

https://api.apify.com/v2/key-value-stores/t7YoTxpZEJOWvw4Ug/records/openfda-adverse-events-scraper.md

What you get

Field	What it holds
domain, endpoint	Which openFDA dataset the record came from
recordId, safetyReportId	The dataset's own identifier: FAERS report number, recall number, MDR report number or label set ID
productName, genericName, productDescription	Brand name, generic or active ingredient, and the full product description where the dataset has one
manufacturer, recallingFirm	Manufacturer as openFDA resolves it, and the firm that issued a recall
reactions	MedDRA reaction terms for drug and food events; report-text types for device events
seriousness	Which seriousness criteria the reporter ticked (Death, Hospitalization, Life-threatening and so on), or the device event type
patientAge, patientSex	Decoded from the FAERS numeric codes into "27 years" and "Male"
outcome	Reaction outcomes, decoded from codes to text
receiveDate, reportDate	ISO dates, converted from openFDA's YYYYMMDD and MM/DD/YYYY forms
recallReason, classification, recallNumber, status	Recall detail: Class I, II or III, ongoing or terminated
country, state, city	Where the firm or reporter is located
narrative	The longest free-text field the dataset offers: MDR narrative, indication, distribution pattern or root cause
productQuery, scrapedAt	Which product name produced the record, and the run timestamp

Every record carries the same key set whichever dataset you query. Fields that do not apply are null.

RUN_SUMMARY in the key-value store holds per-run counts, the filters you used, and any query that failed.

Use cases

• Pharmacovigilance. Sweep FAERS nightly for new reports naming your molecules and page on a seriousness change.
• Competitive recall monitoring. Watch drug/enforcement and device/recall for a list of competitor firms.
• Clinical risk research. Pull every report pairing a drug with a reaction term and count by year.
• Supply chain and food safety. Track Class I food recalls by state and distribution pattern.
• Regulatory intelligence. Diff drug/label set IDs to spot label changes on a watched product.

Quick start

Recent adverse event reports for one drug:

{
 "domain": "drug",
 "endpoint": "event",
 "productNames": ["ibuprofen"],
 "maxResults": 100
}

Two drugs, a specific reaction and a date window:

{
 "domain": "drug",
 "endpoint": "event",
 "productNames": ["metformin", "atorvastatin"],
 "searchQuery": "patient.reaction.reactionmeddrapt:\"nausea\"",
 "fromDate": "2024-01-01",
 "toDate": "2025-12-31",
 "maxResults": 2000
}

Every Class I food recall this year, no product filter:

{
 "domain": "food",
 "endpoint": "enforcement",
 "searchQuery": "classification:\"Class I\"",
 "fromDate": "2025-01-01",
 "maxResults": 5000,
 "apiKey": "YOUR_OPENFDA_KEY"
}

Input

Field	Type	Default	Notes
domain	string	drug	drug, device, food or tobacco.
endpoint	string	event	event, enforcement, label, recall or problem.
productNames	array	[]	One query per name, run against the endpoint's own product field. Empty means the whole dataset.
searchQuery	string	""	Raw openFDA search expression, ANDed with the other filters.
fromDate	string	""	YYYY-MM-DD or YYYYMMDD, on the endpoint's own date field.
toDate	string	""	Same format, upper bound.
maxResults	integer	100	Cap per product name. Hard ceiling 25,100.
apiKey	string (secret)	""	Optional openFDA key.

Only these nine domain and endpoint pairs exist. Anything else stops the run immediately with an error listing the valid ones:

Dataset	Date field used for fromDate/toDate	Product field used for productNames
drug/event	receivedate	patient.drug.medicinalproduct
drug/enforcement	report_date	product_description
drug/label	effective_time	openfda.brand_name
device/event	date_received	device.brand_name
device/enforcement	report_date	product_description
device/recall	event_date_initiated	product_description
food/event	date_created	products.name_brand
food/enforcement	report_date	product_description
tobacco/problem	date_submitted	tobacco_products

Results are always sorted newest first on the dataset's date field.

Output example

{
 "domain": "drug",
 "endpoint": "event",
 "recordId": "26940509",
 "safetyReportId": "26940509",
 "productName": "METFORMIN",
 "genericName": "METFORMIN",
 "manufacturer": "Mylan Pharmaceuticals Inc.",
 "reactions": [
 "Adrenal insufficiency",
 "Hepatocellular injury",
 "Tachycardia",
 "Abdominal pain",
 "Nausea",
 "Vomiting",
 "Diarrhoea",
 "Muscular weakness"
 ],
 "seriousness": "Hospitalization; Other serious",
 "patientAge": "27 years",
 "patientSex": "Male",
 "receiveDate": "2026-06-30",
 "reportDate": "2026-07-09",
 "outcome": ["Recovered/resolved", "Not recovered/not resolved"],
 "recallReason": null,
 "classification": null,
 "recallNumber": null,
 "status": null,
 "recallingFirm": null,
 "productDescription": null,
 "country": "US",
 "state": null,
 "city": null,
 "narrative": "Diabetes mellitus inadequate control",
 "productQuery": "metformin",
 "scrapedAt": "2026-08-06T11:49:45.177Z"
}

A drug/enforcement record fills the recall side of the same shape instead:

{
 "domain": "drug",
 "endpoint": "enforcement",
 "recordId": "D-0390-2026",
 "productName": "IBUPROFEN",
 "manufacturer": "Sun Pharmaceutical Industries, Inc.",
 "receiveDate": "2026-03-02",
 "reportDate": "2026-03-25",
 "recallReason": "Presence of foreign substance: the firm received complaints for a gel-like mass and black particles in the product.",
 "classification": "Class II",
 "recallNumber": "D-0390-2026",
 "status": "Ongoing",
 "recallingFirm": "STRIDES PHARMA INC",
 "productDescription": "Children's Ibuprofen Oral Suspension, USP, 100 mg per 5mL, 4 FL OZ (120 mL) bottle .",
 "country": "United States",
 "state": "NJ",
 "city": "Bridgewater",
 "narrative": "U.S.A. Nationwide"
}

Writing a search expression

searchQuery is passed through to openFDA's Lucene-style syntax and ANDed with the product and date filters. Useful patterns:

Goal	Expression
One reaction term	patient.reaction.reactionmeddrapt:"nausea"
Serious reports only	serious:1
Reports where the patient died	patient.reaction.reactionoutcome:5
One manufacturer	patient.drug.openfda.manufacturer_name:"Pfizer"
Class I recalls	classification:"Class I"
Recalls in one state	state:"CA"
Two terms as an OR	patient.reaction.reactionmeddrapt:("nausea" OR "vomiting")

Field names differ between datasets. The openFDA field reference lists every searchable field per endpoint.

API example

curl -X POST "https://api.apify.com/v2/acts/arman-bd~openfda-adverse-events-scraper/run-sync-get-dataset-items?token=YOUR_TOKEN" \
 -H "Content-Type: application/json" \
 -d '{
 "domain": "drug",
 "endpoint": "event",
 "productNames": ["ibuprofen"],
 "fromDate": "2025-01-01",
 "maxResults": 500
 }'

JavaScript example

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: 'YOUR_TOKEN' });
const run = await client.actor('arman-bd/openfda-adverse-events-scraper').call({
 domain: 'drug',
 endpoint: 'enforcement',
 searchQuery: 'classification:"Class I"',
 fromDate: '2025-01-01',
 maxResults: 1000,
});

const { items } = await client.dataset(run.defaultDatasetId).listItems();
const byFirm = {};
for (const r of items) byFirm[r.recallingFirm] = (byFirm[r.recallingFirm] ?? 0) + 1;
console.log(Object.entries(byFirm).sort((a, b) => b[1] - a[1]).slice(0, 10));

Notes

• 25,000 records per query is the ceiling. openFDA refuses any skip above 25,000, so split a bigger pull into date windows, or into several product names, to get past it.
• A query that matches nothing comes back as a 404. openFDA returns NOT_FOUND rather than an empty list, and the Actor treats that as zero results rather than an error and moves on.
• Codes are decoded, not passed through. FAERS ships sex, age unit, reaction outcome and seriousness as numeric codes, including a seriousness flag where "2" means no, which is easy to misread as a set flag. All four are decoded to text.
• The searched product is the one reported. A FAERS report lists every drug the patient was taking, so the record picks the drug matching your productNames entry and falls back to the reporter's designated suspect drug.
• Rate limits are per IP. Unauthenticated you get 240 requests per minute and 1,000 per day. With a free key it is 240 per minute and 120,000 per day. One page is 100 records, so a 5,000-record pull is 50 requests.
• openFDA is slow. Some device endpoints take four seconds per request, so the request timeout is 60 seconds and the default run timeout is 30 minutes.
• Transient errors get retried. 429 and 5xx responses get three attempts with linear backoff, and one failing product name does not stop the others.

FAQ

Do I need an API key? No, but a free key from open.fda.gov raises the daily cap from 1,000 to 120,000 requests. Paste it into apiKey, which is stored as a secret.

Do I need a proxy? No. Proxy configuration is not required to run this Actor.

Is this data validated? No, and the FDA says so explicitly. Adverse event reports are voluntary submissions, and a report does not establish that the product caused the effect. Duplicate reports of the same case are common, so dedupe on safetyReportId before counting.

Why did a search return nothing? Field names are dataset-specific. patient.drug.medicinalproduct exists in drug/event and nowhere else. Check the endpoint table above and the openFDA field reference.

How current is it? FAERS is refreshed quarterly with rolling additions, and enforcement reports update weekly. The meta.last_updated date openFDA reports is typically within a week or two of today.

Can I schedule it? Yes. Run daily with a fromDate a few days back and dedupe on recordId to build an incremental store.