OpenFDA Animal & Veterinary

Overview

openFDA provides free, public REST API access to FDA CVM (Center for Veterinary Medicine) databases of animal drug adverse events, approved animal products, and veterinary device data. This skill covers API endpoints, query syntax, searchable fields, rate limits, and SDK integration for veterinary AI applications.

When to Use

• User queries adverse event frequency for a veterinary drug
• User searches for FDA-approved animal drug products (NADA database)
• User researches safety signals or patterns in veterinary pharmacovigilance
• User programmatically integrates FDA data into a VetClaw application
• Keywords: openFDA, FDA CVM, adverse event, animal drug, NADA, API, veterinary device, safety signal

API Endpoints

Animal & Veterinary Events (Adverse Event Reports):

https://api.fda.gov/animalandveterinary/event.json

Searchable fields: animal species, drug name, adverse reaction, outcome, date reported

Approved Animal Drug Products (NADA Database):

https://api.fda.gov/animalandveterinary/approved_animal_drug_applications.json

Searchable fields: trade name, generic name, applicant, approval date, species

Veterinary Device Events:

https://api.fda.gov/animalandveterinary/device_event.json

Searchable fields: device type, malfunction, injury, manufacturer

Query Syntax

Basic GET Request:

https://api.fda.gov/animalandveterinary/event.json?search=FIELD:VALUE&limit=10&skip=0

Common Parameters:

• search: Query expression (field-value pairs connected with AND/OR)
• limit: Results per page (1-1000, default 20)
• skip: Offset for pagination (for results >1000)
• count: Aggregate count by field instead of returning individual records
• sort: Order results (field name)

Boolean Operators:

• AND: Both conditions must match
• OR: Either condition matches
• NOT: Exclude results matching condition
• Parentheses for complex expressions: (field1:value1 OR field1:value2) AND field2:value3

Field Query Examples:

drug.name:"amoxicillin" AND animal.species:"canine"
reaction.veddra_term_name:"hepatotoxicity"
date_received:[20200101 TO 20241231]
outcome:"Fatal"

Searchable Fields - Adverse Events

Drug Fields:

• drug.name - Generic or brand name
• drug.dose_value - Numerical dose
• drug.dose_unit - Unit (mg, mL, etc.)
• drug.route - Route of administration (PO, IV, IM, topical, etc.)
• drug.frequency - Dosing frequency

Animal Fields:

• animal.species - Species affected (canine, feline, equine, bovine, avian, etc.)
• animal.breed - Breed (searchable for canine predisposition studies)
• animal.age_value - Numerical age
• animal.age_unit - Unit (year, month, week, day)
• animal.sex - Male, female, unknown
• animal.weight_value - Weight (for dose-response analysis)

Reaction Fields:

• reaction.veddra_term_name - VEDDRA (Veterinary Dictionary for Drug Event Reports and Analysis) standardized term
• reaction.outcome - Death, Euthanasia, Serious, Non-serious, Recovered, Recovering, Unknown
• reaction.severity - Severity classification

Report Fields:

• date_received - Date FDA received report (ISO 8601 format: YYYYMMDD)
• report_id - Unique FDA adverse event ID
• serious - Boolean (whether report indicates serious adverse outcome)

Common Veterinary VEDDRA Terms

Rate Limits

• Without API key: 240 requests per minute, 1,000 requests per day
• With API key (free): 240 requests per minute, 120,000 requests per day
• Rate Limit Headers: Response includes X-Rate-Limit-Limit, X-Rate-Limit-Remaining
• Exceeding Limits: Server returns HTTP 429 (Too Many Requests); implement exponential backoff retry

Register for API Key: https://open.fda.gov/apis/authentication/

VetClaw SDK Integration

VetClaw includes reference clients for this API. See sdk/ in the repo root.

Python (sdk/python/openfda_vet.py):

from openfda_vet import OpenFDAVet

client = OpenFDAVet(api_key="YOUR_KEY")

# Search adverse events by species and drug
events = client.search(species="Dog", drug="Amoxicillin", limit=100)

# Count adverse events by reaction for a specific drug
reactions = client.top_reactions("Cat", drug="Methimazole")

# Count adverse events by outcome
outcomes = client.count("reaction.outcome", species="Dog", drug="Ivermectin")

TypeScript (sdk/typescript/openfda-vet.ts):

import { OpenFDAVet } from "./openfda-vet";

const client = new OpenFDAVet("YOUR_KEY");

// Search adverse events
const events = await client.search({ species: "Dog", drug: "Amoxicillin", limit: 100 });

// Count top reactions
const reactions = await client.topReactions("Cat", "Methimazole");

// Paginate through results
for await (const page of client.paginate({ species: "Dog", drug: "Ivermectin" })) {
  for (const event of page) console.log(event.safetyreportid);
}

Practical Query Examples

Query 1: MDR1 Sensitivity to Ivermectin

Search for ivermectin adverse events in herding breeds (known MDR1 mutation risk):
drug.name:"ivermectin"
  AND animal.breed:("Collie" OR "Australian Shepherd" OR "Sheltie")
  AND reaction.outcome:("Death" OR "Serious" OR "Euthanasia")
Limit: 100

Query 2: Acetaminophen in Cats (Lethal Combination)

drug.name:"acetaminophen" AND animal.species:"feline"
Expected outcome: High rate of hepatotoxicity, death (acetaminophen is lethal to cats)

Query 3: Drug Withdrawal Time Safety

Drug: penicillin, Animal: bovine, Outcome: Recovery with residue concerns
Verify no unusual severe outcomes reported; extract timeline to recovery

Query 4: Breed Predisposition to Drug Reaction

reaction.veddra_term_name:"seizure"
  AND (drug.name:"phenobarbital" OR drug.name:"levetiracetam")
  AND animal.species:"canine"
Group by breed to identify predispositions

Query 5: Time-Based Safety Trend

drug.name:"enalapril"
  AND animal.species:"feline"
  AND date_received:[20150101 TO 20241231]
Count by year to track safety signal emergence over time

Response Structure

Adverse Event Record Example:

{
  "report_id": "123456789",
  "date_received": "2024-03-15",
  "animal": {
    "species": "canine",
    "breed": "Labrador Retriever",
    "age_value": 5,
    "age_unit": "year",
    "sex": "female",
    "weight_value": 30,
    "weight_unit": "kg"
  },
  "drug": {
    "name": "amoxicillin",
    "dose_value": 250,
    "dose_unit": "mg",
    "route": "oral",
    "frequency": "twice daily"
  },
  "reaction": {
    "veddra_term_name": "vomiting",
    "outcome": "Recovered",
    "severity": "moderate"
  },
  "serious": false
}

Use Cases in Veterinary AI

1. Safety Signal Detection: Aggregate adverse events by drug-species-breed to identify unexpected patterns (e.g., increased hepatotoxicity in specific breed using drug X)

2. Pharmacovigilance Monitoring: Track trends in serious adverse events over time; alert veterinarians to emerging safety signals

3. Breed-Specific Dosing Refinement: Identify adverse event clusters by breed; inform precision medicine protocols

4. Drug Interaction Warnings: Cross-reference adverse events to identify potential drug-drug interactions not captured in single-drug studies

5. Clinical Decision Support: When recommending a drug, query recent adverse events to provide evidence-based safety profile to veterinarian

Limitations

• Reporting Bias: Adverse events are voluntarily reported; underreporting of minor side effects, overreporting of serious outcomes
• Incomplete Data: Many adverse event reports lack complete signalment (breed, age), dosing information, or outcome details
• Temporal Lag: FDA data is typically 6-12 months behind current clinical use; emerging signals may not be captured
• Causality Uncertainty: Reported adverse event may not be causally related to drug (confounding variables, concurrent medications)
• Approved Drug Products Database: Limited to FDA-approved products; extralabel use (common in veterinary medicine) not captured
• This skill is reference only: Data interpretation requires clinical expertise; not a substitute for veterinary consultation

Sources

• OpenFDA Documentation: https://open.fda.gov/apis/
• OpenFDA Animal & Veterinary Guide: https://open.fda.gov/apis/animalandveterinary/
• FDA CVM: https://www.fda.gov/animal-veterinary
• VEDDRA Terminology: Available in FDA CVM databases
• VetClaw SDK Documentation: Reference sdk/ folder for Python/TypeScript client libraries

Rate Limit Best Practices

• Implement exponential backoff (wait 2^attempt seconds if rate limited)
• Cache results locally to minimize API calls
• Use count endpoint instead of fetching all records if only aggregate data needed
• Batch queries by date range to reduce total requests
• Monitor X-Rate-Limit-Remaining header; pause if approaching limit