Data & Transparency Gaps in Animal Research

The most consequential transparency failures are rarely about the absence of any data. They are about missing granularity, timeliness, definitions, and auditability — the properties required to independently verify what governments and regulated entities claim. In laboratory animal research, this pattern produces a system that counts bodies but cannot measure burden, tracks species but not breeds, and publishes aggregate statistics while keeping study-level details behind confidential dossiers and redacted FOIA responses.

120K-250K

estimated global dogs/year

In scientific procedures; before hidden categories

Source: Adami et al. 2022; Taylor & Alvarez 2020

~75%

of countries unreported

142 of 179 countries publish no animal-use data

Source: Taylor & Alvarez (2020)

countries track breed use

UK only tracks breed at procedure level

Source: Home Office 2024

142

countries with no data

Only 37 publish national statistics

Source: Peer-reviewed estimate

The Map of Ignorance

“Dogs used in research” is not one number. It varies by what is counted, where, and how. The main denominators — animals used, procedures performed, animals held or bred — are not interchangeable. Converting between them requires raw, animal-level participation histories that no public system provides. What follows is a systematic accounting of what we know, what plausibly exists but cannot be accessed, and what no system currently records. This is not a gap analysis. It is a map of ignorance.

Known (Publicly Reported, Definition-Bound)

+ US research facilities (FY2024): 42,880 dogs across held + used bins; 40,495 in “used” categories (C+D+E)

+ EU + Norway (2022): 8,709 dogs “used for the first time,” down from 14,802 in 2018

+ Great Britain (2024): 2,646 experimental procedures on dogs, 71% for regulatory purposes

+ Canada CCAC-certified (2024): 9,252 dogs (explicitly incomplete national coverage)

+ UK breed split: 2,488 beagles of 2,646 dog procedures (94%) — the only jurisdiction reporting breed

Unknown (Data Plausibly Exists, Not Publicly Joinable)

? Dog use across 142 countries that do not publish standardized national returns

? Industry vs. academic split: UK shows 71% regulatory, but no clean sector breakout elsewhere

? Confidential study dossiers submitted to regulators — the public sees only annual aggregates

? China and Japan — the largest estimated research economies with no standardized national returns

? CRO-supplier relationships: no CRO names its beagle supplier in SEC filings or public disclosures

Unknowable (Not Systematically Recorded in Any Public Dataset)

× Supply-chain attrition: dogs bred but never entering a reportable procedure or facility inventory

× Dogs killed solely for tissue use (excluded from EU “procedure” definition)

× UK excluded populations: non-GA breeding and sentinel/support animals outside headline statistics

× US excluded categories: VCPR clinical trials and field studies not reported in annual summaries

× Cumulative lifetime suffering across multiple studies and reuses — no system records this

× Failed or aborted studies: animals counted administratively but studies never published or registered

The 120,000–250,000 Global Estimate

Two independent constraints anchor the range for true global dog use in scientific procedures:

Floor: Multi-Country Dog Tally

A peer-reviewed review of 36 countries that communicated data found 112,265 dogs used. This is a lower bound: it excludes non-reporting countries and does not resolve scope exclusions within reporting systems.

Envelope: Proportional Extrapolation

Dogs represent ~0.10% of animals in EU data and ~0.25% in Canadian data. Applied to an estimated 79.9 million animals globally (2015), this yields ~80,000–240,000 dogs/year. The low end is too low once non-reporting countries and definitional exclusions are acknowledged.

Working range: 120,000–250,000 dogs/year in procedures. This does not include dogs bred but never entering reportable use, dogs in sentinel/support roles, or dogs in excluded clinical/field contexts. Those populations are additional uncertainty layered on top of procedure-based estimates.

Data Gap

The “true number” of dogs used globally is structurally unknowable from current public data systems. Only 37 countries publish national statistics. The rest — 142 countries, including major research economies like China and Japan — are modeled estimates at best, complete blanks at worst. What is published even in high-transparency systems answers “how many dogs?” but not “in what studies, for how long, with what cumulative harms, and funded by whom?”

USDA Data: What It Captures and What It Misses

What USDA Reports

• Aggregate annual counts by species and pain category (Columns B/C/D/E)

• Inspection reports via APHIS Public Search Tool

• Enforcement actions, civil penalties, and final adjudications

• Facility registration/licensing status

What USDA Does Not Report

• No breed — “dogs” only, no breed field on Form 7023

• No study type — what the dogs were used for

• No post-study outcome — euthanized, adopted, or reused

• No duration — how long animals were in distress

• No cumulative burden — multiple studies collapsed to one category

• No per-facility time series — public data is aggregate

Column E Underreporting

Column E (pain/distress where drugs would adversely affect the study and are withheld) is the most scrutinized category. Only 410 dogs were reported in Column E nationally in FY2024. The reporting packet anticipates confidentiality: Column E attachments may require both a clean and a redacted copy, with the redacted version posted publicly. Prospective reporting — where facilities classify all animals in the highest anticipated category — can paradoxically place animals in higher categories than necessary, while masking the actual distribution of suffering.

The Classification Problem

US pain categories (C/D/E) are analgesia-availability bins, not severity or duration accounts. Column D (“more than momentary pain/distress with appropriate drugs”) does not distinguish a blood draw from a 12-month chronic toxicity study. Animals in multiple studies per year are counted once in the highest category — erasing how many procedures occurred, how long harms lasted, and whether harm was repeated. The EU severity system (mild/moderate/severe/non-recovery) is more granular but still aggregated at the public-statistics layer.

FOIA Battles Over Laboratory Animal Records

Public access to federal animal-welfare records is the primary scalable accountability mechanism for laboratory animal oversight. The modern history of that access is defined by removal, litigation, congressional intervention, and ongoing structural gaps.

The 2017 APHIS Records Purge

On February 3, 2017, APHIS removed thousands of animal-welfare records from its website — inspection reports, enforcement actions, research facility annual reports, and regulatory correspondence. The stated rationale was privacy/redaction concerns. The practical effect: advocacy groups shifted from “search the database” to “file many FOIA requests,” increasing delay and cost and making time-sensitive oversight nearly impossible.

Ninth Circuit: ALDF v. USDA (2019)

Animal Legal Defense Fund led a coalition lawsuit. The Ninth Circuit held that FOIA's “reading room” provision (5 U.S.C. § 552(a)(2)) authorizes courts to enjoin agencies that fail to proactively post frequently requested records online. Standing was based on concrete harm: organizations that used the databases daily experienced real operational harm when records disappeared.

Congressional Intervention: Section 788 (2019)

While litigation was ongoing, Congress enacted 7 U.S.C. § 2146a, mandating restoration of the searchable database as it existed on January 30, 2017, plus all content generated since. The statute requires posting of final inspection reports, animal inventories, enforcement records, and research facility annual reports — with redactions limited to FOIA Exemption 4 (confidential business information) and signatures.

Courtesy Visits Struck Down (2025)

In ASPCA v. APHIS (D.D.C., Oct. 29, 2025), the court held that APHIS's “courtesy visits” policy — which instructed inspectors not to record observed violations during educational visits — violated Congress's Section 756 funding restriction prohibiting non-recording of observed AWA violations. The rule was set aside. Inspectors must now record all observed violations, restoring these visits to the public compliance record.

What Remains Missing

• Whether official warning letters and pre-litigation settlements are treated as “final enforcement records” for posting purposes

• Retention windows: the statute emphasizes current year plus three prior years, creating older-record gaps

• Complaint handling documentation, follow-up records, and internal inspection artifacts beyond posted reports

• The 2025 OIG audit found 57% of inspections missed required elements, 95% did not meet frequency guidelines, and 80% of sampled breeders had continuing noncompliance

Key Finding

The strategic function of records access is pattern detection. Rise for Animals used USDA records to identify that Ridglan Farms had the same VMO (Scott Welch) preparing every inspection report across 28 visits, with a 4% violation rate — while oversight visits found violations 100% of the time. For Marshall BioResources, publicly available inspection reports enabled construction of a longitudinal violation history. These analyses are only possible because of the transparency infrastructure that litigation and congressional action restored.

International Transparency Gaps

Transparency in animal research varies dramatically by jurisdiction. The structural problem is not just “some countries publish less” — it is that reporting systems use incompatible definitions, making cross-border comparison a modeling exercise.

Dimension	US	EU	UK	Canada
Unit of count	Animals held + used	Animals used (first time)	Procedures	Animals used
Period	Fiscal year (Oct-Sep)	Calendar year	Calendar year	Calendar year
Pain/severity	Columns B/C/D/E (analgesia)	Mild/Moderate/Severe/NR	Actual severity	Invasiveness categories
Breed tracking	No	No	Yes (partial)	No
Scope exclusions	Field studies, VCPR trials	Killing for tissue only	Non-GA breeding, sentinels	Non-CCAC institutions
Public access	APHIS Search Tool + FOIA	ALURES database	Home Office annual report	CCAC annual data report

The largest estimated research economies with no standardized public reporting — China and Japan — represent the single biggest gap. Global estimation work for 2015 required modeling for 142 countries because only 37 publish any national statistics.

What We Don't Know

Breed-Level Counts

No country except the UK reports breed. The claim that 95%+ of research dogs are beagles comes from industry sources and expert estimates, not regulatory data. The USDA form has no breed field. The EU ALURES schema reports “dogs” as a species line.

Facility-Level Data

US published data is aggregate. Per-facility time series showing how dog populations change year-over-year at individual labs and breeders are not published. FOIA is required to reconstruct any facility-level picture.

Outcome Data

No jurisdiction systematically reports what happens to dogs after studies. Euthanized, adopted, reused, transferred — these outcomes are not captured in any public dataset. The UK rehoming rate of 0.4% was compiled by independent researchers, not the government.

Global Totals

Reconstructing ~60,000+ dogs from high-transparency systems (US + EU + UK + Canada) requires reconciling mismatched years and incompatible denominators. The remaining 142 countries contribute an unknown additional amount that can only be estimated through proportional modeling.

What Better Data Would Enable

Breed-level tracking

Quantify beagle use directly. Move from '42,880 dogs' to 'X beagles, Y other breeds.' End reliance on industry self-reporting.

Cumulative burden tracking

Replace single-category pain assignments with lifetime exposure accounts. Answer: 'How much suffering did this animal experience?'

Rehoming outcome data

Measure whether beagle freedom laws change outcomes. Compare actual adoption rates against the 0.4% UK baseline.

Breeder mortality reporting

Expose the hidden population: dogs bred but never reaching labs. Enable welfare oversight of the production system.

Inspector rotation data

Detect enforcement capture before it becomes systemic. Flag patterns like 28 visits, same inspector, 4% violation rate.

Global harmonized counts

Enable trend analysis across jurisdictions. Currently impossible due to incompatible units, scope exclusions, and severity definitions.

None of these require new technology. They require political decisions to collect and publish information that already exists within regulated facilities. The data production infrastructure is there. The publication mandate is not.

Why This Matters

Transparency is not an abstract value. It is the precondition for every other reform. Without data on breed use, rehoming outcomes, and enforcement quality, the public cannot evaluate whether the system is working — and reformers cannot identify where it is failing. The current system produces the appearance of accountability (searchable databases, annual statistics, inspection reports) without the substance of transparency (breed-level data, cumulative burden measures, per-facility time series, post-study outcomes, and global comparability).

Sources

1. USDA APHIS FY2024 research facility annual summary: 42,880 dogs (2,385 held/not used; 27,909 Column C; 12,176 Column D; 410 Column E).

2. EU + Norway statistical report on animals used for scientific purposes, 2022: 8,709 dogs used for first time (of 8,385,397 total animals).

3. Home Office, Annual Statistics of Scientific Procedures on Living Animals, Great Britain 2024: 2,646 dog procedures, 71% regulatory.

4. CCAC Animal Data Report 2024: 9,252 dogs at CCAC-certified institutions (9,252 of 3,706,907 total).

5. Taylor & Alvarez (2020), “Estimation of worldwide laboratory animal use in 2015”: 37 reporting countries, 142 modeled. PMID: 32090616.

6. Adami et al. (2022), PMC9252265: 112,265 dogs across 36 reporting countries; global estimate methodology.

7. APHIS Form 7023 / FY25 annual report packet: pain category definitions, Column E redaction procedures, scope exclusions.

8. EU Directive 2010/63/EU, Annex VIII: severity classification criteria including cumulative suffering within procedures.

9. Animal Legal Defense Fund v. USDA, No. 17-16858 (9th Cir. 2019): reading-room enforcement under FOIA.

10. Pub. L. 116-94, Section 788 / 7 U.S.C. § 2146a: Congressional searchable database mandate.

11. ASPCA v. APHIS, No. 1:21-cv-01600-CRC (D.D.C. Oct. 29, 2025): courtesy visits non-recording rule set aside.

12. USDA OIG Report 33601-0001-22 (Feb. 3, 2025): 80% continuing noncompliance, 95% inspection frequency failures, 57% missed elements.

13. Rise for Animals analysis of Ridglan Farms USDA records: single-inspector pattern across 28 visits.

14. Rise for Animals, Marshall BioResources fact sheet (2020): violation history compiled from USDA/APHIS inspection reports.

15. APHIS records removal: Washington Post, Feb. 3, 2017; restoration bulletin, Feb. 18, 2020.

16. OACU/NIH guidance on USDA annual reporting: within-year multiplicity, cumulative burden limitations.

EU Directive 2010/63

The most transparent system — and its limits

Animal Welfare Act

The US baseline and its structural exclusions

AWA Enforcement

Inspection gaps, OIG audits, and the same-inspector problem

Global Trade

Country-by-country data and cross-border flows