How to run a clinical study for a pet supplement

A clinical study for a pet supplement is a planned comparison: animals that get the product against animals that get something else, usually a placebo, with the outcome measured the same way in both groups by people who do not know which animal got which. It needs a written protocol, an ethics review, each owner's informed consent, records that show who entered what and when, and a database that is locked before anyone looks at the result. A company can run one with its own team, a veterinarian who examines the animals, and owners recording at home. What follows is what that takes, with the rule behind each part.

1. Decide what the study has to show

Start from the claim on the label and turn it into one question with a measurable answer. "Supports joint health" becomes, for example, a change in a validated owner-completed mobility questionnaire after twelve weeks, compared with placebo. One primary outcome, chosen before the study starts; a comparator, usually a placebo in the same form; a duration your veterinarian and statistician agree the effect needs; and a number of animals the statistician calculates from the expected difference and the variability of the measure. These four decisions shape everything else, so make them first and write them down.

2. Find out which rules apply

The rules depend on what the product is and where the data are going.

• Intended use decides whether a product is a drug. FDA's regulation on intended use, 21 CFR 201.128, says intent can be shown by the circumstances in which a product is offered or used. A study designed to show that a product treats, prevents or mitigates a disease is one of those circumstances. If the product is an unapproved new animal drug, which includes peptides and compounded products, 21 CFR 511.1 requires a notice to FDA (the INAD file) before the product is shipped for the study, with investigational labelling, shipment records, qualified investigators and monitoring. Settle this with a regulatory adviser before the first animal is enrolled.
• Data meant for FDA follow VICH GL9. 21 CFR 514.117(b)(2) asks for a "standard of conduct acceptable to FDA"; FDA's Guidance #85 (VICH GL9) is that standard. GL9 says studies not intended for regulatory support are outside its scope. Even so, CVM's draft Guidance #282 recommends that all studies in client-owned companion animals use an informed consent form and follow good clinical practice.
• Electronic records kept under an FDA records requirement follow 21 CFR Part 11. For an INAD study the records requirements are 511.1(b)(7) and (b)(8), which FDA's 2024 questions-and-answers guidance names. An observational study with nothing assigned is not under Part 11, though good practice and state veterinary records law still apply.
• Outside the United States the trial itself may need approval. In the European Union, Regulation (EU) 2019/6, Article 9, requires an application to the competent authority of the member state where the trial takes place.

Signalment asks these questions at setup and shows a readiness checklist; it does not answer them for you.

3. Ethics review and owner consent

GL9 does not name a committee, but it expects procedures that independently assure the study animals are protected and that consent has been obtained (section 2.5). Universities route this through their animal care and use committee; a company without one uses an independent veterinary ethics committee or an academic partner. Get the review before enrollment and keep the approval letter with the study file.

Consent is a documented process by which the owner voluntarily agrees to the animal's participation (GL9 1.15), obtained by the investigator before the animal takes part, and GL9 treats the signed form as raw data. Draft Guidance #282 says what the form should do: reach the owner 12 to 24 hours before enrollment; state that the product is investigational and that safety and effectiveness have not been demonstrated; never suggest the product is safe or effective; say that FDA may inspect the records and how the data will be used; and attach written authority when someone signs for the owner. Consent signed on a phone is fine as long as it is bound to the exact version of the text that was read and a copy stays with the owner.

4. Design it so the result can be believed

• Randomize in blocks and conceal the allocation. Permuted blocks keep the groups balanced as animals enroll; concealment means nobody can predict the next allocation. GL9 asks the protocol to state the randomization and masking methods.
• Blind everyone who records. With coded kits, the owner, the examining veterinarian and the person entering data know only an allocation number. A code is broken only for an emergency that needs the information, with the reason on the record.
• Match the placebo. Same form, same look, same dosing, so the blind holds.
• Measure with instruments that exist. Where a validated owner-completed instrument exists for the condition, use it unchanged. Add the veterinarian's examination at scheduled visits. Write the analysis plan before the blind comes off.
• Keep the owner's part short and daily. The same few questions every evening, under a minute; a missed day stays a gap rather than being filled in later.

5. What has to be on the record

GL9 section 8.3 says raw data must be attributable, original, accurate, contemporaneous and legible; that in an electronic system the record is the raw data; that each entry and change is made under the signature of the person making it; and that changes carry the reason and sit in an audit trail. Part 11 adds, for FDA-regulated records, secure computer-generated time-stamped audit trails that never obscure earlier values (11.10(e)) and signatures that show the signer's name, the time and the meaning of the signature (11.50). The study file therefore holds:

1. The protocol, signed and dated by the investigator and the sponsor, with a version identifier; every amendment signed and dated the same way (GL9 6.3).
2. Each owner's consent as signed, with the version of the text.
3. Eligibility, enrollment and the allocation for each animal.
4. Every administration of the product or placebo, and the receipt, storage and reconciliation of the supplies (GL9 3.2.19 to 3.2.25).
5. Adverse events. 21 CFR 514.3 defines an adverse drug experience as any adverse event associated with the use of the drug, whether or not it is considered drug related; a serious one is fatal, life-threatening, requires professional intervention, or causes abortion, stillbirth, infertility, a congenital anomaly, prolonged or permanent disability, or disfigurement. Write the reporting plan before the first animal is enrolled: for an INAD study the sponsor promptly reports findings that suggest significant hazards to FDA and to all investigators (511.1(b)(8)(ii)); Form FDA 1932a covers unapproved products.
6. Protocol deviations, in a signed and dated statement.
7. The monitor's queries and the site's answers. GL9 3.1.5: the monitor is not the investigator.
8. The lock: when, by whom, and a fingerprint of what the database contained.

6. Running it, week by week

The coordinator enrolls each animal, records consent, and sends the owner the link. The owner's first check-in happens the same evening. Scheduled visits follow the protocol's windows; the coordinator watches for missed check-ins and records a deviation when a window is missed. The monitor reads records against the clinic's source documents and raises queries; the site answers them. A serious event is entered the day it is known and reaches the investigators and the sponsor at once. Supplies stay labelled by allocation number so the clinic never learns the arm.

7. Lock, analyze, show

When the last visit is in and every query is closed, the investigator verifies the records with a signature and the database is locked with a fingerprint. The export is one file per form, a data dictionary, and a bundle with the audit trail; the statistician runs the pre-specified analysis; the blind comes off afterwards. If the data go to FDA, Guidance #287 says the audit trail from an electronic data capture system is submitted with the raw data. Whoever reviews the study, whether a monitor, a retailer's scientific adviser or a regulator, should be able to recompute the records from the export without trusting the software that produced it.

What running it yourself changes

A contract research organization adds project management, monitoring and often its own data system, priced per study. Running the study in-house moves the cost to your team's time, the veterinarian's visits, the product and placebo, a statistician, and a data system. Signalment is priced per workspace, not per study: a free pilot plan, then $149 or $499 a month. One study run end to end, from the first form to the locked export, is on film at signalment.app/sponsors.

What the software does, and does not do

Signalment holds the protocol and consent versions, the forms, the owner's daily check-ins, the randomization and blinding, the adverse event form with the 514.3 seriousness criteria, the queries, the signatures, the audit trail, the lock and the exports. It does not write the protocol, review the ethics, decide the product's regulatory status, file an INAD or report an adverse event; those stay with the sponsor and the investigators. For what the data system itself should do, see Veterinary EDC, explained.

Sources

• 21 CFR 201.128, intended use: https://www.ecfr.gov/current/title-21/section-201.128
• 21 CFR 511.1, new animal drugs for investigational use: https://www.ecfr.gov/current/title-21/section-511.1
• 21 CFR 514.117, adequate and well-controlled studies: https://www.ecfr.gov/current/title-21/section-514.117
• 21 CFR 514.3, definitions: https://www.ecfr.gov/current/title-21/section-514.3
• 21 CFR Part 11, electronic records and signatures: https://www.ecfr.gov/current/title-21/part-11
• FDA Guidance #85, VICH GL9, Good Clinical Practice: https://www.fda.gov/media/70333/download
• FDA draft Guidance #282, informed consent forms for studies that enroll client-owned companion animals: https://www.fda.gov/media/172056/download
• FDA Guidance #287, raw data for safety and effectiveness studies: https://www.fda.gov/media/177901/download
• FDA, Electronic Systems, Electronic Records, and Electronic Signatures in Clinical Investigations: Questions and Answers (2024): https://www.fda.gov/media/166215/download