Skip to main content
v2026.11,986 entries · CC-BY 4.0

Learned Intermediary Doctrine: Restatement (Third) of Torts § 6(d), Medical Devices, Pharmaceuticals, and Clinical AI/CDS

An authoritative legal and clinical risk reference on the learned intermediary doctrine under the Restatement (Third) of Torts: Products Liability § 6(d), examining manufacturer failure-to-warn duties for pharmaceuticals, Class II/III medical devices, and clinical AI/CDS software, recognized exceptions, and proximate causation rules.

Written and maintained by CASRAI Editorial Board

Published

On this page

The learned intermediary doctrine is a foundational principle of U.S. tort and medical products liability law holding that a manufacturer of a prescription drug or prescription medical device discharges its legal duty to warn by providing adequate warnings and instructions about foreseeable risks to the prescribing or treating physician—the “learned intermediary”—rather than directly to the ultimate patient consumer. First articulated using the phrase “learned intermediary” by the U.S. Court of Appeals for the Eighth Circuit in Sterling Drug, Inc. v. Cornish (370 F.2d 82, 8th Cir. 1966) and adopted across nearly all U.S. state jurisdictions, the doctrine is formally codified in the American Law Institute’s Restatement (Third) of Torts: Products Liability § 6(d).

In contemporary clinical practice and biomedical engineering, the learned intermediary doctrine governs risk allocation across three interconnected domains: (1) pharmaceutical package inserts and FDA professional labeling (21 CFR § 201.56–201.57), (2) implantable and diagnostic medical device Instructions for Use (IFU), and (3) Software as a Medical Device (SaMD) and artificial intelligence Clinical Decision Support (CDS) systems, where technology errors and omissions (Tech E&O) and medical professional liability intersect.

Codification in the Restatement (Third) of Torts: Products Liability § 6(d)

Under general products liability law, a manufacturer of a consumer product owes a duty to warn the end user directly of non-obvious hazards. However, prescription drugs and medical devices are legally accessible only through the order of a licensed healthcare practitioner. Section 6(d) of the Restatement (Third) of Torts: Products Liability (1998) defines when a prescription drug or medical device is defective due to inadequate instructions or warnings:

“A prescription drug or medical device is not reasonably safe due to inadequate instructions or warnings if reasonable instructions or warnings regarding foreseeable risks of harm are not provided to:
(1) prescribing and other health-care providers who are in a position to reduce the risks of harm in accordance with the instructions or warnings; or
(2) the patient when the manufacturer knows or has reason to know that health-care providers will not be in a position to reduce the risks of harm in accordance with the instructions or warnings.”

Four Policy Rationales Underlying the Doctrine

  1. Clinical Judgment and Risk-Benefit Balancing: Only a trained clinician who knows the patient’s medical history, comorbidities, and concurrent medications can weigh the statistical adverse-event profile of a drug or device against the clinical risk of untreated disease.
  2. Preservation of the Physician-Patient Relationship: Direct manufacturer warnings written in complex technical toxicology or engineering language could confuse patients, induce non-adherence to life-saving therapy, or interfere with the treating clinician’s informed consent dialogue.
  3. Infeasibility of Direct Manufacturer-to-Patient Communication: For hospital-administered parenterals, surgical implants, or EHR-embedded diagnostic algorithms, the manufacturer has no direct point-of-sale contact with the patient.
  4. Alignment with Informed Consent Tort Law: While the manufacturer owes a duty under products liability law to warn the physician of product risks, the physician owes an independent fiduciary and professional duty under medical informed consent law to disclose material risks, benefits, and alternatives to the patient.

The Two-Prong Proximate Causation Standard in Failure-to-Warn Litigation

When an injured patient brings a failure-to-warn products liability claim against a pharmaceutical, medical device, or clinical software manufacturer, the learned intermediary doctrine imposes a rigorous two-prong proximate causation burden on the plaintiff:

  • Prong 1 (Warning Inadequacy): The plaintiff must prove that the manufacturer’s professional labeling, package insert, Dear Healthcare Provider (DHCP) letter, or device IFU failed to adequately warn the prescribing/treating clinician of a known or reasonably scientifically knowable risk.
  • Prong 2 (Prescribing Decision Alteration): The plaintiff must prove that had the manufacturer provided an adequate warning, the specific treating clinician would have altered their clinical decision—either by not prescribing/implanting the product, by selecting an alternative therapy, or by implementing specific physiological monitoring that would have prevented or mitigated the injury.

Consequently, if the treating physician testifies in deposition that they were already independently aware of the specific risk from medical literature and chose to proceed based on clinical benefit-risk judgment—or that they never read the manufacturer’s IFU—the causal chain is broken and the manufacturer is generally entitled to summary judgment.

Recognized Judicial Exceptions to the Learned Intermediary Doctrine

Under Restatement (Third) of Torts § 6(d)(2) and state supreme court jurisprudence, the learned intermediary defense does not apply in four specific contexts where the clinical intermediary is absent, bypassed, or overridden by federal regulation:

Exception Category Leading Legal Authority / Regulation Scope & Rationale
1. Mass Immunization Programs Davis v. Wyeth Laboratories, Inc., 399 F.2d 121 (9th Cir. 1968); Reyes v. Wyeth Laboratories, 498 F.2d 1264 (5th Cir. 1974); National Childhood Vaccine Injury Act (42 U.S.C. § 300aa-22) When vaccines are administered in assembly-line public health clinics without an individualized physician evaluation of the recipient, the manufacturer (or public health program via CDC Vaccine Information Statements) must warn the recipient directly.
2. Oral Contraceptives & FDA-Mandated Patient Labeling MacDonald v. Ortho Pharmaceutical Corp., 475 N.E.2d 65 (Mass. 1985); 21 CFR § 310.501 (Patient Package Inserts) & 21 CFR Part 208 (FDA Medication Guides) Where patients play an active role in elective product selection with periodic refills and FDA regulations mandate direct patient package inserts or Medication Guides for serious and significant public health concerns.
3. Direct-to-Consumer (DTC) Mass Marketing Perez v. Wyeth Laboratories Inc., 734 A.2d 1245 (N.J. 1999) (minority rule; rejected by several other states such as West Virginia after repealing its carve-out) In jurisdictions following Perez, when a manufacturer bypasses the physician-patient relationship by mass-marketing prescription products directly to consumers via television or digital media, it owes a parallel duty to warn consumers directly.
4. Over-Warned / Non-Intervention Contexts Restatement (Third) of Torts § 6(d)(2) (manufacturer knows providers are not in a position to reduce risk) Applies when a product is dispensed or executed autonomously—such as direct-to-consumer wellness wearables or autonomous patient-facing triage apps—where no licensed clinician reviews the output before patient action.

Application to Clinical AI, Black-Box Models, and FDA Non-Device CDS

The deployment of machine learning diagnostics, ambient clinical documentation, and predictive sepsis/deterioration models inside hospital EHRs has created new doctrinal boundaries for the learned intermediary rule and digital health insurance underwriting:

  1. Clinical AI Is the Product, Not the Learned Intermediary: An algorithm or large language model lacks medical licensure, moral agency, and a fiduciary duty of care. Under tort law, the AI/CDS software is treated as a tool or product, while the attending physician or advanced practice provider remains the human learned intermediary.
  2. The “Black-Box” Problem and FDA Section 520(o)(1)(E) Criterion 4: Under Section 3060 of the 21st Century Cures Act (FD&C Act § 520(o)(1)(E)(iii)) and the FDA’s Clinical Decision Support Software final guidance, a software function is excluded from FDA device regulation as Non-Device CDS only if it enables the healthcare professional to “independently review the basis for such recommendations that such software presents so that it is not the intent that such health care professional rely primarily on any of such recommendations.” If a black-box deep learning model outputs a single opaque risk score or treatment directive without disclosing its underlying clinical inputs, logic, validation cohort demographics, and known failure modes, plaintiffs can argue under Restatement § 6(d)(2) that the clinician was placed in a position of uncritical automation bias and could not function as an informed learned intermediary.
  3. Model Cards and IFU Warning Adequacy for SaMD: To preserve learned intermediary protection under § 6(d)(1), SaMD and clinical AI developers provide structured Model Cards and clinician-facing IFUs detailing training dataset exclusion criteria, sensitivity/specificity trade-offs across demographic subgroups, out-of-distribution drift warnings, and explicit contraindications against autonomous unverified use.

Frequently Asked Questions

What is the learned intermediary doctrine?

The learned intermediary doctrine is a products liability rule—codified in the Restatement (Third) of Torts: Products Liability § 6(d)—holding that a manufacturer of a prescription pharmaceutical or medical device fulfills its legal duty to warn of foreseeable risks by providing adequate warnings and instructions to the prescribing or treating healthcare professional (the “learned intermediary”) rather than directly to the patient.

What are the recognized exceptions to the learned intermediary doctrine?

Courts and the Restatement (Third) of Torts § 6(d)(2) recognize specific exceptions where manufacturers owe a direct duty to warn the patient because no individualized physician-patient risk assessment occurs or federal regulations mandate patient labeling: (1) mass immunization clinics (Reyes v. Wyeth Laboratories), (2) oral contraceptives and FDA-mandated patient Medication Guides (21 CFR Part 208), (3) in select jurisdictions (such as New Jersey under Perez v. Wyeth Laboratories), direct-to-consumer (DTC) prescription advertising, and (4) over-the-counter or direct-to-consumer digital health tools that bypass clinician review.

How does the learned intermediary doctrine apply to clinical AI and Clinical Decision Support (CDS) software?

In clinical AI and Software as a Medical Device (SaMD), the software itself is a medical product rather than a licensed learned intermediary. Under Restatement § 6(d)(1) and FD&C Act Section 520(o)(1)(E), software developers must disclose model training boundaries, subgroup error rates, and intended-use limitations to the clinician so the clinician can independently evaluate the recommendation rather than falling into uncritical automation bias.

Follow CASRAI

Research-administration guidance, standards updates and independent tool reviews.

Ask CASRAI · free to try

Ask about Learned Intermediary Doctrine: Restatement (Third) of Torts § 6(d), Medical Devices, Pharmaceuticals, and Clinical AI/CDS

2 free questions a day, no account.

An AI assistant for research: funder rules, compliance, methods and publishing. It links the sources behind its answers and labels answers that come from a web search.

Answers draw on CASRAI's guides and dictionary plus the federal and funder documents we index: Federal Register, Grants.gov, Regulations.gov and UKRI.

Works on this site and inside Claude, Cursor and the AI tools you already use.

Everything CASRAI publishes — this page, the dictionary, the guides and the news — stays open to read, with no account and no card.

Referenced across the research world

University of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logoUniversity of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logo
  • University of Cambridge logo
  • Columbia University logo
  • Crossref logo
  • University of Edinburgh logo
  • Harvard University logo
  • University of Oxford logo
  • Princeton University logo
  • Stanford School of Medicine logo
  • University College London logo
  • ORCID logo

View CASRAI adoption →