T.H., a Minor, etc. v. Novartis Pharmaceuticals Corporation

Under California law, a brand-name drug manufacturer has a duty to warn

of known or reasonably knowable adverse effects arising from an individual’s use

of its drug. (See Stevens v. Parke, Davis & Co. (1973) 9 Cal.3d 51, 65.) In this

case, we examine whether — and if so, under what circumstances — a brandname

drug manufacturer may be sued under a theory of “warning label” liability

when the warning label for its drug was alleged to be deficient, but the plaintiffs

were injured by exposure to a generic bioequivalent drug bearing the brand-name

drug’s warning label.

Plaintiffs’ mother, J.H., was prescribed terbutaline, a generic form of the

brand-name drug Brethine, to suppress premature labor during her pregnancy.

Plaintiffs T.H. and C.H. were born full term, but were diagnosed with

developmental delays at three years of age and autism by the time they turned five.

Through their father as guardian ad litem, the minors allege that those responsible

2

for the terbutaline label knew or should have known — based on studies of the

drug’s effects in rats and in humans — that the drug posed a serious risk to fetal

brain development. They further allege that the drug’s label unreasonably failed

to include a warning about this risk.

Federal law explicitly conveys to the brand-name manufacturer — and only

that manufacturer — the responsibility to provide an adequate warning label for

both generic terbutaline and its brand-name equivalent, Brethine. As explained in

more detail below, only the brand-name drug manufacturer has unilateral authority

to modify the drug’s label by adding to or strengthening a warning. Generic drug

manufacturers are required to follow the brand-name manufacturer’s label to the

letter. Accordingly, the manufacturer of Brethine controlled both the form and

content of the terbutaline warning label.

Plaintiffs brought suit against defendant Novartis Pharmaceuticals

Corporation (Novartis), which manufactured Brethine until December 2001, and

aaiPharma Inc. (aaiPharma), which purchased the rights to and manufactured

Brethine thereafter — using the same label Novartis had used — when plaintiffs’

mother was prescribed the generic bioequivalent in 2007. Plaintiffs claim that

Novartis knew or should have known that its warning label failed to alert pregnant

women or their physicians to the risk Brethine posed to fetal brain development;

that manufacturers of terbutaline were compelled by federal law to include

Brethine’s deficient label on their own products; that it was foreseeable Novartis’s

successor (aaiPharma) would not change or update Brethine’s deficient label; and

that in reliance on the deficient warning label, plaintiffs’ mother was prescribed

terbutaline, which adversely affected plaintiffs’ developing brains in utero. What

Novartis asserts in response is that its duty to provide a safe and adequate warning

label for Brethine did not encompass those who were prescribed terbutaline in

reliance on the Brethine label. Novartis further contends that any such duty should

3

not extend to those who were exposed to terbutaline after Novartis ceased

manufacturing Brethine and sold its rights in the drug to aaiPharma.

Such contentions, and the case in which they arise, reach us at a very early

stage in the litigation. In reviewing a demurrer, we ask only whether the plaintiff

has alleged — or could allege — sufficient facts to state a cause of action against

the defendant. (Schifando v. City of Los Angeles (2003) 31 Cal.4th 1074, 1081.)

In our view, plaintiffs have indeed shown that they could allege a cause of action

against Novartis for warning label liability. Because the same warning label must

appear on the brand-name drug as well as its generic bioequivalent, a brand-name

drug manufacturer owes a duty of reasonable care in ensuring that the label

includes appropriate warnings, regardless of whether the end user has been

dispensed the brand-name drug or its generic bioequivalent. If the person exposed

to the generic drug can reasonably allege that the brand-name drug manufacturer’s

failure to update its warning label foreseeably and proximately caused physical

injury, then the brand-name manufacturer’s liability for its own negligence does

not automatically terminate merely because the brand-name manufacturer

transferred its rights in the brand-name drug to a successor manufacturer. We

therefore affirm the Court of Appeal, which had directed the trial court to enter an

order sustaining Novartis’s demurrer with leave to amend plaintiffs’ negligence

and negligent misrepresentation causes of action.

I. BACKGROUND

From a certain perspective, the claim underlying this lawsuit is quite

straightforward. Plaintiffs T.H and C.H., who are fraternal twins, sued defendant

Novartis for negligence and negligent misrepresentation arising from Novartis’s

failure to warn of the risks of Brethine, an asthma drug sometimes prescribed

“off label” to stop or slow preterm labor. Plaintiffs allege that Novartis knew or

4

should have known that Brethine carried a substantial risk of causing

developmental and neurological damage to the fetus, yet failed to warn of this risk.

What removes this case from the realm of the ordinary is that plaintiffs’

mother was never prescribed Brethine. Rather, she — like many pregnant women

experiencing premature labor — was prescribed terbutaline sulfate (terbutaline),

the generic bioequivalent drug. Moreover, Novartis stopped manufacturing

Brethine and sold all rights to the drug in 2001, six years before plaintiffs’ injury.

During the period it was the brand-name manufacturer, however, Novartis had the

legal duty to disclose Brethine’s known and reasonably knowable risks in the

drug’s warning label. All generic manufacturers, in turn, had a specific legal

responsibility regarding the label: to ensure the terbutaline label was identical to

the Brethine label. We therefore examine plaintiffs’ allegations against the

backdrop of the distinctive legal framework governing labeling for brand-name

and generic pharmaceuticals.

On review of a demurrer, we accept as true all properly pleaded facts.

(Shirk v. Vista Unified School Dist. (2007) 42 Cal.4th 201, 205.) Where particular

facts are set out below, they are those alleged in plaintiffs’ first amended

complaint.

A. Federal Regulation of Drug Labeling



The Food, Drug, and Cosmetic Act (FDCA; 21 U.S.C. § 301 et seq.)

prohibits the marketing of a new brand-name drug unless the manufacturer has

submitted a new drug application (NDA) and the Food and Drug Administration

(FDA) has approved the drug as safe and effective for its intended use. (21 U.S.C.

§ 355(a).) The NDA must include an exemplar of the drug’s proposed label (21

U.S.C. § 355(b)(1)(F)) describing the drug’s indications and usage,

5

contraindications, warnings and precautions, and adverse reactions. (21 C.F.R.

§ 201.56(e)(1).)



In 1984, Congress enacted the Hatch-Waxman Act. (98 Stat. 1585, 1585-

1597, codified as amended at 21 U.S.C. § 355.) This statute allows a prospective

generic drug manufacturer to file an abbreviated new drug application (ANDA)

asserting the generic drug’s bioequivalence to an existing listed drug that has

already been approved by the FDA. (PLIVA, Inc. v. Mensing (2011) 564 U.S. 604,

612 (PLIVA), citing 21 U.S.C. § 355(j).) Such an application is typically filed as

the brand-name drug’s patent is about to expire. The streamlined application

relieves the generic manufacturer of the need to duplicate the clinical trials

previously submitted for the equivalent brand-name drug. (Ibid.) The generic

manufacturer must nonetheless “show that the labeling proposed for the new drug

is the same as the labeling approved for the listed drug.” (21 U.S.C.

§ 355(j)(2)(A)(v).)



So under the federal scheme, “brand-name and generic drug manufacturers

have different federal drug labeling duties.” (PLIVA, supra, 564 U.S. at p. 613.)

It is the brand-name manufacturer that bears responsibility for the accuracy and

adequacy of its label “as long as the drug is on the market.” (Wyeth v. Levine

(2009) 555 U.S. 555, 570-571 (Wyeth).) The generic manufacturer, on the other

hand, is responsible only for “an ongoing federal duty of ‘sameness’ ” — that is,

ensuring that its warning label is the same as the brand-name manufacturer’s.

(PLIVA, at p. 613.)



FDA regulations require the brand-name drug manufacturer to update the

warning label “as soon as there is reasonable evidence of an association of a

serious hazard with a drug; a causal relationship need not have been proved.” (21

C.F.R. § 201.80(e); cf. id., § 314.80(b) [NDA holder “must promptly review all

adverse drug experience information obtained or otherwise received by the

6

applicant from any source”].) A specific warning is required if the drug is

commonly prescribed for a disease or condition, even when the drug has not yet

been approved for that use, where “such usage is associated with serious risk or

hazard.” (Id., § 201.80(e).) Any manufacturer of the drug at issue may request a

change in the label by submitting a “prior approval supplement” to the FDA,

which decides whether to approve the requested change in the warning label. (21

C.F.R. § 314.70(b)(2)(v); FDA, Abbreviated New Drug Application Regulations,

57 Fed.Reg. 17950, 17961 (Apr. 28, 1992).) But a brand-name drug

manufacturer, unlike a generic manufacturer, may unilaterally update a label,

without waiting for FDA preapproval, “[t]o add or strengthen a contraindication,

warning, precaution, or adverse reaction” under the “changes being effected”

(CBE-0) regulation. (21 C.F.R. § 314.70(c)(6)(iii)(A); see Wyeth, supra, 555 U.S.

at p. 568.) By contrast, a generic manufacturer may use the CBE-0 regulation

only to conform its label to an updated brand-name label. (PLIVA, supra, 564

U.S. at p. 614.)



Because federal regulations preclude generic manufacturers from

unilaterally altering the warning labels on their drugs (PLIVA, supra, 564 U.S. at

p. 617), federal law preempts state tort claims against generic manufacturers for

failure to provide adequate warnings. (Id. at p. 609.) State tort claims against a

brand-name manufacturer based on a failure to warn, however, are not preempted.

(Id. at p. 625.)



B. Terbutaline, Brethine, and Novartis

Terbutaline is a beta-adrenergic agonist, acting upon the beta2 receptors in

smooth muscle tissue and causing muscles to relax. The drug was originally

developed by Draco, a Swedish company, and released for use as a bronchodilator

to treat asthma. In 1974, the FDA approved terbutaline as a treatment for asthma

7

in the United States. Astra AB (and later, AstraZeneca LP) licensed the right to

manufacture and market terbutaline in its oral form to Ciba-Geigy (a predecessor

to Novartis) under the brand name Brethine. Novartis owned the NDA for

Brethine until 2001.

In 1976, a Swedish physician with ties to Draco published the results of a

small study indicating that terbutaline was safe and effective as a tocolytic — a

drug to suppress premature labor in pregnant women — on the theory that the drug

could relax uterine smooth muscle tissue. Terbutaline subsequently gained wide

acceptance as a tocolytic, but neither Novartis nor any other manufacturer sought

FDA approval for this off-label use.1 Later studies cast doubt on the safety and

efficacy of terbutaline as a tocolytic.

In 1978, a study published in the British Journal of Obstetrics and

Gynaecology questioned the validity and conclusions of the original Swedish

report. According to plaintiffs’ complaint, the British study warned that the

benefits of this class of drugs on preterm labor was “ ‘not yet established,’ ” that

the evidence was “ ‘too scanty to make conclusions about side effects possible,’ ”

and that other data suggested “ ‘that labor inhibitors are potentially dangerous.’ ”

A year later, a study published in the American Journal of Obstetrics and

Gynecology reported adverse effects in both the pregnant mother and in the fetus

following terbutaline exposure.

A team of American clinical investigators in 1982 sought to replicate the

results of the 1976 Swedish study. They could not. In fact, the investigators were



1 Physicians may, in their professional judgment, prescribe a drug for a

purpose other than that for which it has been approved by the FDA. (Buckman

Co. v. Plaintiffs’ Leg. Com. (2001) 531 U.S. 341, 351, fn. 5 [“ ‘Off-label use is

widespread in the medical community’ ”].)

8

unable to find any benefit among the pregnant mothers who had been prescribed

terbutaline as compared to those who received a placebo. A 1984 study published

in the Journal of Reproductive Medicine similarly failed to confirm any benefits.

In 1985, Dr. Theodore Slotkin and a team of Duke University Medical

Center researchers found that a single dose of terbutaline given to pregnant rats

interfered with an enzyme necessary for neuronal development in the fetal brain.

Dr. Slotkin’s study showed that terbutaline can cross the placenta and fetal brain

barrier in sufficient quantities to affect brain development. Other studies in the

1980s revealed that children born to mothers who had received a different betaadrenergic

agonist had poorer academic achievement and were more likely to have

impairments in vision and language development than children born to mothers

who did not receive such treatment.

In 1989 and 1990, Dr. Slotkin published studies showing that terbutaline

may interfere with the fetus’s neurobehavioral development, presumably through

its effects on receptors in the fetal cerebellum. Shortly thereafter, in 1992,

scientists from the University of Texas undertook a comprehensive and critical

evaluation of the literature relating to terbutaline and concluded, in a study

published in the American Journal of Obstetrics and Gynecology, that the drug

had not been shown to arrest preterm labor and that chronic exposure may

adversely affect the fetus. A 1995 meta-analysis by researchers from the

University of Pennsylvania likewise concluded that the relevant literature did not

support the claimed benefit from maintenance tocolytic therapy. The American

College of Obstetricians and Gynecologists (ACOG) subsequently issued a

“Technical Bulletin on Preterm Labor” to its more than 40,000 members, which

noted the asserted benefit from maintenance tocolytic therapy lacked any

evidentiary basis and warned that the potential risks of such therapy, to both the

mother and the fetus, were well documented. ACOG’s bulletin stated that the risk

9

associated with beta-mimetic agents (such as terbutaline) appeared greater than

that associated with other tocolytic agents. In 1997, the FDA’s Associate

Commissioner for Health Affairs issued a “Dear Colleague” letter, which endorsed

ACOG’s assessment of the benefits and dangers of long-term tocolytic therapy.

In 2001, the German Central Institute of Mental Health issued a report

concluding that children whose mothers had received beta-agonist tocolysis had a

significantly higher rate of psychiatric disorders and psychopathology, and that

such children scored lower on psychometric tests of cognitive development. Dr.

Slotkin’s Duke team released another study in October 2001, which revealed that

beta2 receptors in the fetal brain, unlike those in mature brains, do not desensitize

when exposed to continuous doses of terbutaline. Instead, the fetal receptors

intensify their sensitivity to terbutaline and thus increase their response to the drug

as the dosage increases (and the brain develops).

Over the years, researchers developed –– and companies brought to market

–– newer and more effective bronchodilators and other asthma treatments.

Novartis continued to manufacture and distribute Brethine with the intention that it

be used as a tocolytic. By 2001, nearly half of all prescriptions for terbutaline

were for tocolysis, even though the drug was never approved by the FDA for that

purpose. In December 2001, Novartis transferred the NDA for Brethine to

NeoSan Pharmaceuticals Inc., a wholly owned subsidiary of aaiPharma.

C. The Facts Underlying This Lawsuit

On September 5, 2007, plaintiffs’ mother, J.H., was hospitalized because of

concerns about premature labor. She was prescribed terbutaline, to be

administered every six hours, and was discharged on September 25, 2007. While

in the hospital, J.H. received a generic version that was manufactured by Lehigh

Valley Technologies, Inc.; after discharge, she received a generic version that was

10

manufactured by Global Pharmaceuticals. J.H. continued taking terbutaline as

directed until plaintiffs were born on October 9, 2007. Plaintiffs appeared to be

normal until their pediatrician, during a routine checkup in December 2010,

reported that the twins may have developmental delays. Despite specialized

treatment for both children, a pediatric neurologist diagnosed them with autism in

August 2012.

Plaintiffs’ first amended complaint alleges that terbutaline passed through

the placenta and the blood-brain barrier. As a result, plaintiffs contend, terbutaline

caused them to suffer severe and permanent neurologic injuries, including an

inability to speak and significant limitations and abnormalities in their motor

skills. Plaintiffs further allege that Novartis knew or should have known that

terbutaline was of questionable efficacy as a tocolytic agent, that terbutaline

carried serious risks of side effects for newborns whose mothers received the drug

during pregnancy, and that federal law required Novartis to report this information

to the FDA and to update the warning label — something Novartis could have

done unilaterally. (See 21 C.F.R. § 314.70(c)(6)(iii)(A).) Instead, Novartis falsely

represented that terbutaline was safe and effective and would not cause serious

side effects in newborns, and it intended for pregnant mothers and their physicians

to rely on these representations. The complaint asserted causes of action for

negligence and negligent misrepresentation, as well as strict liability, intentional

misrepresentation, concealment, and medical negligence.

To support and place in factual context their negligence cause of action,

plaintiffs made a variety of specific allegations regarding Novartis. They alleged

that Novartis had a duty to update the label to warn of the drug’s effects on fetal

development, that Novartis knew or should have known of these effects, that

J.H.’s physicians prescribed her terbutaline because of their erroneous belief that

terbutaline was safe to use as a tocolytic, that plaintiffs suffered neurological

11

damage as a result of their exposure to terbutaline in utero, and that plaintiffs’

injuries were foreseeable. Meanwhile, plaintiffs’ negligent misrepresentation

cause of action alleged that Novartis falsely represented that terbutaline was safe

to use as a tocolytic, that Novartis had no reasonable basis for believing terbutaline

was safe to use as a tocolytic, that Novartis intended for pregnant mothers and

their physicians to rely on their false representations concerning the drug’s safety

as a tocolytic agent, that J.H. and her physicians relied on Novartis’s

representations, that plaintiffs suffered neurological damage as a result of their

exposure to terbutaline in utero, and that plaintiffs’ injuries were foreseeable.

Novartis’s core assertion in its demurrer was that it had no duty to

plaintiffs. To justify its position, the company offered two overarching rationales:

It did not manufacture the terbutaline ingested by their mother; and it had

transferred the Brethine NDA to another company in December 2001, nearly six

years before plaintiffs’ mother was prescribed terbutaline. In addition, Novartis

argued that plaintiffs had failed to identify with specificity any misrepresentation

by Novartis or allege that plaintiffs had relied on any such misrepresentation. In

opposition to the demurrer, plaintiffs responded that Novartis had a duty to warn

about the drug’s effects on fetal development during the period it owned the NDA

and manufactured Brethine; that the six-year gap between Novartis’s divestiture of

the NDA and plaintiffs’ in utero exposure is relevant to causation (and not the

existence of the duty); and that the first amended complaint adequately pleaded the

misrepresentations with specificity, given that the specific misrepresentations are

more likely to be within Novartis’s knowledge, and adequately pleaded reliance on

those misrepresentations.

The trial court sustained the demurrer without leave to amend. It concluded

that Novartis owed plaintiffs no duty as a matter of law relating to claims arising

from terbutaline exposure in 2007. Agreeing with Novartis, the court also found

12

that the fraud-based claims suffered from a lack of specificity and that this defect

could not be remedied by allegations about Novartis’s conduct prior to the 2001

NDA divestiture.

The Court of Appeal reversed and directed that the order sustaining the

demurrer be modified to grant plaintiffs leave to amend their causes of action for

negligence and negligent misrepresentation. The appellate court reasoned that if

plaintiffs could allege that Novartis failed to warn about fetal risks it knew or

should have known were associated with terbutaline when used as a maintenance

tocolytic prior to its divestiture of the brand-name drug in 2001, that the warning

would have remained on the label in 2007 had Novartis added a suitable warning

to the label before divestiture in 2001, and that their mother’s physician would not

have prescribed terbutaline as a maintenance tocolytic had the drug been properly

labeled, then their claims for negligence and negligent misrepresentation can

survive demurrer.

We granted Novartis’s petition for review to decide the existence and scope

of warning label liability for brand-name drug manufacturers under California law.

II. DISCUSSION

The sole issue before us is whether the demurrer should have been

sustained with respect to the negligence and negligent misrepresentation claims on

the ground that Novartis owed no duty of care to plaintiffs. In reviewing an order

sustaining a demurrer, we examine the operative complaint de novo to determine

whether it alleges facts sufficient to state a cause of action under any legal theory.

(Lee v. Hanley (2015) 61 Cal.4th 1225, 1230.) Where the demurrer was sustained

without leave to amend, we consider whether the plaintiff could cure the defect by

an amendment. The plaintiff bears the burden of proving an amendment could

cure the defect. (Blank v. Kirwan (1985) 39 Cal.3d 311, 318.)

13

The gist of plaintiffs’ warning label liability claim is that Novartis

negligently failed to warn about the drug’s risk to fetal brain development. They

contend that the deficient label foreseeably and proximately caused harm not only

to the children of women who were prescribed Brethine, but also to the children of

women who were prescribed its generic bioequivalent, which was legally required

to — and did — bear the same deficient label. Among other things, plaintiffs rely

on section 311 of the Restatement Second of Torts (section 311), which addresses

negligent misrepresentation involving physical harm. Under section 311(1),

“[o]ne who negligently gives false information to another is subject to liability for

physical harm caused by action taken by the other in reasonable reliance upon

such information, where such harm results [¶] . . . [¶] to such third persons as the

actor should expect to be put in peril by the action taken.”

Section 311’s theory of liability is intended to be “somewhat broader” than

that for mere pecuniary loss. (Rest.2d Torts, § 311, com. a.) It “finds particular

application where it is a part of the actor’s business or profession to give

information upon which the safety of the recipient or a third person depends.”

(Id., § 311, com. b; see also Prosser, Misrepresentation and Third Persons (1966)

19 Vand. L.Rev. 231, 254 [explaining that one has a duty not to make a false

representation to “[t]hose to whom a public duty is found to have been created by

statute, or pursuant to a statute . . . [and to] [t]hose members of a group or class

whom he has special reason to expect to be influenced by the representation”].)

This court applied and followed section 311 in Randi W. v. Muroc Joint Unified

School Dist. (1997) 14 Cal.4th 1066 (Randi W.). There, we concluded that a

school district’s negligent misrepresentations about a former employee in a letter

of recommendation could render the school district liable for the employee’s

molestation of a third person — a student at the employee’s new school — even

though the student had no special relationship with the former school district and

14

never received the misleading information. (Id. at p. 1081.) In accordance with

the Restatement, we held “that the writer of a letter of recommendation owes to

third persons a duty not to misrepresent the facts in describing the qualifications

and character of a former employee, if making these misrepresentations would

present a substantial, foreseeable risk of physical injury to the third persons.”

(Ibid.) Plaintiffs urge us to hold, in similar fashion, that a brand-name drug

manufacturer owes a duty to third persons not to misrepresent the safety of its

drug, if making those misrepresentations would present a substantial, foreseeable

risk of physical injury to those third persons.

Duty is indeed the cornerstone of every negligence claim. In California, the

general rule governing duty is codified in Civil Code section 1714, subdivision

(a): “Everyone is responsible . . . for an injury occasioned to another by his or her

want of ordinary care or skill in the management of his or her property or person

. . . .” Thus, “each person has a duty to use ordinary care and ‘is liable for injuries

caused by his failure to exercise reasonable care in the circumstances . . . .’ ”

(Parsons v. Crown Disposal Co. (1997) 15 Cal.4th 456, 472.) Whether a party has

a duty of care in a particular case is a question of law for the court, which we

review independently on appeal. (Kesner v. Superior Court (2016) 1 Cal.5th

1132, 1142 (Kesner).)

The conclusion that a duty exists in a particular case “ ‘is not sacrosanct in

itself, but only an expression of the sum total of those considerations of policy

which lead the law to say that the particular plaintiff is entitled to protection.’ ”

(Dillon v. Legg (1968) 68 Cal.2d 728, 734, quoting Prosser, Law of Torts (3d ed.

1964) pp. 332-333.) We invoke the concept of duty to limit “ ‘ “ ‘the otherwise

potentially infinite liability which would follow from every negligent act,’ ” ’ ” yet

we do so only where public policy clearly supports (or a statutory provision

establishes) an exception to the general rule of Civil Code section 1714. (Kesner,

15

supra, 1 Cal.5th at p. 1143.) When considering whether to depart from the general

rule, we balance a number of considerations, including “the foreseeability of harm

to the plaintiff, the degree of certainty that the plaintiff suffered injury, the

closeness of the connection between the defendant’s conduct and the injury

suffered, the moral blame attached to the defendant’s conduct, the policy of

preventing future harm, the extent of the burden to the defendant and

consequences to the community of imposing a duty to exercise care with resulting

liability for breach, and the availability, cost, and prevalence of insurance for the

risk involved.” (Rowland v. Christian (1968) 69 Cal.2d 108, 113 (Rowland).)

In the context of prescription drugs, a manufacturer’s duty is to warn

physicians about the risks known or reasonably known to the manufacturer.

(Carlin v. Superior Court (1996) 13 Cal.4th 1104, 1112 (Carlin); see generally

Finn v. G.D. Searle & Co. (1984) 35 Cal.3d 691, 699-700.) The manufacturer has

no duty to warn of risks that are “merely speculative or conjectural, or so remote

and insignificant as to be negligible.” (Carlin, at p. 1116.) If the manufacturer

provides an adequate warning to the prescribing physician, the manufacturer need

not communicate a warning directly to the patient who uses the drug. (Ibid.)

In this case, plaintiffs allege that the terbutaline label failed to warn about

the risks to fetal brain development and falsely represented that the drug was safe

for use by pregnant women. They further claim that Novartis’s control over the

Brethine label rendered it responsible for any deficiencies in the terbutaline label,

given that generic drug manufacturers are legally obligated to use the label crafted

by the brand-name drug manufacturer. Novartis contends that it owed no duty to

plaintiffs to update or maintain an accurate label because (1) it did not

manufacture the terbutaline that caused plaintiffs’ injuries; and (2) it had divested

ownership of Brethine, the name-brand drug, several years before plaintiffs’

mother was prescribed terbutaline.

16

To determine whether to create an exception to a brand-name drug

manufacturer’s duty to warn, we balance the constellation of factors set out in

Rowland, supra, 69 Cal.2d at page 113. Three of those factors — foreseeability,

the certainty of the injury, and the closeness of the connection between the

plaintiff and the defendant — address the foreseeability of the relevant injury.

(Kesner, supra, 1 Cal.5th at p. 1145.) The remaining four — moral blame, the

policy of preventing future harm, the burden on the defendant and the general

public, and the availability of insurance — focus on the public policy justifications

for or against carving out an exception to the general duty in this category of

cases. (Ibid.) Our task is to determine whether a brand-name manufacturer owes

a duty of ordinary care to those who may be injured by deficiencies in its warning

label, not whether Novartis acted reasonably under the particular circumstances

here. (See Cabral v. Ralphs Grocery Co. (2011) 51 Cal.4th 764, 772-774

(Cabral).)

A. Whether Plaintiffs Exposed to the Generic Bioequivalent Drug Can

Assert Warning Label Liability Against Novartis, the Brand-name Drug

Manufacturer

The first case to recognize warning label liability was Conte v. Wyeth, Inc.

(2008) 168 Cal.App.4th 89 (Conte). In Conte the court concluded that a brandname

drug manufacturer’s common law duty of care when warning of the dangers

of its drug extended not only to consumers of the brand-name drug, “but also to

those whose doctors foreseeably rely on the name-brand manufacturer’s product

information when prescribing a medication, even if the prescription is filled with

the generic version of the prescribed drug.” (Id. at p. 94.) The Court of Appeal’s

holding predated by more than two years the United States Supreme Court’s ruling

that federal law requires generic drug manufacturers to conform their warning

label to the label used by the brand-name manufacturer (PLIVA, supra, 564 U.S. at

17

p. 613), and its analysis referenced some — but not all — of the Rowland factors.

(Conte, at pp. 105-107.)

Only a handful of courts have followed Conte. (See, e.g., Dolin v.

SmithKline Beecham Corp. (N.D.Ill. 2014) 62 F.Supp.3d 705; Chatman v. Pfizer,

Inc. (S.D.Miss. 2013) 960 F.Supp.2d 641, 654; Kellogg v. Wyeth, Inc. (D.Vt.

2010) 762 F.Supp.2d 694, 704; Wyeth, Inc. v. Weeks (Ala. 2014) 159 So.3d 649

(Weeks).) But our careful review of the federal regulatory scheme and analysis of

all the Rowland factors persuades us that a brand-name drug manufacturer has the

duty under California law to warn of the risks about which it knew or reasonably

should have known, regardless of whether the consumer is prescribed the brandname

drug or its generic “bioequivalent.” (See 21 U.S.C. § 355(j)(2)(A)(iv).)

1. Foreseeability and related factors

In determining whether to create an exception to the general statutory duty

of care, the “major” (Cabral, supra, 51 Cal.4th at p. 771, fn. 2), and ultimately

“most important” (Kesner, supra, 1 Cal.5th at p. 1145), consideration under

California law is the foreseeability of physical harm. Novartis could reasonably

have foreseen that deficiencies in its Brethine label could mislead physicians about

the safety of terbutaline, Brethine’s generic bioequivalent, which was legally

required to bear an identical label.

A brand-name pharmaceutical manufacturer has a duty under federal law to

draft, update, and maintain the warning label so that it provides adequate warning

of the drug’s potentially dangerous effects. (21 U.S.C. § 352(f)(2).) The FDA, as

part of its premarket review process, must approve the text of the proposed label.

(21 U.S.C. § 355; Wyeth, supra, 555 U.S. at p. 568.) Although the brand-name

manufacturer generally must obtain FDA approval before making any change to

the label, this category of manufacturers may use the “changes being effected”

18

(CBE-0) regulation (21 C.F.R. § 314.70(c)(3)) to “add or strengthen a

contraindication, warning, precaution, or adverse reaction” immediately upon

filing a supplemental application, without waiting for FDA approval. (Id.,

§ 314.70(c)(6)(iii)(A).)

The duty for a manufacturer of generic drugs, on the other hand, is to

ensure that its warning label is identical to the label of the brand-name drug.

(PLIVA, supra, 564 U.S. at p. 613.) In other words, generic manufacturers “have

an ongoing federal duty of ‘sameness.’ ” (Ibid.) A generic manufacturer may use

the CBE-0 regulation to change its label only to match a revised brand-name label

or otherwise comply with FDA instructions. (Id. at p. 614.)

What a brand-name manufacturer thus knows to a legal certainty is that any

deficiencies in the label for its drug will be perpetuated in the label for its generic

bioequivalent. A brand-name manufacturer will also be aware that although the

warnings communicated in its drug label are designed for physicians — and are

intended to influence a physician’s decision whether to prescribe the drug (see

Stevens v. Parke, Davis & Co., supra, 9 Cal.3d at pp. 64-65) — it is often the

pharmacist who actually decides whether the patient receives the brand-name drug

or its generic bioequivalent. (Bus. & Prof. Code, § 4073.) Moreover, many

insurance companies require the substitution of a generic drug for the brand-name

drug as a matter of course, unless the physician justifies use of the branded drug.

(PLIVA, supra, 564 U.S. at p. 628, fn. 2 (dis. opn. of Sotomayor, J.).)

Accordingly, it is entirely foreseeable that the warnings included (or not included)

on the brand-name drug label would influence the dispensing of the generic drug,

either because the generic is substituted by the pharmacist or the insurance

company after the physician has prescribed the brand-name drug, or because the

warning label on the generic drug is legally required to be identical to the label on

19

the brand-name drug. (Conte, supra, 168 Cal.App.4th at p. 105; accord, Weeks,

supra, 159 So.3d at p. 670.)

Under the second Rowland factor, we assess the degree of certainty that the

plaintiff suffered injury. This factor, too, strengthens the case for finding a duty of

care in these circumstances. Plaintiffs allege that they suffer from global

neurological impairment, including autism and pervasive developmental delays.

These are indisputably injuries and are compensable under the law. (See Kesner,

supra, 1 Cal.5th at p. 1148.)

The third Rowland factor implicates the closeness of the connection

between the defendant’s conduct and the plaintiff’s injury. The label for a generic

drug is (and must be) the same as the label for the brand-name drug, so any

deficiency in the brand-name label will be reflected in the generic label. Plaintiffs

allege that the deficient Brethine label led their mother’s physician to prescribe

terbutaline, which caused their neurological injuries. This scenario describes a

close connection between Novartis’s allegedly negligent conduct and plaintiffs’

injuries.

Novartis, meanwhile, relies on O’Neil v. Crane Co. (2012) 53 Cal.4th 335

(O’Neil). This is a case we can distinguish. There, a naval seaman developed

mesothelioma caused by asbestos exposure. Following his death, his family filed

a wrongful death action asserting strict liability and negligence claims against

several defendants, including the manufacturers of valves and pumps that were

used in warships. (Id. at p. 346.) At the close of evidence, the defendant

manufacturers moved for nonsuit, pointing out the plaintiffs’ failure to show that

the decedent had been exposed to asbestos from any of their products. Plaintiffs

responded that even if the decedent was never exposed to asbestos from the

defendants’ products themselves, it was foreseeable that the defendants’ valves

and pumps would need to be replaced with new asbestos-containing components,

20

and that asbestos could be released into the air during the repair and replacement

process. (Ibid.) In reinstating the trial court’s judgment of nonsuit, we invoked

the Rowland factors and noted, in particular, that the connection between the

defendant manufacturers’ conduct and the decedent’s injury was “extremely

remote” (id. at p. 365): Although component parts of the defendants’ valves and

pumps had been replaced “during routine maintenance” (id. at p. 344), the

decedent did not begin to work in the vicinity of these valves and pumps until

more than 20 years after they were installed — and did not suffer an injury for

another 40 years. In addition, the defendant manufacturers did not produce, sell,

or supply any of the asbestos-containing products that could have caused his

mesothelioma. Because the defendants’ asserted misconduct, according to the

plaintiffs, was simply that they failed to warn about the potential dangers in

replacement parts sold by other manufacturers — and there was “no reason to

think a product manufacturer [would] be able to exert any control over the safety

of replacement parts or companion products made by other companies” — we

found that the connection between the alleged misconduct and the injury was too

“attenuate[d]” to warrant imposition of a duty of care. (Id. at p. 365.)

Here, by contrast, federal regulations granted the brand-name drug

manufacturer — and no other manufacturer — control over the active ingredients

in the generic drug and the content of the warnings included in the generic’s

label.2 In addition, the temporal connection between Novartis’s allegedly



2 The FDA has been considering for some time a rule that would effectively

abrogate PLIVA and enable generic drug manufacturers to update a drug’s warning

label unilaterally, even if the brand-name manufacturer had not yet done so. (See

FDA, Supplemental Applications Proposing Labeling Changes for Approved

Drugs and Biological Products, 78 Fed.Reg. 67985 (Nov. 13, 2013); see Dept. of

Health and Human Services, Regulatory Agenda, 82 Fed.Reg. 40277, 40279 (Aug.

24, 2017).) If adopted, the new rule “would create parity between NDA holders

21

negligent conduct, on the one hand, and plaintiffs’ exposure to harm and

subsequent injury, on the other, is much closer than was the case in O’Neil.

2. Considerations of public policy

Foreseeability alone, however, is not sufficient to justify a duty of care in

every instance. (Erlich v. Menezes (1999) 21 Cal.4th 543, 552.) We will not

recognize a duty of care even as to foreseeable injuries “where the social utility of

the activity concerned is so great, and avoidance of the injuries so burdensome to

society, as to outweigh the compensatory and cost-internalization values of

negligence liability.” (Merrill v. Navegar, Inc. (2001) 26 Cal.4th 465, 502.)

Novartis contends that the circumstances here present such an exceptional case.

We disagree.

Time and again we have recognized how “ ‘[t]he overall policy of

preventing future harm is ordinarily served, in tort law, by imposing the costs of

negligent conduct upon those responsible.’ ” (Kesner, supra, 1 Cal.5th at p. 1150,

quoting Cabral, supra, 51 Cal.4th at p. 781.) A brand-name drug manufacturer is

not only in the best position to warn of a drug’s harmful effects (Sindell v. Abbott

Laboratories (1980) 26 Cal.3d 588, 611): It is also the only manufacturer with the

unilateral authority under federal law to issue such a warning for the brand-name

drug or its generic bioequivalent. Although federal regulations impose a

continuing duty on the brand-name manufacturer to update and maintain an



and ANDA holders with respect to submission of CBE-0 supplements for safetyrelated

labeling changes based on newly acquired information” (78 Fed.Reg.,

supra, at p. 67989) and may conceivably justify reweighing of the Rowland factors

and some reconsideration of the brand-name manufacturer’s duty in this category

of cases.

22

adequate warning label (see 21 C.F.R. § 201.80(e)), a brand-name manufacturer’s

incentive to comply with that duty declines once the patent expires and generic

manufacturers enter the market, since the market share for the brand-name drug at

that point “may drop substantially.” (78 Fed.Reg., supra, at p. 67988 [“Among

drugs for which a generic version is available, approximately 94 percent are

dispensed as a generic”].) The possibility that any consumer injured by a deficient

drug label, including those who were dispensed the generic bioequivalent drug,

could assert a claim of warning label liability restores the brand-name

manufacturer’s incentive to update the warning label with the latest safety

information, even as the brand-name drug’s market share declines.

If the policy of preventing harm has special relevance to any particular

endeavor, surely prescription drug labeling is one. (Sindell v. Abbott

Laboratories, supra, 26 Cal.3d at p. 611.) A substantial body of state law serves

to protect California consumers from the dangers posed by false, misleading, and

inadequate labeling of prescription medications. (See, e.g., Bus. & Prof. Code,

§§ 4070-4078.) The United States Supreme Court, too, has recognized the pivotal

role of state tort actions “as a complementary form of drug regulation” with

respect to drug labeling. (Wyeth, supra, 555 U.S. at p. 578; see id. at p. 579

[“State tort suits uncover unknown drug hazards and provide incentives for drug

manufacturers to disclose safety risks promptly. They also serve a distinct

compensatory function that may motivate injured persons to come forward with

information. Failure-to-warn actions, in particular, lend force to the FDCA’s

premise that manufacturers, not the FDA, bear primary responsibility for their

drug labeling at all times”]; accord, Stevens v. Parke, Davis & Co., supra, 9 Cal.3d

at p. 65 [recognizing that federal warning-label regulations alone may be

insufficient to protect patient safety].)

23

The brand-name drug manufacturer is the only entity with the unilateral

ability to strengthen the warning label. So a duty of care on behalf of all those

who consume the brand-name drug or its bioequivalent ensures that the brandname

manufacturer has sufficient incentive to prevent a known or reasonably

knowable harm. In O’Neil, by contrast, we found “no reason” to believe that the

defendant valve and pump manufacturers would have any control over the safety

of other companies’ replacement parts or companion products (or even the Navy’s

purchasing choices or specifications). (O’Neil, supra, 53 Cal.4th at p. 365.) Our

no-duty conclusion also rested explicitly on the fact that the replacement parts’

“dangerous feature” — i.e., the asbestos — “was not integral to the product’s

design.” (Id. at p. 343.) Here, on the other hand, the brand-name drug

manufacturer exercised complete control over the contents of the generic drug

label at the time of its alleged negligence, and the generic drug was legally

required to be the brand-name drug’s bioequivalent. We therefore conclude that

warning label liability is likely to be effective in reducing the risk of harm to those

who are prescribed (or are exposed to) the brand-name drug or its generic

bioequivalent.

Against the public interest in preventing harm, we must balance the

defendant’s burden and the consequences to the community of imposing a duty of

care. The burden that matters, though, is not the cost of compensating individuals

for injuries that the defendant has actually and foreseeably caused. As we recently

explained in Kesner, “shielding tortfeasors from the full magnitude of their

liability for past wrongs is not a proper consideration in determining the existence

of a duty. Rather, our duty analysis is forward-looking, and the most relevant

burden is the cost to the defendants of upholding, not violating, the duty of

ordinary care.” (Kesner, supra, 1 Cal.5th at p. 1152.)

24

Strictly speaking, then, the burden on brand-name drug manufacturers of

satisfying a common law duty of care to those who are prescribed the generic

version of the drug is zero. Brand-name manufacturers already have a continuing

duty to warn of potential risks “as soon as there is reasonable evidence of an

association of a serious hazard with a drug; a causal relationship need not have

been proved.” (21 C.F.R. § 201.80(e).) A brand-name manufacturer’s burden to

maintain an adequate warning label persists without regard to the happenstance

that a given prescription for a brand-name drug may — because of insurance

company cost-savings rules (Meijer, Inc. v. Warner Chilcott Holdings Co. III Ltd.

(D.D.C. 2007) 246 F.R.D. 293, 297), or a pharmacist’s discretion (Bus. & Prof.

Code, § 4073, subd. (c)) — be filled with a generic bioequivalent. And where the

brand-name manufacturer provides an adequate label, then it necessarily has also

fulfilled its duty with respect to the generic bioequivalent.

Novartis complains that unless the ordinary duty of care is narrowed, the

brand-name drug manufacturer would end up an insurer for the entire market.

This would occur, Novartis contends, even though the brand-name manufacturer

may hold only a small fraction of the combined sales of the brand-name drug and

its generic bioequivalent. We disagree. A brand-name drug manufacturer would

not be liable where, for example, the injury arose from a defect in the

manufacturing process of the generic drug (see, e.g., Fisher v. Pelstring (D.S.C.

2012) 817 F.Supp.2d 791, 818), the generic manufacturer failed to conform its

label to the brand-name drug’s label (Fulgenzi v. PLIVA, Inc. (6th Cir. 2013) 711

F.3d 578, 582, 584; Huck v. Wyeth, Inc. (Iowa 2014) 850 N.W.2d 353, 356 (plur.

opn. of Waterman, J.)), or the generic manufacturer was promoting a use that was

inconsistent with the FDA-approved label (Arters v. Sandoz, Inc. (S.D. Ohio 2013)

921 F.Supp.2d 813, 819-820). Under warning label liability, the brand-name drug

manufacturer is liable only in a narrow circumstance — when deficiencies in its

25

own label foreseeably and proximately caused injury. If instead tort law simply

carved out those who were given the generic bioequivalent from obtaining

otherwise available compensation for injuries attributable to the brand-name drug

manufacturer’s defective warning label, then consumers would insist on the brandname

drug over the cheaper bioequivalent, inflating health costs with no

corresponding increase in safety and in contradiction to the stated federal policy of

making low-cost generic drugs more available. (See H.R.Rep. No. 98-857, 2d

Sess., p. 14 (1984), reprinted in 1984 U.S. Code Cong. & Admin. News, p. 2647.)

Novartis nonetheless predicts that unless we carve out an exception for

those taking generic drugs, warning label liability will lead to overwarning — i.e.,

inclusion of a slew of speculative risks in the warning label — which would dilute

the effectiveness of any individual warning. But why this would occur is far from

clear. To recognize that the duty of care includes all those who would foreseeably

be affected by a deficient brand-name drug label merely preserves the brand-name

manufacturer’s duty as it existed when its patent excluded all competitors from the

market. Nor has Novartis identified any surge in overwarning since 2008, when

Conte recognized warning label liability. (Cf. Carlin, supra, 13 Cal.4th at p. 1116,

fn. 6 [“[T]here is no evidence that any such [overwarning] problem has emerged

or that patients have suffered any detriment, despite the fact that strict liability has

long been the rule in California”].) Plaintiffs further suggest that the

consequences of overwarning on physicians’ prescription decisions is uncertain

(Steven Garber, RAND Institute for Civil Justice, Economic Effects of Product

Liability and Other Litigation Involving the Safety and Effectiveness of

Pharmaceuticals (2013) p. xiv [“That claim is controversial within the medical

community, and there is no direct empirical evidence about it”]) and, in any event,

can be solved through the FDA’s power to reject a labeling change it deems

unnecessary or counterproductive. (See Wyeth, supra, 555 U.S. at p. 571.)

26

Novartis cautions that warning label liability could perversely incentivize a

brand-name manufacturer to withdraw its drug from the market, rather than expose

itself to the risk of suit by those who — in reliance on the brand-name

manufacturer’s label — were prescribed the generic bioequivalent and suffered

injury. Yet Novartis fails to explain why brand-name manufacturers would find it

economically advantageous to withdraw drugs from the market rather than simply

modify the warning labels to include the newly discovered risks. Nor does it offer

any evidence that brand-name manufacturers have accelerated their withdrawal

from the market in the nine years since Conte was decided. Moreover, a brandname

drug manufacturer cannot avoid its duty to update and maintain its warning

label simply by unilaterally exiting the market. Under FDA regulations, a brandname

drug manufacturer’s duty to update and maintain the warning label

continues, even if the brand-name drug has been withdrawn from the market, until

the FDA (having assured itself that the drug is safe, effective, and correctly

labeled) withdraws approval of the NDA. (21 C.F.R. § 314.150(a)(2), (b)(3) &

(c); FDA, supra, 78 Fed.Reg. at p. 67993.) A brand-name manufacturer that

sought to exit the market but was unsure whether the FDA would determine that

the drug was withdrawn “for reasons other than effectiveness or safety” thus

would presumably go ahead and update the label. (Lasker et al., Taking the

“Product” Out of Product Liability: Litigation Risks and Business Implications

of Innovator and Co-promoter Liability (July 2015) 82 Def. Counsel J. 295, 306.)

Novartis complains next that it is unfair to subject a brand-name drug

manufacturer to liability for harm caused by a competitor’s product — a product

from which the brand-name manufacturer derives no revenues or profit. But the

plaintiffs’ claim here is not that terbutaline is defectively designed or inherently

dangerous. It is that terbutaline’s warning label failed to mention the risk to fetal

brain development, and that Novartis was responsible for the deficient label. So

27

the alleged fault here lies with Novartis, not with its generic competitors. The

brand-name drug manufacturer’s burden to adequately label its drug as a means of

ensuring adequate warnings for the generic bioequivalent is more than offset by

the substantial benefits federal law confers on the brand-name manufacturer: a

monopoly over the market for the life of the patent, which can be extended for the

time consumed by FDA review of the NDA (see 35 U.S.C. §§ 154, 156(a), (c)); an

additional five-year exclusivity period if the brand-name drug contains a new

chemical entity or an additional three years for a new use of a previously approved

drug (see 21 U.S.C. § 355(c)(3)(E); 21 C.F.R. § 314.108(b)(2), (4), (5)); and the

higher prices the brand-name drug can continue to command even after the

exclusivity period expires. (See Conte, supra, 168 Cal.App.4th at p. 110.)

Because federal law bundles –– and indeed, only makes available –– those

benefits along with the responsibility to maintain an adequate warning label, it is

as logical as it is reasonable for state common law to ensure the brand-name

manufacturer holds up its end of the deal. (See Wyeth, supra, 555 U.S. at pp. 578-

579; see generally Struve, The FDA and the Tort System: Postmarketing

Surveillance, Compensation, and the Role of Litigation (2005) 5 Yale J. Health

Pol’y L. & Ethics 587, 605-606 [“The problem of insufficient resources persists,

as does the [structural] concern that the FDA may be loath to move swiftly to

address emerging safety issues”].) The public interest in adequate drug warnings,

in short, is just as acute when the brand-name drug manufacturer has an effective

monopoly over the warning label as it was when the brand-name manufacturer

had a monopoly over the entire market for the drug. (See Wyeth, supra, 555 U.S.

at p. 571 [noting that federal regulations “plac[e] responsibility for postmarketing

surveillance on the manufacturer”].)

We are equally unpersuaded by Novartis’s contention that warning label

liability would stifle innovation by substantially raising drug costs and chilling the

28

development and marketing of new drugs. The logic buttressing this argument is

far from self-evident. Warnings about a product’s efficacy or danger may indeed

risk diminishing its value to the manufacturer. Less obvious is the manufacturer’s

response to this predicament. One might just as easily assert that a drug company,

after adding a new warning, will be incentivized to develop new and safer

alternatives to the drug so that it can recapture the market for treatment of that

disease. (See Carlin, supra, 13 Cal.4th at p. 1117.)

Indeed, the pharmaceutical industry raised a similar objection in Carlin to

the imposition of strict liability for the failure to warn about the known or

reasonably scientifically knowable dangers of a drug. We found “no clear or

sufficient basis for concluding that research and development will inevitably

decrease” as a consequence of imposing liability for failure to warn of known or

knowable risks (Carlin, supra, 13 Cal.4th at p. 1117) — nor has Novartis offered

any evidence that drug innovation has declined in the 21 years since Carlin was

decided. Carlin therefore saw “no reason to depart from our conclusion . . . that

the manufacturer should bear the costs, in terms of preventable injury or death, of

its own failure to provide adequate warnings of known or reasonably scientifically

knowable risks.” (Ibid.)

The same is true here. When it comes to choosing whether the cost of an

injury involving prescription medication should be borne by an innocent plaintiff

or a negligent defendant, our case law has routinely held that the latter should bear

the cost. (Sindell v. Abbott Laboratories, supra, 26 Cal.3d at pp. 610-611.) A

brand-name drug manufacturer is in the best position to discover and cure

deficiencies in its warning label, to bear the cost of injury resulting from its failure

to update and maintain the warning label, to insure against the risk of liability, and

to spread any increased cost widely among the public. (Id. at p. 611.) After all,

the fault (if any) for a deficient label lies with the brand-name manufacturer alone.

29

(Cf. Groll v. Shell Oil Co. (1983) 148 Cal.App.3d 444, 449 [manufacturer of bulk

fuel owed no duty to the ultimate consumer where the manufacturer provided

adequate warnings to the distributor, “who subsequently packages, labels and

markets the product,” and the manufacturer thus “did not have the ability to

prepare the warning”].) The balance of preventing harm and avoiding an undue

burden on drug manufacturers and the public generally thus tips in favor of

warning label liability.

Neither of the two remaining Rowland factors weighs in favor of an

exception to the general duty of care. To wit: Significant moral blame attaches

where a brand-name drug manufacturer fails to warn about the unsafe effects of its

drug, when those effects are known or reasonably should have been known to the

manufacturer. (See Peterson v. San Francisco Community College Dist. (1984) 36

Cal.3d 799, 814.) Blameworthiness would not depend on whether the pregnant

woman, in reliance on the brand-name drug manufacturer’s label, was dispensed

the brand-name drug or its generic bioequivalent. Even those women who were

prescribed the brand-name drug may nonetheless have received the generic

version, either because the insurance company required it or because the

pharmacist chose it. Moreover, both the pregnant woman and her physician would

have relied on the brand-name drug manufacturer to warn of any serious hazards

that were “associated” with the drug. (21 C.F.R. § 201.80(e).) Indeed, under

federal law, no other manufacturer could have advised them of the drug’s risks. In

these circumstances, potential plaintiffs — the unborn — would be “particularly

powerless,” while the defendant brand-name drug manufacturer would have the

best information about the drug’s risks. (Kesner, supra, 1 Cal.5th at p. 1151.)

Although we declared in O’Neil that “little moral blame can attach to a

failure to warn about dangerous aspects of other manufacturers’ products and

replacement parts” (O’Neil, supra, 53 Cal.4th at p. 365), the context of that

30

statement was a situation in which the valve and pump manufacturers had no

control or influence over the design, manufacturing, or safety of those parts; the

warning attached to them; or the consumer’s decision whether to purchase such

products. (Ibid.) Here, by contrast, the brand-name manufacturer legally

controlled the label on the generic bioequivalent drug, and thus had significant

influence on the decision whether to prescribe it.

Finally, Novartis offers no reason why a brand-name drug manufacturer

would be unable to insure against the risk of warning label liability. Presumably, a

brand-name manufacturer already insures against the risk of liability arising from

a deficient warning label when a drug is introduced and the manufacturer has a

monopoly over that market. It is far from clear why the brand-name drug

manufacturer’s exposure would become fatally uncertain merely because the

brand-name manufacturer is sharing the market with generic manufacturers. (Cf.

O’Neil, supra, 53 Cal.4th at p. 365 [“it is doubtful that manufacturers could insure

against the ‘unknowable risks and hazards’ lurking in every product that could

possibly be used with or in the manufacturer’s product”].)

3. Out-of-state authorities

Novartis (and its amici curiae) rely in substantial part on what they call the

“overwhelming” majority of courts that have declined to recognize warning label

liability owed to those who were prescribed a generic version of the drug in

reliance on the brand-name drug label. Although the decisions of our sister states

and the lower federal courts may be instructive to the extent we find their analysis

persuasive, they are neither binding nor controlling on matters of state law.

(People v. Gonzales and Soliz (2011) 52 Cal.4th 254, 296.) We have respectfully

considered the authorities cited by Novartis. We do not find them persuasive in

analyzing California law.

31

The “ ‘leading case’ ” (Strayhorn v. Wyeth Pharms., Inc. (6th Cir. 2013)

737 F.3d 378, 401) for the proposition that a brand-name drug manufacturer owes

no duty to warn patients who were dispensed the generic bioequivalent is Foster v.

American Home Products Corp. (4th Cir. 1994) 29 F.3d 165 (Foster) — so we

examine that case in some detail. In Foster, the decedent’s pediatrician prescribed

Phenergan, a brand-name antihistamine manufactured by the defendant which was

sometimes used to treat colic. The pharmacist substituted promethazine, a generic

bioequivalent. After being given promethazine several times over the next few

days, six-week-old Brandy was found dead in her crib. A pediatrician specializing

in sudden infant death syndrome at the University of Maryland opined that

Brandy’s death was caused by promethazine. (Id. at pp. 167-168.) The district

court found the plaintiffs (Brandy’s parents) had stated a claim for negligent

misrepresentation, despite the fact that the defendant had not manufactured the

drug ingested by Brandy, but subsequently granted summary judgment because of

the plaintiffs’ failure to demonstrate that their pediatrician had relied on the

defendant’s representations. When the plaintiffs appealed the grant of summary

judgment and the defendant cross-appealed the district court’s initial

determination that a negligent misrepresentation claim could lie against the brandname

manufacturer for harm arising from the generic drug, the Fourth Circuit

sustained the cross-appeal. (Ibid.)

Foster reasoned first that the negligent misrepresentation cause of action

was in essence a claim of product liability, but “without meeting the requirements

[Maryland] law imposes in products liability actions” — i.e., “that the defendant

manufactured the product at issue.” (Foster, supra, 29 F.3d at p. 168.) The court

next addressed the peculiarities of the regulated pharmaceutical market, under

which “any representations [the defendant] makes when advertising Phenergan

also apply to generic promethazine”; a warning “will simply not be made” if the

32

brand-name manufacturer does not issue one; and a patient who is prescribed

Phenergan “may actually receive generic promethazine.” (Id. at p. 169.) In

rejecting liability nonetheless, the Foster court assumed that although generic

manufacturers “must include the same labeling information as the equivalent name

brand drug, they are also permitted to add or strengthen warnings and delete

misleading statements on labels, even without prior FDA approval. . . .

Manufacturers of generic drugs, like all other manufacturers, are responsible for

the representations they make regarding their products.” (Id. at p. 170.) The court

also concluded that “to impose a duty in the circumstances of this case would be to

stretch the concept of foreseeability too far” under Maryland law, which had

recognized the tort of negligent misrepresentation only where “ ‘one party has the

right to rely for information upon the other, and the other giving the information

owes a duty to give it with care.’ ” (Id. at p. 171.) In the court’s view, no such

relationship could ever exist because Brandy “was injured by a product that [the

defendant] did not manufacture.” (Ibid.)

At the core of the Foster court’s analysis is an erroneous assumption: that

generic drug manufacturers may “add or strengthen warnings and delete

misleading statements on labels, even without prior FDA approval,” and that they

can be sued for their failure to do so. (Foster, supra, 29 F.3d at p. 170.) In reality,

generic drug manufacturers are legally obligated to conform their drug label to the

brand-name manufacturer’s label (PLIVA, supra, 564 U.S. at p. 613) and, so long

as they fulfill their “duty of ‘sameness’ ” (ibid.), cannot be sued in tort for

deficiencies in the label. (See id. at p. 624.) Fortunately, the Fourth Circuit has

since recognized its error. Despite its categorical rejection of any duty in Foster,

the court recently certified to the Supreme Court of Appeals of West Virginia the

question “Whether West Virginia law permits a claim of failure to warn and

negligent misrepresentation against a branded drug manufacturer when the drug

33

ingested was produced by a generic manufacturer.” (McNair v. Johnson &

Johnson Corp. (4th Cir. May 30, 2017, No. 15-1806) 2017 WL 2333843, *1.)3

Even on its own terms, though, Foster’s reasoning proves unhelpful in

construing California law, and finds no support in it. First, California law does not

conflate negligent misrepresentation and strict liability in the manner Foster

believed was true of Maryland law. (Conte, supra, 168 Cal.App.4th at p. 108.)

Under our state’s law, there is no per se requirement in negligent

misrepresentation actions that the misrepresentation be made by the product

manufacturer. Consider Hanberry v. Hearst Corp. (1969) 276 Cal.App.2d 680,

where the plaintiff alleged that defective shoes caused her injuries. (Id. at p. 682.)

The Court of Appeal allowed the negligent misrepresentation claims to go forward

against a nonmanufacturer — the publisher of Good Housekeeping magazine,



3 There is a sad coda to Foster. The Fourth Circuit’s ruling relieved the

brand-name manufacturer of any duty to warn of known or knowable risks of the

drug when a plaintiff had been given the generic equivalent — and (contrary to

Foster’s key assumption) the generic manufacturer had no ability to deviate from

the brand-name manufacturer’s label. As a result, it took until 2000 –– six years

after Foster was decided –– for the FDA to modify the warning to recommend that

promethazine not be given to children younger than two years old. (Starke et al.,

Boxed Warning Added to Promethazine Labeling for Pediatric Use (2005) 352

New Eng. J. Med. 2653, 2653.) Four years thereafter, following further review of

all adverse events that had been reported, the FDA added a boxed warning — the

strongest type of warning (21 C.F.R. § 201.57(c)(1)) — stating that the drug

should not be given to children younger than two years old because of the

potential for fatal respiratory depression. (Starke et al., supra, at p. 2653; Rostron,

Prescription for Fairness: A New Approach to Tort Liability of Brand-name and

Generic Drug Manufacturers (2011) 60 Duke L.J. 1123, 1146-1147.) This

example underscores the reality that the FDA depends heavily on the brand-name

drug manufacturer exercising its own unilateral ability to strengthen its warning

label. (Wyeth, supra, 555 U.S. at p. 579; see generally Weeks, supra, 159 So.3d at

p. 676 [“The FDA has limited resources to monitor the approximately 11,000

drugs on the market”].) Yet Foster’s rule seriously undermines the brand-name

manufacturer’s incentive to do so.

34

which had given the shoes its renowned seal of approval. (Id. at p. 683.) This seal

appeared not only in the pages of its own magazine, but was used by the shoe

manufacturer in its advertising as well as on the product and its packaging. (Ibid.)

The court acknowledged that the defendant publisher was neither the seller nor the

manufacturer of the shoes, but nonetheless recognized a duty of care because of

the allegations that the publisher “held itself out as a disinterested third party

which had examined the shoes, found them satisfactory, and gave its

endorsement”; and the plaintiff reasonably relied on the endorsement and

“purchased the shoes because of [it].” (Id. at pp. 686, 683.) As to the plaintiff’s

claim under strict liability, however, the court affirmed the trial court’s dismissal

— declining to extend strict liability “to a general endorser who makes no

representation it has examined or tested each item marketed.” (Id. at p. 688; see

also Conte, supra, 168 Cal.App.4th at pp. 101-102 [similarly distinguishing

between strict liability and negligent misrepresentation theories].)4



4 Novartis suggests that we recently conflated strict liability and negligence

in Webb v. Special Electric Co., Inc. (2016) 63 Cal.4th 167 when we observed that

“ ‘there is little functional difference between the two theories in the failure to

warn context.’ ” (Id. at p. 187.) Not so. Webb’s observation was merely that the

sophisticated user and sophisticated intermediary defenses applied to both theories

of liability. (Ibid.) We did not categorically alter our longstanding recognition

that “California law recognizes the differences between negligence and strict

liability causes of action.” (Johnson v. American Standard , Inc. (2008) 43 Cal.4th

56, 71; see Saller v. Crown Cork & Seal Co., Inc. (2010) 187 Cal.App.4th 1220,

1239 [“ ‘Negligence and strict products liability are separate and distinct bases for

liability that do not automatically collapse into each other because the plaintiff

might allege both when a product warning contributes to her injury’ ”].)

O’Neil did not erase the distinction between strict liability and negligence,

either. Far from conflating the two theories, we said simply that “[t]he same

policy considerations that militate against imposing strict liability in this situation

apply with equal force in the context of negligence.” (O’Neil, supra, 53 Cal.4th at

p. 366, italics added.) As we have demonstrated above, O’Neil is soundly

distinguishable from the situation here.

35

Second, California law places greater weight on the element of

foreseeability in the duty analysis than does Maryland law. Indeed, this state

treats foreseeability as “[t]he most important factor” (Kesner, supra, 1 Cal.5th at

p. 1145), and we do not narrowly circumscribe the kinds of relationships that must

exist between the plaintiff and the defendant as a predicate to imposing a duty on

the defendant to prevent injuries arising from its own conduct. (Id. at p. 1163; see

Randi W., supra, 14 Cal.4th at p. 1077 [one who negligently provides false

information to another can owe a duty of care to a third person “who did not

receive the information and who has no special relationship with the provider”].)5

By contrast, Foster found that a “duty . . . arises” under Maryland law only “when

there is ‘such a relation that one party has the right to rely for information upon the

other, and the other giving the information owes a duty to give it with care.’ ”

(Foster, supra, 29 F.3d at p. 171, quoting Weisman v. Connors (Md. 1988) 540

A.2d 783, 790.) Foster then summarily concluded that “[t]here is no such

relationship between the parties to this case, as Brandy Foster was injured by a

product that [defendant] did not manufacture.” (Foster, at p. 171.) Even this

explanation, though, seems to overlook the fact that there is never a direct

relationship between a prescription drug manufacturer and the ultimate consumer.

A consumer may obtain a prescription medication only through the physician as a



5 We therefore do not find persuasive those out-of-state cases discounting the

role of foreseeability (see, e.g., Huck v. Wyeth, Inc., supra, 850 N.W.2d at p. 376

(plur. opn. of Waterman, J.) [“ ‘foreseeability should not enter into the duty

calculus’ ”]) or requiring the existence of a specific type of relationship between

the plaintiff and the defendant (see, e.g., Moretti v. Wyeth, Inc. (9th Cir. 2014) 579

Fed. Appx. 563, 564 [construing negligent misrepresentation, under Nevada law,

to “ ‘require[], at a minimum, some form of relationship between the parties’ ”];

Schrock v. Wyeth, Inc. (10th Cir. 2013) 727 F.3d 1273, 1282 [“Oklahoma courts

have also required a relationship between the defendant company and the product

at issue for other theories of liability, including negligence”]).

36

learned intermediary. (See Carlin, supra, 13 Cal.4th at pp. 1116, 1126.) A

physician, in turn, typically relies on the drug’s warning label, the contents of

which (regardless of whether the medication ultimately dispensed is the brandname

or generic bioequivalent) are controlled by the brand-name manufacturer. It

is difficult to understand why the relationship between the brand-name

manufacturer and the physician must be deemed to evaporate simply because an

insurance company or pharmacist subsequently decides to dispense a generic

version of the drug that bears the warning label crafted by the brand-name

manufacturer.

Third, in one crucial respect, Foster is like the vast majority of the out-ofstate

cases on which Novartis relies: it arose in federal court under diversity

jurisdiction. Federal courts sitting in diversity are “extremely cautious” about

recognizing innovative theories under state law (Combs v. Int’l Ins. Co. (6th Cir.

2004) 354 F.3d 568, 578) and are bound to “apply the applicable state law as it

now exists.” (Foster, supra, 29 F.3d at p. 171; see generally Gluck, Intersystemic

Statutory Interpretation: Methodology as “Law” and the Erie Doctrine (2011)

120 Yale L.J. 1898, 1939 [federal courts “pick the narrowest possible answer,

usually the one that does the least to change the status quo, regardless of its

predictions of what the state court would do”].) Because only a handful of

jurisdictions have adopted the duty recognized in Conte, supra, and followed by

the Court of Appeal here, it is not surprising that federal courts have been reluctant

to interpret the law of various states to embrace it. But the task of this court is not

to “ ‘opt for the interpretation that restricts liability, rather than expands it’ ” until

someone else tells us otherwise. (Travelers Indem. Co. v. Dammann & Co., Inc.

(3d Cir. 2010) 594 F.3d 238, 253; see also Germain v. Teva Pharmaceuticals,

USA, Inc. (In re Darvocet, Darvon, & Propoxyphene Products Litigation) (6th Cir.

2014) 756 F.3d 917, 937 [“federal courts must be cautious”].) It is instead

37

emphatically the province of this court to declare what the law is. By contrast,

Novartis’s collection of federal decisions merely attempt to predict the law of

other states, while operating under a presumption against expanding liability. (See

Schrock v. Wyeth, Inc., supra, 727 F.3d at p. 1290 [“As a federal court . . . we have

limited authority to correct this potential injustice. It is for the state courts, rather

than this panel, to engage in the delicate policy considerations predicate to the

expansion of the scope of state tort law”].) They are of little use in discharging

our task.

We likewise discount decisions from those jurisdictions that differ from

California by categorically excluding from liability certain defendants (see, e.g.,

Huck v. Wyeth, Inc., supra, 850 N.W.2d at p. 371 (plur. opn. of Waterman, J.)

[“the tort of negligent misrepresentation does not apply to sellers of products but

rather is limited to those in the business or profession of supplying information for

the guidance of others”]) or certain injuries (see, e.g., Flynn v. American Home

Products Corp. (Minn.Ct.App. 2001) 627 N.W.2d 342, 351 [“the Minnesota

Supreme Court has recognized negligent misrepresentation involving damages

only for pecuniary loss, and has expressly declined to recognize the tort of

negligent misrepresentation involving the risk of physical harm”]) from the tort of

negligent misrepresentation. And we find unhelpful the views of those

jurisdictions that (federal courts predict) will recharacterize under their product

liability act or similar rule all claims against a product manufacturer, no matter the

theory, as product liability actions, which can be asserted only against the

manufacturer of the product. (See, e.g., Germain, supra, 756 F.3d at pp. 941-954

[construing the laws of Arkansas, Connecticut, Florida, Georgia, Illinois,

Kentucky, Louisiana, Maryland, Mississippi, Nebraska, New York, North

Carolina, Ohio, Texas, Washington, and West Virginia]; Phelps v. Wyeth, Inc.

(D.Or. 2012) 857 F.Supp.2d 1114, 1121 [Oregon law]; Stanley v. Wyeth, Inc.

38

(La.Ct.App. 2008) 991 So.2d 31, 33-34 [noting the “numerous cases where the

negligent misrepresentation claims were . . . preempted by . . . a state’s enactment

of products liability law”].)

At core, what Novartis seems to want is more than just an exception to the

general duty of care applicable in California — an exception constructed to avoid

liability where a biologically equivalent product is sold and the warning label used

is required by federal law to be the label that the brand-name manufacturer

controls. Perhaps because there is no logical basis to justify such an exception,

Novartis instead seeks a more categorical result, though one no easier to justify —

i.e., an unequivocal declaration that California law relieves a manufacturer of any

failure-to-warn liability relating to another manufacturer’s products. True: An

exception to California’s general duty of care is ordinarily applicable to relieve a

manufacturer of the duty to warn consumers about a product’s risks where the

product is that of another manufacturer. (O’Neil, supra, 53 Cal.4th at pp. 364-

366.) For good reason: A product manufacturer ordinarily will have no control

over the design or safety of another manufacturer’s product, the other

manufacturer’s use of dangerous materials, or any warnings the other

manufacturer might place on the product. (Id. at pp. 350, 365.) Nor would there

be any reason to think that a manufacturer has the ability to influence a customer’s

decision to buy another manufacturer’s product. (Id. at p. 365.) Without such

predicate connections between one manufacturer and another, it would be difficult,

if not impossible, for a manufacturer to foresee the dangers lurking in the

seemingly limitless number of other products that might be used with or in its

product. (Ibid.) But prescription drug markets are different. They present the

unusual situation where one entity’s misrepresentations about its own product

foreseeably and legally “contributed substantially to the harm” caused by another

entity’s product (i.e., the generic drug bearing the warning label drafted by the

39

brand-name manufacturer). (O’Neil, at p. 362.) That key circumstance

distinguishes the situation here from those involving the general run of products.

The negligence causes of action are potentially viable because of the

allegedly deficient representations in Novartis’s warning label. Novartis is not

being sued for dangers inherent in the generic terbutaline manufactured by some

other entity. Nor do plaintiffs claim that any product manufactured by Novartis

caused them harm. They claim instead that allegedly deficient representations and

omissions in Novartis’s warning label caused them harm. The fact that Novartis

also manufactured a product is extrinsic to the analysis and does not insulate it

from liability for its alleged misrepresentations. (See Conte, supra, 168

Cal.App.4th at pp. 109-110; accord, Weeks, supra, 159 So.3d at p. 672 [“the [tort]

claims are not based on the manufacturing of the product but instead allege that

the label — drafted by the brand-name manufacturer and required by federal law

to be replicated verbatim on the generic version of the medication — failed to

warn”].)

B. Whether Warning Label Liability Was Extinguished as a Matter of Law

When Novartis Divested Ownership of Brethine

We have determined that Novartis owed a duty of care to those who were

prescribed Brethine or its generic bioequivalent in reliance on Novartis’s warning

label. Novartis claims that the demurrer should nonetheless be sustained without

leave to amend on the ground that it sold the Brethine NDA to aaiPharma in 2001

and no longer had control over its warning label in 2007, when plaintiffs’ mother

was prescribed terbutaline. So we now consider whether Novartis should be

relieved of all liability for its allegedly negligent failure to update the label as a

matter of law, despite the fact that aaiPharma continued using the label Novartis

had crafted.

40

Plaintiffs fault Novartis. But they do not claim the company was

responsible for any negligent acts or omissions after the transfer of ownership.

After all, under FDA regulations, only the current NDA holder has the authority to

unilaterally modify the drug’s warning label. Plaintiffs claim instead that they

were harmed by Novartis’s failure to update the label prior to transferring the

NDA to aaiPharma, in that aaiPharma continued to use the same warning label

until at least 2007, when their mother was prescribed terbutaline. In effect,

plaintiffs claim that the Brethine warning label was deficient at the time Novartis

transferred the NDA –– and it was reasonably foreseeable that it would remain

deficient, given the incentives facing any successor manufacturer.

To address this aspect of plaintiffs’ claim, we must determine whether to

recognize an exception to a brand-name manufacturer’s duty to warn. Is a brandname

drug manufacturer’s duty to warn extinguished simply because the

deficiency in the label caused the injured plaintiff to be exposed to the drug after

the manufacturer had transferred the NDA to a successor? Foreseeablity of harm

is the touchstone of our duty analysis. (Kesner, supra, 1 Cal.5th at p. 1148.)

Plaintiffs allege (or claim they can allege) that Novartis negligently provided

inaccurate and incomplete warnings about the safety of its drug, that it was

foreseeable the new NDA holder (aaiPharma) would continue to use Novartis’s

warning label without modification, that their mother’s physician relied on the

deficient warning label drafted by Novartis and reiterated by aaiPharma in

prescribing terbutaline, and that they were harmed in utero by the terbutaline

ingested by their pregnant mother. Whether aaiPharma would also be liable for

any deficiencies in its warning label should not, in plaintiffs’ view, automatically

negate Novartis’s culpability.

41

1. Foreseeability and related factors

We explained above why it was foreseeable that Novartis’s failure to

update the Brethine warning label could affect fetuses exposed to the generic

version of the drug in utero. And there is no dispute that plaintiffs have alleged

injury. Although Novartis was no longer responsible for updating the warning

label at the time plaintiffs’ mother was prescribed the drug, aaiPharma was using

the same label it had inherited from Novartis. According to plaintiffs, neither

NDA holder had sufficient financial incentive to update the label: Nearly half of

all prescriptions for Brethine or its generic equivalent were to slow preterm labor.

Under the circumstances, it was certainly foreseeable that aaiPharma would be no

more conscientious about updating the warning label than Novartis allegedly had

been.

Novartis contends the connection between its alleged negligence and

plaintiffs’ injury was extremely remote, as it had ceased producing the drug six

years before the injury. But it is not clear why liability should turn on Novartis’s

role in the manufacturing process. What warning label liability stems from is

Novartis’s failure to warn about a drug’s risks, not its production of a defective

drug. The complaint alleges that Novartis and aaiPharma were concurrent

tortfeasors whose liability stemmed from failure to warn, because each negligently

failed to update the warning label.

We agree that Novartis’s failure to update the warning label could

foreseeably cause harm to plaintiffs. Under the circumstances arising from the

federal regulatory regime for prescription drugs, a successor manufacturer’s

negligent failure to update the warning label is foreseeable. According to federal

regulatory rules, a successor brand-name drug manufacturer has no choice but to

use the former manufacturer’s warning label — or a warning label at least as

strong as the one used by the previous brand-name manufacturer — unless

42

directed otherwise by the FDA. (Wyeth, supra, 555 U.S. at p. 568 [“Generally

speaking, a manufacturer may only change a drug label after the FDA approves a

supplemental application”]; see 21 C.F.R. §§ 314.70(b)(2)(v), (c)(6)(iii),

314.72(b).) Unlike other product manufacturers (cf. conc. & dis. opn., post, at pp.

2-3), a brand-name drug manufacturer knows that, without FDA action, a

successor manufacturer will produce a drug identical to the original in ingredients

and design, and bearing an identical warning label (or a label that is at least as

strong as the one used by the former manufacturer). (Cf. Cadlo v. Owens-Illinois,

Inc. (2004) 125 Cal.App.4th 513, 516 [affirming summary judgment in favor of a

former asbestos insulation manufacturer where there was no evidence the

manufacturer “had an actual connection with the design, manufacture or

distribution” of the product causing harm]; id. at p. 520 [distinguishing cases

where “the maker of the misrepresentation reasonably foresaw that the

intermediary would repeat the misrepresentation to another person”].)6 Because

nearly half of all terbutaline prescriptions at the time of sale were written to

prevent premature labor, it was also reasonably foreseeable that aaiPharma would

be reluctant to add warnings about the risks to fetal brain development. In sum, a

successor drug manufacturer’s negligent conduct can be “ ‘derivative of [the

brand-name drug manufacturer’s] allegedly negligent conduct’ ” and thus

foreseeable. (Kesner, supra, 1 Cal.5th at p. 1148.)



6 Nor can the concurring and dissenting opinion derive any support from the

scattering of federal district court cases involving a challenge to the adequacy of a

medical device label. Federal law preempts state tort actions based on deficient

warnings for medical devices. (Riegel v. Medtronic, Inc. (2008) 552 U.S. 312,

329; cf. Wyeth, supra, 555 U.S. at p. 574 [“despite its 1976 enactment of an

express pre-emption provision for medical devices, [citation], Congress has not

enacted such a provision for prescription drugs”].)

43

Novartis highlights the six years that elapsed between its surrender of the

NDA for the drug at issue in this case and the decision to prescribe terbutaline to

plaintiffs’ mother. Yet the gap between the transfer of this particular NDA and the

time at which plaintiffs’ mother was prescribed terbutaline does not bear on the

question of duty, “which must be addressed at a higher level of generality.”

(Kesner, supra, 1 Cal.5th at p. 1158.) In determining whether to create an

exception to a brand-name drug manufacturer’s duty of care, we do not evaluate

“ ‘whether a particular plaintiff’s injury was reasonably foreseeable in light of a

particular defendant’s conduct,’ ” but “ ‘whether the category of negligent

conduct at issue is sufficiently likely to result in the kind of harm experienced that

liability may appropriately be imposed . . . .’ ” (Cabral, supra, 51 Cal.4th at p.

772.) So the relevant inquiry is whether a successor drug manufacturer is

sufficiently likely to continue using the warning label it inherited from the prior

brand-name manufacturer, even when that label was deficient at the time the NDA

was transferred.

It is true enough that a successor drug manufacturer has an obligation,

under state as well as federal law, to ensure adequacy of the warning label. But

the scenario at issue here implicates whether a successor drug manufacturer is

sufficiently likely –– as a matter of law –– to modify the warning label when the

brand-name manufacturer, which labored under an identical obligation,

negligently failed to do so. In such circumstances, it is at least plausible that a

successor manufacturer may choose to undertake only a cursory investigation of

the medical literature, on the assumption that the prior manufacturer must have

done a more thorough inquiry during the period that it was responsible for

maintaining the warning label. This option will seem especially attractive when

the prior manufacturer has greater resources or expertise than its successor. A

successor manufacturer may also undertake an adequate inquiry but make no

44

changes to the label in close cases, partially or entirely trusting the judgment of the

prior manufacturer. Or a successor manufacturer, like the prior manufacturer, may

fear an adequate warning would damage the market share for the drug and balance

its lost revenue and potential exposure in the same way as the prior manufacturer.

Indeed, plaintiffs claim that neither NDA holder wanted to jeopardize Brethine’s

use as a tocolytic, which accounted for almost half of the drug’s market share.

Any or all of these factors could explain why a drug’s warning label may prove

“stickier” than what is optimal to protect public safety at a reasonable cost, and

why a successor drug manufacturer would not be categorically distinguishable

from the prior manufacturer in its likelihood of being conscientious about its

obligations to disclose relevant risks.

Under the “general” rule, “ ‘ “every person has a right to presume that

every other person will perform his duty and obey the law.” ’ ” (Webb v. Special

Electric Co., Inc., supra, 63 Cal.4th at p. 191.) But we have never allowed a

defendant to excuse its own negligence as a matter of law simply by asserting that

someone else should have picked up the slack and discharged the duty at issue.

(See Stewart v. Cox (1961) 55 Cal.2d 857, 864 [“The fact that a third person does

not perform his duty to protect the plaintiff from harm, either because he makes no

effort or through his negligence does not succeed, is not a superseding cause”].)

Nor have we permitted a negligent actor to evade liability simply because another

party may also be liable for a similar tort. (See, e.g., Beacon Residential

Community Assn. v. Skidmore, Owings & Merrill LLP (2014) 59 Cal.4th 568, 583

(Beacon); accord, Humble Oil & Refining Co. v. Martin (Tex. 1949) 222 S.W.2d

995, 1001 [“there is a distinction between the general axiom that a person is not

bound to anticipate the negligence of others and the idea that one may always

discharge a duty of due care to the public by relying on performance by another of

the same duty owed by the latter”].) So while “ ‘[a] person who, himself, is

45

exercising ordinary care, has a right to assume that others, too, will perform their

duty under the law’ ” — and thus may not be negligent in failing to anticipate

injury that results “ ‘ only from a violation of law or duty by another’ ” — the

general rule does not apply when “ ‘it is reasonably apparent to one, or in the

exercise of ordinary care would be apparent to him, that another is not going to

perform his duty.’ ” (Stickel v. San Diego Elec. Ry. Co. (1948) 32 Cal.2d 157,

166, first and second italics added; see id. at pp. 166-167 [“It is but a statement as

to that common type of negligence, the unreasonable failure to observe what is

going on about one, including the negligence of others. ‘One may not continue to

assume that the law is being observed after knowing or having an opportunity, by

the use of reasonable care, to know that it is not being observed’ ”]; Harris v.

Johnson (1916) 174 Cal. 55, 58 [“ ‘The general rule . . . that every person has a

right to presume that every person will perform his duty’ ” applies only “ ‘in the

absence of reasonable ground to think otherwise’ ”].) Few if any entities would be

in a position to know better that the law “ ‘is not being observed’ ” (Stickel, at p.

167) than a brand-name drug manufacturer that itself had negligently failed to

update the label. So the assumption underlying the brand-name drug

manufacturer’s duty is not at all “that successor corporations will routinely ignore

th[eir] duty.” (Conc. & dis. opn., post, at p. 5, italics added.) It’s that when a

brand-name drug manufacturer has ignored its own duty, there is a risk that the

successor manufacturer will adopt the same strategy. Under these circumstances,

categorically justifying a manufacturer’s neglect of that risk requires heroic, and

ultimately unreasonable, assumptions distinguishing an original brand-name

manufacturer’s behavior from that of its successors. For these reasons, we find it

reasonably foreseeable that a successor drug manufacturer could continue to use

the same label it inherited, even when the label was deficient.

46

2. Considerations of public policy

According to Novartis, the policy of preventing future harm would not be

advanced by subjecting the brand-name drug manufacturer to liability after it has

already divested itself of the drug and no longer has control over the warning

label. But in examining the prevention of future harm, we undertake the duty

analysis “look[ing] to the time when the duty was assertedly owed.” (Kesner,

supra, 1 Cal.5th at p. 1150.) It is during the time Novartis owned the drug that

both its legal duty and its power to discharge that duty converge. At that point,

Novartis did have control over the warning label and could have modified it,

without waiting for FDA approval, to warn of the risks to fetal brain development.

Recognizing a brand-name drug manufacturer’s potential responsibility for

injuries proximately caused by deficiencies in its warning label –– regardless of

whether the injury occurred before or after divestment — provides a further

incentive to the brand-name manufacturer to update the label as soon as it knows

(or should have known) of the unwarned risks. Consider, on the other hand, the

implications of allowing the brand-name manufacturer to shield itself from

liability as soon as it transfers ownership to another manufacturer, as Novartis

proposes. What such a rule would do is encourage an economically rational

brand-name manufacturer to transfer the NDA, rather than add a warning to the

label, since an updated label would diminish the utility (and thus the value) of the

drug.7 Such a scenario obviously poses greater risks to consumer safety relative to

the alternative.



7 This case does not present the question, and we do not decide, whether a

brand-name manufacturer would remain liable for deficiencies in its warning label

when the FDA has formally withdrawn its approval of the NDA and has

determined “that the drug was voluntarily withdrawn from sale for reasons other

than effectiveness or safety.” (Lasker, supra, 82 Def. Counsel J. at p. 306; see 21

C.F.R. § 314.150; 78 Fed.Reg., supra, at p. 67993.)

47

Novartis counters with a different scenario. It claims that under plaintiffs’

regime, a successor brand-name drug manufacturer would have an incentive to

maintain the identical label without change so that the former brand-name

manufacturer would be forced to share in any liability. We are skeptical. When it

is economically rational for the manufacturer to update the label, it will update the

label –– regardless of the prospect that a prior manufacturer might share in the

liability for its own negligent failure to update the label. Even in a marginal case,

though, it does not seem especially likely that a successor drug manufacturer

which knows or should know of an unwarned risk would choose to leave a

warning label unchanged simply to preserve the possibility that –– if the label

remained the same as under the former manufacturer –– the former manufacturer

could be a codefendant in a future tort action. It seems implausible that a

successor manufacturer would take that gamble if its proportional share of fault

would be ever increasing as medical research became more confident about the

link between the drug and some adverse effect. After all, the successor

manufacturer could avoid liability altogether by updating the label to warn about

the risk.

The more substantial danger is that neither manufacturer will have

sufficient incentive to update the label. Unless there is warning label liability, it

will be economically rational in some circumstances for a brand-name

manufacturer to offload the drug to a successor rather than update the warning

label. And if the brand-name manufacturer fails to update the label to disclose a

known or knowable risk, economic interests and simple inertia may lead the

successor manufacturer to the same strategy. (See ante, at pp. 44-45.) The better

rule is to provide appropriate incentives for the brand-name manufacturer to

update the warning label at the earliest possible time, given that the successor

manufacturer cannot remove any aspect of the warning without FDA approval.

48

To determine how best to incentivize a drug manufacturer to provide

prompt warnings, we turn to the very factors on which Novartis trains its attention:

the extent of the duty’s burden on the defendant and the consequences to the

community. Novartis complains first that plaintiffs’ proposed rule would lead to

immeasurable and perpetual liability for brand-name drug manufacturers. This

appears to be an overstatement. Only during the time it holds the NDA does the

brand-name drug manufacturer have a duty of care. Although a breach of that

duty can have enduring effects — effects that do not magically disappear merely

because the brand-name manufacturer no longer holds the NDA — a plaintiff

would still need to prove that the injury was foreseeable at the time the brandname

manufacturer held the NDA, that the brand-name manufacturer’s deficient

label proximately caused the injury, and that the prescribing physician relied on

the brand-name manufacturer’s misrepresentations or omissions. The passage of

time would naturally tend to undermine a plaintiff’s ability to prove that an injury

was foreseeable at that earlier stage,8 that the physician actually relied on the

defendant’s warning label, or that the defendant’s negligence proximately caused

injury. (See Beacon, supra, 59 Cal.4th at p. 583.) An extended period of time

also presupposes a lengthy latency period before an injury (or its connection to the

drug) manifested itself, further complicating the showing of foreseeability,

reliance, or causation. (Cf. PLIVA, supra, 564 U.S. at p. 625, fn. 9 [“the FDA



8 Indeed, approximately half of the studies cited in the first amended

complaint to demonstrate a link between terbutaline exposure in pregnancy and

fetal brain development postdated Novartis’s sale to aaiPharma. To avoid the

distortion caused by hindsight bias, trial courts should be careful to protect the jury

from needlessly being exposed to or considering scientific studies connecting a

drug to some harm where those studies postdate transfer of the NDA.

49

informs us that ‘[a]s a practical matter, genuinely new information about drugs in

long use . . . appears infrequently’ ”].)

Yet the question before us involves neither causation nor these other

elements of negligence. What role the six-year gap between Novartis’s transfer of

the NDA to aaiPharma and plaintiffs’ exposure to terbutaline might play in

plaintiffs’ ability to prove these remaining elements is beyond the scope of this

proceeding. We granted review to decide only the threshold question of a brandname

drug manufacturer’s duty of care and therefore have no occasion to address

other arguments Novartis might advance to defeat liability. (See Kesner, supra, 1

Cal.5th at p. 1157.)9 But we reject Novartis’s contention that a finding of duty

will result in perpetual liability for brand-name drug manufacturers as well as the

burden of a trial to address every claim of harm. Time’s effect on causation, while



9 Recognizing a brand-name drug manufacturer’s duty of care in these

circumstances does not prevent the manufacturer from arguing in a given case that

it did not breach its duty given the scientific knowledge at the time; that its label

could not have proximately caused the harm given the passage of time between the

transfer of the NDA and the plaintiff’s exposure to the drug, as well as the

successor’s exclusive responsibility for promoting the assertedly dangerous offlabel

use of the drug (see Lyman v. Pfizer, Inc. (D.Vt. July 20, 2012, No. 2:09-CV-

262) 2012 WL 2970627, *17); or that its disclosure of the unwarned risks to the

successor manufacturer severed any link between its own label and the harm. But

our task here is not to decide whether there should be “an exception to the general

duty of reasonable care on the facts of the particular case before us, but whether

carving out an entire category of cases from that general duty rule is justified by

clear considerations of policy.” (Cabral, supra, 51 Cal.4th at p. 772.)

The concurring and dissenting opinion finds “perhaps most troubling” the

court’s unwillingness to “predict” when the gap between transfer of the NDA and

exposure to the drug will be so remote as to preclude a finding of proximate cause.

(Conc. & dis. opn., post, at p. 9.) But neither party has briefed the issue of

proximate cause, nor is proximate cause fairly encompassed within the issue

presented –– indeed, the issue presented involves exclusively the tort law element

of duty. Novartis remains free to contest the existence of proximate cause — as

well as any of the other elements of negligence and negligent misrepresentation.

50

ordinarily a question of fact, becomes a question of law “ ‘where the facts are such

that the only reasonable conclusion is an absence of causation.’ ” (State Dept. of

State Hospitals v. Superior Court (2015) 61 Cal.4th 339, 353; see id. at p. 357

[sustaining demurrer where the theory of causation was “conjectural, depending

on a long series of determinations”]; accord, Lyman v. Pfizer, Inc., supra, 2012

WL 2970627 at p. *17 [affirming grant of summary judgment to a former brandname

drug manufacturer on causation grounds].) Similarly, the question of breach

can be decided as a matter of law where “no reasonable jury could find the

defendant failed to act with reasonable prudence under the circumstances.”

(Cabral, supra, 51 Cal.4th at p. 773.) The burden of a potential trial on a brandname

drug manufacturer that, under the facts presented, acted unreasonably in

failing to update the warning label before transferring the NDA — and whose

negligence proximately caused harm to those exposed to the drug — is not a

compelling justification for carving out an entire category of cases from the

general duty of reasonable care. (See id. at p. 772.)

Moreover, the greater the gap between transfer of the NDA and the

plaintiffs’ exposure to the drug, the greater the likelihood that the NDA would

have been transferred to yet another manufacturer, which would multiply the

number of potential defendants available to share responsibility for damages.

Because a defendant’s liability for noneconomic damages is not joint but several

(Civ. Code, § 1431.2, subd. (a)), a negligent brand-name manufacturer would be

liable for noneconomic damages only in an amount that was directly proportional

to its percentage of fault. (Ibid.)

Indeed, a brand-name manufacturer could entirely avoid the prospect of

extended exposure by including an indemnification provision when it transferred

ownership of the NDA. (See, e.g., Conte, supra, 168 Cal.App.4th at p. 95, fn. 1.)

This might lower the sales price of the brand-name drug in the transaction, but not

51

in any way that fails to reflect the true costs and benefits of being the NDA holder

or that is unfair to the seller. Meanwhile, an indemnification provision may have

the salutary effect of focusing both the seller and the purchaser, at a critical time,

on the existence of any known or knowable risks not reflected in the warning

label. And, as before, Novartis identifies no reason why it could not insure against

the effects of any negligence related to the warning label for its drug. (See

Vasilenko v. Grace Family Church (2017) 3 Cal.5th 1077, 1091.) Commercial

general liability insurance policies cover injuries that accrue from multiple

occurrences over a period of years (see Montrose Chemical Corp. v. Admiral Ins.

Co. (1995) 10 Cal.4th 645), and tail coverage is available for injuries caused by

the insured that did not manifest themselves until well after the manufacturer

either sold the product or shut down its operations. (See State of California v.

Continental Ins. Co. (2012) 55 Cal.4th 186, 195-196.)

A somewhat analogous situation lay at the heart of a case the court

addressed recently. In Centinela Freeman Emergency Medical Associates v.

Health Net of California, Inc. (2016) 1 Cal.5th 994 (Centinela Freeman), we

considered the circumstances under which a health care service plan could transfer

its financial responsibility to pay for its enrollees’ emergency medical services to

its contracting medical providers. Under state and federal law, licensed hospitals

are required to provide emergency medical care to anyone, regardless of the

patient’s ability to pay. A health care service plan, in turn, is required to

reimburse a noncontracting emergency service provider for necessary services, but

may delegate this responsibility to another entity, such as an individual practice

association (IPA). (Id. at pp. 1000-1001.) The health care service plans in

Centinela Freeman delegated their financial responsibility for emergency services

to their contracting IPAs, which were (or became) financially insolvent and

eventually went out of business. (Id. at p. 1001.) When the IPAs failed to

52

reimburse the plaintiff emergency service providers for the care they had provided

to enrollees of the defendant health care services plans, the plaintiffs sued the

health plans for payment. (Ibid.)

The defendant health care service plans made an argument that echoed what

Novartis argues here: that they had lawfully transferred their legal responsibilities

to another entity and had therefore terminated any duty of care. We unanimously

rejected the argument that the delegation of financial responsibility to an IPA

necessarily relieved the health care service plans of any obligation to pay for its

enrollees’ covered emergency care. (Centinela Freeman, supra, 1 Cal.5th at p.

1001-1002.) What we held instead was that a health care service plan owes

certain duties to noncontracting emergency service providers: first, a duty at the

outset not to delegate its financial responsibility to an IPA “that it knew or should

have known would not be able to pay for emergency service and care provided to

the health plan’s enrollees” (id. at p. 1002); and second, a duty not to continue or

renew a delegation to an IPA “when it knows or should know that there can be no

reasonable expectation that its delegate will be able to reimburse noncontracting

emergency service providers for their covered claims.” (Ibid.)

Centinela Freeman tends to undermine Novartis’s absolutist position that a

lawful transfer of its duty to another entity necessarily terminated its liability for

its own negligence. Under our ruling in Centinela Freeman, a health care service

plan remains responsible for the costs of its enrollees’ emergency medical care,

despite a lawful delegation of that financial responsibility, if the plan knows or

should know the IPA would be unable to fulfill that financial responsibility.

(Centinela Freeman, supra, 1 Cal.5th at pp. 1001-1002.) Here, we find that a

brand-name drug manufacturer can be liable for the effects of its deficient warning

label, despite transferring the NDA to a successor, if the harm is reasonably

foreseeable and is proximately caused by the label.

53

Novartis’s argument echoes the position we rejected in Centinela Freeman.

Although some differences exist between these two scenarios, they do not

undermine our conclusion that a brand-name drug manufacturer owes a duty to all

those who may foreseeably and proximately be harmed by its deficient warning

label. Unlike the duty we recognized in Centinela Freeman, warning label

liability does not constitute a continuing duty of care. (See Centinela Freeman,

supra, 1 Cal.5th at p. 1019 [“We agree that a health care service plan has a

continuing duty of care to noncontracting emergency service providers”].) The

conduct giving rise to a brand-name drug manufacturer’s liability can occur only

during the period that it holds the NDA. Negligent conduct during that period

may have effects that extend beyond the transfer of the NDA, but a brand-name

manufacturer is not subject to liability for any of its actions that occur after

transfer of the NDA. Moreover, in this case, unlike in Centinela Freeman, we are

analyzing a duty to prevent physical harm. Such a duty is broader than the duty to

prevent pecuniary loss. (Rest.2d Torts § 311, com. a.)

Novartis renews its claim that warning label liability would severely chill

both the innovation and marketing of new drugs if imposed after the brand-name

manufacturer exits the market. Yet once again, it offers neither evidence nor a

persuasive rationale to support its contention –– and no reason for us to prefer

some unknown increment of drug development over the urgent need to

compensate a victim whose injury was foreseeably and proximately caused by a

brand-name manufacturer’s negligence. (See Carlin, supra, 13 Cal.4th at p. 1117

[“Upjohn offers no clear or sufficient basis for concluding that research and

development will inevitably decrease” because of failure-to-warn claims]; id. at p.

1116, fn. 6 [discounting the risk of overwarning because of the lack of evidence];

cf. Kesner, supra, 1 Cal.5th at p. 1156 [noting the defendants “cite no evidence to

suggest such [preventive] measures would have been unreasonably costly”].)

54

After all, the duty imposed here merely reinforces the brand-name drug

manufacturer’s existing duty to update and maintain the warning label. It does not

require a brand-name drug manufacturer to do anything new.

We explained earlier why significant moral blame attaches to the failure to

warn about a drug’s risks when the brand-name drug manufacturer knew or should

have known about those risks. The fact that the brand-name manufacturer has

since exited the market does not alter the calculus. Under plaintiffs’ theory, the

actionable conduct occurred while the manufacturer still had control over the

warning label. Had Novartis updated the warning label before surrendering the

NDA, the federal regulations make it very likely that the warning would have

remained on the label in 2007. (See Wyeth, supra, 555 U.S. at p. 568.) Although

it can be difficult to assess the full extent of moral blame before a factual record

has been developed (Kesner, supra, 1 Cal.5th at p. 1151), concealment of a drug’s

effects on the fetal brain for the purpose of preserving the drug’s share of the

premature labor drug market and thus inflating the sales price of the NDA would

be especially objectionable. So Novartis fails to show how “ ‘clear considerations

of policy’ ” justify a categorical exception to the duty of care. (Kesner, supra, 1

Cal.5th at p. 1144.) What the Rowland factors support instead is the conclusion

that Novartis had a duty to warn about the potential risks of its drug, regardless of

whether the consumer received the brand-name or generic bioequivalent, and that

liability for the asserted breach of that duty did not end as a matter of law at the

moment Novartis sold its rights to aaiPharma, an allegedly concurrent tortfeasor.

A contrary rule would convey to Novartis and to similarly situated drug

manufacturers the unjustified benefit of an exception to the general duty of care,

incentivizing brand-name drug manufacturers that know or should know of

unwarned risks to unload a problematic drug on another entity instead of

modifying the drug’s warning label to include those hazards.

55

The concurring and dissenting opinion makes much of the fact that no other

jurisdiction has yet recognized a brand-name drug manufacturer’s duty to maintain

a warning label in these circumstances. The legal landscape was just as bare when

the Court of Appeal recognized a brand-name drug manufacturer’s duty to

consumers of the generic bioequivalent drug (see Conte, supra, 168 Cal.App.4th at

p. 101) — a duty we unanimously affirm here. Rarely, if ever, do jurisdictions

face precisely the same jurisprudential questions at the same time, nor is our

system premised on the idea that law congeals across jurisdictions. To the

contrary, the common law incorporates the possibility of change as a foundational

premise: “[t]he law of torts is anything but static, and the limits of its

development are never set. When it becomes clear that the plaintiff’s interests are

entitled to legal protection against the conduct of the defendant, the mere fact that

the claim is novel will not of itself operate as a bar to the remedy.” (Prosser &

Keeton, Torts (5th ed. 1984) § 1, p. 4.) Indeed, even if we acknowledge the value

of reducing uncertainty where possible, what is critical in common law

adjudication is not that all jurisdictions rapidly converge on a particular

understanding of tort liability. Instead a court must carefully weigh whether an

existing rule should apply in a particular context under current conditions.

Applying the Rowland factors to address that context, we conclude that brandname

drug manufacturers owe a duty to those whose injuries are foreseeably and

proximately caused by the manufacturer’s deficient warning label.

We do not doubt the wisdom of crowds in some settings. But the value of

an idea conveyed by or through a crowd depends not on how loudly it is

proclaimed or how often it is repeated, but on its underlying merit relative to the specific issue at hand. Despite the impressive case authority Novartis has

collected on its behalf, none of it purports to interpret California law. Yet it is California law that we must construe and apply in this case.

In doing so, we find that brand-name drug manufacturers have a duty to use

ordinary care in warning about the safety risks of their drugs, regardless of

whether the injured party (in reliance on the brand-name manufacturer’s warning)

was dispensed the brand-name or generic version of the drug. We also conclude

that a brand-name manufacturer’s sale of the rights to a drug does not, as a matter of law, terminate its liability for injuries foreseeably and proximately caused by deficiencies present in the warning label prior to the sale. We therefore affirm the Court of Appeal.