Black Box Warnings in Prescription Drug Labeling: Results of a Survey of 206 Drugs
Prescription drugs are unavoidably associated with adverse effects, but the ben-
efits of using such drugs outweigh the accompanying risks so long as the expert judg-ment of a licensed practitioner is first applied in the decision to use a drug for aparticular patient. Most risks for prescription drugs appear in descending order ofimportance in the drug’s labeling, i.e., the package insert. The Food and Drug Ad-ministration (FDA), however, may require special health risks (particularly those thatmay lead to death or serious injury) to be identified in the labeling in a prominentlydisplayed box or “black box” warning. FDA generally reserves boxed warnings forserious or life-threatening risks that best can be minimized by conveying critical in-formation to the prescribing doctor in a highlighted manner. A decision by FDA to setapart a particular drug with a black box warning has serious implications for thelicensed practitioner, the pharmacist, the patient, the pharmaceutical manufacturer,and the distributor. Nevertheless, FDA has not articulated specifically the scope ofstudies it relies on or the special circumstances in which the agency would impose thisspecial warning.
To identify FDA’s previous practices in order to describe the agency’s rationale
and bases for imposing the requirement of black box warnings, the authors surveyed206 black box warnings in package inserts in an attempt to determine both the kind ofinformation the agency sought to impart to the physician and the types of humanstudies relied upon by the agency.
Under the Federal Food, Drug, and Cosmetic Act (FDCA),1 a drug is misbranded
if it lacks “adequate directions for use.”2 Medications that cannot be used safely byconsumers without the diagnosis and supervision of a physician are designated asprescription drugs and can be dispensed only on the order of a licensed medical prac-titioner. FDA’s view is that “the primary objective of prescription drug labeling is toprovide the essential information the practitioner needs to use the drug safely and
* Dr. Beach is an Associate in the law firm of Hyman, Phelps & McNamara, P.C., Washington, D.C. ** Dr. Faich is a consultant with Pharmaceutical Safety Assessments, Inc., Narberth, PA. *** Ms. Bormel is Associate Counsel for Scientific Affairs for U.S. Pharmacopeia. **** Mr. Sasinowski is a Director in the law firm of Hyman, Phelps & McNamara, P.C., Washington, D.C.
The authors wish to acknowledge Craig B. Mendelsohn, M.D., J.D., and Christina Ammerman for their
contribution in compiling and organizing the data.
1 Pub. L. No. 75-717, 52 Stat. 1040 (1938) (codified as amended 21 U.S.C. §§ 301 et seq. (1994)). 2 21 U.S.C. § 352(f)(1) (FDCA § 502(f)(1)).
effectively in the care of patients”3 such as “the directions for use and cautionarystatements, if any.”4
Thus, the manufacturer of a prescription drug is required to provide adequate
labeling for practitioners licensed by law to be able to administer or dispense the drugsafely and for the purposes for which it is intended.5 Therefore, the manufacturermust provide information to the health professional or the so-called learned interme-diary,6 whose burden it is to decide whether to prescribe the medicine for a particularpatient, knowing the risk and benefits associated with the drug.7
FDA may determine that a drug is misbranded if the agency finds the labeling as
false or misleading8 for failing to reveal certain facts material to the customary orusual conditions for use of the drug.9 FDA asserts that the FDCA requires that “[a]nadequate warning of possible danger must appear on all [drug] labeling . . . or theproduct is misbranded.”10
Thus, in the labeling for any prescription drug, FDA requires that cautionary
information be categorized according to the relative severity of the hazard and thedegree to which the risk has been substantiated. Accordingly, drug package insertscontain many detailed paragraphs of information about side effects to assist physi-cians in making prescription decisions. Depending on the relative severity of the haz-ard, topic headings in prescription drug labeling are set forth in descending order ofimportance as “Contraindications,” “Warnings,” “Precautions,” and “Adverse Reac-tions.”11 Therefore, “as known adverse side effects increase in intensity and severity,the manufacturer’s warning in respect to the drug’s potential for harm should accord-ingly ascend to a higher category.”12
B. “Boxed” or “Black Box” Warnings
Although the large majority of prescription drugs have risks reported in the
Contraindications, Warnings, Precautions and Adverse Reactions sections of theirlabeling,13 “[s]pecial problems, particularly those that may lead to death or seriousinjury, may be required by [FDA] to be placed in a prominently displayed box”14 — aso-called “boxed warning” or “black box warning.” FDA reserves boxed warnings forthose risks that best can be minimized by conveying the information to the prescribingdoctor in this highlighted manner.15 Further, boxed warnings are “ordinarily . . . based on clinical data.”16 Even though FDA may require boxed warnings if there arespecial problems associated with a drug, particularly those that may lead to death orserious injury,17 the agency rejected a comment in rulemaking that information re-
3 40 Fed. Reg. 15,392 (Apr. 7, 1975). 4 21 U.S.C. § 353(b)(2) (FDCA § 503(b)(2)). 5 21 C.F.R. § 201.100 (1997). 6 United States v. Vital Health Prod., Ltd., 786 F. Supp. 761, 786 (E.D. Wis. 1992). 7 United States v. Evers, 643 F.2d 1043, 1052 (5th Cir. 1981). 8 21 U.S.C. § 352(a) (FDCA § 502(a)). 9 21 U.S.C. § 321(n) (FDCA § 201(n)). 10 39 Fed. Reg. 33,229 (Sept. 16, 1974). 11 21 C.F.R. §§ 201.56, 201.57 (1997). 12 McFadden v. Haritatos, 448 N.Y.S.2d 79, 81 (N.Y. App. Div. 1982). 13 44 Fed. Reg. 37,434 (June 26, 1979); 21 C.F.R. § 201.57(d)-(g). 14 21 C.F.R. § 201.57(e). 15 Lars Noah, The Imperative To Warn: Disentangling the “Right To Know” From the “Need To Know”About Consumer Product Hazards, 11 YALE J. ON REG. 293 (1994).
16 21 C.F.R. § 201.57(e). See also 44 Fed. Reg. at 37,447; 40 Fed. Reg. at 15,397. 17 21 C.F.R. § 201.57(e).
lated to clinical benefit or use restrictions for accelerated approval drugs always re-quires a boxed warning.18 The agency may refer questions of whether a boxed warn-ing should be included in the labeling of a drug to an FDA Advisory Committee forthe particular drug.19 Although FDA is not obligated to follow recommendations fromthe Advisory Committee,20 “it usually does.”21
For example, FDA has imposed a black box warning in the package insert for
Abbott Laboratories’ anti-epileptic drug Depakene® (valproic acid). The warning isbased on postmarketing reports of adverse reaction in humans indicating that thedrug could cause fatalities, and FDA required the sponsor to warn physicians to moni-tor patients in order to minimize the risks. Specifically, the black box warning pro-vides in part that:
HEPATIC FAILURE RESULTING IN FATALITIES HAS OCCURRED INPATIENTS RECEIVING VALPROIC ACID . . . . PATIENTS SHOULD BEMONITORED CLOSELY FOR APPEARANCE OF THESE SYMPTOMS.22
The further details of serious drug-associated risks as reported in the black box,
such as “[t]he frequency of these serious adverse reactions, and, if known, the ap-proximate mortality and morbidity rates for patients sustaining the reaction, whichare important to safe and effective use of the drug,” generally are reported elsewhere,such as in the “‘Adverse Reactions’ Section of the labeling.”23
Although decisions are made for each new drug independently at the time of its
approval, the agency has attempted to achieve some consistency in the labeling forsimilar prescription drugs. For example, FDA has instituted class-wide labelingchanges, including a black box warning for all angiotensin converting enzyme (ACE)inhibitors, such as Accupril® (quinapril), Altace® (ramipril), Capoten® (captopril),Lotensin® (benazepril), Monopril® (fisinopril), Prinivil® (lisinopril), Vasotec® I.V. (enalaprilat I.V.), Univasc® (moexipril) and Zestril® (lisinopril).24 The standardizedboxed warning for these drugs states that, “[w]hen used during the second and thirdtrimesters of pregnancy, ACE inhibitors can cause injury and even death to the devel-oping fetus,”25 adding that the drug should be discontinued when pregnancy is de-tected.
Although FDA’s imposition of a black box warning for prescription drugs has
important repercussions for the licensed practitioner, pharmacist, patient, the phar-maceutical manufacturer, and the distributor, FDA has not articulated clearly the
18 Preamble to New Drug, Antibiotic, and Biological Drug Product Regulations; Accelerated Approval,
57 Fed. Reg. 58,942 (Dec. 11, 1992).
19 FOOD AND DRUG ADMIN., POLICY & GUIDANCE HANDBOOK FOR FDA ADVISORY COMMITTEES, ch. XI, Center-
Specific Guidance, CDER: Specific Guidance for Topics to Be Presented Before Advisory Committees (1994)(“Consideration of special labeling features, such as boxed warnings, special limitations on use, monitoringrequirements, or patient package inserts.”)
20 44 Fed. Reg. at 37,434. 21 K. Gilpin, Company News: Panel Backs Tagamet Use In Treatment of Heartburn, N.Y. TIMES, Mar.
22 PHYSICIANS’ DESK REFERENCE 414 (49th ed. 1995). 23 21 C.F.R. § 201.57(e). 24 ACE Inhibitors Will Be Required to Carry “Boxed Warning,” Food and Drug Admin, Press Release
25 Dangers of ACE Inhibitors During Second and Third Trimesters of Pregnancy,22 FDA MED. BULL.
2 (1992); Katie Rodgers, Michael C. Vinson & W. Marvin Davis, Breakthroughs: New Drug Approvals of1995 — Part I, 140 DRUG TOPICS 84 (Feb. 5, 1996).
agency’s rationale and bases for setting apart a particular drug with this distinction.26Therefore, in order to determine the agency’s criteria for requiring a black box warn-ing, boxed warnings in package inserts were surveyed and analyzed. In the surveypresented in this article, the types of studies on which FDA relies and the kind ofinformation the agency seeks to convey to the prescribing physician in order to mini-mize serious hazards to patients were noted.
The authors analyzed all package inserts with black boxes found in the 1995
Physicians’ Desk Reference.27 First, the authors summarized the type of serious haz-ards or warnings described in the black box warnings. Next, the nature of the data thatsupported the statement in the black box was classified.
Based on the analysis, the authors devised a classification scheme for black box
where early detection of a side effect by a physician may result in interventionthat may reverse the adverse reaction (e.g., valproic acid and hepatic failure);
where a well-defined subset of patients are at higher risk for the treatment (e.g.,pregnant women and isotretinoin);
where the risks from the treatment with the particular drug may outweigh thebenefits in particular circumstances (e.g., aplastic anemia or hepatic failure andfelbamate);
where the dosing/drug interaction is critical to the risk (e.g., fatalities fromterfenadine when taken in conjunction with ketoconazole);
where the physical setting of administration of the drug or the experience andtraining of the physician is critical (e.g., because of severe toxic reactions, allpatients receiving floxuridine should be hospitalized for first course therapy);and
where the method of administration is critical (e.g., avoiding or treating extrava-sation of Alkeran).
It was assumed that the black box warning was required by FDA for some drugs basedon more than one of these special problems/risks.
Next, the authors classified the type of evidence for supporting a boxed warning
(1) mainly from controlled clinical trials that were part of the new drug application
(NDA) (e.g., HIVID (zalcitabine)); and/or
(2) by patterns of postmarket reporting (e.g., fatalities from Felbatol® (felbamate));
(3) from epidemiological surveys (e.g., cigarette smoking and oral contraceptive use);
(4) occasionally, some other miscellaneous bases.
26 Without clarification, FDA states only that it may require a boxed warning for “[s]pecial problems,
particularly those that may lead to death or serious injury . . . .” 21 C.F.R. § 201.67(e). Seealso 44 Fed. Reg. at37,447-48; 40 Fed. Reg. at 15,397. The agency has issued no guidance relevant to when and why it requiresboxed warnings.
27 PHYSICIANS’ DESK REFERENCE, supra note 22.
These other miscellaneous bases include: drugs determined to have potential for abuse;drugs included in a category that generally are recognized to warrant a specific warn-ing (such as a warning that antihypertensive drugs need periodic reevaluation in eachpatient and thyroid replacement drugs are ineffective for weight loss); drugs not stud-ied adequately in children or in pregnant women; and drugs shown to be toxic inanimals. Because “the boxed warning ordinarily . . . [is] based on clinical data,”28 theauthors excluded animal toxicity data in this analysis.
As stated, FDA-imposed black box warnings were grouped into six categories
(see table 1). Overall, the most frequent warning (n=95) was for the identification andavoidance of high-risk patients (e.g., cigarette smokers and oral contraceptive users). This category was followed by information on dosing or drug interactions (n=74). Next were warnings about the need for special training or a special setting (n=66)(these were found particularly for cancer chemotherapeutic agents).
Because several types of warnings often were found in a given black box, the
distribution was analyzed by the total number of such warnings. There were a total of375 warnings contained in the black boxes for the 206 drugs studied, or an average of1.82 black box warnings for each drug. The distribution of these warnings is given intable 1. Again, avoidance of high-risk patients and dosing/drug interaction informa-tion represented the highest relative proportions (25% and 20%, respectively) of allwarnings.
Only seventy of the 206 drugs with black box warnings had just one of the six
types of warnings. The black box had two warning types for seventy-five drugs, threewarning types for thirty-one drugs and four or more warning categories for sixty drugs. For example, Cerubidine® (daunorubicin HCl), an anticancer drug, meets four ofthese warning categories: the drug’s boxed warning states that it poses a higher riskfor patients with impaired hepatic or renal function; the drug must be given into arapidly flowing intravenous infusion — never by the intramuscular or subcutaneousroute; it should be administered only by physicians familiar with leukemia chemo-therapy in facilities with laboratory and supportive resources; and, the physician andinstitution must be capable of responding rapidly and completely to severe hemor-rhagic condition and/or overwhelming infection.
A number of drugs have risks for a well-defined subset of patients and dosing and
drug interaction issues. For example, for Toradol® (ketorolac tromethamine), anonsteroidal anti-inflammatory drug (NSAID), FDA requires a boxed warning thatits use is contraindicated in nursing mothers because of potential adverse effects onneonates, surgical candidates and patients with suspected or confirmed cerebrovascu-lar bleeding due to platelet inhibition, and patients with advanced renal impairmentdue to volume depletion. The boxed warning also cautions that dosage should beadjusted for patients sixty-five years or older and for patients under 110 pounds. Inaddition, the drug is contraindicated in patients currently receiving other NSAIDsbecause of the cumulative risk of inducing serious NSAIDs-related side effects.
Some of the warning classifications are related closely and may be difficult to
distinguish from one another. The following example illustrates the subtle differences
between a high-risk subset and a question of risks outweighing benefits. Koate®-HP(antihemophilic factor, human) is a product prepared from pooled human plasma, andis indicated for the treatment of classical hemophilia. Hemophiliacs who have hadlittle exposure to blood products, as a group, have a higher risk of developing hepatitisfrom this product. On the other hand, depending on the severity of the hemophilia,the risk for some hemophiliacs may be greater than the benefits of the product. Theblack box warns that the physician should weigh the risk/benefits for this productwith the severity of the patient’s condition.
Of the 206 prescription drugs with black box warnings, the warning appeared to
be based on a combination of sources in many instances (see table 2). Thus, there were254 sources of evidence drawn upon for the 206 warnings, or 1.23 sources per blackbox (table 2). The most frequent source was postmarketing reports (52%) followed byclinical trial data (29%). Epidemiological studies (9.4%) were cited particularly forhormonal products, while other sources accounted for 9.4% of the evidence used forthe basis of the warning given.
The study indicates that FDA seems to have established a pattern of practice for
imposing black box warnings on drugs. Of the 206 drugs examined with black boxwarnings, the vast majority were supported by the incidence of adverse events in con-trolled clinical trials as part of the NDA, or were based on postmarketing surveillancereports, or both. Those drugs with black box warnings supported solely by large-scaleor multiple epidemiological studies were limited to special situations such as cigarettesmoking and oral contraceptive use.
The authors have determined that six criteria describe when black boxes have
been used by FDA to convey critical information to the prescribing physician. In themajority of cases, the survey indicates that FDA imposes a black box warning whenthere is more than one message or concern. From the analysis, and as described in theMethods section, the concerns are summarized as follows:
(1) early detection of a side effect by physicians may result in intervention that may
(2) a well-defined subset of patients are at higher risk for the treatment;(3) the risk from the treatment of the particular drug may outweigh the benefits in
(4) the dosing or drug interaction is pivotal to the risk;(5) the training of the physician or the setting is crucial; and(6) there are other special requirements for administering the drug.
Whether FDA imposes a requirement that certain information be included in a
black box has important ramifications for the licensed practitioner, the pharmacist,the patient, the pharmaceutical manufacturer, and the distributor. The labeling of aprescription drug “plays a role in influencing medical judgment” and is considered“the most reliable guide for health care professionals using a drug in the daily practiceof medicine.”29 The practitioner, when deciding to use a drug, should be able to look
29 Thomas Scarlett, The Relationship Among Adverse Drug Reaction Reporting Drug Labeling, Prod-uct Liability, and Federal Preemption, 46 FOOD DRUG COSM. L.J. 31, 33 (1991).
to the boxed warnings for information about life-and-death risks associated with drugsand for ways to minimize these risks. It is therefore critical that FDA’s decision torequire a black box be fair, reasoned, and based on its established criteria. Otherwise,the physician’s decision may be prejudiced against a drug that, in fact, may be theideal choice for treating a particular patient’s disease or condition.
It is imperative that the prominence of a warning be proportionate to the risk and
supported by data. If the seriousness of the information in the boxed warning is exag-gerated, practitioners and pharmacists may become skeptical so that their confidenceand reliance on such information will diminish. Indeed, concerned that the impor-tance of the information in a black box does not become diluted, FDA has emphasizedthat, “to ensure the significance of boxed warnings in drug labeling, they are permit-ted in labeling only when specifically required by FDA.”30 Even when present, ablack box warning may not always be heeded. For example, despite a class-wide boxedwarning stating that synthetic thyroid products were ineffective and potentially dan-gerous for weight loss, its use for this purpose persisted.31
There are also marketing considerations for the pharmaceutical manufacturer.
The addition of a black box warning to a drug’s labeling can have a profound effect onthe sales of the drug. For example, during the year after FDA required a black boxwarning for Seldane® (terfenadine), cautioning against its use in certain settings,particularly with drugs that inhibit its metabolism in the patient,32 the product’s “salesdropped from around $700 million to $450 million.”33
With patients’ increased access to drug information in the Physicians’ Desk Ref-erence and elsewhere, it is essential that warnings highlighted in a black box areindeed serious enough to warrant such prominence. Indeed, there are several knownadverse consequences to patients made aware of drug-associated risks, such as in-creasing patients’ reporting of side effects, but not the accuracy of the reporting,34 and“compromis[ing] rather than enhanc[ing] patient compliance.”35
A black box has implications for product liability as well. Some courts have de-
termined that a black box warning sufficiently warns a physician of a drug-associatedrisk, protecting the manufacturer from product liability pursuant to the “learned inter-mediary rule.”36 With such significant effects of black box warnings, it is critical thatFDA carefully consider the ramifications of imposing such an important warning.
30 44 Fed. Reg. at 37,434. 31 S.C. Kaufman, T.P. Gross & D.L. Kennedy, Thyroid Hormone Use: Trends in the United States From1960 Through 1988, 1 THYROID 285 (1991).
32 FDA Request Marion Merrell Dow to Warn Doctors, FDA Press Release P92-22 (July 7, 1992); Th-
ompson and G. Oster, Use of Terfenadine and Contraindicated Drugs, 275 JAMA 1339 (1996).
33 Jennifer Mann Fuller, Drug Stung by Rivals, New Label; Competitors Moved in as Warning WasAdded to Seldane,KAN. CITY STAR, May 30, 1996, at B1.
34 Ivan Barofsky, Potential Effects on the Health Care System - 1, Joint Symposium on Drug Informa-tion for Patients: The Patient Package Insert, 11 DRUG INFO. J. 71S, 73S (Jan. 1977) (“We can never have thesituation where we improve our ability to recognize dysfunction . . . without also increasing the chances of a falsealarm.”). Also, “evidence indicates that listing possible side effects of drugs increases the probability that pa-tients will report them, even when they are in fact taking placebos.” Richard Dorsey, The Patient PackageInsert: Is It Safe and Effective?, 238 JAMA 1936, 1938 (1977) (citing K. Rickels and R.W. Downing, SideReactions in Neurotics: II. Can Patients Judge Which Symptoms Are Caused by Their Medication?, 10 CLIN. PHARMACOL. 298 (1970)).
35 See B. Blackwell, Drug Therapy: Patient Compliance, 289 N. ENG. J. MED. 249 (1973). 36 Hammock v. Hoffmann-LaRoche, Inc., N.J. Super. Ct. A-124 Sept. Term 1994, DDS No. 52-1-6305
(decided Aug. 2, 1995) (determining that Hoffmann-LaRoche’s black box warning sufficiently informed physi-cians prescribing Accutane® of the drug’s potential to cause birth defects); Margaret Gilhooley, Learned Inter-mediaries, Prescription Drugs, and Patient Information, 30 ST. LOUIS U. L.J. 633 (1986).
Several limitations in the methods used in this survey should be noted. The au-
thors did not examine the NDA data and have used “best judgment” to deduce thesource of the evidence used to support the warnings given in the black boxes studied. No attempt was made to separate the initial labeling from revisions appearing subse-quently and so comment cannot be made on the evolution of individual labels overtime. Lastly, the authors did not attempt to analyze the differences in FDA use ofblack boxes in recent years compared with earlier years.
The results indicate that FDA reserves black box warnings generally for those
situations in which 1) there is a strong clinical database to define the risk or hazard,and 2) the medical practitioner’s attentiveness to the highlighted risk has importantclinical significance that requires the judgment of that practitioner. Although FDAhas not declared explicitly the nature and scope of the agency’s bases and rationale forrequiring black box warnings, from their analyses, the authors have determined thatthe agency seems to have established patterns of practice for imposing the highlightedwarnings in certain cases. FDA seems to require such warnings for serious or life-threatening drug-associated risks that are well established and well defined, supportedin nearly all cases by the incidence of such events in controlled clinical trials that arepart of the NDA, and/or based on patterns of postmarket reporting.
FDA imposes the warnings in order to convey life-and-death information to the
prescribing doctor,37 who is in a position to minimize the risks to the patient. Fromthe analyses, one reaches the conclusion that the agency has multiple rationales forimposing the boxed warning:
to ensure that the prescribing doctor has the necessary information to performrisk/benefit analyses for the specific patient in particular circumstances;
to ensure safe and effective dosing of the drug;
to secure crucial methods and settings for administration of the drug;
to reduce dangerous drug interactions; and
to promote the monitoring of patients for serious, but possibly reversible, adversereactions.
Placing a warning for a prescription drug into a black box, if unwarranted, di-
lutes the importance of all such warnings, and creates inconsistency with other mar-keted products. With patients’ increased access to drug information, it is essential thatwarnings highlighted in a black box are indeed serious enough to warrant such promi-nence. Because of several adverse consequences to patients made aware of drug-asso-ciated risks, it is imperative that these prominent warnings are sanctioned by theagency’s established patterns of practice as described herein. Distribution of 375 Warnings contained in Black Boxes by Type for 206 Drugs* * There was an average of 1.83 black box warnings in each blackbox for each of the 206 drugs. Evidence Used for Black Boxes for 206 Drugs * 1.23 sources per black box
necesarios trabajos de mantenimiento e inspección, el comprador deberá efectuarlos oportunamente,El comprador no podrá pignorar ni transferir a título de garantía la mercancía bajo reserva de dominio. El comprador deberá comunicar al proveedor, sin dilación alguna, cualquier caso de embargo, prenda-Adjudicación del pedido, condiciones especialesciones o demás actos de disposición o
[· · · ] a transformational programming method-ology that includes a fully operational set-the-oretic proof checker [· · · ][· · · ] examples of two moderately difficult pro-gram derivations. One of these derives a highlevel form of an algorithm to compute the bisim-ulation equivalence relation [· · · ] the other de-rives an algorithm to minimize the number ofstates in a deterministi