Postmarketing Surveillance of Prescription Drugsota.fas.org/reports/8220.pdfPostmarketing Surveillance of Prescription Drugs ... This report describes the drug approval process, the

Postmarketing Surveillance of PrescriptionDrugs

November 1982

NTIS order #PB83-173625

Library of Congress Catalog Card Number 82-600652

For sale by the Superintendent of Documents,U.S. Government Printing Office, Washington, D.C. 20402

Foreword

Before a drug can be prescribed for use in the United States, it must meet minimumstatutory requirements for proof of its efficacy and safety as these have been establishedby the U.S. Food and Drug Administration (FDA). In premarketing testing, the numbersand types of patients exposed to a drug are necessarily limited compared with the numbersand types of patients who will eventually be prescribed the drug after it is marketed.New uses, contraindications, and side effects of drugs will then inevitably be discovered.Thus, various kinds of postmarketing surveillance have been proposed over the pastdecade.

A background paper on postmarketing surveillance of prescription drugs was origi-nally being prepared by the Office of Technology Assessment for the project on strategiesfor medical technology assessment, as requested by the House Committee on Energyand Commerce and its Subcommittee on Health and the Environment. At the furtherrequest of that committee and its subcommittee, that background paper was expandedinto this full report, Postmarketing Surveillance of Prescription Drugs.

Current interest in drug regulation is also focused on the premarketing approvalprocess, because the process has been criticized as unnecessarily delaying the releaseof valuable drugs in this country. As a result of such criticism, efforts are underwayto shorten the approval process through administrative changes within FDA’s Officeof Drugs, and through revisions of the regulatory interpretations of the statutory require-ments for “adequate tests” of a drug’s safety and “substantial evidence” of its effectiveness.

This report describes the drug approval process, the history and objectives of post-marketing surveillance, the methods employed to accomplish it, and current activitiesin postmarketing surveillance. The report provides guidelines to determine whethershortening the drug approval process by various means would diminish its ability todetect adverse drug reactions prior to a drug’s release for marketing. The report alsoidentifies oversight issues and options for increased postmarketing surveillance bothin the case that Congress decides to relax premarket approval requirements and in thecase that it does not.

Director

,..///

Health Program Advisory

Michael

Stuart H. AltmanFlorence Heller SchoolBrandeis UniversityWaltham, Mass.

Kurt DeuschleMount Sinai School of MedicineNew York, N.Y.

Carroll Estes*School of NursingUniversity of CaliforniaSan Francisco, Calif.

Rashi Fein

Committee

Sidney S. Lee, ChairReese Hospital and Medical Center

Chicago, Ill.

C. Frederick MostellerDepartment of Health Policy and

ManagementSchool of Public HealthHarvard UniversityBoston, Mass.

Mitchell RabkinBeth Israel HospitalBoston, Mass.

Dorothy P. Rice*Department of Social and

Behavioral SciencesSchool of Nursing

Center for Community Health andMedical Care

Harvard Medical SchoolBoston, Mass.

Melvin A. GlasserCommittee for National Health InsuranceWashington, D.C.

Patricia KingGeorgetown Law CenterWashington, D.C.

Joyce C. LashofSchool of Public HealthUniversity of CaliforniaBerkeley, Calif.

Mark LepperRush-Presbyterian-St. Luke’s Medical CenterChicago, Ill.

Margaret MahoneyThe Commonwealth FundNew York, N.Y.

University of CaliforniaSan Francisco, Calif.

Richard K. Riegelman’School of MedicineGeorge Washington UniversityWashington, D.C.

Walter L. Robb*Medical Systems OperationsGeneral ElectricMilwaukee, Wis.

Frederick C. RobbinsInstitute of MedicineWashington, D.C.

Rosemary StevensDepartment of History and Sociology

of ScienceUniversity of PennsylvaniaPhiladelphia, Pa.

Kerr L. WhiteRockefeller FoundationNew York, N.Y.

‘Appotnted summer of 1982

iv

OTA Postmarketing Surveillance of Prescription Drugs Project Staff

H. David Banta, Assistant Director, OTAHealth and Life Sciences Division

Clyde J. Behney, Health Program Manager

Lawrence H. Miike, Project Director

Kerry Britten Kemp, Editor

Mary E. Harvey, Secretary

Pamela J. Simerly, Secretary

Virginia Cwalina, Administrative Assistant

contractor

Renée G. Ford

OTA Publishing Staff

John C. Holmes, Publishing Officer

John Bergling Kathie S. Boss Debra M. Datcher Joe Henson

Doreen Cullen Donna Young

Contents

Chapter PageGlossary of Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

1. SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3The Drug Approval Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4History and Objectives of Postmarketing Surveillance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5Methods of Surveillance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6Issues and Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2. THE DRUG APPROVAL PROCESS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13Notice of Claimed Investigation for a New Drug . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

IND Application Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13Compassionate or Treatment ID. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

The New Drug Application Process ...,...,...,,. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16Abbreviated New Drug Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Requirements Following Approval .,....,...,.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

3. HISTORY AND OBJECTIVES OF POSTMARKETING SURVEILLANCE . . . . . . . 23

4. METHODS OF DRUG EVALUATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31Types of Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31Detection and Association . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32Case Studies in Drug Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

5. CURRENT ACTIVITIES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

6. ISSUES AND OPTIONS . . . . . . . . . . . . . . . . . . . . . . . . . ! . . . . . . . . . . . . . . . . . . . . . . . . . 49

Appendixes Page

A. Selected Excerpts From the Statutes Governing Drugs and Medical Devices . . . . . . . 57

B. Conclusions and Recommendations of the Joint Commission on PrescriptionDrug Use, Jan.23, 1980 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

List of Tables

Table No. Page1. Guidelines for Duration of Animal Toxicity Studies for Oral and

Parental Drugs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142. Studies Required in FDA’s Premarketing Drug Approval Process..,.. . . . . . . . . . . . . 153. FDA’s Drug Classification System ....,.....,.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164. Percentage of Reports of Suspected ADRs in Great Britain by Class of Reporter

and Method Used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255. Likelihood of Observing an ADR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336. Number of Patients Required To Detect One, Two, or Three ADRs With No

Background Incidence of Adverse Reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347. Number of Patients Required in Drug-Treated Group To Detect One ADR With

Background Incidence of Adverse Reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

vii

8. Number of Patients Required in Drug-Treated Group To Allow for Examinationof 100 Adverse Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

9. Potential Uses by FDA of Medicaid Data From Michigan and Minnesota . . . . . . . . . 4310. Data Sources Available to FDA for Estimating Actual Populations of Drug Users.. 4311. Number and Percentage of FDA’s Adverse Event Reports by Source, 1972-78 . . . . . 45

Figures

Figure No. PageI. Yellow Card Report Form Used in Great Britain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242. Comparison of Additional Drug-Induced Effects of Decreasing Incidences . . . . . . . . . . 353. Drug Experience/Epidemiologic Sources Available to FDA for Postmarketing

Surveillance and Risk Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

. .Vlll

Glossary of Acronyms

ADAMHA

ANDAADRCDCCDSCFRDEADESDHHS

DRAETIP

FDAHMOIND

Alcohol, Drug Abuse, and MentalHealth Administration (PHS)

abbreviated new drug applicationadverse drug reactionCenters for Disease ControlCenter for Drug SurveillanceCode of Federal RegulationsDrug Enforcement AgencydiethylstilbestrolDepartment of Health and Human

Servicesdrug regulatory authorityExperimental Technology Incentives

Program (NBS)Food and Drug Administrationhealth maintenance organizationinvestigational new drug

NBS

NCHS

NCINDANIDA

NIHNHLBI

PHSPMSSOAR

U.S.C.WHO

National Bureau of Standards(Department of Commerce)

National Center for Health Statistics(DHHS)

National Cancer Institute (NIH)new drug applicationNational Institute on Drug Abuse

(ADAMHA)National Institutes of Health (DHHSINational Heart, Lung, and Blood

Institute (NIH)Public Health Service (DHHS)postmarketing surveillancescreening of adverse reactions

(method)United States CodeWorld Health Organization

ix

Summary

Summary

INTRODUCTION

To market a drug, the manufacturer must pro-vide evidence of its efficacy and safety to the U.S.Food and Drug Administration (FDA). Once thesepremarketing requirements are met and the drughas been released, FDA can remove a drug fromthe market—after giving due notice and an oppor-tunity for a hearing—because of new evidence onthe drug’s efficacy and safety, the discovery thatthe drug was approved on the basis of any un-true statement of a material fact, or the failureof the drug to meet manufacturing standards. Incases where a drug may be an “imminent hazardto the public health, ” FDA can suspend the drug’sapproval immediately, giving prompt notice ofthe action and offering the opportunity for an ex-pedited hearing.

In premarketing testing, the numbers and typesof patients used to demonstrate a drug’s efficacyand safety are limited compared with the numbersand types of patients who will eventually be pre-scribed the drug after it is marketed. The initialdecision to approve a drug for use, however, mustbe made on the basis of the available knowledge.

Although postmarketing surveillance cannotprovide knowledge of the safety or efficacy ofdrugs at the time of their introduction on the mar-ket, various kinds of postmarketing surveillancehave been proposed over the past decade to mon-itor and aid in modifying the use of drugs. Theprincipal focus of postmarketing surveillance pro-posals has been on the safe use of prescriptiondrugs, even though the range of issues has encom-passed both efficacy and safety considerations,e.g., concern over refinements in use as well asbetter definition of drug risks.

Current interest in prescription drug evaluationand monitoring is focused on the premarketingapproval process and the length of time it takesfor a drug to be approved by FDA; postmarketingsurveillance appears to have waned as a policyissue. Thus, policy formulation and implementa-tion for the premarketing approval process is be-ing pursued without parallel efforts for the post-marketing period.

However, postmarketing surveillance deservesattention as a policy issue for both short- andlong-term objectives. Regarding short-term ac-tion, if current testing requirements for the pre-marketing approval process are reduced, pharma-ceutical manufacturers could be required to main-tain their drug evaluation responsibilities by in-creasing postmarketing surveillance. Regardinglong-term action, postmarketing surveillance re-mains a policy issue irrespective of current interestin the premarketing approval process: it is onlyafter marketing that a drug’s full therapeutic andharmful potentials can be determined.

One way to shorten the premarketing periodof the drug approval process would be by reinter-preting the regulations for assessing safety and ef-ficacy. This report provides theoretical and ex-periential criteria for evaluating how such changesmay affect the ability of current guidelines todetect a drug’s harmful and beneficial effects. Italso discusses the kinds of qualitative changes inthe evidence required for drug approval that FDAis implementing. Finally, the report identifies op-tions relating to FDA’s postmarketing surveil-lance. These options could be implemented re-gardless of whether there is a change in currentpremarketing drug approval requirements.

3

4 . Postmarketlng surveillance of Prescription Drugs

THE DRUG APPROVAL PROCESS

A drug’s sponsor must provide: 1) “adequatetests by all methods reasonably applicable to showwhether or not such drug is safe for use under theconditions prescribed, recommended, or sug-gested;” and 2) “substantial evidence that the drugwill have the effect it purports or is representedto have” (2 I U. S. C., sec. 355 (d)). This statutorylanguage has led in practice to FDA’s establishinga premarketing phase of drug testing that consistsof two parts: 1) the investigational new drug(IND) application process, and 2) the filing of anew drug application (NDA).

The IND application describes the investigators’qualifications and the planned clinical trials, thechemical composition of the drug, and data onthe pharmacology and toxicology of the new drugcollected in animal studies and in prior humanstudies, if any, such as those conducted in othercountries.

The clinical investigations in the IND processare divided into three phases (24):

Phase I: Clinical Pharmacology is that phasein which a drug is first used on humans toconfirm dose ranges and pharmacologic ef-fect. The number of subjects in phase I variesdepending on the drug, but is usually in therange of 20 to 80 (excluding control patients).Pharmacodynamic and metabolic studies, inwhichever stage of investigation they are per-formed, are considered to be phase I clinicalpharmacologic studies.Phase II: Clinical Investigation consists ofcontrolled clinical trials to demonstrate adrug’s effectiveness and relative safety. Theseare performed on closely monitored patientsof limited number, usually 100 to 200 pa-tients, with equal numbers of control pa-tients.Phase 111: Clinical Trials are expanded con-trolled and uncontrolled trials to gather addi-tional evidence of a drug’s effectiveness forspecific indications and to more preciselydefine its adverse effects. Phase 111 studiesobserve a total of 500 to 3,000 patients inmore natural settings—in clinics, outpatienthospital facilities, and private practice. Phase

111 usually consists of more than two con-trolled trials.

After completion of the testing required underthe IND application, the sponsor may file anNDA. At least two well-controlled studies estab-lishing each indication for which the drug is in-tended are required. More than one indication canbe established in a single study. (These require-ments are under review; see chs. 3 and 6.)

All INDs are classified by chemical type andtherapeutic potential, so that those drugs consid-ered by FDA to be of particular therapeutic im-portance can receive priority review. The highestclassification is given to drugs that are new molec-ular entities (type 1) and that may represent im-portant therapeutic gains (type A)—type 1Adrugs.

Several mechanisms are available to FDA to ob-tain information about drugs once they have beenapproved for marketing. Once the NDA has beenapproved, the sponsor is required to monitor in-formation and submit reports about the drug.Other information on adverse drug reactions(ADRs) is monitored by FDA in a number of

ways:

●

●

●

●

●

●

the Spontaneous Reaction Reporting Pro-gram, in which information on ADRs is sentto FDA by physicians, pharmacists, and hos-pitals;a monthly review of the medical literatureon ADRs (reports and letters to the editorsof medical journals, etc.);intensive surveillance and epidemiologicstudies of ADRs in selected hospitalized andambulatory populations;several specialized registries that collect andanalyze possible ADRs;in-house monitoring and research studies ofsuch data bases as those of the MedicaidMedical Information Systems of some Statesand those of commercial sources of drug usedata; andthe World Health Organization, which ex-changes reports with FDA, each summariz-ing the ADRs added to their systems in theprevious year.

——

Ch. l—Summary ● 5——— — — — . . — — — — . —— —

This postmarketing information is useful for answered by the phase 111 studies, but which dc~two purposes. First, it may provide the grounds not warrant delaying the release of what promisesfor FDA to remove a drug from the market, when to be a useful new product (24). Although FDAsuch action is appropriate. Second, it is used by has no explicit authority to require such studies,FDA to ensure that limits are placed on advertis- these “phase IV” studies are almost always per-ing and promotional claims and that the drug’s formed, as the alternative would be nonapprovallabeling is appropriate. of the drug.

FDA may request further studies when there arequestions about a drug that were not sufficiently

HISTORY AND OBJECTIVES OF POSTMARKETING SURVEILLANCE

As a result of 1974 hearings before the SenateCommittee on Labor and Human Resources’ Sub-committee on Health, the Department of Health,Education, and Welfare formed a Review Panelon New Drug Regulation. The panel issued itsreport in May 1977 (16).

A bill was subsequently introduced in the Sen-ate in early 1978 to revise the drug provisions ofthe Food, Drug, and Cosmetic Act. A revised bill,S. 1075, the Drug Regulation Reform Act of 1979,passed the Senate in September 1979. A similarbill, H.R. 4258, was not acted on by the Houseof Representatives. Included in the Senate billwere the following specifications: 1) drug spon-sors could be required to conduct postmarketingsurveillance of a drug for up to 5 years; 2 ) aprescription drug could have its distributionlimited if the drug could not otherwise be foundto be safe and effective; 3) the standard for adrug’s immediate removal from the market wouldbe changed from the drug being an “imminenthazard to the public health” to the less stringentstandard of “unreasonable risk of illness or injuryto any segment of the population;” and 4)establishment of a “National Center for DrugScience. ”

During this period, in a speech to the Pharma-ceutical Manufacturers Association, Senator Ed-ward Kennedy (D-Mass. ) suggested that a bettersystem was needed for monitoring the use and ef-fects of prescription drugs after they were mar-keted. As a result, the Joint Commission on Pre-scription Drug Use was established in 1976,funded largely by the drug industry, with themandate to design a postmarketing surveillance

system to detect, quantify, and describe the antic-ipated and unanticipated effects of marketeddrugs, and to recommend a means by which in-formation on the epidemiology of prescriptiondrug use in the United States could be distributedregularly to interested parties. The Joint Commis-sion issued its report in January 1980 (42), but bythis time, interest in postmarketing surveillancehad waned, and the commission’s report and rec-ommendations were little noticed.

In 1976, the year in which the Joint Commis-sion was formed, an interagency agreement wassigned between FDA and the Experimental Tech-nology Incentives Program (ETIP) of the NationalBureau of Standards in the Department of Com-merce. The purpose of ETIP was to provide incen-tives or reduce barriers to technological innova-tion through changes in the regulatory process.ETIP’s agreement with FDA was to jointly funda program to determine if improvement in post-marketing surveillance could help reduce theregulatory requirements of the premarketin~period, principally those of phase III of the INDprocess and those of the NDA process. The spe-cific experiment was to develop postmarketingsurveillance systems and a method of managingand evaluating the reform (11). The project con-centrated on collecting the information requiredto design these systems (12). By 1982, FDA hadassumed most of the funding, as ETIP was to bephased out that year.

A Commission on the Federal Drug ApprovalProcess was convened in mid-1981 to examinehow FDA’s procedures for the approval of newdrugs could be expedited without compromising

6 ● Postmarketing Surveillance of Prescription Drugs— — —

public safety and to make recommendations onthe development of cost-effective postmarketingsurveillance to guarantee the quick withdrawalfrom the market of drugs that cause significantadverse effects. The commission had its genesisin a joint hearing held in April 1981 by the HouseScience and Technology Committee’s Subcommit-tee on Natural Resources, Agriculture Research,and Environment and its Subcommittee on Inves-tigations and Oversight. The first meeting washeld in July 1981. The commission completed itswork and announced its general findings in thespring of 1982, and its printed report was to bereleased in late 1982.

FDA is examining specific ways to speed up thedrug approval process. It is reviewing past phase111 trials to see if longer trials or those with largesamples have contributed useful information be-yond that obtained in phase 11 and early phaseIII testing. Past postmarketing studies that FDArequired are also being reviewed to see if they pro-vided the information that they were designed toobtain. Data on FDA approval time are being re-viewed to see what other factors may slow theapproval process. And, as a pilot test, an FDAcommittee is reviewing the pharmacologic andclinical data on selected drugs at the end of phase11 testing, and will make recommendations aboutthe best time for gathering additional information(e.g., phase III v. the postmarketing period) (11).

METHODS OF SURVEILLANCE

The primary objective of postmarketing studiesis to develop information about drug effects undercustomary conditions of drug use. Initial cluesabout a drug’s potential effects come from the ex-perimental studies carried out with both animalsand humans in the premarketing period. Spon-taneous or voluntary reporting (e. g., in letters tothe editors of medical journals) is the oldest, andto date, the most productive source of new in-formation about a drug’s possible effects once adrug is marketed. Other types of studies are usedto examine in more detail the possible effects ofa drug. In general, these other types of studies useeither cohort or case-control methods.

In March 1982, the FDA Commissioner begana related organization by merging the Bureau ofDrugs with the Bureau of Biologics, and replac-ing the Director of the New Drug Evaluation Divi-sion. The merged bureaus have since been desig-nated the National Center for Drugs and Bio-logics.

Finally, in a related development, the Senatepassed by a voice vote, in the first session of the97th Congress, the Patent Term Restoration Actof 1981 (S. 255). The bill would restore to the termof a patent the time lost in complying with theGovernment’s premarketing testing and review re-quirements, up to a maximum of 7 years, Patentedproducts eligible for extension would not be lim-ited to human drugs, but would include “humandrugs and biological, antibiotic drugs, animaldrugs and biological, food additives, color ad-ditives, pesticides, other chemical substances,medical devices, and any other product subjectto Federal premarket requirements” (72). In Sep-tember 1982, the House of Representatives votedon the bill under suspension of its rules. Undersuch conditions, a two-thirds vote was requiredfor passage, and although the bill received a ma-jority of the votes, it fell just short of the two-thirds majority needed.

Thus, four types of studies are generally usedto identify drug effects: 1) controlled clinical trials,2) spontaneous or voluntary reporting, 3) cohortstudies, and 4) case-control studies (23,50,61,77).

Controlled clinical trials match treatment andcontrol groups as closely as possible, minimizebias through such methods as randomization and“double-blinding,” and directly monitor patientsfor the duration of the study. Controlled clinicaltrials are considered the most definitive methodfor evaluating a drug’s efficacy and safety, butthey are often costly or impractical in specific sit-uations, for example, when a drug’s effects are

Ch. l—Summary ● 7

rare, or appear only after long-term use or a longlatency period.

Voluntary reporting by physicians and otherhealth providers, hospitals, and consumers mayact to alert FDA and pharmaceutical firms to pos-sible adverse effects of drugs, so that the inferenceof an association between a drug and an observedhealth condition may be further studied by cumu-lative, careful reporting, and confirmed or discon-firmed by more vigorous methods. Underreport-ing may be a serious deficiency of this method.A drug may also be erroneously associated withan adverse effect until the suspected associationfails to show up in repeated, statistically validatedstudies.

Cohort studies follow a defined group of pa-tients (the cohort) for a period of time. In thismethod, patients are not randomly assigned togroups, and there is no blinding. Cohort studiesare usually prospective and observe the cohortfrom the beginning of drug use. A group of pa-tients taking the drug of interest is assembled andfollowed to see, for example, if adverse reactionsoccur. A second group of patients (the controls)with the same medical condition, who are not tak-ing the drug and who may be receiving alternativetreatment, but who are otherwise matched asclosely as possible with the cohort, may also bestudied in parallel. The control group is used toidentify the frequency of occurrence of any con-dition observed in the drug-exposed group whichis due to causes other than the drug (i. e., the“background incidence” of the condition). In thismethod, patients can be directly monitored to en-sure they take the drug appropriately, and toobserve the drug’s effects; or monitoring can beless controlled. With less control, a larger cohortcan be followed, but bias is thus increased.

Case-control studies identify patients with theadverse effects to be studied (the cases), and com-pare them with a sample (the controls), drawnfrom the same cohort that gave rise to the cases.Controls are matched as closely as possible withthe cases, except with regard to the drug’s sus-pected adverse effect, to examine whether expo-sure to the drug is the cause. Patients with con-ditions suspected of being associated with a cer-tain drug would have their medical records re-

viewed or be interviewed concerning the use ofthat drug. The histories of the controls would alsobe studied for information about drug use in thegeneral population. By comparing the proportionof drug users among the cases with the propor-tion of drug users in the general population, it ispossible to infer the relative frequency with whichadverse reactions occur in users of certain drugsas compared with nonusers. A sufficient numberof appropriate cases must be identified and accu-rate histories of exposure to drugs must be ob-tained.

Controlled clinical trials and prospective cohortstudies can be used to determine a drug’s beneficialas well as adverse effects. Case-control studies areusually used to trace adverse effects back to priordrug use. Voluntary reporting can uncover addi-tional uses of drugs as well as their adverse ef-fects, but reporting of adverse effects is muchmore common.

The ability of a particular surveillance methodto detect a drug’s effect depends on two factors:1) the time that transpires between use of that drugand the occurrence of the drug’s effect (the laten-cy period), and 2) how often the effect occurs (itsfrequency). There are many other determiningfactors, such as accuracy of observation, and ac-curacy and completeness of medical records, butthese factors present more of a problem in thedesign of a study’s details.

Controlled clinical trials, because of their rela-tively short duration, will detect only acute orsubacute effects. Long-term cohort studies candetect delayed effects, but the data bases necessaryfor such long-term, large studies are still sparse.Voluntary reporting is usually the way in whichlong-term effects are first identified. Long-termeffects are usually confirmed through retrospec-tive case-control studies, but such studies’ relianceon historical data such as medical records can limittheir accuracy.

The chance that a particular study will discovera drug effect also depends on the study’s samplesize and the frequency of the drug effect, For ex-ample, in a cohort study, if a drug causes blind-ness in 1 out of every 100 users (1/100), how manyusers must be observed to find one case of blind-

8 ● Postrnarketing Surveillance of Prescription Drugs

ness? If there are 1 million users of the drug, therewould be 10,000 users blinded. But in a small sam-ple of only 100 users, the probability of findingone or more cases of blindness would only be 63percent. If the sample were 200 users, the prob-ability of finding one or more cases would increaseto 86 percent. With a sample of 500, the prob-ability would be 99 percent that at least one caseof blindness would be found in the observed users.

To state it another way, what number of userswould have to be observed to be 95 percent sureof finding one or more cases of blindness whenthey occur at a frequency of 1 in 100 users? Theanswer is 300 users, and the general rule is thatthe number of users in the sample must be threetimes the reciprocal of the frequency; e.g., for afrequency of 1 in 1,000, the sample would haveto be 3,000 to be 95 percent sure of observing atleast one case.

Except for some effects that are unique to a spe-cific drug, many drug effects (e. g., stroke, bleed-ing, skin rashes) are indistinguishable from condi-tions due to other causes. The “background inci-dence” of a condition must be known before pur-ported drug effects observed in a study can rightlybe attributed to a drug.

Larger sample sizes are needed to determine adrug’s effect as the background incidence of a con-dition increases and as the frequency of a drug’scontribution to a condition decreases. For exam-ple, given a background incidence of 1/100, asthe incidence of a drug’s added effect decreasesfrom 1/100 to 1/10,000, the sample size wouldhave to increase from 1,600 to 11 million to re-main 95 percent sure of observing at least one caseof the added effect. The relationship between

ISSUES AND OPTIONS

Issue 1:

Revising premarketing tests and short-ening the drug approval process.

The efficacy and safety tests in animals and hu-mans specified in FDA regulations for premarket-ing approval are based on broad statutory lan-guage. Efforts to shorten the drug approval proc-

background and added incidence is also revealedin considering sample sizes at the extremes. Fora known background incidence of 1/1,000 and anadded incidence of 1/100, the sample size neededto observe at least one case of the added effectis only 500. But when the background incidenceis 1/10 and the added incidence is only 1/10,000,the sample size must be 98 million. These illustra-tions merely indicate what sample size is requiredto observe an effect when background incidenceis known.

Controlled clinical trials are used primarily forevaluating drug efficacy, not safety, because theyare carried out on hundreds, or, at the most, afew thousand drug users. Their use for evaluatingdrugs already on the market is also limited bytheir high cost and logistical problems. In fact,the use of controlled clinical trials for determin-ing efficacy alone is already constrained by thesetwo factors (9,46).

These limitations of controlled clinical trials inevaluating the safety of marketed drugs have ledto relying on cohort and case-control methods forpostmarketing studies. While these latter methodscan only indicate an association between a drugand observed conditions, not that the relation iscausal (49,77), the cumulative experience of multi-ple cohort and case-control studies showing con-sistent associations between a drug and such aneffect can lead to a high degree of confidence thatthe relationship is causal. The most prominentexamples of drug studies showing consistentassociations are those on oral contraceptives andthe risks of cardiovascular disease; similar ex-amples of nondrug studies are those on the risksof smoking.

ess have focused not on the statutory languagebut on the regulations issued by FDA to imple-ment the law. Thus, the focus here is on oversightissues, not on legislative changes.

Proposals to curtail or eliminate phase 111 pre-marketing tests, or shift them to the postmarketingperiod, can be evaluated both theoretically andexperientially.

Ch.1—summary ● 9

Theoretically, phase III testing is significantlymore sensitive than phase II testing. Adverse ef-fects with an incidence of 1 /100 or more are morelikely than not to be detected in the 100 to 200patients given a drug in phase II. But the theoreti-cal sensitivity of detection rises in phase III to1/500 with 500 patients and to 1/1,000 with 1,000to 3,000 patients (see ch. 4, table 5).

These observations are relevant to the detectionof adverse reactions, but they are not so relevantto the detection of therapeutic effects. Since a drugthat helps only 1 in 100 patients would not be veryeffective, efficacy should be established in phaseII. Phase III is intended to gather additional evi-dence on a drug’s effectiveness for specific indi-cations.

If phase 111 testing were curtailed or eliminated,there is also the question of whether premarket-ing evaluations would test sufficient numbers ofpatients to reasonably ensure a drug’s safety orgive substantial evidence of its efficacy. Evenunder current regulations, the use of a drug onhuman subjects is very limited before the drug isreleased for market: 20 to 80 patients in phase I;100 to 200 patients in phase II; and 500 to 3,000patients in phase III—a range of only 620 to 3,280patients per drug (excluding controls).

In addition to theoretical criteria, experientialcriteria could be applied in considering proposalsto curtail or eliminate phase 111 tests. The dimin-ished power to observe adverse drug effects thatsuch changes theoretically entail may not in factbe found, judging on the basis of actual experiencein phase 111 testing, or if it is, it may only con-cern infrequent, minor effects. Agreement of theexperiential data with the differences theoreticallyexpected would strengthen the hypothesis thatcurtailing phase 111 would lower the capacity ofcurrent premarketing tests to identify adversereactions. If the experiential data fail to detect thetheoretical differences, then a better case can bemade for curtailing phase 111, with or withouttransfer of some of its testing to the postmarketingperiod.

Current interpretations of the statutory require-ments for “adequate tests” of safety and “substan-tial evidence” of efficacy emphasize methodology,as reflected in the requirement that each indica-

tion for which a drug is intended be supportedby at least two well-controlled clinical trials. ButFDA can alter the criteria by which it approvesdrugs. For example, propanolol, the first beta-blocking drug approved for use in the UnitedStates, was approved by an advisory committeeon the basis of all the evidence presented to FDA,even though no one study was found to be ade-quate and well controlled (21). And in late 1981,timolol, another beta-blocker, was approved, onthe basis of evidence from a foreign study, for usein preventing death and recurrent heart attacksin patients who have survived initial heart attacks(26).

The approval of propanolol and timolol illus-trates that FDA can grant exceptions to its usualrequirement of two well-controlled U.S.-basedclinical trials. In such cases, expert judgment relieson qualitative, not quantitative, criteria in ap-proving a drug, and such an approach falls out-side the theoretical and experiential guidelines out-lined above. If FDA is to rely increasingly on suchqualitative criteria through increased use of advi-sory committees, it will be necessary for FDA todevelop general guidelines to aid the advisorycommittees in their deliberations. Otherwise, ina case-by-case analysis, evidence of the same qual-ity may lead to approval for one drug and nonap-proval for another.

Issue 2:

Improving postmarketing surveillanceand its role in the drug approval process.

Even if phase 111 testing were not curtailed oreliminated, FDA’s powers in the postmarketingperiod could be strengthened to enhance its sur-veillance role.

Postmarketing surveillance “systems” that havebeen advocated are not systems in the formalsense, but a series of related activities orientedtoward several purposes, with the regulatory ap-proval process being only one. Three activities aremost frequently mentioned. First is the buildingof a resource base through training of additionalexperts and improving epidemiologic tools suchas methods for cohort and case-control studies.Second, unless a drug effect has a sufficient fre-

10 ● Postmarketing Surveillance of Prescription Drugs

quency of occurrence (usually identified as I/1,000) and for delayed effects of, for example,greater than I year, strengthened voluntary re-porting is the most realistic method of identify-ing possible adverse drug reactions. Once suchreactions are suspected, clinical trials, case-control, and cohort studies could be used to deter-mine whether an association with drug use in factexists. Third is the development of an efficientmethod for monitoring selected drugs after theirrelease into the market. The most frequently men-tioned mechanism is formation of prospectivecohorts of drug users.

These aforementioned components of a post-marketing surveillance “system” and FDA’s rolein supporting and using them are oversight issues.

There are also several legislative options thatcould strengthen FDA’s powers in the postmarket-ing period. The following legislative options arepresented for congressional consideration.

Option 1: Give FDA the power to require post-marketing studies.

A variation of this option is for FDA to use itsexisting regulatory powers over advertising andpromotional practices to “certify” an industry-sponsored postmarketing study.

Option 2: Give FDA the power to restrict thedistribution, dispensing, and administration of adrug.

A variation of this option is for FDA to use itsexisting regulatory powers to develop a parallelapproval process for the use of a limited groupof drugs during phase III testing, such as for drugsof unusual need and promise.

Option 3: Change the standard for a drug’s re-moval from the market from “imminent hazardto the public health” to “unreasonable risk of ill-ness to any segment of the population” or someother less stringent standard.

2The Drug Approval Process;

2.The Drug Approval Process

A drug’s sponsor must provide: 1) adequatetests by all methods reasonably applicable to showwhether or not such drug is safe for use underthe conditions prescribed, recommended, or sug-gested; and 2) substantial evidence that the drugwill have the effect it purports or is representedto have. “Substantial evidence” means “evidenceconsisting of adequate and well-controlled investi-gations, including clinical investigations, by ex-perts qualified by scientific training and experienceto evaluate the effectiveness of the drug involved,on the basis of which it could fairly and reason-

ably be concluded by such experts that the drugwill have the effect it purports or is representedto have” (21 U. S. C., sec. 355(d)). (See app. A forselected sections of the act. )

This statutory language has led the Food andDrug Administration (FDA) in practice to estab-lish a premarketing phase of drug testing that con-sists of two parts: 1) the investigational new drug(IND) application process, and 2) the filing of anew drug application (NDA).

NOTICE OF CLAIMED INVESTIGATION FOR A NEW DRUG

IND Application Process

A new drug is defined as any drug: 1) that isnot generally recognized by experts to be safe andeffective for the use described in the drug’s label-ing (except for certain so-called “grandfatherdrugs, ” i.e., those approved prior to the 1962amendments); or 2) that has been shown to besafe and effective in clinical investigations, but hasnot been used to any material extent or for a ma-terial time.

A drug is considered to be new for any of thefollowing reasons: 1) it is composed in whole orin part of a new substance (this includes activecomponents and inert ones, such as a coating orcarrier); 2) it is a new combination of approveddrugs; 3) it is an approved combination in a newratio; 4) it is an approved drug with a proposednew use (i. e., a use for which the drug has notbeen approved); or 5) it is an approved drug witha proposed new dose or new method or durationof administration (21 CFR 310.3(g)).

A sponsor is the entity responsible for the en-tire investigation of a new drug. The sponsor canbe an individual, a partnership, a corporation, oranother agency of the Government (e. g., the Na-tional Cancer Institute). In testing a new drug, a

sponsor may use a number of different investi-gators.

A sponsor wishing to investigate a new drugby means of clinical tests in humans must firstcarry out various studies in animals (see table 1).Such studies examine acute and chronic drug tox-icity at different dose levels, by different routesof administration, and in different species. Bio-chemical data are also obtained on the drug’s ab-sorption, distribution, metabolism, and excretion.The data from chronic animal studies, which cantake over 2 years to collect and analyze, are notordinarily required for permission to proceed tohuman trials. Long-term animal tests must alsobe undertaken at the same time that the drug isbeing tested in humans, particularly for drugs in-tended for use over long periods of time, as forchronic diseases and oral contraception. The pur-pose of long-term animal tests is to investigate thedrug’s toxicity (e.g., carcinogenicity) when takenchronically, and its effects on fertility, reproduc-tion, and fetal development (e.g., teratogenicity).

The point at which FDA becomes involved inthe development process for a new drug is whena sponsor desires to investigate the drug’s safetyand effectiveness via clinical tests in humans, Be-

13


Table 1 .—Guidelines for Duration of Animal Toxicity Studies for Oraland Parenteral Drugs

Expected duration of continuous Phase of clinical Duration of subacute or chronicadministration to humans investigation toxicity studies in animals

Several days. . . . . . . . . . . . . . . . . . . I, II, Ill, NDAa Two species; 2 weeksUp to 2 weeks . . . . . . . . . . . . . . . . . I Two species; 2 weeks

II Two species; 2 monthsIll, NDAa Two species; up to 3 months

Up to 3 months . . . . . . . . . . . . . . . . 1, II Two species; 4 weeksIll Two species; 3 months

NDAa Two species; up to 6 months b

6 months to unlimited . . . . . . . . . . 1, II Two species; 3 monthsIll Two species; 6 months or longer

NDA a Two species; 12 monthsb (nonrodent)18 monthsb (rodent)

aNew drug applicationbAlthough as yet there has been no formal updating of these guidelines, they have been expanded to include 2.Year animal

toxlclty and carc!nogenlclty stud!es for those drugs that would be administered chronically or intermittently (n a large population(e g , contraceptives) These studies are presently being required on a drug-by .drug basis as investigational new drugs arereviewed

SOURCE U S Food and Drug Administration

fore proceeding, the sponsor must file an IND ap-plication with the Office of Drugs in FDA’s NationalCenter for Drugs and Biologics. The 1962 amend-ments to the Food, Drug, and Cosmetic Act, byempowering the Secretary of Health and HumanServices to write specific requirements, effectivelyrequire that all results from testing a new drugin humans be submitted by the sponsor of thedrug and approved by FDA under an IND. In fil-ing the IND application, the sponsor agrees torefrain from beginning studies for 30 days, butmay begin them after that time unless FDA asksthe sponsor to continue to avoid or to restrict useof the drug in humans. The 30-day delay can bewaived upon a showing of good reason.

The IND application describes the qualificationsof the investigators and the planned trials, andincludes a chemical description of the drug andavailable data on its pharmacology and toxicol-ogy as collected in animal studies and in priorhuman studies, if any (e.g., those conducted inforeign countries). If the necessary animal testshave been carried out and give evidence for thesafety of the proposed human use, if the drug isadequately characterized so that the tests will bemeaningful, and if the proposed human studiesappear reasonably safe, the IND application isusually approved. The sponsor may then proceedwith clinical testing if FDA does not raise objec-tions within 30 days.

Once the IND application is approved, addi-tional protocols and investigations can be added.FDA sets no time limit on the IND process as longas annual reports in the form of summaries aresubmitted and serious adverse reactions arepromptly reported. All data on drug effectiveness,including that from clinical studies in patients, aswell as chemical and animal data, are consideredto be the sponsor’s property and subject to pro-tection as trade secrets. If, at any time during thetests on human subjects, the continuance of thosetests is determined to endanger public health, theycan be stopped immediately.

The clinical investigations of a new drug—i.e.,studies in humans—are divided into three phasesthat in actual practice are not so distinctly sepa-rated (see table 2) (24).

Phase I: Clinical Pharmacology is the initial useof the drug on humans. The purpose of this phaseis to determine levels of tolerance (toxicity), tobegin to ascertain safe dose ranges, and, in somecases, to study drug efficacy in selected patients.The total number of healthy volunteer subjectsand patients administered the drug ranges fromabout 20 to 80. At this stage, many drugs arescreened out because their safety is found to beseriously questionable or because they are foundto be inactive in humans. If the drug appears tobe well tolerated, it may go on to the next stageof testing.

. —-—

Ch. 2— The Drug Approval Process ● 15

Table 2.—Studies Required in FDA’s PremarketingDrug Approval Process

Phase 1:— Studies in normal volunteers or relatively healthy

patients to determine safety and pharmacologiceffects.

— Small studies in patients to determine clinicaleffectiveness.

— Total number of subjects—up to 80 administeredthe investigational drug.

Phase 11:— Controlled clinical trials to determine appropriate

doses, safety, and effectiveness.— Total number of patients—about 200 administered

the investigational drug.

Phase Ill:— Controlled and uncontrolled clinical trials to deter-

mine safety and effectiveness and to support label-ing claims.

— Total number of patients—about 500 to 3,000administered the investigational drug.


Phase II: Clinical Investigations are the earliestinvestigations specifically designed to demonstrateeffectiveness and relative safety, and to includecontrolled studies. In this phase, the drug is ad-ministered to 100 to 200 patients, under rigid pro-tocols and close monitoring. If the therapeutic val-ue of the drug has been demonstrated, and it ap-pears to have no serious adverse effects, it maythen enter the final stage of testing.

Phase 111: Clinical Trials are expanded con-trolled and uncontrolled trials. They are carriedout on 500 to 3,000 patients in situations similarto those of actual clinical practice—in clinics, out-patient hospital facilities, and private practice.These studies are performed after a drug’s efficacyhas been established, at least to some degree, andare intended to gather additional evidence of drugeffectiveness, to discover rarer drug effects or ef-fects that develop after longer periods, and to bet-ter define the frequency and severity of more com-mon effects as well as the proper use of the drug(e.g., by identifying best dose, dose interval, andthe drug’s interactions with other drugs). Ade-quate and well-controlled trials that give evidenceof a drug’s effectiveness, accompanied by com-plete case records for each patient, are sometimes

termed pivotal studies. FDA usually requires atleast two independent well-controlled studies toapprove an NDA, the following stage of the drugapproval process, though more than one drug in-dication can be evaluated in a single study. (Theserequirements are currently under review; seech. 3.)

Compassionate or Treatment IND

FDA has recognized that under special circum-stances, when a patient has exhausted all othertherapies for a life-threatening disease, a drug thatmight be of value but is still unapproved shouldbe made available. Investigational drugs can bemade available under a “compassionate” or “treat-ment” IND. The drugs made available are general-ly in phase III of clinical testing.

Physicians can obtain an investigational drugif they have a patient with a disease that is life-threatening or significantly impairs the quality oflife, and the patient is allergic or resistant to ex-isting methods of treatment. Under such circum-stances, FDA usually recommends that the physi-cian contact the medical director of the companyinvestigating the drug to inquire whether the com-pany will accept the physician as an investigatorunder its IND application for that particular pa-tient. The company may also supply the drug tothe physician, who files his or her own IND appli-cation.

Sometimes these treatment uses can becomequite extensive, especially when available therapyis unsatisfactory, for example, in the treatmentof serious cardiac arrhythmia and angina pectoris.According to Robert Temple, recently appointeddirector of the Office of Drugs’ New Drug Evalua-tion Division, promising new antiarrhythmics andantianginal drugs have been given to thousandsof patients under these circumstances. But thecompassionate IND procedure may be inadequatefor providing a needed drug that is still in theprocess of being approved when patients are dis-tant from a medical center or when physicians arenot familiar with FDA procedures.

76 Ž Posfmarketlng Surveillance Of Prescnptlon Drugs——— . ——————— — — . — — — —

THE NEW DRUG APPLICATION PROCESS

After the completion of required testing underthe IND, if the sponsor believes that the drug’ssafety and effectiveness have been proved and thatthe drug has commercial potential, the sponsormay file an NDA, a request for FDA’s permissionto market the drug in interstate commerce. Thisapplication includes everything the sponsor con-siders necessary for meeting the statutory require-ments: 1) full reports of animal and clinical studiescarried out to determine whether the drug is safeand effective; 2) a statement of the drug’s compo-sition; 3) a description of the methods, facilities,and controls used in the drug’s manufacturing,processing, and packaging; 4) samples of the drugand its components as may be required; and 5) acopy of the proposed labeling. The labeling de-scribes what is known about the drug: the ap-proved uses, dosages, the indications for whichits effectiveness is approved, and its known ad-verse side effects. The final wording of the label-ing is negotiated between FDA and the sponsor,and must be formally approved as part of theNDA.

All INDs are classified by chemical type andtherapeutic potential so that those drugs consid-ered by FDA to be of particular therapeutic im-portance can receive priority review (see table 3).The highest classification is for a drug that is botha new molecular entity (type 1) and that mightrepresent an important therapeutic gain (type A)—a type 1A drug. The next highest classificationis given to a new molecular entity that representsa modest therapeutic gain (a type IB drug). Butskepticism has been expressed by some industryrepresentatives as to whether a correct determina-

tion of the potential benefits of a drug can be madeat the time of NDA submission,

After the NDA is filed, a team of FDA review-ers analyzes the sponsor’s summaries of the dataor, when needed, the actual data. The review teamincludes a physician, who reviews the clinical testresults; a pharmacologist, who reviews the animaltest results; and a chemist, who reviews the chem-ical data and manufacturing controls and proc-esses; supported by a biopharmaceutic specialist,a biometrician, and, when applicable, a microbi-ologist. The main objective of the process is toensure that the data from the clinical experimentssupport the claims for the drug’s safety and effi-cacy in the labeling the sponsor submitted.

NDAs may be presented for consideration toadvisory committees composed of experts (mostlynongovernmental) in the various subspecialties ofmedicine, in clinical pharmacology, and in bio-metrics. The committees recommend whether ornot an NDA should be approved to market a drugand, if the drug is approved, what wording shouldappear in its labeling. They also may recommendwhether the sponsor should be requested to carryout additional studies after the drug is marketed.If the committees recommend against approval,they identify deficiencies and may suggest newstudies that need to be done by the sponsor to fur-ther investigate the drug’s safety and efficacy.

FDA invites a sponsor to confer with FDAabout important new therapeutic drugs during thedrugs’ investigational phases, Usually, such con-ferences are arranged at the end of phase II, whena drug’s degree of efficacy and safety has been

Table 3.—FDA’s Drug Classification System

Chemical type

Type 1 New molecular entityType 2 New salt, ester, or derivativeType 3 New formulationType 4 New combinationType 5 Duplicate of an already marketed

drugType 6 Already marketed product by same

firm—primarily used for newindications

SOURCE U S Food and Drug Administrat~on –

Therapeutic type

Type A Important therapeutic gainType B Modest therapeutic gainType C Little or no therapeutic gainType D Decreased safety or efficacy

compared with other drugs buthas some compensating virtue

——.. —-—

Ch. 2— The Drug Approval Process ● 17— — ——.

largely established. The purpose of the confer-ences is to discuss whether the studies to date areacceptable, in view of the drug’s proposed indi-cations and the claims made for the drug, andwhether the additional controlled studies pro-posed for phase III will be adequate for the NDA’sapproval.

FDA must approve or disapprove a submittedor resubmitted NDA within 180 days. It may takelonger if the sponsor and FDA agree on an addi-tional period of time. An FDA study of the reviewtime for approved NDAs found that in 1976, itwas 25.9 months, in 1977, 26.6 months, and in1978, 33.3 months (12). In the first half of 1979,however, the average time had been reduced to20 months (30). These approval times includeseveral resubmissions of the NDAs either to cor-rect deficiencies noted in them during FDA re-view, or because the sponsor obtained a largeamount of additional data. Furthermore, the totalamount of information required for an NDA hasincreased markedly since 1938, when the 180-daylimit was imposed. In 1938, the required infor-mation consisted of data from short-term animaltoxicology studies, along with that from a fewstudies in humans. Today, an NDA must containthe reports of numerous short- and long-termanimal studies along with the reports of variousclinical trials to demonstrate the drug’s safety andeffectiveness in humans.

Should FDA decide to refuse approval, thesponsor receives a “nonapprovable” letter explain-ing why the NDA fails to fulfill statutory re-quirements. An applicant’s approval can be re-fused for any of the following reasons:

Drug safety has not been studied by all rea-sonably applicable tests.The drug is not safe for the intended use.The drug’s manufacturing processes are notadequate to ensure its identity, strength,quality, and purity.Substantial evidence of the drug’s effective-ness is lacking; i.e., the clinical investigationswere not adequate or well controlled or theirresults do not adequately support the claimsmade.The labeling is false or misleading in any par-ticular.

Ž The application is missing data, e.g., on bio-availability or bioequivalence or on the envi-ronmental impact of the manufacturing proc-ess.

If the clinical trials establish that a drug is effec-tive for its intended indications, FDA must decidewhether the drug’s benefits outweigh its risks.FDA does not require a sponsor to prove that adrug is safer than available drugs, nor more effec-tive, nor even as effective as other treatments inorder to receive permission to market the drug.In some cases, a drug may be less safe than analternative therapy. Although most manufacturersnormally would not be interested in marketing adrug whose benefit/risk ratio is less than that ofa treatment already available, FDA recognizesthat there are times when such drugs should bemade available.

When presented with an NDA in which the dataclearly show a drug to have an inferior benefit/risk ratio, FDA considers the indications for whichthe drug is offered. Thus, such drugs when in-tended to treat a life-shortening condition mightreceive approval, while such drugs when intendedfor lesser indications (e.g., mild analgesics thatare less safe and no more effective than aspirin)would not. Obviously, the decisions that causethe most difficulty are those concerning drugs be-tween these extremes (22).

Abbreviated New Drug Application

FDA has established an abbreviated new drugapplication (ANDA) for generic versions of drugsfirst marketed between 1938 and 1962. Thesedrugs require less testing for approval than dooriginal versions of a drug. The amount of infor-mation required for approval of such genericdrugs varies, depending on the nature of the drug.The ANDA policy does not apply to drugs mar-keted after 1962. Approval of generic copies ofthese recent drugs requires a standard NDA, butFDA has been willing to accept published reportsdemonstrating the safety and efficacy of thesedrugs. NDAs that rely on published reports aresometimes referred to as “paper” NDAs.

18 ● Postmarketing Surveillance of prescription Drugs

REQUIREMENTS FOLLOWING APPROVAL

Once the NDA has been approved, the spon-sor is required to keep records and submit reportsabout the drug. This information is used to: 1)maintain the procedures and safeguards for manu-facturing established during the approval process,2) ensure that there are limits placed on advertis-ing and promotional claims and that the drug’slabeling is appropriate, and 3) provide the basisfor FDA’s removal of a drug from the market,when such action is appropriate.

The studies carried out before a drug is releasedhave a number of inherent limitations with respectto the amount and kind of information they gen-erate, limitations that include the following:

●

●

●

●

The patients in premarketing studies (evenin phase III) do not represent all those whowould ultimately take the drug. Thus, adrug’s effects on special patient populationsnot specifically studied in premarketing testsmay not be known. Such special populationsinclude patients who are taking severalmedications concurrently, those havingdiseases in addition to the one treated by thedrug, and those who suffer more severelyfrom the disease being treated than the pa-tients in the study groups. Similarly, whenphase 111 studies are conducted on children(e.g., when they are to be a drug’s chief re-cipients), usually only very small groups ofpatients are studied, and the entire age rangeof concern (e.g., including newborns) is notstudied for ethical and other reasons.The total number of people exposed to thedrug in premarketing studies is relativelysmall; therefore, uncommon adverse reac-tions (i.e., those less frequent than 1/1,000)are unlikely to be detected.The duration of exposure to the drug in pre-marketing studies is relatively brief (12 to 24months at most); therefore, adverse effectsthat only appear after long-term use or thatrequire a latent interval after exposure todevelop (e.g., cancer) cannot be detected.Clinical trials must be conducted accordingto strict protocols (regarding dosages, dura-tion of treatment, etc.), and they are usuallycarried out by specialists in large medical

●

centers where such research can be done.Thus, the effects a drug might have when ad-ministered by a regular physician in an of-fice or outpatient clinic, when patient com-pliance to a treatment regimen is less con-trolled, cannot be fully assessed.In premarketing studies, a drug is oftenevaluated for only one purpose (e.g., treat-ment of hypertension), but it may have an-other use (e.g., treatment of angina pectoris).(In its labeling, the only drug indications thatmay appear are those that have been explicit-ly tested and approved).

FDA has several mechanisms to obtain infor-mation about drugs once they have been ap-proved for marketing.

FDA may request further studies when ques-tions about a drug remain unanswered by phase111 studies but do not warrant delaying the releaseof what may be a useful new product. These stud-ies are referred to as “phase IV” studies. Althoughthis designation is not defined in FDA regulations,it is discussed in the guidelines (24). The studies’nature depends on the question to be resolved,and their design is negotiated between the drug’smanufacturer and FDA through its New DrugEvaluation Division in the Office of Drugs.

Phase IV postmarketing studies can be of sev-eral

●

●

●

●

types (24):

Additional studies to elucidate the incidenceof adverse reactions, to explore a specificpharmacologic effect, or to obtain informa-tion of a circumscribed nature.Large-scale, long-term studies to determinethe effect of a drug on morbidity and mortal-ity.Additional clinical trials similar to those inphase III, to supplement premarketing datawhere it has been deemed in the public inter-est to release a drug for more widespread useprior to acquisition of all data which wouldordinarily be obtained before marketing.Clinical trials in a patient population not ade-quately studied in the premarketing phase;e.g., children.

Ch. 2—The Drug Approval Process ● 19

• Clinical trials for an indication for which itis presumed that the drug, once available,will be used.

In general, phase IV studies have been requestedwhen a drug is likely to be widely used, and im-portant safety and efficacy questions about it re-main. Phase IV studies may be requested, for ex-ample: 1) when there are suspected or known ad-verse drug reactions (ADRs) associated with thedrug, in order to confirm the ADRs and to deter-mine their true incidence; 2) when a drug belongsto a class of drugs known to be associated witha serious ADR, but its incidence may not be highenough to observe in the limited number of pa-tients in phase 111 studies; 3) when the drug is onethat will be used with children, and it was nottested on them in the premarketing studies; 4) ifthe drug is likely to be used therapeutically incombination with another drug, and there is rea-son to be concerned about the toxicity of the com-bination; and 5) if a drug approved for one indi-cation is very likely to be used for several otherindications. (Studies may be required for each ofthe other uses. For example, most beta-adrenergicblocking agents, such as propanolol, are used inthe treatment of angina, hypertension, and ar-rhythmia, but an NDA may be approved whenthe use of a drug for only one of these indicationshas been documented. Even though the drug hasonly one approved use, it will likely be prescribedfor the other two uses. )

Phase IV studies may be typical clinical trialssimilar to those carried out in the premarketingperiod. They may also use other surveillancemethods. (See ch. 4 and its case studies of strep-tokinase and cimetidine. )

Finally, FDA receives information on marketeddrugs through various kinds of monitoring of ad-verse drug reactions carried out by the Divisionof Drug Experience in the Office of Drugs:

●

●

●

●

●

the Spontaneous Reaction Reporting Pro-gram, in which ADRs are reported to FDAby physicians, pharmacists, hospitals, andmanufacturers (the last kind of reporting ismandatory);a monthly review of the medical literatureon ADRs (reports to medical journals, lettersto the editor, etc.);intensive surveillance and epidemiologicalstudies of ADRs in selected hospitalized andambulatory populations;several specialized registries that collect andanalyze possible ADRs; andthe World Health Organization (WHO),which sends reports to FDA summarizing theADRs added to its system in the previousyear. FDA reciprocates by providing U.S.data to WHO.

This report focuses on the regulatory uses ofsuch postmarketing surveillance. Past studies ofthis issue have also focused on the need for sys-tematic evaluations in the postmarketing periodregardless of their importance for regulation, forexample, in order to build a resource base of scien-tists, evaluation methods, and data sources forthe better understanding and use of drugs oncethey are marketed. The evolutionary context ofpostmarketing surveillance is summarized in chap-ter 3.

3History and Objectives of

Postmarketing Surveillance

—

3.History and Objectives of

Postmarketing Surveillance

In the 1960’s, at least two serious drug reactionswere observed in many patients. The drug thalid-omide, taken worldwide, led to limb deformities(phocomelia) in the newborns of those motherswho took the drug while pregnant. Less known,and almost exclusively observed in Japan, was theoptic nerve damage (subacute myelo-optic-neuro-pathy) and other adverse effects from the drugclioquinol, over which almost 4,000 civil suitswere still pending in 1979 (68). And in Great Brit-ain in the early 1970’s, more than 4 years afterthe drug had been introduced there, the “practololsyndrome” was uncovered. Practolol, a drug usedto treat cardiovascular disease, was eventuallyfound to cause skin rashes, eye lesions, hearingimpairment, and sclerosing peritonitis (56), withdeaths occurring in about 2 percent of reportedcases (37).

Great Britain, with its national health system,already had a voluntary reporting system (37).The national health system had instituted the useof “yellow cards” for reporting suspected adversedrug reactions (see fig. 1). As a guide to report-ing, certain drugs are marked the first 4 years afterthey are marketed with an inverted black trianglein a booklet, the Monthly Index of Medical Spe-cialties, which is distributed to physicians andused as a source of information for prescribingdrugs more frequently than any other publication.In 1976, a slip of yellow paper was inserted intoprescription pads to remind physicians to reportreactions, leading to a large and consistent in-crease in the rate of reporting. Many British drugcompanies now use the yellow card, and theyellow card system and reports from drug com-panies together yield 90 percent of all reports ofsuspected adverse drug reactions (ADRs) (see table4).

The delayed discovery of practolol’s adverse ef-fects spurred efforts to improve postmarketingsurveillance, and several international meetingsquickly followed in Sestri Levante, Italy (20),Honolulu, Hawaii (33), and London (53). In Great

Britain, efforts focused on “early detection ofadverse drug reactions by recording all adverseevents occurring in a specified number of patientsfor an appropriate period of time; endeavoringto avoid collecting masses of unusable data andminimizing costs” (75). Thus, the early impetuswas toward monitoring new drugs for adverse ef-fects through some type of program that wouldhelp fill the gap between identifying those adverseeffects sufficiently common to be detected in thepremarketing trials, and identifying those so rarethat voluntary reporting after marketing is theirmost feasible form of monitoring. The proposedmethods all centered around the prescribing prac-tices of physicians, with the experience of theirpatients on new drugs being examined periodicallythrough questionnaires to the prescribing physi-cians. Such methods of monitoring include regis-tered release (17), recorded release (36), and mon-itored release (45,79).

More recently, the objectives of postmarketingsurveillance in Great Britain have been expanded,though not implemented (35):

The need for PMS [postmarketing surveillance]is not restricted to new drugs. Some of those al-ready marketed for many years may increase therisk of chronic disease or may have long-delayedcarcinogenic effects, as illustrated in the UnitedStates by the cases of vaginal adenocarcinoma inthe adolescent female children of women whotook diethylstilbestrol during pregnancy.

PMS should also include assessment of effi-cacy, especially of long-term treatment. Very lit-tle is known of the relative merits of members ofgroups of drugs such as hypotensive or anti-dia-betic agents, anti-rheumatics or psychotropic.Prescribers need to know the most effective treat-ment available just as much as they need to knowthe risks involved, but government drug regula-tory authorities (DRAs) are reluctant to becomeinvolved in relative efficacy, and drug companiesare not naturally inclined to invest in compari-sons of closely related compounds which may notshow their own product to be the best.

23

< -. 2 . -


Figure 1 .—Yellow Card Report Form Used in Great Britain

1.

2.3.

4.

IN CONFIDENCE–REPORT ON SUSPECTED ADVERSE DRUG REACTIONSPlease report all reactions to recently introduced drugs and serious or unusual reactions to other drugs. (Vaccines should be regardedas drugs).Record on the top line the drug you suspect of causing the adverse reaction.Record all other drugs, including self-medication, taken in the previous 3 months. With congenital abnormalities, record all drugs takenduring pregnancy.Do not be deterred from reporting because some details are not known.

NAME OF PATlENT SEX AGE OR DATE OF WEIGHT(To allow linkage with other re- BIRTHports for same patient. Also giverecord number for hospital patient)

DRUGS* (Give brand name if known) ROUTE DAILY DATE INDICATIONSDOSE S T A R T E D E N D E D

● For Vaccines give Batch No.

REACTIONS STARTED ENDED OUTCOME (e.g. fatal, recovered)

ADDITIONAL NOTES REPORTING DOCTOR

Name

Address

Tel. No.

Signature

Date

— .— -

Ch. 3—History and Objectives of Postmarketing Surveillance . 25

Figure 1 .—Yellow Card—Continued

REPORTING ADVERSE REACTIONS: GUIDELINES

DO NOT REPORT 1. Deliberate or accidental overdose2. Overdose due to errors of prescribing or administration3. Excessive but otherwise ‘normal’ effects of drugs whose dose has to be carefully titrated (e.g., insulin,

hypotensives, anticoagulants)4. Familiar relative overdose—excessive effect of a normal dose due to known predisposing factors (e. g., digoxin

toxicity in presence of hypokalaemia; toxicity in patients with impaired renal function)5. Inevitable side effects produced by known pharmacological activities of a drug (e.g., dry mouth with anticholinergic

drugs, tricyclic antidepressants, etc.; and hyperuricaemia with diuretics)

REPORT ALL 1, Drug interactions—known or suspected2, Reactions to new drugs marketed for less than 3 years3. Totally unexpected or unexplained events (including death) which could be drug induced4. Congenital abnormalities5. Infrequent reactions causing significant morbidity even if well known

CHECK LIST FOR REPORTING (not comprehensive)

General a. anaphylaxisb. all serious skin rashesc. all blood dyscrasiasd. thrombosis associated with

oestrogens or oral contraceptivesGastro-lntestinal a, jaundice

b. malabsorptionc, intestinal ulcerationd. severe bleedlng

Cardiovascular a. myocardial toxicity, e.g., unexpectedarrhythmias (exclude digoxin)

b. hypertensive reactionsRespiratory a. brohchospasm

b. non-infect!ve lung disease, e.g.,pneumonitis, fibrosis

Nervous System a. convulsionsb. unexpected confusional states

(hallucination or psychotic reactions)c. unexpected extrapyramidal effects (exclude

phenothiazines and butyrophenones)

Renal

Joints

d. peripheral neuropathye. neuromuscular blockadef. myopathy and myalgia

Eye & Ear all reactionsEndocrine a. unexpected reactions

b. amenorrheac. infertilityExclude gynaecomastia, fluid retention,hypokalaemiaemia, hyperkalaemia,hypoglycemia, hyperglycemia,hyperuricaema porphyria (if produced bydrugs known to have these effects)ail nephrotoxicityExclude urinary retention induced by diureticsor due to anticholinergic drugsa. all arthropathiesb. D. LE. syndrome

SOURCE Committee on Safety of Medicines. London

Table 4.— Percentage of Reports of Suspected ADRs in Great Britainby Class of Reporter and Method Useda

Percentage of reports by class of reporter

General Hospital Hospital Others (e.g,,Method used practitioner consultant junior coroner) All reporters

Yellow card. . . . . . . . . . . 58.3% 6.30/o 9.7% 1 .7%0 76.00/oDrug company . . . . . . . . 6.2 4.5 2.9 — 13.6Correspondence . . . . . . 1.7 0.8 0.4 4.6 7.5Death certificates . . . . . 0.2 — 1.3 0.9 2.4Medical journal . . . . . . . — 0.2 0.3 — 0.5

All methods . . . . . . . . 66.40/, 11 .80/0 14.60/o 7.20/, 100.0%0aBased on random 10 percent sample of approximately 12,000 recent reports UP to June 1978

SOURCE W H W Inman (ed ), “The United Kingdom, ” In &for? /torirtg for Drug Safety (Philadelphia’ J B. Llppincott Co , 1980)


Furthermore, an earlier suggestion linked post-marketing surveillance with restricted release ofdrugs (75):

A case could be made for an immediate or re-stricted release system for the introduction of newdrugs before their widespread use, half way be-tween clinical trials and the monitored releaseproposals. Because of inevitable delays, it is pos-sible that by the time 5,000-10,000 cases havebeen fully monitored by any of the aboveschemes, many more patients will have been ex-posed to the drug so that any serious adversereactions in the monitored group are duplicatedin those patients not fully monitored. The dura-tion of the “restricted release” phase would de-pend on the drug and disease concerned. Moni-toring of patients would continue, when appro-priate, after the drug became generally available.

In the United States, the Food and Drug Admin-istration’s (FDA’s) drug approval process was al-ready under intense scrutiny in the early 1970’S.As a result of 1974 hearings before the Subcom-mittee on Health, Senate Committee on Labor andHuman Resources, chaired by Senator EdwardKennedy (D-Mass.), the Department of Health,Education, and Welfare formed a Review Panelon New Drug Regulation. This panel, which con-vened in February 1975 and issued its report inMay 1977 (16), addressed two issues: 1) whetherthe drug law requirements for premarketing test-ing unnecessarily delayed the availability ofvaluable prescription drugs, and 2) whether thedrug industry exerted undue influence on FDAdecisions. The panel concluded that there was in-sufficient evidence on the first question and nowidespread improper influence. It also identifiedfour categories of deficiencies in the regulation ofdrugs: 1) openness and public accountability,2) FDA’s science environment, 3) standards andprocedures for premarketing approval, and4) FDA’s role in the postmarketing period (16,18).

Senators Kennedy, Javits, and others then intro-duced a bill in early 1978 to revise the drug pro-visions of the Food, Drug, and Cosmetic Act. Arevised bill, S. 1075, the Drug Regulation ReformAct of 1979, passed the Senate in September 1979.However, a similar bill, H.R. 4258, was not actedon by the House of Representatives. Included inthe Senate bill were these proposed changes in ex-isting law: 1) drug sponsors could be required to

conduct postmarketing surveillance of a drug forup to 5 years; 2) a prescription drug could haveits distribution limited if it could not otherwisebe found to be safe and effective; 3) the standardfor a drug’s immediate removal from the marketwould be changed from the drug being an “immi-nent hazard to the public health” to the lessstringent standard of “unreasonable risk of illnessor injury to any segment of the population;” and4) establishment of a “National Center for DrugScience. ”

During this period, in a speech to the Pharma-ceutical Manufacturers Association, Senator Ken-nedy suggested that a better system was neededfor monitoring the use and effects of prescriptiondrugs after they were marketed. As a result, theJoint Commission on Prescription Drug Use wasestablished in 1976, funded largely by the drugindustry, with the mandate to design a postmar-keting surveillance system to detect, quantify, anddescribe the anticipated and unanticipated effectsof marketed drugs, and to recommend a meansby which information on the epidemiology of pre-scription drug use in the United States could bedistributed regularly to interested parties. TheJoint Commission issued its report in January 1980with the following five conclusions and recom-mendations identified as its most important ones(see

1.

2.

3.

4.

app. B for the complete list) (42):

A systematic and comprehensive system ofpostmarketing drug surveillance should be de-veloped in the United States.Such a system should be able to detect impor-tant adverse drug reactions that occur morefrequently than once per thousand uses of adrug, to develop methods to detect less fre-quent reactions and to evaluate the beneficialeffects of drugs as used in ordinary practice.New methods will have to be developed for thestudy of delayed drug effects, including boththerapeutic and adverse effects.An integral function of the postmarketing sur-veillance system should be to report the usesand effects of new and old prescription drugs.Recognizing the progress that FDA has madein the area of postmarketing drug surveillancein the last 3 years, the Commission recom-mends that PMS [postmarketing surveillance]should be a priority program of the FDA andthat the FDA should continue to strengthen itsprogram in this area.

Ch. 3—History and Objectives of Postmarketing Surveillance ● 2 7.— —

5.

In

sion

A private, nonprofit Center for Drug Surveil-lance (CDS) should be established to furtherthe development of a postmarketing surveil-lance system in the United States. This centershould foster cooperation among existing post-marketing surveillance programs, develop newmethods for carrying out surveillance, trainscientists in the disciplines needed for doingpostmarketing surveillance, and educate bothproviders and recipients of prescription drugsabout the effects of these drugs.

1976, the year in which the Joint Commis-was formed, an interagency agreement was

signed between FDA and the Experimental Tech-nology Incentives Program (ETIP) at the NationalBureau of Standards of the Department of Com-merce. The purpose of ETIP was to provide incen-tives or reduce barriers to technological innova-tion through changes in the regulatory process.ETIP’s agreement with FDA was to jointly funda program to determine if improvement in post-marketing surveillance could help reduce theregulatory requirements of the premarketingperiod, principally those of phase III of the in-vestigational new drug process and those of thefollowing new drug application process. Thespecific experiment was to develop postmarketingsurveillance systems and a method of managingand evaluating the reform (11).

The project concentrated on collecting data todesign such systems, and issued a status reportin 1981 (12). Another report will be issued by theproject in 1982. FDA has assumed most of thefunding, and the Department of Commerce wasto phase out ETIP in 1982. FDA is continuing theactivities originated or stimulated by the program(see ch. 5).

A Commission on the Federal Drug ApprovalProcess was convened in mid-1981 to examinehow FDA’s procedures for the approval of newdrugs can be expedited without compromisingpublic safety; it is to also make recommendationson the development of cost-effective postmarket-ing surveillance to guarantee the quick withdrawalfrom the market of drugs that cause significantadverse effects. The commission had its genesisin a joint hearing held in April 1981 by the HouseScience and Technology Committee’s Subcommit-tee on Natural Resources, Agriculture Research,

and Environment and its Subcommittee on Inves-tigations and Oversight. The commission’s firstmeeting was held in July 1981, and its report wasto be released in late 1982.

FDA is also examining specific ways to speedup the drug approval process. It is reviewing pastphase 111 trials to see if longer trials or those withlarge sample sizes have contributed useful infor-mation beyond that obtained in phase II and earlyphase III testing. Past postmarketing studies thatFDA has required are also being reviewed to seeif they provided the information that was origi-nally sought. FDA data on approval time, vali-dated by interviews with FDA and the manufac-turers, are being reviewed for factors that mayslow the approval process. And, as a pilot test,an FDA committee is reviewing the pharmacologicand clinical data on selected drugs at the end ofphase II testing, and will make recommendationsabout the best time to gather additional safety in-formation (e.g., phase III v. the postmarketingperiod) (11). FDA Commissioner Hayes has beenquoted as saying that one possible methodologicalchange is to accept foreign data in the premarketapproval process (67). These activities haveresulted in proposed new rules for new drugregulation (47 Federal Register, pp. 46622-66, Oct.

19, 1982). In March 1982, the FDA Commissionerbegan a related reorganization by merging theBureau of Drugs with the Bureau of Biologics, andreplacing the Director of the New Drug Evalua-tion Division. The merged bureaus have sincebeen designated the National Center for Drugsand Biologics.

Finally, in a related development, the Senatepassed by a voice vote, in the first session of the97th Congress, the Patent Term Restoration Actof 1981 (S. 255). The bill would restore to the termof a patent the time lost in complying with theGovernment’s premarket testing and review re-quirements, up to a maximum of 7 years. Patentedproducts eligible for extension would not belimited to human drugs, but would include“human drugs and biological, antibiotic drugs,animal drugs and biological, food additives,color additives, pesticides, other chemicalsubstances, medical devices, and any other prod-uct subject to Federal premarket requirements”


(72). In September 1982, the House of Represent-atives voted on the bill under suspension of itsrules. Under such conditions, a two-thirds votewas required for passage, and although the billreceived a majority of the votes, it fell just shortof the two-thirds majority needed.

Thus, in the United States, the issue of post-marketing surveillance has involved more thanidentifying the serious adverse effects of newly in-troduced drugs. It has also led to the recommen-dation to monitor all drugs for their effectivenessand appropriate use by patients and physicians.In addition, improvements in postmarketing sur-

veillance have been linked to changes in the drugregulatory process. Proponents of more rapiddrug approval claim that phase III testing (e.g.,the chronic trials) adds little to the data collectedin phase II, and that, as a consequence, phase 111testing could be curtailed or shifted to the post-marketing period. On the other hand, those con-cerned about drug deficiencies discovered afterdrug approval point out that FDA has limited op-tions once a drug has been released. Since FDAlacks authority to limit drug distribution or use,it can only try to remove a drug from the market,if such action appears appropriate.

4Methods of Drug Evaluation;

4Methods of Drug Evaluation;

TYPES OF STUDIES

The primary objective of postmarketing studiesis to develop inforrnation about drug effects undercustomary conditions of drug use.

The initial clues about a drug’s potential effectscome from the experimental studies carried outwith both animals and humans in the premarket-ing period. Spontaneous or voluntary reporting(e.g., in letters to the editors of medical journals)is the oldest, and to date, the most productivesource of new information about a drug’s possi-ble effects once a drug is marketed. Other typesof studies are used to examine in more detail thepossible effects of a drug. In general, these othertypes of studies use either cohort or case-controlmethods. Thus, four types of studies are generallyused to identify drug effects: 1) controlled clini-cal trials, 2) spontaneous or voluntary reporting,3) cohort studies, and 4) case-control studies (23,50,61,77).

Controlled clinical trials match treatment andcontrol groups as closely as possible, minimizebias through such methods as randomization and“double-blinding,” and directly monitor patientsfor the duration of the study. For example, pa-tients can be randomly assigned to either the con-trol or treatment group. The control group re-ceives a placebo or an active comparison drug thatlooks exactly like the drug being tested, and boththe investigators and patients do not know whois receiving the real drug. (Personnel not directlyinvolved in the tests would of course know whatsubstance each patient was receiving). In thismethod, possible drug effects, both therapeuticand adverse, are closely monitored, so that theyare discovered as they occur.

The controlled clinical trial is considered themost definitive method for evaluating a drug’s ef-ficacy and safety, but the use of rigorous criteriafor patient selection usually means that the pa-tients tested represent only a special class of theanticipated users of the drug(s), and the careful-

ly controlled conditions allow for study of fewerpatients than in the other methods, Thus, for ex-ample, to observe drug effects that are rare or thatappear only after long-term use, controlled clinicaltrials might be impractical or too expensive.

Voluntary reporting may be spontaneous, suchas in a “letter to the editor” of a medical journalabout an unusual condition observed in a patienton a particular drug, or it may be more organized,as with the “yellow card” system in Great Britain.Most of the reporting to the Food and Drug Ad-ministration (FDA) is by pharmaceutical manufac-turers, who are required by law to report adversereactions. In practice, most of the information ob-tained by the manufacturers originates from phy-sicians and other health professionals. Such obser-vations serve as warnings of possible adverse drugeffects, so that the inference of an association be-tween a drug and an observed health conditionmay be further studied by cumulative, carefulreporting, and confirmed or disconfirmed by morevigorous methods. Underreporting is a seriousdeficiency of this voluntary method, and a drugmay also be wrongly associated with an adverseeffect until the suspected association fails to showup in repeated, statistically validated studies.

Cohort studies follow a defined group of pa-tients (the cohort) for a period of time. In thismethod, patients are not randomly assigned togroups, and there is no blinding. Cohort studiesare usually prospective, and observe the cohortfrom the beginning of drug use. A group of pa-tients taking the drug of interest is assembled andfollowed to see, for example, if any adverse reac-tions occur. A second group of patients (the con-trols) with the same medical condition, who arenot taking the drug and who may be receivingalternative treatment, but who are otherwisematched as closely as possible with the cohort,may be studied in parallel. The control group isused to identify the frequency of occurrence of

31

32 Ž Postmarketing Surveillance of Prescription Drugs

any condition observed in the drug-exposedgroup, but which must be due to causes other thanthe drug (the “background incidence” of the con-dition). In this method, patients can be directlymonitored to ensure they take the drug appropri-ately and to observe the drug’s effects; or monitor-ing can be less systematic. With less monitoring,a larger cohort can be followed, but bias is thusincreased.

Although uncommon, a retrospective cohortstudy may also be conducted when purporteddrug-induced effects have already been observedat the time the study is started. A retrospectivecohort study must accurately ascertain patients’past drug use. In principle, this could be done bya “closed” pharmacy record system, in which usershave been restricted to a single supplier or pay-ment source. In practice, the use data on individ-uals needs to be connected to data on their out-come, as through computer files. Medicaid, Medi-care, the military, and some prepaid health main-tenance organizations could be employed for thispurpose (76).

Case-control studies identify patients with theadverse effects to be studied (the cases), and com-pare them with a sample (the controls), drawnfrom the same cohort that gave rise to the cases.Controls are matched as closely as possible withthe cases, except with regard to the drug’s sus-pected adverse effect, to examine whether expo-sure to the drug is the cause. Patients with condi-tions suspected of being associated with a certaindrug would have their medical records reviewedor be interviewed concerning the use of that drug.The histories of the controls would also be studiedfor information about drug use in the general pop-

ulation. By comparing the proportion of drugusers among the cases with the proportion of drugusers in the general population, it is possible toinfer the relative frequency with which adversereactions occur in users of certain drugs as com-pared with nonusers. A sufficient number of ap-propriate cases must be identified and accuratehistories of exposure to drugs must be obtained.

Among the advantages of retrospective case-control studies compared with prospective cohortstudies are the smaller number of patients requiredin retrospective case-control studies, the relativeease of carrying out the study, the lower cost, andthe shorter time needed. A disadvantage of theretrospective case-control method is that a con-dition must have been already identified and sus-pected as the effect of drug use. It is also harderto reduce bias in a retrospective study than to doso in a prospective one (65).

Bias is equally possible in cohort and case-con-trol studies, though each kind of study is liableto a different kind of bias. For example, bias inthe observations can arise with respect to identify-ing the effects of the drug in cohort studies andwith respect to identifying the exposure in case-control studies.

Controlled clinical trials and prospective cohortstudies can be used to determine a drug’s beneficialas well as adverse effects. Case-control studies areusually used to trace adverse effects back to priordrug use. Voluntary reporting can uncover addi-tional uses of drugs as well as their adverse ef-fects, but reporting of adverse effects is muchmore common.

DETECTION AND ASSOCIATION

The ability of a particular surveillance method these factors are more a problem in the design ofto detect a drug’s effect depends on two factors: a study’s details,1) the time that transpires between use of that drugand the occurrence of the drug’s effect (the laten- The latency of some drug effects presents a seri-cy period), and 2) how often the effect occurs (its ous problem for their study. Some effects occurfrequency). There are many other determining immediately, or within days or weeks after drugfactors, such as accuracy of observation, and ac- use, or with continued use of a drug. But othercuracy and completeness of medical records, but effects may occur long after a drug has been dis-

.

Ch. 4—Methods of Drug Evaluation ● 3 3

continued, or only when another drug is takensimultaneously, or only in patients with certainpredisposing conditions. Other effects may notbe manifest in the patients themselves, but ratherin their children. The use of DES (diethylstilbes-trol), in pregnant women, for example, has beenassociated with vaginal cancer in their daughters,but only after the daughters have reached adoles-cence.

Controlled clinical trials, because of their rela-tively short duration, will detect only acute orsubacute effects. Long-term cohort studies candetect delayed effects, but the data bases neces-sary for such long-term, large studies are stillsparse. Voluntary reporting is usually the way inwhich long-term effects are first identified. Long-term effects are usually confirmed through retro-spective case-control studies, but their reliance onhistorical data such as medical records can limitthe accuracy of these studies.

The chance that a particular study will discovera drug effect also depends on the study’s samplesize and the frequency of the drug effect. For ex-ample, in a cohort study, if a drug causes blind-ness in 1 out of every 100 users (1/100), howmany users must be observed to find one case ofblindness? If there were 1 million users of the drug,there would be 10,000 users blinded. But in a smallsample of only 100 users, the probability of find-ing one or more cases of blindness in the sample

would only be 63 percent. If the sample were 200users, the probability would increase to 86 per-cent. With a sample of 500, the probability wouldbe 99 percent that at least one case of blindnesswould be found in the observed users.

To state it another way, what number of userswould have to be observed to be 95 percent sureof finding one or more cases of blindness whenthey occur at a frequency of 1 in 100 users? Theanswer is 300 users, and the general rule is thatthe number of users in the sample must be threetimes the reciprocal of the frequency; e.g., for afrequency of 1 in 1,000, the sample would haveto be 3,000 to be 95 percent sure of observing atleast one case. Table 5 summarizes: 1) the proba-bilities of observing an adverse drug reaction(ADR) for different sample sizes and frequencies,and 2) the sample size required for various fre-quencies of an adverse reaction to be 95 percentsure of observing that reaction. The numbers inthe bottom row are the sample sizes needed to be95 percent sure of observing at least one ADR.In table 6 are shown the sample sizes needed tobe 95 percent sure of observing one, two, or threeadverse reactions for the frequencies shown.

In a sample of 3,000, effects occurring at anyrate higher than 1/1,000 (e.g., 1/10, 1/50, 1/200,etc. ) should be observed, and there will be a 95percent probability of observing at least one ef-fect that occurs at a frequency of 1/1,000. How-

Table 5. —Likelihood of Observing an ADR (95% likelihood)

Threshold for an ADR

1 /100 1 /500a 1/1,000b 1/5,000c 1/10,000 d 1/50,000e

Number of patients inADR study:

100 ., . . . . . . . . . . . . ... . 0,63 0.18 0.10 0,02 0.01 0.002200 . . . . . . . . . . . . . . . . . . . . 0.86 0.33 0.18 0.04 0.02 0.004500 . . . . . . . . . . . . . . . . . . . . 0.99 0.63 0.39 0.10 0.05 0.011,000, . . . . . . . . . . . . . . . . . . 0.99 0.86 0.63 0.18 0.10 0.022,000 . . . . . . . . . . . . . . . . . . . . 0,99 0.98 0.86 0.33 0.18 0,045,000 .., . . . . . . . . . . . . . . . . . 0.99 0.99 0.99 0.63 0.39 0.1010,000 . . . . . . . . . . . . . . ., . . . 0.99 0.99 0,99 0.86 0.63 0.18

Number of patients requiredto be 95°/0 likely toobserve an ADR . . . . . . . . . 300 1,500 3,000 15,000 30,000 150,000—. ——.

aFor r example, Iymphoma from azothioprinebF or ~ xample, eye damage from practololCFor example, heart attack I n older women frp, oral contraceptivesdFor example, anaphylaxls from penlclllane For example aplastlc anemia from chloramphenlcol

SOURCE D L Sackett, et al ‘Compliance,”’ I n Monftorfng for Drug Safety W H W I n man (ed ) (Phi I adel ph la J B LI ppI ncot tCO 1980)

34 . Postmarketing Surveillance of Prescription Drugs

Table 6.—Number of Patients Required To DetectOne, Two, or Three ADRs With No Backgroundincidence of Adverse Reaction (950/0 likelihood)

Required number of

Expected incidence of adverse reactions

adverse reaction 1 ADR 2 ADRs 3 ADRs1 in 100 . . . . . . . . . . . . . . . 300 480 6501 in 200 . . . . . . . . . . . . . . . 600 960 1,3001 in 1,000 . . . . . . . . . . . . . 3,000 4,800 6,5001 in 2,000 . . . . . . . . . . . . . 6,000 9,600 13,0001 in 10,000 . . . . . . . . . . . . 30,000 48,000 65,000SOURCE J A. Lewis, “Post-Marketing Surveillance How Many,” Trends in Phar.

macological Sciences 2.93, 1981

ever, except for some effects that are unique toa drug, many drug effects (e. g., stroke, heart at-tack) are indistinguishable from conditions dueto other causes. This “background incidence” ofa condition must be known before purported drugeffects observed in a study can rightly be attrib-uted to a drug.

The sample size needed to be reasonably surethat an observed condition is the drug’s effectdepends on the following factors: 1) the back-ground incidence of the condition, 2) the addi-tional incidence due to the drug, and 3) the sizeof the control group, if the background incidenceis unknown (47). The quantitative relationshipsamong those factors are summarized in table 7,which presents the sample sizes required to be 95percent sure that an observed condition is due tothe drug and not to some other cause.

Larger sample sizes are needed to determine adrug’s effect as the background incidence of a con-dition increases and as the frequency of the drug’s

added contribution to a condition decreases. Thisis best explained graphically, as in figure 2. In thisexample, the background incidence in every caseis 1/100. As the incidence of the added effectdecreases from 1 in 100 to 1 in 10,000, the sam-ple size has to increase from 1,600 to 11 millionto remain 95 percent sure of observing the addedeffect (see the rows in table 7 representing aknown background incidence of 1 in 100).

The relationship between background andadded incidence is also revealed in consideringsample sizes at the extremes. For a known back-ground incidence of 1 in 1,000 and an added inci-dence of 1 in 100, the sample size needed toobserve at least one case of the added effect is only500. But when the background incidence is 1 in10 and the added incidence is only 1 in 10,000,the sample size must be 98 million (table 7).

Table 7 also illustrates that another factor couldincrease the sample size that a study requires. Ifthe background incidence is not known and hasto be estimated through observation of controlgroups, the smaller the size of the control group,the larger the sample size of drug users must befor the same degree of confidence in the results(compare the sample sizes required for controlgroups equal to the treated group and for thosefive times its size).

If there is a background incidence, the samplesize needed to observe a drug-induced effect risesdramatically. Recall that a sample size of 3,000was needed to be 95 percent sure of observing atleast one drug-induced effect with a frequency of

Table 7.–Number of Patients Required in Drug-Treated Group To Detect One ADRWith Background Incidence of Adverse Reaction (95% likelihood)

Number of patients requiredwith additional incidence of

Background incidence of adverse reaction to drug

Size of control group adverse reaction 1 in 100 1 in 1,000 1 in 10,000

“Infinite” (background incidence known) . . . . . . . . . . . 1 in 10 10,000 980,000 98,000,0001 in 100 1,600 110,000 11,000,0001 in 1,000 500 16,000 1,100,000

Five times as big as treated group . . . . . . . . . . . . . . . . 1 in 10 12,000 1,200,000 120,000,0001 in 100 1,900 130,000 13,000,0001 in 1,000 700 19,000 1,400,000

Equal to treated group . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 in 10 20,000 2,000,000 200,000,0001 in 100 3,200 220,000 22,000,0001 in 1,000 1,300 32,000 2,300,000

SOURCE J A Lewis, “PostMarketing Surveillance How Many, ” Trends m Pharmacological Sciences 2“93, 1981.

.

Ch. 4—Methods of Drug Evaluation ● 3 5

Figure 2.—Comparison of Additional Drug-InducedEffects of Decreasing Incidence

1 in 100Added incidence

Background incidence

1 in 1,000

I in 10(

SOURCE Office of Technology Assessment

1 in 10,000

I in 10(

—

1 in 1,000 and with no background incidence (ta-ble 6). If the background incidence is also 1 in1,000, the sample size required rises to 16,000. Abackground incidence higher than the drug’s

added incidence increases the sample size requiredeven more dramatically (to 110,000 and 980,000for background incidence of 1 in 100 and 1 in10 in table 7).

Many studies monitor for several effects, notjust one, and some apparent drug effects may haveoccurred by chance. Minimizing these chance rela-tionships also increases the sample size required.Table 8 illustrates the hypothetical case where 100effects will be examined in one study, and the sam-ple size reflects a 95 percent chance that the ob-served effects will be related to use of the drug.

These sample size illustrations reflect, at the 95percent confidence level, that an effect will beobserved and whether that effect can be attributedto drug use. They do not provide a good estimateof the added incidence of the effect from the drug(recall from table 6 that a sample size of 3,000 isneeded just to detect at least one effect that hasan incidence of 1 in 1,000).

Controlled clinical trials are used primarily forevaluating drug efficacy, not safety, because theyare carried out on hundreds, or, at the most, afew thousand drug users. Their use for evaluatingdrugs already on-the market is also limited bytheir high cost and logistical problems, In fact,the use of controlled clinical trials for determin-ing efficacy alone is already constrained by thesetwo factors (9,46).

Table 8.—Number of Patients Required in Drug-Treated Group To Allow forExamination of 100 Adverse Reactions (950/’ likelihood)

Number of patients requiredwith additional incidence of

Background incidence of adverse reaction to drugSize of control group adverse reaction 1 in 100 1 in 1,000 1 in 10,000“Infinite” (background incidence known) . . . . . . . . . . . 1 in 10 23,000 2,200,000 220,000,0%

1 in 100 3,100 250,000 24,000,0001 in 1,000 800 321000 2,500,000

Five times as big as treated group . . . . . . . . . . . . . . . . 1 in 10 27,000 2,600,000 260,000,0001 in 100 4,000 300,000 29,000,0001 in 1,000 1,300 40,000 3,000,000

Equal to treated group . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 in 10 46,000 4,400,000 440,000,0001 in 100 7,200 510,000 48,000,0001 in 1,000 2,900 73,000 5,100,000

SOURCE” J A Lewis, “Post-Marketing Surveillance How Many, ”Trends m Pharmacological Sciences 2:93, 1981.


These limitations of controlled clinical trials inevaluating the safety of marketed drugs have ledto relying on cohort and case-control methods forpostmarketing studies. While these latter methodscan only indicate an association between a drugand observed conditions, but not that the rela-tionship is causal (49,77), the cumulative ex-perience of multiple cohort and case-control

studies that show consistent associations betweena drug and a suspected effect can lead to a highdegree of confidence that the relationship iscausal.

The following three case studies illustrate theuse of the four methods of drug evaluation: con-trolled clinical trials, voluntary reporting, cohortstudies, and case-control studies.

CASE STUDIES IN DRUG EVALUATION

Oral contraceptives, the most investigateddrugs in use, were first associated with cardiovas-cular disease in 1966 by British researchers. Earlierthat year, they had noticed that Britain’s yellowcard system had revealed that relatively more re-ports of thromboembolism (heart attacks, strokes,deep vein thrombosis of the legs with or withoutemboli to the lungs) were associated with one typeof oral contraceptive than with another. A case-control study later that year established that therewas a significant association between oral contra-ceptive use and thromboembolism, but it foundno significant difference between the two contra-ceptive types (38). Publicity over these adverseeffects increased voluntary reporting, so that with-in 3 years more than 1,300 reports of thromboem-bolism had been received. By 1969, researchersestablished that it was the total dose of estrogens,not the different types of estrogen, that was re-sponsible for any differences (39). As a result,high-dose estrogen oral contraceptives were re-moved from the market in 1970. The associationbetween oral contraceptives and thromboembol-ism has been confirmed in other studies (15,51,52,73,74).

It is now known that the risk of thromboem-bolic disease from oral contraceptives increaseswith age. Young women have a chance of 1 in20,000 of dying from the cardiovascular effectsof oral contraceptives (62); but a woman takingoral contraceptives from age 35 to 45 has a riskof death of about 1 in 1,000 over the 10-year peri-od (32). The risk of heart attack from oral contra-ceptives has also been found to increase for pa-tients who smoke cigarettes (52), so that olderwomen who smoke have the greatest risk.

Other conditions associated with oral contra-ceptives include gall bladder disease (6, 7), livertumors (l), yeast infections of the vagina (27), andincreased bacteria in the urine (78). Oral contra-ceptives have also been associated with jaundice,but only in patients with a rare genetic conditionin which there is impaired biliary excretion of bili-rubin, the Dubin-Johnson syndrome (14). De-creased effectiveness of oral contraceptives hasbeen associated with the use of anticonvulsantsand other drugs (54), but oral contraceptives havealso been associated with reduced incidence ofnonmalignant breast disease (8) and ovarian cysts(58). Most, if not all, of these other conditionsassociated with oral contraceptive use have beenobserved in only a few studies, and the role ofcontraceptives in producing them is not so clearas it is in the case of the occurrence of cardiovas-cular conditions.

To illustrate that an association between a drugand an observed condition may be too casuallyinferred, it should be noted that in 1970, aSwedish physician reported Down’s syndrome (ortrisomy 21, the most obvious symptoms of whichare mental deficiency and mongoloid features) intwo children whose mothers had taken oral con-traceptives before pregnancy. At the same time,one hospital reported increased numbers of chil-dren born with the syndrome. Sweden’s monitor-ing system therefore compared the incidence ofthe syndrome during a period (1968-70) when oralcontraceptives were widely used with earlierperiods when oral contraceptives were not avail-able. The incidence of the syndrome turned outto be lower in the period of oral contraceptive use(48). The public did not know of the investiga-

Ch. 4—Methods of Drug Evaluation ● 3 7— — . ——

tion until it was completed. “One can easily imag-ine the worldwide alarm such a suspicion couldhave caused” (3).

Streptokinase, a drug derived from group Cbeta-hemolytic streptococci for use in dissolvingblood clots, was released for marketing in theUnited States in November 1977, following its usein Europe for many years. Bleeding, allergic reac-tions, and fever had been found in the premarket-ing clinical trials carried out on 535 patients.Hoechst-Roussel Pharmaceuticals, manufacturersof the drug, implemented a postmarketing surveil-lance study at the request of FDA. The studylasted until May 1980, at which time FDA liftedthe postmarketing study requirement.

Hoechst-Roussel included a reporting form ineach package of the drug vials sent to hospitalpharmacies, and requested that the pharmacisthave treating physicians complete the forms.Based on sales, the company estimated that re-ports were returned for about 20 percent of treatedpatients, a total of 306 patients of 260 physiciansin 44 States and the District of Columbia,

Physicians were asked to rate treatment out-comes as “successful, ” “partially successful, ” or“unsuccessful. ” Thirty-nine percent were reportedas “successful, ” 32 percent as “partially success-ful, ” and 30 percent as “unsuccessful” (percentagesrounded to nearest point).

Fifty percent of the patients (153/306) had a to-tal of 208 adverse reactions, with 49 patients re-ported to have 2 or 3. The reactions reported werethe same as those seen in the clinical trials—aller-gic reactions, fever, and bleeding. Fifteen patientssuffered severe bleeding, and 48 less severe bleed-ing. Three patients with severe bleeding died, andone had nerve damage secondary to bleeding(femoral hematoma). Most patients recoveredfrom the drug reaction, but a total of 25 of the306 patients died during or within a few weeksof therapy. The investigators concluded that noadverse reactions were discovered that were notalready known from the premarketing clinical tri-als and that their incidence and severity were sim-ilar (71).

This postmarketing study is instructive for anumber of reasons. First, it was necessary to select

the sample from marketing sources and to rely

on voluntary reporting, so there was a bias in pa-tient selection, there were no controls, and thecriteria for effectiveness were vague. The low re-porting rate of 20 percent is also typical of volun-tary methods. Second, the purpose of the studywas “to determine whether the incidence and se-verity of adverse reactions with widespread usewould differ from that seen during clinical trials. ”The investigators concluded that incidence, sever-ity, and type of reaction were similar, but notedthat the final sample size was only 306, smallerthan that of the 535 patients in the premarketingclinical trials. Adverse reactions with incidenceless than those observed in the clinical trials werenot likely to be discovered in the smaller postmar-keting study. The strongest conclusion that canbe reached on the basis of the later study is thatthe types of drug reactions in an uncontrolled pa-tient population were similar to those of the morecarefully selected patients in the clinical trials.

Finally, this study illustrates a problem that isoften encountered in evaluating drugs. In contrastto patients taking oral contraceptives, these pa-tients were being treated for underlying diseases,which can complicate the interpretation of whatis observed. For example, 25 of the 306 patientsdied, 15 from pulmonary emboli and 4 from deepvein thrombosis, both conditions that are relatedto the diseases being treated, The investigatorsconcluded that these deaths occurred about as fre-quently as they did during the premarketing trials.

Cimetidine, an anti-ulcer drug that blocks therelease of stomach acid, was approved for market-ing in August 1977. Because it was a new classof therapeutic agent and was expected to be wide-ly used—it was an alternative to surgery in reduc-ing stomach acidity—a postmarketing study wasrequested of the drug’s sponsor—Smith, Kline &French Laboratories. The drug had become avail-able a few years earlier in some other countries,including Great Britain.

The sponsor established a cohort of about10,000 patients by using its sales representativesto enroll over 1,000 physicians who were askedto complete case report forms on at least 10 pa-tients for whom they would prescribe cimetidineduring a 3-month period starting in March 1978.

38 ● Postmarketing Surveillance of Prescription Drugs— —— — — — —— . — —

A second phase (not reported by early 1982) wasto cover the following 6-month period. The ex-pected 10,000 patient cohort would provide a 90-percent probability of observing a drug effect withan incidence of 1/4,348 and an 80-percent proba-bility of observing one with an incidence of 1/6,250. The physician response rate was 85 per-cent, with 9,907 case reports used in the study.A control group was not included because of de-sign and cost problems.

A total of 577 adverse reactions were reportedin 442/9,907 (4.4 percent) cases. The investigatorsconcluded that these adverse reactions did not dif-fer in type or incidence from those observed inthe premarketing controlled clinical trials (31).

The most common drug reactions observed inthe patients were diarrhea and nausea or vomit-ing, Previous reports in the literature (i. e., sponta-neous reporting) from Great Britain had postu-lated an association between cimetidine use andmental confusion, blood dyscrasias, and endo-crine effects. Some patients did develop suchsymptoms, but the investigators concluded thatonly the 18 cases of gynecomastia (breast swell-ing in males) and one case of blood abnormalitywere related to cimetidine use.

There were 65 patient deaths during the 3-month review period, none considered to becimetidine-related. Previous spontaneous reportshad also led to postulating an association betweencimetidine use and stomach cancer (19,34,60,70).However, the hypothesis was viewed with skepti-cism, as people who have ulcers also have a higherchance of developing stomach cancer. Though theinvestigators did not address this possible effectof cimetidine, they reported that the 19 patientswith tumor-related deaths were diagnosed for cor-responding conditions before cimetidine therapybegan.

The year after this study was published, Britishresearchers reported experiments in humans onthe basis of which they speculated about how cim-etidine could produce stomach cancer (59). Theinvestigators pointed out that the toxicologicaldata from long-term studies in rats and dogs did

not uncover any evidence of the drug’s carcino-genicity, but they contrasted these findings withtheir experimental findings and with the spontane-ous reports associating cimetidine use and stom-ach cancer in humans.

In contrast to the streptokinase study’s enroll-ment of a smaller cohort than the number of pa-tients evaluated in the premarketing tests, the cim-etidine postmarketing study was able to enroll itsgoal of a cohort of nearly 10,000 patients. Thecimetidine postmarketing study, however, is ofinterest for other reasons.

The published report covering only the first 3months of cimetidine use found no difference fromthe premarketing controlled clinical trials in thetypes or incidence of adverse reactions. Oneproblem with postmarketing studies, however, isto detect adverse reactions without prejudiceabout what will be found. The findings of the pre-marketing clinical trials can provide guides forwhat reactions to anticipate, but such guidesmight cause the observer to overlook other ad-verse reactions associated with the drug understudy.

In contrast to the published preliminary find-ings, FDA’s review of this study, not yet completeas of early 1982, suggests a lower rate of adversereactions than that observed in the premarketingtrials, although the types of adverse reactions andtheir relative rates in the two are similar (10). Thislast observation is in agreement with at least onepharmaceutical company’s experience (66):

Some people might think postmarketing sur-veillance can sharpen up estimates of incidenceof adverse reactions seen in phase III trials. PhaseIII trials, however, use trained investigators andthey ordinarily are treating the more severe pa-tients. Both of these facts contribute to higheradverse reaction rates in phase III studies thanone expects to get and actually gets in postmar-keting surveillance studies.

Thus, in designing postmarketing studies, theinvestigator must both be guided by the resultsof the premarketing clinical trials and not be over-ly influenced by them.

5.Current Activities

I

5.Current Activities

The following discussion suggests only some ofthe kinds of activities and resources that can con-tribute to postmarketing surveillance.

The Food and Drug Administration (FDA) ob-tains information on the postmarketing status ofdrugs from four main sources: 1) pharmaceuticalmanufacturers, 2) FDA contracts and grants,3) other governmental agencies, and 4) interna-tional centers such as the World Health Organi-

zation (WHO). Figure 3 summarizes the datasources available to FDA in its monitoring ofdrugs.

Pharmaceutical manufacturers are required toreport all adverse reactions regardless of whethera causal connection is believed to exist betweenthe drug and the symptom (CFR 21.310.301).Most of these adverse events are reported byphysicians, and many manufacturers have routine

Figure 3.— Drug Experience/Epidemiologic Sources Available to FDA forPostmarketing Surveillance and Risk Assessment

- — — — — — — —— - ——-—————— - — — — — . — — — — — — — — - — — — — — — — . — — — —; FDA contracts/grants

I

P P m

Boston ?uget Boston BostonI

Collaborative Sound Drug Children’sI Drug Study HMO Epidemiology HospitalI Unit SurveillanceI

— — — — — — — — — — — —— —————— ———

MedicaidDrug-Event

Data

— I ——I

Adverse reactionregistries● Hepatic● E y e● Radio contrast● Tissue (pathology)● Radiopharmaceut ical

iIIIIII

● Dermatologici I II

I ● prescription Data Services I] ● Medicaid

LI

. — — — — — — — — — — — — - iOther Federal agency data sources(partial list)

— ● Drug abuse warning network (DEA, NIDA)

l_-~ c Nat iona l morb id i ty , mor ta l i ty data (NCHS)

Ž NIH-funded studies (e. g., NC I epidemiologystudies, etc.)

● N IDA drug abuse surveys

Pharmaceutical industry

SOURCE U S Food and Drug Adm(nlstratlon

47

42 ● Postmarketing Surveillance of prescription Drugs_.—

procedures for responding to such spontaneousreports. A number of companies also carry out“product support studies” to interest physiciansin their products, and to defend a drug if it issuspected of causing an adverse effect, or if ques-tions are raised about its therapeutic value (12).

Manufacturers may also have to conduct a“phase IV” postmarketing study as a condition ofmarketing approval. Two examples of such re-search, that on streptokinase and cimetidine, weredescribed in chapter 4.

Some manufacturers are also attempting to de-velop their own surveillance programs. For ex-ample, Upjohn used group medical practices toform a pharmacy-based cohort for prospective ob-servational studies of selected drugs in order toestablish a large registry of patients to identify andanalyze important medical events through follow-up patient questionnaires (5).

FDA grants and contracts cover a number ofactivities. Voluntary reporting by physicians tomedical journals and directly to FDA is aug-mented by several specialized registries:

●

●

●

●

The National Registry of Drug-Induced Ocu-lar Side-Effects collects data from U.S. oph-thalmologists on drugs and their effects.The Registry of Patients Exposed to Radio-pharmaceutical Drugs collects informationfrom a random sample of hospitals licensedto practice nuclear medicine. All patients ex-posed to radiopharmaceuticals are registered,and adverse effects are reported.The Registry of Hepatic Toxicity to Drugscollects data on drug reactions that affect theliver.The Registry of Dermatological Reactions toDrugs, administered by the American Acad-emy of Dermatology, was recently initiatedto collect data on reactions affecting the skinand on reactions to dermatological drugs.

FDA purchases data from private organizationsto estimate the actual population using a drug.One use of this data might be to determinewhether patients exposed to a drug have an in-creased risk of experiencing an adverse effect.Another use might be to estimate the number ofpatients who would be using a new drug to help

decide whether a postmarketing study of the drugwould be needed.

For example, FDA’s Division of Drug Experi-ence contracted to use the Medicaid Medical In-formation Systems of Michigan and Minnesotato link individual prescription and diagnosticclaim files for 1.2 million patients, to access andtabulate the data, and to provide FDA with on-line terminal access to the data, to update the pa-tient profiles on a quarterly basis, and to providespecified reports. The Division of Drug Experienceplans to use this Medicaid data base for the fol-lowing purposes: 1) investigation of what FDAconsiders high-priority issues on adverse reac-tions, 2) development of data on drug utilization,3) development of a screening system for hypoth-esis generation of unsuspected adverse effects, and4) studies to describe and validate the usefulnessof the data base (25). Table 9 summarizes the an-ticipated uses of this data base. Table 10 identifiesthis and other data sources available to FDA forestimating actual populations of drug users.

In addition to using the Medicaid data fromMichigan and Minnesota to develop cohorts forstudies of drug use, FDA contributes to the sup-port of several ongoing surveillance programs,some of which include data collected from othercountries (e. g., Great Britain, New Zealand,Israel).

The Boston Collaborative Drug SurveillanceProgram (44) has accumulated data on the pasthospitalizations of more than 50,000 patients. Thisdata base is still useful for evaluating older drugs.The program also uses current data from the280,000 members of the Group Health Coopera-tive of Puget Sound (Washington State) and thedata files of the Commission on Professional andHospital Activities-Professional Activity Study inAnn Arbor, Mich. These data sources are usefulfor evaluating new drugs.

The Drug Epidemiology Unit at the Boston Uni-versity Medical Center conducts case-controlstudies (e.g., on birth defects), and also intensivelymonitors pediatric hospital patients in collabora-tion with the Children’s Hospital Medical Centerin Boston.

—

Ch. 5—Current Activities ● 4 3

Table 9.— Potential Uses by FDA of Medicaid Data From Michigan and Minnesota

General descrlpt!on of study General use Example Anticipated future impact

Drug utilization1. Overall characterization of drug

use by Medicaid population -

2.

3.

4.

Demographics of drug use

Drug use combinat ions

Drug use in diseasedindividuals,

Drug adverse effects1. Relative prevalence/incidence

of known adverse effects.

2, Rate of recognized drugadverse effects.

3. Identification of drug adverseeffects by means of rate ratiosfrom case control or cohortmodules.

4. Detection of acute drug-induced disease from temporalclustering of events followingexposure.

Cross-validation with data ondrug use.

Identification of potentialhigh drug exposure prob-lem areas,

Inappropriate drug use alongdemographically definedhigh-risk groups.

Inappropriate use of interact-ing drugs in individuals.

Inappropriate use of drugsin certain risk groups de-fined by disease

Current rates of ADRs areknown only for commondrugs— this data base al-lows definition of relativerisk for most ADRs (exceptrarest)

Supplement to spontaneousreporting system.

Source of information onperception and reportingof ADRs.

Identify possible drug-induced disease for furtherinvestigation.

Identify possible drug-induced disease for furtherinvestigation.

.

Oral contraceptive use.

Use of isoxsuprine in pregnancy.High codeine use,

Association of high-dose estrogenoral contraceptives with olderpremenopausal women.

Use of steroids in individuals onoral hypoglycemic agents.

Use Of indomethacin in individ-uals with history of GI bleeding.

Relative rate of hepatic disordersin erythromycin (estolate v, othersalt users).

Relative rate of heart failure indisopyramide v. quinidine or pro-cainamide users (alone and/or incombination).

Estimates of rate of reporting (orphysician recognition) of ADRssuch as hepatitis, etc.

Association of Iaryngospasm withNSAID.

Association of phenylpropanola-mine with CVA.


Table 10.—Data Sources Available to FDA forEstimating Actual Populations of Drug Users

(current and future)

1. IMS America, Ltd.● National Prescription Audit—survey of prescriptions

dispensed at pharmacies● National Disease and Therapeutic Index—survey of

physician-patient contacts listing diagnoses andtherapies (not prescriptions)

● U.S. Hospital —survey of bulk sales to hospitals● U.S. Drug Store— survey of sales to drug storesŽ Audatrex —sample of physicians and their prescrib-

ing habits over time2. Prescriptlon Card Service (PCS)—data on third-party

paid prescriptions, usually union, with nationalestimates

3. Medicaid (exploratory) -data will be available on alldrug use, linked to clinical events, for 1.2 million pa-tients in Michigan and 0.25 million patients inM inneso ta

Other than bulk sales to hospitals, there are no cur-rent national samples of hospital drug use.

SOURCE: US Food and Drug Administration

Confirmation of suspected prob-lems (e.g., unlabeled uses).

Label changes.Publication in medical literature,ADR Highlights.

Labeling changes to indicatedemographically defined high-risk groups, ADR Highlights.

Labeling changes and publicationsto warn against untoward drugcombinations that appear prev-alent. ADR Highlights.

Publication to warn of apparentprevalent misuse of drugs. Label-ing changes as indicated. ADRHighlights.

Labeling changes,Publications (ADR Highlights and/or medical literature).

General ability to place in perspec-tive ADRs due to disease v. drug.

Also serves as information basefor insight into perceived drug-induced disease.

Stimulate further research usingother data bases.

Labeling change/warning.

Stimulate further research usingother data bases.

Other governmental agencies also collect datathat may be of use to FDA:

●

●

●

The National Center for Health Statistics col-lects a wide variety of health data. FDA hasasked the center to collect drug data on deathcertificates, and also to investigate the pos-sibility of collecting drug information in itsNational Ambulatory Medical Care Survey.The National Centers for Disease Controlhave a Birth Defects Monitoring Project toalert for possible increases in the incidenceof birth defects. This project could help detectbirth defects due to drugs taken during preg-nancy.The National Cancer Institute’s Environmen-tal Epidemiology Branch sponsors case-con-trol studies to investigate associations be-tween environmental agents (including drugs)

44 • Postmarketing Surveillance of Prescription Drugs. —

●

and the development of specific types of can-cer, using data from prepaid health plans.The Drug Enforcement Agency and the Na-tional Institute on Drug Abuse-survey emer-gency rooms and coroners through theirDrug Abuse Warning Network to monitordrugs that are abused and their associationwith adverse reactions.

International monitoring activities includeWHO’s Program for International Monitoring ofAdverse Reactions, which has 23 participatingcountries. FDA’s Division of Drug Experience isthe designated National Monitoring Center for theUnited States. It exchanges reports with WHO,each organization summarizing the adverse drugreactions (ADRs) added to their systems in theprevious year. The purpose of WHO’s monitor-ing program is to increase the probability of de-tecting effects that might be overlooked by in-dividual countries. Its chief value to the UnitedStates is in providing information about drugs thatare available elsewhere in the world, but are notyet marketed here.

Drugs may be introduced and used abroad fora number of years before they are approved inthe United States. In fact, the more rapid approvaland introduction of drugs in other countries hasbeen one of the reasons behind the charge thatthe U.S. drug approval process is too slow, keep-ing valuable drugs from the market. However,this earlier introduction of drugs in other coun-tries also uncovers adverse effects and helpsmodify the indications for use before the drug isreleased in this country, because the internationalcommunity can pool and share information onthe effectiveness, safety, and use of drugs.

Currently, FDA’s Division of Drug Experiencereceives about 12,000 adverse event reports an-nually from industry (reports that are man-datory), the medical community, and others.About three-quarters (60 to 80 percent) of thesereports come from manufacturers, who, in turn,receive most of their reports from physicians.Table 11 summarizes the sources of adverse eventreports for 1972-78, and table 12 identifies in-dividual sources for fiscal year 1978. The relation-ship between these reported effects and drug usevaries greatly. For example, in 1977, approximate-ly 150 reports were on deaths that were “definite-ly” or “probably” drug related, but more than 400other reports were not clearly drug related or dueto overdoses (63).

The reports of the Division of Drug Experiencehave been computerized and, depending on theiraccuracy and completeness, can be screened orretrieved in the following classifications (63):

by drugs or drug class related to the suspectedadverse reactions;by adverse reaction related back to suspecteddrugs;by one of four cause-and-effect relationships:definite, probable, possible, and remote;by alert “yes” or “no,” meaning that the reac-tion is or is not novel, unanticipated, previ-ously unreported, or not mentioned in theproduct labeling;by demographic data;by source of report;by possible interactions; andby outcome of reaction.Retrieval can also extend to original reportson microfilm.

Table 11.— Number and Percentage of FDA’s Adverse Event Reports by Source, 1972-78

Source 1972 1973 1974 1975 1976 1977 1978

Manufacturers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9,750 6,700 6,807 8,097 7,664 8,945 9,143( 7 6 . 0 % ( 6 2 . 2 0 / o ) ( 7 0 . 1 0 / 0 ) (7!j.2°/0) ( 6 3 . 8 0 / , ) (71.80/.) (81.2”/.)

Physicians/other health professionals . . . . . . . . . . . . . . . . . 2,189 985 699 2,138 1,862 1,335(1.8°/0) (20.30/,) (10.1 ”Io) (6.6%) (17.8%) (15.0%) (11.90\o)

Hospitals. . . . . . . . . . . . . . . . . . . . . ... , . . . . . . . . . . . . . . . . 2,559 1,690 1,306 859 423(20.0%) (15.8%) (13.5°/0) (8.7%) (8.7%)~ ,6 5 ( 6 . 9 % ) ( 4 . 0 ° / 0 )

All other , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297 182 611 917 358( 2 . 2 % ) ( 1 . 7 % ) ( 6 . 3 % ) ( 8 . 5 % ) ( 9 . 7 % ) ( 6 . 3 % ) (3.7°/0)— — — —

Totals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12,837 10,761 9,709 10,756 12,012 12,459 11,259SOURCE A Ruskin and C. Anello, “The United States of America, ”In Monitoring for Drug Safety, W. H, W, Inman (cd.) (Philadelphia’ J B. L}ppincolt Co , 1980)

Ch. 5—Current Activities . 45

Table 12.—Number of FDA’s Adverse Event Reportsby Specific Sources, Fiscal Year 1978

--Number of

Source reports

Manufacturers . . . . . . . . . . . . . . . . . . . . . . . . . . .Physicians/other health professionals:

Physicians . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Pharmacists . . . . . . . . . . . . . . . . . . . . . . . . . .Dentists . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Hospitals:Federal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Non-Federal . . . . . . . . . . . . . . . . . . . . . . . . . . . .

All other:Consumer . . . . . . . . . . . . . . . . . . . . . . . . . . . . .District of Columbia, . . . . . . . . . . . . . . . . . . . .Methadone Centers . . . . . . . . . . . . . . . . . . . . .

SOURCE US Food and Drug Administration

9,143

1,199105

31

182241

2468

266

These reports are supplemented by a separatebut parallel literature search by FDA of l40 jour-nals (in English) that focuses on serious adverseeffects not mentioned in the drug labeling. Thisliterature search is published monthly in FDA’s“Current Drug Experience Literature.”

FDA has begun to use the computerized file ofADRs in a screening method (screening of adversereactions, or SOAR) to assist its clinical reviewersin identifying previously unsuspected adversereactions that might warrant in-depth clinical in-vestigations. The SOAR method compares adrug’s proportional share of specific adverse reac-tions (relative to its therapeutic class) with itsrespective proportional share of drug use for aparticular time period. The latter data are derivedfrom national estimates of the total amount of aparticular drug dispensed during a 3-monthperiod, divided by the amount that would be usedper day. The resulting number would representthe potential number of days a patient was ex-posed to the drug. An in-depth review may beconsidered when the proportional share of re-ported adverse events relative to the proportionalshare of drug use is greater than that predicted.

FDA’s postmarketing surveillance of ADRs isnot a surveillance “system” per se, but rather agroup of potentially related activities. Many ofthese activities are not exclusively carried out forFDA’s monitoring program. For example, the in-

tensive monitoring programs to which FDA con-tributes support also collaborate in large-scaleclinical trials of drug efficacy that are sponsoredby the National Institutes of Health. Many of thevoluntary reporting systems, such as the special-ized registries (e. g., those for ocular side effects,dermatological reactions, and hepatic toxicity),also have multiple purposes.

FDA’s monitoring is based primarily on volun-tary reporting. The agency is just beginning toestablish the data bases needed to confirm or re-ject those possible associations between drugs andadverse reactions that are first identified throughthe various kinds of voluntary reporting.

These observations point to the following areasof inquiry:

First, how can FDA’s coordination of its post-marketing surveillance activities be improved?Currently, it is difficult to evaluate the monitor-ing activities in which FDA is involved, becausetheir emphasis so far has been on collection ofdata, not on its final use; thus, the potential usesof the data are largely unrealized.

Second, are FDA’s own voluntary reportingand data-gathering programs and the other, morespecialized drug reaction registries it supports ap-propriately targeted and sufficiently utilized? Arecent General Accounting Office reevaluation ofFDA’s voluntary reporting system criticizes FDA’stardiness in evaluating and using these reports(29). The relative value of the various specializedregistries to FDA’s monitoring responsibilities isnot known.

Finally, should FDA be responsible for testingthe hypothesis that a drug maybe associated withan adverse reaction, or should that be the respon-sibility of the drug’s manufacturer? FDA couldsupport the formation of various prospective co-horts and coordinate that activity with the infor-mation FDA receives from voluntary reporting.But FDA’s role could also be limited to helpingidentify the most important drugs to monitor,with the actual monitoring performed by thedrug’s manufacturers.

6.Issues and Options

— .

Current interestand monitoring isapproval process,

Issues and Options

n prescription drug evaluationfocused on the premarketingwhile postmarketing surveil-

lance has waned as a policy issue. The recommen-dations of the Joint Commission on PrescriptionDrug Use, issued in January 1980, have not beenimplemented, and the linchpin of its recommenda-tions, a national Center for Drug Surveillance,reached its zenith before the commission issuedits report when the center concept (renamed asthe “National Center for Drug Science”) was in-cluded in the 1979 bill that passed the Senate butwas not acted on by the House of Representatives.In contrast, in the first session of the 97th Con-gress, the Senate passed the Patent Term Resto-ration Act of 1981, in large part as a directresponse to the length of the drug approval proc-ess. As mentioned earlier, the House of Represent-atives also voted on the bill, but under suspen-sion of its rules. Although the bill received a ma-jority of the votes, a two-thirds vote was requiredfor passage under such conditions, and the billfell just short of the two-thirds majority needed.Shortly prior to publication of this report, thereport of the Commission on the Federal Drug Ap-proval Process and the current Food and DrugAdministration {FDA) review will both also becomplete. Both focus on methods to hasten drugapproval by FDA.

Thus, policy formulation and implementationfor the premarketing approval process is beingpursued without parallel efforts for the postmar-keting period. As one person has remarked (43):

I don’t really see that any significant shorten-ing of approval time can be engaged in as a resultof a tradeoff in regard to postmarketing surveil-lance, although this was originally thought to bea possibility when a former FDA commissionersuggested that such was the case.

Others do see a linkage between the approvalprocess and postmarketing surveillance, althoughthey agree that there are no direct tradeoffs in thekinds of information obtained (4):

Postmarketing evaluation studies can be con-ducted much more cheaply than clinicaltrials . . . As a motivator for industry, it isdesirable to have the drug marketed sooner witha return on investment while studies are being

conducted. At the same time, much larger obser-vational studies can be done, at the same cost,[to] evaluate drugs in their customary use situa-tion. It is important to recognize that phase IIIclinical trials and postmarketing drug evaluationstudies are not alternatives. They address dif-ferent issues and are complementary. The ap-propriate question is whether it would not be bet-ter to reduce the size and cost of phase III withlimited likelihood of losing meaningful informa-tion and conduct much larger studies aftermarketing.

Hence, some relationship does exist betweenproposed changes in the premarketing approvalprocess and the monitoring activities of the post-marketing period. This relationship can be clari-fied by the answers to two questions. Can the sizeand cost of phase III clinical trials be reduced withlimited likelihood of losing meaningful informa-tion? And, should pharmaceutical manufacturersbe required to maintain the level of their drugevaluation responsibilities by increasing post-marketing surveillance?

Issue 1:Revising premarketing tests and short-ening the drug approval process.

The efficacy and safety tests in animals andhumans specified in FDA regulations for premar-keting approval are based on broad statutory lan-guage (21 U. S. C., sec. 355 (d)). “Adequate testsby all methods reasonably applicable” are neces-sary to show that a drug is safe for use. Theremust also be “substantial evidence that the drugwill have the effect it purports or is representedto have, ” where “substantial evidence” is definedas “evidence consisting of adequate and well-controlled investigations, including clinical in-vestigations, by experts qualified by scientific

49

50 . Postmarketing Surveillance of Prescription Drugs

training and experience to evaluate the effec-tiveness of the drug involved, on the basis ofwhich it could be fairly and responsibly concludedby such experts that the drug will have the effectit purports or is represented to have. ”

Efforts to shorten the drug approval processhave focused not on the statutory language buton the regulations issued by FDA to implementthe law. Thus, the focus here is on oversightissues, not on legislative changes.

Proposals to curtail or eliminate phase III pre-marketing tests, or shift them to the postmarketingperiod can be evaluated both theoretically andexperientially.

Theoretically, phase III testing is significantlymore sensitive than phase 11 testing. Adverse ef-fects with an incidence of 1/100 or more are morelikely than not to be detected in the 100 to 200patients in phase II. But the theoretical sensitiv-ity of detection rises in phase 111 to l/500 with500 patients, and 1/1,000 with 1,000 to 3,000 pa-tients (see ch. 4, table s).

These observations are relevant to the detec-tion of adverse reactions, but they are not so rele-vant to the detection of therapeutic effects. A drugthat helps only 1 in 100 patients would not be veryeffective, so effectiveness should be establishedin phase II. Phase 111 is intended to gather addi-tional evidence on a drug’s effectiveness for spe-cific indications.

If phase 111 testing were curtailed or eliminated,there is also the question of whether premarketingevaluations would test sufficient numbers of pa-tients to reasonably ensure a drug’s safety or givesubstantial evidence of its efficacy. Even undercurrent regulations, the use of a drug on humansis very limited before the drug is released formarket: 20 to 80 patients in phase I; 100 to 200patients in phase II; and 500 to 3,OOO patients inphase III—a range of only 620 to 3,280 patientsper drug (excluding controls). Curtailing the largerphase 111 tests would lower the range of patientstested to 620 to 780, and eliminating phase III testsaltogether would reduce that range further to 120to 280 patients who would be tested with a drugbefore it is released for general use.

In addition to theoretical criteria, experientialcriteria could be applied in considering proposalsto curtail or eliminate phase III tests. The dimin-ished power to observe adverse drug effects thatsuch changes theoretically entail may not in factbe found, or if it is, it may concern only infre-quent minor effects. As was mentioned previous-ly, FDA is reviewing past phase III tests to see ifthe trials for chronic effects or those with largesample sizes have contributed useful informationbeyond that obtained in phase II and early phaseIII. This review should indicate whether or notactual experience reflects the theoretical dif-ferences discussed above between phase 11 andphase III tests involving 500 versus l,500 to 3,000patients. Agreement of the experiential data withthe differences theoretically predicted wouldstrengthen the hypothesis that curtailing phase 111tests would lower the capacity of current premar-keting tests to identify adverse reactions. If theexperiential data fail to reflect the theoretical dif-ferences, then a better case can be made for cur-tailing phase III, with or without transfer of someof its testing to the postmarketing period.

Current interpretations of the statutory require-ments for “adequate tests” of safety and “substan-tial evidence” of effectiveness emphasize method-ology, as reflected in the requirement that eachindication for which a drug is intended be sup-ported by at least two well-controlled clinical tri-als. This is the reason for the preceding discus-sion of statistical guidelines and the complemen-tarity of normative guidelines in evaluating howFDA revises its drug approval regulations andprocedures. But FDA can alter the criteria bywhich it approves drugs. For example, pro-panolol, the first beta-blocking drug approved foruse in the United States, was approved by an ad-visory committee on the basis of all the evidencepresented to FDA, even though no one study wasfound to be adequate and well controlled (21).And in late 1981, timolol, another beta-blocker,was approved for use in preventing death and re-current heart attacks in patients who have sur-vived initial heart attacks. This approval wasbased on a foreign study—a 3-year Norwegianstudy showing that the risk of death or a secondheart attack following a first heart attack was

. .

Ch. 6—issues and Options “ 51

reduced by about one-third when timolol therapybegan within 28 days and continued for up to 33months (26). Although the study was Norwegian,the results were accepted for publication by themost prestigious medical journal in the UnitedStates, the New England Journal of Medicine (57).(Approval for another indication, high bloodpressure, was based on U.S. studies. )

Approval of timolol may also have been influ-enced by a similar study of propanolol in theUnited States by the National Heart, Lung, andBlood Institute (NHLBI). In the clinical trial ofpropanolol, NHLBI’s Policy and Data MonitoringBoard took the unusual step of curtailing the trialwhen data indicated that patients receiving pro-panolol experienced a 26 percent lower mortal-ity from all causes than did a control group. Ac-cording to the Beta-Blocker Heart Attack StudyGroup (2):

The results . . . strengthen and extend theconclusions of previous studies of beta-blockersin survivors of acute myocardial infarction. Thislarge study of a noncardioselective agent is in ac-cord with the results of the recent trial of timololmaleate.

The criteria for FDA’s acceptance of timolol,therefore, closely approximated the requirementof two well-controlled clinical trials, notwith-standing the fact that one of these trials was per-formed in another country.

The approval of propanolol and timolol, how-ever, do illustrate that FDA can grant exceptionsto its usual requirement of two well-controlledU.S.-based clinical trials. In the case of timolol,the validity of the Norwegian study was con-firmed by its acceptance for publication in a pres-tigious U.S. medical journal, and even though ap-proval was based on this one study, the resultsof the NHLBI trial on propanolol, another beta-blocker, must surely have influenced the FDAdecision to approve timolol for prevention ofheart attacks. In the case of propanolol, the firstbeta-blocker drug to be approved by FDA, ap-proval was based on the preponderance of theevidence as judged by an advisory committee.

In such a case, expert judgment relies on quali-tative, not quantitative, criteria in approving adrug, and such an approach falls outside the theo-

retical and experiential guidelines outlined above.If FDA is to rely increasingly on such qualitativecriteria through the increased use of advisorycommittees, it will be necessary for FDA to devel-op general guidelines to aid the advisory commit-tees in their deliberations. Otherwise, in a case-by-case approach, evidence of the same quality

may lead to approval for one drug and nonap-proval for another.

Issue 2:

Improving postmarketing surveillanceand its role in the drug approval process.

Controlled clinical trials, the most accepted sci-entific means of identifying and confirming adrug’s effectiveness and safety, are used in drugtesting in the premarketing stage, but the evidencethey yield is necessarily limited because their sam-ple sizes are small and the patients they test rep-resent only a fraction of the kinds of patients whowill eventually use the drug. Other shortcomingsof small controlled clinical trials are that rare butserious adverse effects or effects with a long laten-cy will not be observed, and that average condi-tions of use are not duplicated. These limitationsof premarketing testing can only be addressed inthe wider use that comes with marketing the drug.Thus, even if phase III testing were not curtailedor eliminated, FDA’s powers in the postmarketingperiod could be strengthened to enhance its sur-veillance role.

Generally, postmarketing surveillance “sys-tems” that have been advocated are not systemsin the formal sense, but a series of related activitiesoriented toward several purposes, with the regu-latory approval process being only one use. Theactivities most frequently mentioned include thefollowing three.

First is the building of a resource base throughtraining of additional experts and improving epi-demiological tools such as methods for case-con-trol and cohort studies. The concept of a nationalCenter for Drug Surveillance, advocated in thereport of the Joint Commission on PrescriptionDrug Use, was one such attempt. Others believe,in contrast, that the resources are already in hand.According to Jick (40):

52 ● Postmarketing Surveillance of Prescription Drugs— — .

Contrary to the views stated in the report [ofthe Joint Commission on Prescription Drug Use],I believe that methods to efficiently performpostmarketing surveillance are known and welltested, and that a vast amount of data hasalready been collected to evaluate drug toxicity.

Remington states (61):

IM]ethods for estimating characteristics oflarge populations, although available since the1940’s, have not been assimilated to any appre-ciable extent into the field of drug evaluation . . .I think we must begin to apply modern masspopulation methods to modern problems of drugevaluation, both at the pre- and post-marketinglevels. Such methods, however, are particularlyappropriate to the evaluation of marketed drugs.

Second is strengthening voluntary reporting toidentify possible adverse drug reactions. Oncesuch reactions are suspected, clinical trials, case-control, and cohort studies could be used to deter-mine whether an association with drug use in factexists. In this regard, FDA has come under criti-cism insofar as its adverse drug reaction report-ing activities are concerned. For example, the ac-tivities listed in chapter 4 have been criticized asbeing little more than a catalog, with no assess-ment of the relative values of the various activities(80). Furthermore, in a recently released followupto a study conducted in 1974, the U.S. GeneralAccounting Office concluded (29):

Many adverse reaction reports do not get tothe Division maintaining the system and manyothers require a long time to get into the system.Some of the missing or late reports involved seri-ous reactions which were not discussed in thedrug labeling. Reporting by non-manufacturersources, such as hospitals or physicians, couldalso be increased.

Third is developing an efficient method formonitoring selected drugs after their release intothe market. The most frequently mentioned mech-anism is formation of prospective cohorts of drugusers, utilizing existing data bases such as thosepreviously identified, i.e., Medicaid, Medicare,the military health systems, and some healthmaintenance organizations. In the opinion of oneexpert, appropriate large-scale systems are avail-able, but only drug companies currently use them(41).

These components of a postmarketing surveil-lance “system,” and FDA’s role in supporting andusing them, are oversight issues.

There are also several legislative options thatcould strengthen FDA’s powers in the postmar-keting period. The following legislative optionsare presented for congressional consideration.

Option 1: Give FDA the power to requirepostmarketing studies.

Currently, FDA has no express power to requirea drug’s sponsor to conduct postmarketing studiesof the types summarized earlier under “phase IV”testing. Drug companies have agreed to such re-quests in the past, however, because refusal mightmean nonapproval of the drug.

FDA is examining these studies to see if theyprovided the kinds of information identified in itsobjectives for “phase IV” testing. Additional ques-tions are to determine when adverse reactions be-came specified on the package insert, and whetherthe source of the identification of adverse reac-tions was spontaneous reports, the postmarketingstudy, or both.

However, a broader analysis is needed, one thatdoes not focus only on these specific studies. Itwould be helpful if FDA also assessed a sampleof drugs that have been marketed for several yearsto see if significant additional adverse reactionswere later discovered that were not uncoveredduring premarketing trials. Or a study could beconducted on the significant adverse events thatwere discovered during the postmarketing period.In either of these latter types of studies, the assess-ment focus should be on whether the adverse ef-fects could have been discovered through studiesof the kinds performed by the manufacturers atFDA’s request. This assessment could provide ad-ditional information for deciding whether formalpostmarketing monitoring would be valuable.

The larger postmarketing studies carried out bymanufacturers at the request of FDA have cost$1 million to $3 million each (11), and the industryis certain to resist giving FDA explicit authorityto require them. Industry might be more willingto conduct postmarketing studies if the drug ap-proval process were shortened, however, since it

——. — — —

could obtain a quicker return on its investments.In the absence of such a tradeoff, Congress couldconsider increasing FDA’s appropriations to fi-nance selected postmarketing studies on a drug-by-drug basis.

A variation of this option is for FDA to use itsexisting regulatory powers over advertising andpromotional practices to “certify” an industry-sponsored postmarketing study. Although initial-ly reluctant to conduct postmarketing studies, theindustry now sees them as part of its marketingstrategy. Physicians may cease to cooperate, how-ever, if they feel they have been used to promotea company’s product. FDA certification could beused to distinguish between postmarketing moni-toring and “product support studies” that are usedto interest physicians in a manufacturer’s productsand to defend a drug’s safety and therapeuticvalue (69).

Option 2: Give FDA the power to restrict thedistribution, dispensing, and administration of adrug.

FDA has considerable power in withholdir,g ap-proval for a drug, but once a drug is approved,there is a significant shift in the burden of proofand in the amount of evidence required to rescindapproval. If FDA had the authority to restrictdrug marketing or less burden of proof in rescind-ing approval, it might give approval more freelyin the first place. Restrictions on drug use and alesser burden of proof for FDA to rescind approv-al would be particularly pertinent in the first sev-eral years of marketing, when adverse effects arestill quite likely to be identified. The therapeutic-toxic ratio of some new drugs is so narrow thatit seems reasonable to provide the means wherebyFDA can restrict the use of certain prescriptiondrugs to groups of specially trained or experiencedphysicians.

S. 1075, the bill introduced in 1979 and passedby the Senate, would have given restrictive pow-ers to FDA, provided that: 1) the drug presentedsignificant risks to patients or the public health;2) a drug could not be determined to be safe unlessrestrictions on distribution and dispensing wereimposed; 3) the restrictions could reasonably beexpected to reduce the risks while permitting itsuse in appropriate patients; and 4) no other ad-

Ch. 6—Issues and Options ● 5 3— —

ministrative or educational actions permissibleunder the law could reasonably be expected toreduce the risks. The bill also would have alloweda drug to be restricted to practitioners with specialtraining or experience in its use or to practitionersfor use in certain facilities, but only if this werenecessary to determine drug safety; and no practi-tioners could be excluded solely on the basis thatthey were not eligible for certification in a medicalspecialty. The 1979 Senate bill also would haverequired FDA to review any imposed restrictionsevery 2 years.

The Medical Device Amendments of 1976 (Pub-lic Law 94-295) contains somewhat similar lan-guage in its section on “restricted devices. ” Ac-cording to this amendment, a device could be re-stricted in its sale, distribution, or use if, “becauseof its potentiality for harmful effect or the collater-al measures necessary to its use, the Secretarydetermines that there cannot otherwise be reason-able assurance of its safety and effectiveness. ” Thelaw also contains a prohibition against excludingpractitioners solely because of their ineligibilityfor certification in a medical specialty (21 U. S. C.,sec. 360j(e)).

In a limited or phased distribution, drugs couldbe introduced into different geographic regionsrather than into the whole country at once. Theregions could be chosen with attention to the factthat some drugs are used more in one part of thecountry than another. Introduction of drugs ona regional basis could also provide some insuranceagainst unexpected adverse effects in that less ofour population would be at risk, but different re-gions could be used for the first marketing of dif-ferent drugs so that no one region would be the“guinea pig. ” By specifying the regions well, com-parable regions could be compared for reportedside effects (.5.5).

A variation of this option is to develop a paral-lel approval process for the use of a limited groupof drugs during phase III testing. This specialphase III testing would be considered only as arelatively exceptional procedure restricted to drugsof unusual need and promise. For drugs of appar-ent unusual therapeutic value compared with al-ternative therapies and with acceptable risk, alimited number of physicians could be permitted

54 ● Postmarketing Surveillance of prescription Drugs

to use a drug without the fully detailed record-keeping requirements of phase III (28).

Option 3: Change the standard for a drug’s re-moval from the market from “imminent hazardto the public health” to “unreasonable risk of ill-ness or injury to any segment of the population”or some other less stringent standard.

Such a change was contained in the 1979 Senatebill. However, under the present law, the “immi-nent hazard” standard is to be used only in caseswhere a drug’s harmful effect would be so imme-diate and severe as to justify suspension of dueprocess until after the drug has been removedfrom the market. The “imminent hazard” stand-ard applies only to cases where FDA suspends ap-proval of a drug first, then gives the drug spon-sor prompt notice and an opportunity for an expe-dited hearing. FDA otherwise can remove a drugfrom the market on the basis of new evidence onsafety or effectiveness or for other reasons suchas discovering that approval of the drug was basedon an untrue statement of a material fact. To takesuch action, however, FDA must give due noticeand an opportunity for a hearing before it can pro-ceed.

Substituting “unreasonable risk” for “imminenthazard” in the standard would blur the presentdistinction between those cases when due noticeand opportunity for a hearing should be requiredbefore a drug could be taken off the market andthose cases when it would be justified to removea drug from the market prior to notice and an op-portunity for a hearing. In other words, “unrea-sonable risk” is already the standard for thosecases when, in order to protect the drug sponsor’seconomic interests, FDA must give due notice andan opportunity for a hearing before taking action.

If such a change in the standard were approved,FDA would be able to remove a drug for any ofthe currently accepted reasons to question safetywithout first giving notice and an opportunity fora hearing.

In the Medical Device Amendments of 1976,no such “imminent hazard” standard is specified.There is, however, a slight difference in the word-ing of the law. It states that, if a drug does notrepresent an “imminent hazard, ” withdrawal ofthe drug’s approval can proceed “after due noticeand opportunity for a hearing” (21 U. S. C., sec.355(e)). For devices, withdrawal of approval canproceed “after due notice and opportunity for in-formal hearing” (emphasis added) (21 U. S. C., sec.360(e)).

In sum, efforts to shorten the drug approvalprocess in the premarketing period could takeplace through reinterpreting the guidelines for as-sessing safety and efficacy. This report has pro-vided theoretical and experiential criteria for eval-uating how such changes could affect the detec-tion capabilities of the current guidelines. It hasalso discussed the desirability of guidelines for thekinds of qualitative changes FDA is implementingregarding the evidence required for drug approval.These changes include accepting foreign data andcumulative evidence (as opposed to the require-ment of at least two well-controlled clinical trials).Finally, the report has identified options relatingto FDA’s postmarketing surveillance. These op-tions could be pursued independently of any revi-sions in the premarketing drug approval process,but they could also be implemented to requiredrug sponsors to maintain their level of drug eval-uation responsibilities if there is a change in cur-rent premarketing approval requirements.

Appendixes

Appendix A.— Selected Excerpts From theStatutes Governing Drugs and Medical Devices——

Drugs

Grounds for refusing application; approval of applica-tion; “substantial evidence” defined (21 U. S. C., sec.355(d)).

If the Secretary finds after due notice to the appli-cant in accordance with subsection (c) of this sectionand giving him an opportunity for a hearing, in ac-cordance with said subsection, that (1) the investiga-tions, reports of which are required to be submittedto the Secretary pursuant to subsection (b) of this sec-tion, do not include adequate tests by all methods rea-sonably applicable to show whether or not such drugis safe for use under the conditions prescribed, recom-mended, or suggested in the proposed labeling thereof;(2) the results of such tests show that such drug is un-safe for use under such conditions or do not show thatsuch drug is safe for use under such conditions; (3) themethods used in, and the facilities and controls usedfor, the manufacture, processing, and packing of suchdrug are inadequate to preserve its identity, strength,quality, and purity; (4) upon the basis of the infor-mation submitted to him as part of the application,or upon the basis of any other information before himwith respect to such drug, he has insufficient informa-tion to determine whether such drug is safe for useunder such conditions; (5) evaluated on the basis ofthe information before him with respect to such drug,there is a lack of substantial evidence that the drug willhave the effect it purports or is represented to haveunder the conditions of use prescribed, recommended,or suggested in the proposed labeling thereof; or(6) based on a fair evaluation of all material facts, suchlabeling is false or misleading in any particular; he shallissue an order refusing to approve the application. If,after such notice and opportunity for hearing, the Sec-retary finds that clauses (1) through (6) do not apply,he shall issue an order approving the application. Asused in this subsection and subsection (e) of this sec-tion, the term “substantial evidence” means evidenceconsisting of adequate and well-controlled investiga-tions, including clinical investigations, by expertsqualified by scientific training and experience toevaluate the effectiveness of the drug involved, on thebasis of which it could fairly and responsibly be con-cluded by such experts that the drug will have the ef-fect it purports or is represented to have under the con-ditions of use prescribed, recommended, or suggestedin the labeling or proposed labeling thereof.

Withdrawal of approval; grounds; immediate suspen-sion upon finding imminent hazard to public health(21 U. S. C., sec. 355(e)).

The Secretary shall, after due notice and opportuni-ty for hearing to the applicant, withdraw approval ofan application with respect to any drug under this sec-tion if the Secretary finds (1) that clinical or other ex-perience, tests, or other scientific data show that suchdrug is unsafe for use under the conditions of use uponthe basis of which the application was approved;(2) that new evidence of clinical experience, not con-tained in such application or not available to theSecretary until after such application was approved,or tests by new methods, or tests by methods notdeemed reasonably applicable when such applicationwas approved, evaluated together with the evidenceavailable to the Secretary when the application wasapproved, shows that such drug is not shown to besafe for use under the conditions of use upon the basisof which the application was approved; or (3) on thebasis of new information before him with respect tosuch drug, evaluated together with the evidence avail-able to him when the application was approved, thatthere is a lack of substantial evidence that the drug willhave the effect it purports or is represented to haveunder the conditions of use prescribed, recommend-ed, or suggested in the labeling thereof; or (4) that theapplication contains any untrue statement of a materialfact: Provided, That if the Secretary (or in his absencethe officer acting as Secretary) finds that there is animminent hazard to the public health, he may suspendthe approval of such application immediately, and givethe applicant prompt notice of his action and affordthe applicant the opportunity for an expedited hear-ing under this subsection; but the authority conferredby this proviso to suspend the approval of an applica-tion shall not be delegated. The Secretary may also,after due notice and opportunity for hearing to the ap-plicant, withdraw the approval of an application withrespect to any drug under this section if the Secretaryfinds (1) that the applicant has failed to establish asystem for maintaining required records, or has re-peatedly or deliberately failed to maintain such recordsor to make required reports, in accordance- with aregulation or order under subsection ( j ) of this sectionor to comply with the notice requirements of section360(j)(2) of this title, or the applicant has refused topermit access to, or copying or verification of, suchrecords as required by paragraph (2) of such subsec-

57

58 ● Postmarketing Surveillance of Prescription Drugs— — — — — ——.— ——

tion; or (2) that on the basis of new information beforehim, evaluated together with the evidence before himwhen the application was approved, the methods usedin, or the facilities and controls used for, the manufac-ture, processing, and packing of such drug are inade-quate to assure and preserve its identity, strength,quality, and purity and were not made adequate with-in a reasonable time after receipt of written notice fromthe Secretary specifying the matter complained of; or(3) that on the basis of new information before him,evaluated together with the evidence before him whenthe application was approved, the labeling of suchdrug, based on a fair evaluation of all material facts,is false or misleading in any particular and was notcorrected within a reasonable time after receipt of writ-ten notice from the Secretary specifying the mattercomplained of. Any order under this subsection shallstate the findings upon which it is based.

Devices

Restricted devices (21 U. S. C., sec. 360j(e)).The Secretary may by regulation require that a

device be restricted to sale, distribution, or use—(A) only upon the written or oral authorization of

a practitioner licensed by law to administer or use suchdevices, or

(B) upon such other conditions as the Secretary may

prescribe in such regulation, if, because of its poten-tiality for harmful effect or the collateral measures nec-essary to its use, the Secretary determines that therecannot otherwise be reasonable assurance of its safe-ty and effectiveness. No condition prescribed undersubparagraph (B) may restrict the use of a device topersons with specific training or experience in its useor to persons for use in certain facilities unless theSecretary determines that such a restriction is requiredfor the safe and effective use of the device. No suchcondition may exclude a person from using a devicesolely because the person does not have the trainingor experience to make him eligible for certification bya certifying board recognized by the American Boardof Medical Specialties or has not been certified by sucha Board. A device subject to a regulation under thissubsection is a restricted device.

Withdrawal of approval of application for premarketapproval (21 U. S. C., sec. 360e(e)).

The Secretary shall, upon obtaining, where ap-propriate, advice on scientific matters from a panel orpanels under section 360c of this title, and after due

notice and opportunity for informal hearing to theholder of an approved application for a device, issuean order withdrawing approval of the application ifthe Secretary finds—

(A) that such device is unsafe or ineffective under

the conditions of use prescribed, recommended, or sug-gested in the labeling thereof;

(B) on the basis of new information before him withrespect to such device, evaluated together with theevidence available to him when the application wasapproved, that there is a lack of a showing of reason-able assurance that the device is safe or effective underthe conditions of use prescribed, recommended, or sug-gested in the labeling thereof;

(C) that the application contained or was accom-panied by an untrue statement of a material fact;

(D) that the application (i) has failed to establish asystem for maintaining records, or has repeatedly ordeliberatel y failed to maintain records or to makereports, required by an applicable regulation undersection 360i(a) of this title, (ii) has refused to permitaccess to, or copying or verification of, such recordsas required by section 374 of this title, or (iii) has notcomplied with the requirements of section 360 of thistitle;

(E) on the basis of new information before him withrespect to such device, evaluated together with theevidence before him when the application was ap-proved, that the methods used in, or the facilities andcontrols used for, the manufacture, processing, pack-ing, or installation of such device do not conform withthe requirements of section 360j(f) of this title and werenot brought into conformity with such requirementswithin a reasonable time after receipt of written noticefrom the Secretary of nonconformity;

(F) on the basis of new information before him, eval-uated together with the evidence before him when theapplication was approved, that the labeling of suchdevice, based on a fair evaluation of all material facts,is false or misleading in any particular and was notcorrected within a reasonable time after receipt of writ-ten notice from the Secretary of such fact; or

(G) on the basis of new information before him,evaluated together with the evidence before him whenthe application was approved, that such device is notshown to conform in all respects to a performancestandard which is in effect under section 360d of thistitle compliance with which was a condition of ap-proval of the application and that there is a lack ofadequate information to justify the deviation fromsuch standard.

. —

Appendix B.— Conclusions and Recommendations of theJoint Commission on Prescription Drug Use, Jan. 23,1980

1. Postmarketing surveillance (PMS) can be used todevelop information about prescription drugs that isunavailable from premarketing studies but necessaryand, or useful for the clinical practice of medicine orongoing regulation of drugs. Existing postmarketingsurveillance programs should be coordinated with newprograms to form a comprehensive system.

2. The purpose of postmarketing surveillance is todetect important drug effects earlier than would other-wise be possible. The surveillance, per se, or its resultswill not and cannot be used to change the biologicalproperties or effects of drugs, but they can be used tominimize the harmful consequences and maximize theoptimal use of drugs.

3. Certain risks posed by a PMS system must be rec-ognized, accepted, and addressed, although they arejudged by the Commission to be far outweighed bypotential benefits.

4. Highest priority in a PMS system should be givento surveillance of new chemical entity prescriptiondrugs, delayed or slowly developing drug effects,commonly used drugs, populations in which drug ef-fects are not well documented, certain important med-ical events (e. g., births, deaths, etc. ) and their rela-tionship to drug use, and patterns of prescription druguse. Other, but secondary, priorities should includethe study of: certain important non-prescription drugs,the drug-taking practices of patients, the frequency ofgiven drug effects, dose-response relationships, thecharacteristics of patients who experience certain ef-fects as compared to those who do not, the relativerisks and benefits of individual drugs, changes in fre-quencies of drug effects over the course of time, andthe comparative efficacy and safety of different drugswithin the same class.

5. PMS must operate, to the maximum extent possi-ble, in a setting of actual medical practice. In the ma-jority of instances, PMS should be concerned primarilywith drugs marketed without special restrictions be-yond those already enforced for drugs of certainclasses. Data collection for PMS should not interferewith the normal delivery of health care.

6. A PMS system will require both an alerting (hy-pothesis-generating) mechanism and a confirming orrejecting (hypothesis-testing) mechanism. Hypothesesmust be recognized as suggestions about cause-effectrelationships and not as established fact.

7. An hypothesis-generating mechanism should

ideally have access to all animal and clinical data abouta drug.

8. A great deal of selectivity will be required forjudicious decisions about which of the available hy-potheses to test.

9. PMS must insure that great attention is paid toresearch design and to the sequence of studies. Par-ticular attention should be paid to the validity of themethodology and to the sample size required. Possi-ble types of study are prospective or retrospective, ex-perimental or non-experimental, controlled or uncon-trolled, cohort or case-control, studies of drugs or ofevents, and studies of prophylactic, therapeutic, ordiagnostic drugs.

10. Development of new methodology for study ofdrug effects must be a high priority for the PMSsystem.

11. The drug effects requiring study will be expectedtoxicity, unexpected toxicity, intended efficacy and un-intended efficacy.

12. Currently available methodology can be usedto study expected or unexpected toxicity and unin-tended efficacy, successfully detecting events that oc-cur with a frequency of at least one event per 1000 ex-posures. Less frequent events (e.g., one in 10,000 uses)can be detected less reliably.

13. The Commission strongly recommends againstPMS that ignores the determination of intended ef-ficacy and of long-term drug effects.

14. An intensive system of surveying the medicalliterature should be established and particular atten-tion given to a systematic review of data comparingdrug responses in various countries.

15. A PMS system should develop methods of seek-ing and receiving brief reports from large numbers ofhealth professionals.

16. Prospective, non-experimental cohorts that arelong-term and lifelong, if possible, should be estab-lished for PMS studies.

17. Liaison between the various components of aPMS system will be necessary. This would requiredevelopment of a standardized terminology for de-scribing PMS and for use in PMS.

18. Reports must be prepared, published and dis-tributed to aIl parties involved in or affected by PMS.Multiple types of reports will be necessary, tailoredto the differing needs and interests of the recipients.

19. For a prescription drug surveillance system to

59

60 Ž Postmarketing Surveillance of Prescription Drugs

be effective, it must have the confidence of the publicat large, including health care providers and patients.This confidence must extend not only to the validityof the information generated, but also to the mannerin which the information will be used.

20. In order to maintain trade secret protection formanufacturers without prejudicing the PMS system’spublic and academic accountability, only the FDAshould have the power to require the disclosure of suchsecrets. Because of this restriction, the FDA mustassume responsibility for reviewing this trade secretdata for any hints of drug effects in man that it maycontain.

21. In order to protect patient confidentiality, in-dividually identifiable information should be keptstrictly confidential by any PMS system unless patientsspecifically authorize release, with only the followingexceptions:

a.

b.

c.

Such information could be released to organiza-tions engaged in similar research if an expressfinding with supporting written statements is pre-pared documenting that such disclosure is nec-essary and identifying the individual receiving theinformation. Such organizations must have com-parable guarantees of confidentiality as the or-ganization releasing the data. Redisclosure by thereceiving organization would be prohibited with-out written approval of the original organization.Disclosure could be made to a properly identifiedrecipient pursuant to a bona fide medical emer-gency.Both patient and provider must have access totheir own identifiable data and the ability tomake corrections or amendments.

22. Given the fact that the law is developing on theissue of provider liability for disclosing patient iden-tifiable medical records, the Commission recommendsthat the organizations concerned with PMS review theissue of liability as they begin to undertake PMSfunctions.

23. In order to assure adequate security for datagathered, a PMS organization should:

a. Maintain only that information which is relevantand necessary to accomplish the purpose.

b. Maintain all records used in making any decisions

about an individual with such accuracy, rele-vance, timeliness, and completeness as is reason-ably necessary to assure fairness to the individualin the determination.

c. Establish administrative, technical, and physicalsafeguards to insure the security and confiden-tiality of records.

24. The value for PMS of a limited shield law is rec-ognized. Such a law would specify that identifiable in-formation of the patient or provider submitted volun-tarily to a PMS organization could not be admissibleas evidence in a medical or product liability action,and might remove a deterrent to voluntary reporting.Even without such a law however, the PMS functionsof the CDS [Center for Drug Surveillance] should com-mence and the results should be used to define whetherevidence needs to be gathered to support the need forand formation of shield law in order to have an op-timal PMS system.

25. A national Center for Drug Surveillance (CDS)is necessary in the United States.

26. The objectives and functions of the CDS shouldbe to educate scientists, prescribers and the public, toperform and encourage research into drug effects andto promote cooperation among all existing PMS pro-grams.

27. In using its resources, the CDS must be strictlyaccountable for fairness, scientific accuracy, andhonesty.

28. The activities of the Center for Drug Surveil-lance (CDS) should not unnecessarily duplicate thefunctions of other organizations engaged in PMS ac-tivities, whether public or private.

29. The CDS should be a private, non-profit, non-regulatory organization with a full-time staff and aphysical facility located in an environment that cansupport and be supported by academic endeavors.

30. After five years, there should bean external re-view of the effectiveness of the CDS. If the projectedbenefits from the CDS are not realized, the CDSshould be abandoned.

31. Support for the CDS should be in addition to,not instead of, added support for other worthwhilePMS activities,

—

References

References—— — —. —1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

Armed Forces Institute of Pathology, HepaticBranch, and the Centers for Disease Control,Bureau of Epidemiology, Family Planning Evalua-tion Division, “Increased Risk of HepatocellularAdenoma in Women With Long-Term Use of OralContraceptives, ” Morbidity and MortaIity WeeklyReport 26:293, 1977.Beta-Blocker Heart Attack Study Group, “TheBeta-Blocker Heart Attack Trial, ” ./. A.h4. A. 246:2073, 1981.Boman, G., “The Nordic Countries, ” in A40nitor-ing for Drug Safety, W. H. W. Inman (cd. ) (Phil-adelphia: J. B. Lippincott Co., 1980).Borden, E, K., The Upjohn Co., personal com-munication with the Office of Technology Assess-ment, February 1982.Borden, E. K,, The Upjohn Co., “PostmarketingSurveillance of Outpatient Drugs Using a Phar-macy-Based Registry System—Report of TwoStudies” presented at the Drug Information Asso-ciation Workshop, Williamsburg, Va., December1981.Boston Collaborative Drug Surveillance Program,“Oral Contraceptives and Venous Thromboem-bolic Disease, Surgically Confirmed GallbladderDisease and Breast Tumors, ” Lancet I :1399, 1973.Boston Collaborative Drug Surveillance Program,“Surgically Confirmed Gallbladder Disease,Venous Thromboembolism and Breast Tumors inRelation to Post Menopausal Estrogen Therapy, ”N. Eng. J. Med. 290:15, 1974.British Medical Journal, “Oral Contraceptives andBreast Neoplasia” (leading article), Br. Meal, J.1:545, 1976.Chalmers, T. C., “Who Will Fund Clinical Trials?”The Sciences 22:6, 1982.Clarren, S. N., Director, The Regulatory Center,Performance Development Institute, Washington,D. C., personal communication with the Office ofTechnology Assessment, 1982.Clarren, S. N., and Jones, J., “Using Post-Marketing Experience To Improve the Drug Re-view and Approval Process, ” presented at the An-nual Meeting of the American Association for theAdvancement of Science, session on “RegulatoryReform as It Affects Science and Technology, ”Washington, D. C., January 1982.Clarren, S. N., et al., The Evaluabdity Assessmentof the Developing Experiment in Post-MarketingSurveillance of Prescription Drugs, prepared forthe Experimental Technology Incentives Program,

13.

14.

15.

16,

17

18,

194

20.

21.

22.

23.

24.

25.

26.

National Bureau of Standards (contract No. NBS-GCR-ETIP-81-96) (Washington, D. C.: Perform-ance Development Institute, 1981).Cohen, J., Statistical F’ower Analysis for theBehavioral Sciences (New York: Academic Press,1967).Cohen, L., et al., “Pregnancy Oral Contraceptivesand Chronic Familial Jaundice With Predominant-ly Conjugated Hyperbilirubinemia (Dubin-John-son Syndrome), ” Gastroenterology 62:1182, 1972.Collaborative Group for the Study of Stroke inYoung Women, “Oral Contraceptives and Strokein Young Women, ” J. A.M.A. 231:718, 1975.Department of Health, Education, and Welfare,Review Panel on New Drug Regulation, FinalReport (Washington, D. C.: U.S. GovernmentPrinting Office, 1977).Dollery, C. T., and Rawlins, M. D., “MonitoringAdverse Reactions to Drugs, ” Br. Med. J. 1:96,1977.Dorsen, M., and Miller, J. M., “The Drug Regula-tion Process and the Challenge of Regulatory Re-form,” Ann. intern. Med. 91:908, 1979.Elder, J. B., et al., “Cimetidine and GastricCancer, ” L2ncet 1:1005, 1979.European Workshop (Report of), “Towards aMore Rational Regulation of the Development ofNew Medicines, ” Europ. J . Clin. Pharmacol.11:233, 1977.F-D-C Reports (“The Pink Sheet”), “’SubstantialEvidence’ Requirement Should Be Reduced byFDA Through Regs, McMahon Commission Urgesin Draft of Report, “ F-D-C ’Reports 44(10):3, 1982.Finkel, M. J., “FDA’s Criteria for Safety and Ef-fectiveness of Drugs, ” presented at the First Work-shop on Drug Control in the Region of the Ameri-cas, Pan American Health Organization, Wash-ington, D. C., May 1, 1979.Finney, D. J., “Statistical Logic in the Monitoringof Reactions to Therapeutic Drugs, ” in Monitor-ing for Drug Safety, W. H. W. Inman (cd. ) (Phil-adelphia: J, P. Lippincott Co., 1980).Food and Drug Administration, “General Con-siderations for the Clinical Evaluation of Drugs, ”HEW(FDA)77-3040 (Washington, D. C.: U.S.Government Printing Office, 1977).Food and Drug Administration, “Supplementary

Report to Contracts and Grants Committee onMedicaid, ” from the Division of Drug Experience,Bureau of Drugs (mimeo. ), 1982.Food Drug Cosmetics Law Reports,

—

“Timolol Ap-

63

64 “ Postmarketing Surveillance of Prescription Drugs

27.

28.

29,

30,

31.

32.

33.

34.

35

36

37

38

39<

40,

41,

proved as First Drug for Prevention of Death AfterHeart Attack, ” HHS News, dated Nov. 25, 1981,sec. 41,560.Friedman, G. D., et al., “Experience in Monitor-ing Drug Reactions in Outpatients, ” J. A, M. A.217:567, 1971.Frommer, P. L., Acting Director, National Heart,Lung, and Blood Institute, personal communica-tion with the Office of Technology Assessment,1982.General Accounting Office, U.S. Congress, FDACan Further Improve Its Adverse Drug ReactionReporting System (Washington, D. C.: GAO,1982).General Accounting Office, U.S. Congress, FDADrug Approval: A Lengthy Process That Delaysthe Availability of Important New Drugs (Wash-ington, D. C.: GAO, 1980).Gifford, L. M,, et al., “Cimetidine PostmarkedOutpatient Surveillance Program, ” J.A. M.A.243:1532, 1980.Gillie, O,, “The Journalist, ” in Monitoring forDrug Safety, W, H, W. Inman (cd, ) (Philadelphia:J. B. Lippincott Co., 1980).Gross, F. H., and Inman, W. H. W. (eds. ), DrugMonitoring, proceedings of an InternationalWorkshop held in Honolulu, January 1977 (Lon-don: Academic Press, 1977).Hawker, P, C., et al., “Gastric Cancer AfterCimetidine in Patient With Two Negative Pre-treatment Biopsies, ” Lancet 1:709, 1980.Inman, W. H. W. (cd.), Monitor@ for Drug Sak-ty (Philadelphia: J. B. Lippincott Co., 1980).Inman, W. H. W., “Recorded Release,” in DrugMonitoring, F. H. Gross and W. H. W. Inman(eds. ), Proceedings of an International Workshopheld in Honolulu, January 1977 (London: Aca-demic Press, 1977).Inman, W. H. W., “The United Kingdom, ” inMonitoring for Drug Safety, W. H. W. Inman(cd. ) (Philadelphia: J. B. Lippincott Co., 1980).Inman, W. H. W., and Vessey, M. P., “Investiga-tion of Deaths From Pulmonary, Coronary andCerebral Thrombosis and Embolism in Women ofChild-Bearing Age, ” Br. Med. J. 2:193, 1968.Inman, W . H . W . , e t a l . , “ThromboembolicDisease and the Steroidal Content of Oral Con-traceptives, ” Br. Med. J. 2:203, 1970.Jick, H., Boston Collaborative Drug SurveillanceProgram, Boston University Medical Center, per-sonal communication with the Office of Technol-ogy Assessment, 1981.Jick, H., Boston Collaborative Drug SurveillanceProgram, Boston University Medical Center, per-

42.

43

44.

45.

46.

47.

48.

49,

50.

51.

52.

53.

54.

55.

56.

sonal communication with the Office of Technol-ogy Assessment, 1982.Joint Commission on Prescription Drug Use,Report of the Joint Commission on PrescriptionDrug Use, Final Report (Rockville, Md.: JointCommission on Prescription Drug Use, Inc., 1980).Lasagna, L., Professor of Pharmacology and Tox-icology and Professor of Medicine, University ofRochester School of Medicine, personal commu-nication with the Office of Technology Assess-ment, 1982.Lawson, D. H., “Intensive Monitoring in Hos-pitals—l: Boston Collaborative Drug SurveillanceProgram (BCDSP), ” in Monitoring for Drug Safk-ty, W. H, W. Inman (cd. ) (Philadelphia: J. B. Lip-pincott Co., 1980).Lawson, D. H., and Henry, D. A., “MonitoringAdverse Reactions to New Drugs: ‘Restricted Re-lease’ or ‘Monitored Release’, ” Br. Med. J. 1:691,1977.Levy, R., and Sondik, E., “Decision-Making inPlanning Large-Scale Comparative Studies, ” Ann.N. Y. Acad. Sci. 304:441, 1978.Lewis, J, A., “Post-Marketing Surveillance: HowMany, ” Trends in Pharmacological Sciences 2:93,1981.Lindsjo, A., “Down’s Syndrome in Sweden, ” Ac-ta Paediatr. Scand. 63:571, 1974 .MacMahon, B., et al., Epidemiologic Methods(Boston: Little, Brown, 1960),Mann, J. I., “Principles and Pitfalls in DrugEpidemiology,” in Monitoring for Drug Salkty, W.H. W. Inman (cd. ) (Philadelphia: J.B. LippincottCo., 1980),Mann, J. I., and Inman, W. H. W., “Oral Contra-ceptives and Death From Myocardial Infarction inYoung Women, ” Br. Med. J. 2:245, 1975.Mann, J. I., et al., “Myocardial Infarction inYoung Women With Special Reference to OralContraceptive Practice, “ Br. Med. J. 2:241, 1975.Medico-Pharmaceutical Forum, Post-MarketingSurveillance of Adverse Reactions to New Medi-cines, publication No. 7 (London: Medico-Phar-maceutical Forum, 1977).Meyboom, R. H. B., “Drug Interactions With the‘Pill’, ” in Drug Interactions, D. G. Grahame-Smith(cd, ) (London: MacMillan, 1977).Mosteller, F., Chairman, Department of HealthPolicy and Management, School of Public Health,Harvard University, personal communication withthe Office of Technology Assessment, 1982.Nicholls, J. T., “The Practolol Syndrome—A Ret-rospective Analysis, ” in Postmarketing Surveil-lance of Adverse Reactions to New Medicines,

References ● 6 5

57.

58,

59<

60.

61<

62<

63

64.

65.

66.

67.

68,

69,

publication No. 7 (London: Medico-Pharma-ceutical Forum, 1978).Norwegian Multicenter Study Group, “Timolol-Induced Reduction in Mortality and Reinfarctionin Patients Surviving Acute Myocardial Infarc-tion, ” N. Eng. J. Med. 304:801, 1981.Ory, H., “Function of Ovarian Cysts and OralContraceptives, ” J. A.M.A. 228:68, 1974.Reed, P. I., et al., “Effect of Cimetidine on GastricJuice N-Nitrosamine Concentration, ” Lmcet 2:553,1981.Reed, P. I., et al., “Gastric Cancer in Patients WhoHave Taken Cimetidine, ” Izmcet 1:1234, 1979.Remington, R. D., “Post-Marketing Drug Surveil-lance: A Comparison of Methods, ” Amer. J. Phar-macy 150:72, 1978.Royal College of General Practitioners’ Oral Con-traception Study, “Mortality Among Oral-Contra-ceptive Users, ” Lancet 2:727, 1977.Ruskin, A., and Anello, C , “The United Statesof America, ” in Monitoring for Drug Safety, W.H. W. Inman (cd. ) (Philadelphia: J. B. LippincottCo., 1980).Sackett, D. L., et al., “Compliance,” in Monitor-ing for Drug Safety, W. H. W. Inman (cd. ) (Phil-adelphia: J. B. Lippincott Co., 1980).Sartwell, P. E., “Retrospective Studies: A Reviewfor the Clinicianr ” Ann. Lntern. Med. 81:381, 1974.Schor, S., Executive Director, Clinical Biostatisticsand Research Data Systems, Merck Sharp &Dohme Research Laboratories, personal com-munication with the Office of Technology Assess-ment, 1982.Sherrid, P., “Drugs,” Forbes magazine, Jan. 4,1982, p. 214.Shimomura, T., “Compensation for Drug-InducedIllness in Japan, “ in Monitoring for Drug Safety,W. H. W. Inman (cd. ) (Philadelphia: J. B. Lippin-Cott co., 1980).Strom, B., Clinical Epidemiology Unit, School of

70.

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

Medicine, University of Pennsylvania, personalcommunication with the Office of TechnologyAssessment, 1982.Taylor, T. Jr., et al., “Gastric Cancer andCimetidine,” J, Royal Co]]. Surg. Edinburgh 26:34,1981.Thayer, C. F., “Results of Postmarketing Surveil-lance Program on Streptokinase, ’f Current Thera-peutic Research 20:129, 1981.U.S. Senate, Committee on the Judiciary, reporton The F’atent ~erm Restoration Act of 1981, r e -port No, 97-138 (Washington, D. C.: U.S. Govern-ment Printing Office, 1981).Vessey, M. P., and Doll, R., “Investigation of Re-lation Between Use of Oral Contraception andThromboembolic Disease, ” Br. Med. J. 2:199,1968.Vessey, M. P., et al., “Postoperative Thromboem-bolism and the Use of Oral Contraceptives, ” B.Med. J. 3:123, 1970.Walden, R. J., and Prichard, B. N. C., “Post-Marketing Drug Surveillance, ” M. J. C/in. Phar-mace]. 6:191, 1978.Walker, A. M., Sidney Farber Cancer Institute,School of Public Health, Harvard University, per-sonal communication with the Office of Technol-ogy Assessment, 1982.Wardell, W. M., et al., “Postmarketing Surveil-lance of New Drugs: I. Review of Objectives andMethodology, ” J. CZin. Pharmacol. 19:85, 1979.Wardell, W. M., et al., “Postmarketing Surveil-lance of New Drugs: II. Case Studies, ” J. Clin.Pharmacol. 19:169, 1979 .Wilson, A. B., “Post-Marketing Surveillance ofAdverse Reactions to New Medicines, ” Br. Med.J. 2:1001, 1977.Wolfe, S., Director, Health Research Group,Washington, D. C., personal communication withthe Office of Technology Assessment, 1982.

Index

Index

abbreviated new drug applicationadverse drug reactions (ADRs), 4,

37, 41, 42, 43, 45, 50, 51, 52in Great Britain, 23-25, 38in Japan, 23international monitoring, 44screening of (SOAR), 4 5

1719, 24, 25, 33, 34/ 35,

American Academy of Dermatology, 42authority for refusing application; approval of

application, 57-58

Beta-Blocker Heart Attack Study Group, 51Birth Defects Monitoring Project (National Centers for

Disease Control), 43Boston Collaborative Drug Surveillance Program, 41, 42Boston University Medical Center Drug Epidemiology

Unit, 41, 42

case studies in drug evaluation, 36-38Center for Drug Surveillance, 60Children’s Hospital Medical Center (Boston, Mass.), 41,

42cimetidine, 19, 37, 38, 42Commission on Professional and Hospital Activities—

Professional Activity Study (Ann Arbor, Mich. ), 42Commission on the Federal Drug Approval Process, 5,

27, 49compassionate IND, 15conclusions and recommendations of the Joint

Commission on Prescription Drug Use, 59-60Congress:

acts of (see legislation)House Committee on Energy and Commerce,

Subcommittee on Health and Environment, iiiHouse Committee on Science and Technology:

Subcommittee on Natural Resources, AgricultureResearch and Environment, 6, 27

Subcommittee on Investigations and Oversight,6, 27

Senate Committee on Labor and Human Resources,Subcommittee on Health, 5, 26

current activities that contribute to postmarketingsurveillance, 41-45

Department of Commerce, 5, 2 7Department of Health, Education, and Welfare (now

DHHS), 5Review Panel on New Drug Regulation, 26Secretary of, 14, 57-58

diethylstilbestrol, 23, 33Down’s syndrome, 36drug approval process, 4-5, 13-19drug classification, 16Drug Enforcement Agency, 44Dubin-Johnson syndrome, 36

Experimental Technology Incentives Program (ETIP),5, 27

Food and Drug Administration (FDA) (Department ofHealth and Human Services), 3, 4, 5, 6, 7, 8, 9, 10,13, 14, 15, 16, 17, 18, 26, 27, 31, 37, 38, 41, 42,43, 44, 45, 49, 50, 51, 52, 53, 54

authority, 3, 4, 26, 27, 28, 49, 51, 52, 53, 54Bureau of Biologics, 6, 27Bureau of Drugs, 6, 27Current Drug Experience Literature, 45Division of Drug Experience, 42, 44National Center for Drugs and Biologics, 6, 27New Drug Evaluation Division, 6, 27monitoring questions, 45refusing application, 57-58removal of drugs, 53, 54

General Accounting Office (GAO), 45, 52glossary of acronyms, ixGreat Britain:

adverse drug reactions, 23-25oral contraceptives, 36postmarketing surveillance, 23practolol syndrome, 23voluntary reporting system, 23yellow card, 23, 24-25, 31, 36

Group Health Cooperative of Puget Sound, 41, 42

history and objectives of postmarketing surveillance,23-28

Hoechst-Roussel Pharmaceuticals, 37

improving postmarketing surveillance, 9-10, 51-54IND application process, 13investigational new drug (IND), 4, 5, 14, 15, 16

clinical investigations:Phase I: Clinical pharmacology, 4, 9, 13, 14, 15, 17Phase II: Clinical investigations, 4, 6, 9, 13, 14, 15

27, 50Phase III: Clinical trials, 4, 5, 6, 8, 9, 10, 15, 17,

18, 19, 27, 38, 49, 50, 53, 54issues and options, 8-10, 49-54

Javits, Senator Jacob, 26Joint Commission on Prescription Drug Use, 5, 26, 49,

51, 52conclusions and recommendations, 59-60

Kennedy, Senator Edward, 5, 26

legislation:Drug Regulation Reform Act of 1979 (S. 1075) (H. R.

4258), 5, 26excerpts from statutes governing drugs and medical

devices, 57-58

69


Food, Drug, Cosmetic Act, 5, 14, 26Medical Device Amendments of 1976 (Public Law

94-295), 53, 54Patent Term Restoration Act of 1981 (S. 255),

6, 27, 49Senate Bill S. 1075, 5, 26, 53, 54

Medicaid, 32, 42, 43, 52Medicaid Medical Information Systems, 4, 42Medicare, 32, 52methods of surveillance, 31-38Monthly Index of Medical Specialties, 23

National Bureau of Standards (Department ofCommerce), 5, 27

National Cancer Institute, 13, 43National Center for Drugs and Biologics, 14

Office of Drugs, 14, 15, 18, 19National Center for Drug Science, 5, 26, 49National Center for Health Statistics, 43

National Ambulatory Medical Care Survey, 43National Heart, Lung, and Blood Institute (NHLBI), 51National Institutes of Health, 45National Institute on Drug Abuse, 44National Registry of Drug-Induced Ocular

Side Effects, 42new drug application process, 16New England Journal of Medicine, 51

oral contraceptives, effects of 36, 37in Great Britain, 36in Sweden, 36

Pharmaceutical Manufacturers Association, 5, 26practolol, 23propanolol, 9, 19, 50, 51

Registry of Dermatological Reactions to Drugs, 42Registry of Hepatic Toxicity to Drugs, 42Registry of Patients Exposed to Radiopharmaceutical

Drugs, 42Remington, R. D., 52requirements following new drug approval, 18-19Review Panel on New Drug Regulations, 5revising premarketing tests, 8-9, 49-51

shortening the drug approval process, 8-9Smith, Kline & French Laboratories, 37

Spontaneous Reaction Reporting Program, 4, 19streptokinase, 19, 37, 42surveillance:

abbreviated new drug application (ANDA), 17background incidence, 8, 32, 34, 35benefit/risk ratio, 17clinical investigations (see investigational new drug)evidence of efficacy and safety, 3, 6grandfather drugs, 13history, 5imminent hazard to the public health, 3, 5, 10, 26,

54, 57international monitoring, 44manufacturer’s programs, 41, 42methods, 6-8

case-control studies, 6, 7, 8, 9, 10, 31, 32, 33, 36, 51cohort studies, 6, 7, 8, 9, 10, 31, 32, 33, 36, 51controlled clinical trials, 6, 7, 8, 9, 10, 17, 31, 32,

33, 35, 36, 38, 51voluntary reporting, 6, 7, 10, 31, 32, 33, 37, 42, 45

new drug application (NDA), 4, 5, 13, 14, 15, 16, 17,18, 19

nonapprovable letter, 17objectives, 5postmarketing, 3, 5, 6, 7, 9, 10, 18, 23, 26, 27, 28,

31, 37, 38, 41, 45, 49, 50, 51, 52, 53, 54, 59-60Phase IV, 18, 19, 42, 52principal focus, 3

premarketing testing, 3, 4, 6, 8, 9, 13, 14, 18, 19, 26,27, 28, 31, 37, 38, 49, 50, 52, 54

removal of drug from market, 3sampling, 7, 8, 9, 33, 34, 35, 37, 38sponsor of drug, 4, 5, 13, 14, 16, 17, 18, 26, 37, 52substantial evidence, 13timolol, 9, 50, 51

Norwegian study, 50-51

Temple, Robert, 15thalidomide, 23

World Health Organization (WHO), 4, 19, 41Program for International Monitoring of Adverse

Reactions, 44

yellow card (fig. 1), 24-25