The chemical and biological functions of the human body are specifically adapted to maintain each individual in a state of homeostasis or biological equilibrium. Disease occurs when this balance is upset by some internal or external force. The physician treating diseases may administer drugs to try to restore the equilibrium. If the appropriate medication exists, it must be given at the right time by a suitable route, and in a proper amount and dosage form to obtain the desired response.

Even with properly administered drugs, biological variation being what it is, particular sets of conditions existing in some individuals may produce untoward events, that is, unfavorable responses not intended by the prescribing physician, at least not to the degree they occurred.

These untoward events may occur relatively often with administration of the drug, as a rash following treatment with penicillin, or with relative infrequency, as blood dyscrasia after administration of chlor

The FDA has the responsibility of keeping a constant vigil to uncover adverse drug effects, to estimate the degree of risk, and to take remedial regulatory action where indicated. Our objective here is to review the kinds of adverse reaction data currently available to the FDA, to describe some additional kinds of adverse reaction data needed by the Agency to safeguard the public health, and to summarize some of health, and to summarize some of the major problems that must be resolved to obtain such additional data.

Manufacturers' Reports. Reports of adverse experience with drugs for human use are required as part of a New Drug Application under Section 130.4, of the New Drug Regulations, and as a part of a supplemental new drug application under Section 130.9. For the pur

poses of the New Drug Regulations the term "adverse experience" refers to "any adverse experience associated with the use of the drug, whether or not considered drugrelated," and includes "any side effect, injury, toxicity, or sensitivity reaction, or significant failure of expected pharmacological action."

The information, except for investigational study Phase I and II reports (early pharmacological and clinical studies of a drug) and literature reports, should be reported on Standard Form FD-1639, Drug Experience Report, or, alternatively, in an approved computer-generated report. The identical information elements are required in either case; these elements include the date sent to FDA and designation as initial or follow-up report, and the patient's initials, identification number, sex, height, weight, date of birth, racial origin, the date of reaction onset, source of report and source's address, description of the suspected reaction, reaction outcome, listing of all drugs in order of suspicion to

gether with dosage form, total daily dose, route of administration, duration of therapy and dates of administration, reason for use of the drug, substantiating laboratory studies, potentially noxious or environmental factors, existing or prior disorders, past drug reaction or allergic history, and reproductive history.

The time requirements for reporting depend on the content of the reports. The general requirements, if the drug is intended for administration to man, are that routine reports shall be submitted at intervals of three months during the first year that begins with the date of approval of the application; at intervals of six months during the second year; and at yearly intervals thereafter. The regulations require, however, within 15 days of receipt by the applicant, complete records or reports of information concerning any unexpected side effect, injury, toxicity, or sensitivity reaction or any unexpected incidence or severity of such conditions associated with clinical uses, studies, investigations, or tests, whether or not determined to be attributable to the drug. For the purposes of this regulation, "unexpected" refers to conditions or developments not previously submitted as part of the New Drug Application or not encountered during clinical trials of the drug, or conditions or developments occurring at a rate higher than shown by information submitted as part of the New Drug Application.

Spontaneous, Voluntary Reporting. Besides the reports from the manufacturer required by regulation, there exist several other sources of adverse drug effect information which depend on the spontaneous, voluntary, direct reporting of untoward events believed to be drugrelated. These sources generally include clinical investigators, practicing physicians, other medical workers, participants in paid hospital-reporting programs and patients themselves. These additional "spontaneous voluntary reporting" sources serve as an alerting system for those untoward events in which the effect can be easily related to the drug

responsible. Usually such untoward events are acute and quickly follow administration of the drug, anaphylaxis to penicillin, for example. A simplified reporting form is being developed that will be convenient to use and will encourage reporting of adverse drug experiences.

Major deficiencies exist in the voluntary reporting programs, whether hospital-based or based on individual physicians. These include:

1. Reports are likely to identify only those events already familiar;

2. Diagnosis of drug reactions may be mistaken and misleading;

3. Reports do not establish the incidence of untoward events;

4. Delayed effects tend to remain unattributed to the drug; such effects include, for example, potentially serious hazards, such as phocomelia as related to thalidomide, or aplastic anemia as related to chloramphenicol.

The British use of voluntary reporting systems seems more advanced than our own. W. H. W. Inman, reporting at the International Conference on Adverse Reaction Reporting Systems in October 1970, noted that the basic elements of the British approach include: (a) a computer based printout, (b) motivation of the physicians to report suspected untoward events, (c) a highly qualified team of medical investigators to validate the report, (d) a good system of drug utilization data, and (e) a feedback procedure to the practicing physicians and other professionals to encourage the medical community to participate in the national drug monitoring programs.

But this system, like its American counterparts, is subject to bias and a substantial degree of underreporting of untoward events, since many events that occur can go unrecognized. Particularly likely to be overlooked are those subtle drug-related events that occur in an outpatient environment and those for which the time between drug administration and effect is long. For assurance that we have comprehensive surveillance, spontaneous reporting must be supplemented by intensive sur

veillance programs and controlled clinical trials.

Existing Intensive Surveillance Systems. Hershel Jick, at the recent International Conference on Adverse Reactions, defined the term "intensive drug surveillance" as a study in which a defined population is continuously monitored to determine:

1. the frequency with which certain events are associated with drug administration;

2. whether a causal relationship between the drugs and event is likely; and

3. whether there are subpopulations at greater risk, namely, more susceptible to reaction to the drug than is the general population.

The essential criteria for an intensive surveillance program, SO defined, is that: (a) the target population is clearly defined; and (b) standardized and routine procedures exist for the collection of information on patient characteristics, drugs administered or discontinued, and changes in the patient's condition (events).

Intensive hospital monitoring has already revealed that the incidence of adverse reactions is higher than suspected from voluntary reporting studies. Depending upon the institution and the investigational methods and the definitions used, the incidence of adverse reactions in patients during hospitalization has been reported in a range of values including 13.6 percent, and 30.9 percent (some of which were judged nontrivial).

The Food and Drug Administration, the National Institutes of Health (National Institute of General Medical Sciences), and others over the past few years have sponsored programs designed to improve the Nation's monitoring and drug surveillance capabilities. These programs have included both outpatient and inpatient populations. The requirements for the outpatient system are somewhat more complex and this system will take longer to perfect. Both of these types of intensive surveillance programs bear resemblance to the model suggested by

D. J. Finney in a proposal for an "International Drug Safeguard Plan," published in 1964. This model calls for the collection of drug use and medical history data on a total population of patients without regard to whether a particular medical condition is believed to be a consequence of a drug. The medical history data used for drug monitoring should also include identification of the therapeutic indication for starting the drug and the reasons for stopping a particular treatment regimen. These two facts appear to be helpful for the proper interpretation of the untoward effects of drugs. Finally, in this proposed model, information is compiled and analyzed by a central computer, which determines incidence rates or suspicious correlations.

Three monitoring programs that have been developed over the past few years represent prototypes of potentially expandable expandable intensive surveillance programs. The in-hospital intensive programs primarily cover patients in selected hospital medical wards. In the Tufts University program, pertinent information is recorded by nurse monitors, while in the University of Florida program pharmacists record the data. The outpatient program developed at Kaiser-Permanente in California has employed on-line entry of pharmacy records of prescriptions filled, optical scanning of forms containing physician diagnoses obtained during outpatient visits, and other modes of putting in data as directly recorded in patient care activities.

These programs differ substantially in "intensity" of monitoring, reliability of the information obtained, duration over which surveillance is possible, and capability for searching the computer records for unsuspected drug reactions. Even taken together, and not denying their merit, these programs fall far short of filling the total national needs for surveillance. W. H. W. Inman, in his report mentioned earlier, has pointed out that the majority of serious or fatal drug reactions are rare, with an incidence

ranging from perhaps 1 in each 1,000 to 1 in each 1,000,000 courses of treatment. The existing intensive surveillance systems do not have adequate numbers of patients under treatment to demonstrate reactions of such incidence.

General Problems in Design. The fundamental issues related to design of drug surveillance arise with the selection of an appropriate population to be monitored, the selection of drugs that because of their extensive use or relative recent entry into the physicians armamentarium require intensive or special monitoring, and the selection of those variates which need to be observed and screened to avoid serious consequences to the drug-consuming public.

To date little attention has been given to the characteristics which would make a population appropriate for a national drug surveillance program. Size and proximity to quality medical institutions have been viewed as essential characteristics, but these are not sufficient in identifying a target population. Most individuals take medication at some point in their lifetimes, and the specific drugs taken and duration of consumption vary greatly.

Hospital populations are convenient for surveillance because they can be intensively monitored and their medical records are readily accessible. In the hospital situation, however, the administration of drugs is carefully controlled and the choice of drugs is not representative of the pattern of drug administration in an outpatient environment. A comprehensive drug surveillance program must ensure that most sectors of the drug-consuming population are represented. The population under surveillance should include those subjected to less than the highest level of medical care. Attention should be given to the effects of drugs on the outcome of pregnancy.

Since it may not be economically feasible to monitor all drugs over an extended period of time, some selection of drugs may be necessary. It is clear that those drugs which will be used by patients for an extended

period of time, those which are taken just prior to or during pregnancy, and those which are given to a large segment of the population must be kept under surveillance. The recent studies with oral hypoglycemic agents have shown that many years may elapse before one can establish the existence of trends or identify possible adverse effects attributed to a specific drug. There should also be a procedure for terminating specific surveillance once confidence in a drug's safety has been established. This, however, is difficult to identify.

Considerable thought must be given to the particular biological effects that must be monitored as a minimum. A list of such effects could be almost unlimited. It includes mortality, selected morbidity, signs, diagnoses, symptoms, and laboratory findings. Not only must a decision be made concerning which variables are to be observed, but a decision also is needed on the timing of an observation. In the past, potentially drug-related events were recorded as part of the standard medical practice and little attention was given to drug surveillance objectives. The recent trend suggests that future programs, inpatient or outpatient, will provide for more intensive monitoring specifically designed to detect untoward events that may be shown to be related to drug intake. The outpatient situation, more complex than the inpatient situation, will require considerable thought and study. This is particularly true because, unlike hospital situations, the health status of an outpatient cannot always be ascertained. The most that can be expected of a drug surveillance program is that it provide sufficiently intense screening to detect significant morbidity and mortality which would not be detectable through other programs, such as spontaneous and voluntary reporting.

Several important issues need to be considered, assuming agreement can be reached on the population to be brought under surveillance, the variates that should be observed as a minimum, and the drugs that

must be included in the monitoring programs. The major issues include: (a) quality of the data, (b) size of population, (c) estimate of attributable risk, (d) use of retrospective and prospective studies, and (e) role of sampling.

First, as with all of science, the quality of information used to make inferential statements about a population must be defined and its credibility established; otherwise, there can be no confidence in the findings or conclusions. To this general scientific requirement, however, we must add that findings relating drugs to adverse effects can have important medical, legal, and economic consequences. Therefore, the usefulness of a comprehensive surveillance program will depend to a large extent on the reliability and the defensibility of the data and the subsequent findings. Every effort must be made to eliminate known or potential sources of bias and unsubstantiated subjective inferences. Intensive surveillance programs augmented by special clinical or animal studies should produce a data base acceptable to the medical profession, the academic community, and the drug-consuming public.

Another issue concerns the size of the population needed to provide an adequate basis for inference. Those trained in biometry know that this depends on the size of the population at risk, the expected frequency of an adverse effect, and significance attaching to a possible error. Intensive hospital monitoring programs have not covered large populations with diverse drug exposures. They have for the most part been restricted to medical wards. Clearly, any expansion in existing programs must include a greater variety of drug exposures. The outpatient environment, like the program at Kaiser-Permanente, offers an opportunity to keep large populations under surveillance (specifically, subscribing members of a prepaid medical program). But "under surveillance" is not the same as "exposed to drugs," and it is the size of the population exposed to the drug that must be considered when one is

attempting to specify surveillance programs.

An important objective for any comprehensive drug surveillance program, in addition to identifying whether a specific untoward effect may be related to the administration of a particular drug, concerns the need to estimate the magnitude of the excess risk attributable to that drug. The attributed risk can be expressed in a variety of ways (for example, a relative risk or an excess risk measured from a suitable control group). In addition, a surveillance system should be capable of providing estimates of the benefit that can be attributed to the drug. An estimation of risk without an estimation of benefit is an insufficient basis for making regulatory decisions on marketed drugs.

Recent evidence of drug effects has been generated by both retrospective studies (thromboembolism and oral contraceptives) and prospective controlled studies (the recently reported University Group Diabetes Program). Once a surveillance program has alerted us to the potential hazard of a drug, confirmational investigation must be initiated. Of course the reverse could also possibly occur: the finding could first be generated by controlled experiments, and confirmational evidence might be obtained through the intensive monitoring or other surveillance programs. This cross-checking of findings is necessary to assure the scientific objectivity of research in this difficult

area.

Finally, we should consider anew the role of survey sampling in support of a national drug surveillance

D

system. Good information on drug utilization appears to be essential if the magnitude of any drug-related benefit or hazard is to be viewed in proper perspective. Interpretation of drug usage data is complicated because some patients do not take drugs as prescribed, and tend to ingest many drugs based on their own evaluation of the therapeutic effect. A complete monitoring program will take into account the totality of drugs and devices and view them in the light of actual levels of consumption.

Those making efforts to deal with adverse reaction hazards should recognize that some such hazards will inevitably persist. They will persist even though the appropriate medication has been administered at the right time by a suitable route, in a proper amount and dosage form, because of the intrinsic variability of biological response to potent drugs. But adverse reactions can be minimized by foreknowledge of the risks. This knowledge, if it cannot always be established prior to approval of the drug for marketing, can be derived from post-marketing drug experience data. Such data is presently provided to the FDA in the form of spontaneous, voluntary reports, either transmitted by the manufacturer or submitted directly and voluntarily to the Agency from the observer of the reaction. These isolated reports in themselves do not constitute an adequate method of surveillance; intensive surveillance programs and controlled clinical trials are also required. Our objective is a comprehensive drug monitoring program to safeguard the public health.

THE

LABORATORY SCIENTIST AND IDIP

by Matthew H. Lewis and Anthony Duran

One of FDA's most ambitious field activities is the Intensified Drug Inspection Program, popularly called the IDIP. This program had its beginnings with the 1962 Kefauver-Harris Drug Amendments to the Federal Food, Drug, and Cosmetic Act. The goal of the IDIP is to ensure that manufacturers produce quality drug products through adherence to the Current Good Manufacturing Practices regulations published in the U.S. Code of Federal Regulations, Title 21, Part 133. The IDIP is intended to achieve compliance, when possible, through a cooperative effort by the FDA and the drug industry.

The new concept of intensified inspections was initiated in fiscal 1969. Unlike the more limited establishment inspections normally conducted in FDA operations, the Intensified Drug Inspection may last several months. Another significant difference is that the inspection is made by a team of FDA experts. This team consists of one or more inspectors (one of whom is the team leader), a chemist experienced in drug chemistry, and, when the firm deals with sterile products, a microbiologist. For the chemist and the microbiologist to be fully effective members of the team, they participate with the inspectors in the pre-IDIP meeting between