Professional Documents
Culture Documents
28 Premarketing Applications of Pharmacoepidemiology: Harry A. Guess
28 Premarketing Applications of Pharmacoepidemiology: Harry A. Guess
Premarketing Applications of
Pharmacoepidemiology
HARRY A. GUESS
Merck Research Laboratories, Uest Point, PA, and University of North Carolina, Chapel Hill, NC, USA
tions of drug safety during premarketing clinical Harmonisation of Technical Requirements for
studies. Next, we will briefly mention information Registration of Pharmaceuticals for Human Use
technology requirements and statistical (ICH)7, 8 and by the Uorld Health Organization’s
methods. Finally, we will discuss some Council for International Organizations of Med-
interesting case studies in which the analyses ical Sciences (CIOMS).9, 10 This is part of an
resulted in publica- tions. Because premarketing ongoing process of international standardization
applications of phar- macoepidemiology of definitions and reporting procedures used in
normally require access to premarketing, as yet drug safety evaluation.
unpublished, clinical data, such studies are of According to current regulations,6 a ‘‘serious’’
greatest importance and interest to those in the adverse drug experience is any adverse drug
pharmaceutical industry or regulatory agencies, experience occurring at any dose that results in
and the perspective taken in this chapter reflects any of the following outcomes„ death, a life-
this. However, the topic should also be of threatening adverse drug experience, inpatient
interest to others in the field of hospitalization or prolongation of existing hospi-
pharmacoepidemiology, including academic talization, a persistent or significant disability/
researchers and clinical investigators. incapacity, or a congenital anomaly/birth defect.
Important medical events that may not result in
death, be life threatening, or require hospitaliza-
CLINICAL PROBLEMS TO BE tion may be considered a serious adverse drug
ADDRESSED BY experience when, based upon appropriate medical
PHARMACOEPIDEMIOLOGY judgment, they may jeopardize the patient or
RESEARCH subject and may require medical or surgical
intervention to prevent one of the outcomes listed
The need for premarketing epidemiologic assess- in this definition. Examples of such medical events
ments of drug safety may arise in several situa- include allergic bronchospasm requiring intensive
tions, including evaluation of serious adverse treatment in an emergency room or at home, blood
events in noncomparative clinical trials, preparing dyscrasias or convulsions that do not result in
integrated summaries of safety for a New Drug inpatient hospitalization, or the development of
Application (NDA) or international marketing drug dependency or drug abuse.
application, and answering inquiries from regula- An ‘‘unexpected’’ adverse drug experience is
tory agencies or advisory committees. For clinical any adverse drug experience, the specificity or
trials conducted under US Food and Drug severity of which is not consistent with the
Administration (FDA) regulations for Investiga- current investigator brochure; or, if an
tional New Drugs (IND), the Code of Federal investigator bro- chure is not required or
Regulations (21 CFR 312.32) currently (January available, the specificity or severity of which is
1999) mandates reporting to FDA and all partici- not consistent with the risk information described
pating investigators within 15 calendar days any in the general investiga- tional plan or elsewhere
adverse experience which simultaneously meets the in the current application, as amended. In
three conditions„ (i) ‘‘serious,’’ (ii) ‘‘unexpected,’’ reporting IND safety reports, the regulations
and (iii) having a reasonable possibility of having indicate that the sponsor shall identify all safety
been caused by the study drug.6 If, in addition to reports previously filed with the IND concerning
the above three conditions, the event is also fatal a similar adverse experience and shall analyze
or life threatening, a report to FDA by telephone the significance of the adverse experience in
or facsimile transmission within seven calendar light of the previous similar reports.
days is required (6). The publication of these An epidemiologic approach to the preparation
regulations in the Federal Register6 noted that of an IND safety report can occasionally be
they were issued to implement definitions, report- useful when the report concerns a rare adverse
ing periods, formats, and standards as recom- event for which appropriate comparative
mended by the International Conference on incidence data are available. For example
cancer incidence and
mortality data by age, gender, and race are the drug as well. Uhile hypothesis testing for
available online from a number of countries. 11−13 effectiveness outcomes generally is done within
To produce an epidemiologic assessment it is individual studies, it usually is not appropriate to
necessary to perform, review, and report the proceed with hypothesis testing procedures for
safety outcomes. Rather, the approach to safety
analysis in time to meet the 15 calendar day data may be viewed more as exploration and
regulatory reporting deadline. This naturally puts estimation.
a constraint on how comprehensive such an
assessment can be.
The FDA has made available a number of This discussion of the goals and approaches of a
guidelines to assist manufacturers in preparing pre-approval safety evaluation makes it clear that
NDAs and FDA medical officers in reviewing epidemiologic reasoning and methods are clearly
them. A recent draft guideline for medical useful. The need to evaluate data from all trials,
reviewers outlines the goals of a safety review of with different kinds of patient population, differing
an NDA„14 designs, and differing durations makes
epidemiolo- gic approaches especially appropriate.
In general, the goals of a safety review are (1) The approach required is one that recognizes the
to identify important adverse events that are limited nature of the data and avoids
causally related to the use of the drug, (2) to overinterpretation in either direction.
estimate incidence for those events, and (3) to
identify factors that predict the occurrence of
those events, e.g., patient factors such as age,
gender, race, comorbid illnesses, and drug METHODOLOGIC PROBLEMS TO BE
factors such as dose, plasma level, duration of ADDRESSED BY
exposure, concomitant medications. The safety PHARMACOEPIDEMIOLOGY
review is useful not only in making a RESEARCH
risk/benefit decision, but also in drafting
labeling for a drug that is going to be approved.
EPIDEMIOLOSIC APPROACHES TO
The guideline also emphasizes the difference
ASSESSIUS CAUSALITY OF ADVERSE
between the formal, prespecified efficacy
EVEUTS IU PRE-APPROVAL CLIUICAL
evalua- tion and the more exploratory
TRIALS
approaches used in safety evaluation in pre-
approval clinical trials„ Clinical investigators and clinical monitors con-
tinually review reports of adverse events
Approaches useful for evaluating the safety of a occurring in pre-approval clinical trials.
drug under development generally differ
Investigators are asked to provide a clinical
substan- tially from those useful in evaluating its
effective- ness. In fact, most of the studies in assessment as to the causality of each event.
phases 2 − 3 of a development program focus Such assessments are a necessary part of
on establishing a drug’s effectiveness. In safety monitoring in clinical research, although
designing these trials, critical efficacy endpoints they are known to be subjective and imprecise14
are identified in ad- vance and sample sizes are
estimated to permit an adequate test of the null
(see also Chapter 32). Criteria to help guide
hypothesis. For the most part, phase 2 − 3 trials causality assessments have been pub- lished by
are not designed to test hypotheses about FDA.15 These criteria are most useful when there
safety. In fact, the safety end- points are are well defined differences in clinical features of
generally not known prior to the conduct of drug related and non-drug-related cases of the
these trials, and for many of the observed safety
outcomes, one can assume that the available
adverse event. However, for serious, uncommon
studies are underpowered. The usual approach adverse events where the clinical features of the
is to screen broadly and sensitively for adverse drug related cases could be similar to those of
events, and it is hoped that this approach non-drug-related cases, it can some- times be
should reveal the common adverse event profile helpful to supplement clinical causality
of a new drug and detect some of the less
common and more serious adverse events assessments of individual cases by
associated with epidemiologic
assessment criteria based on comparisons It is also important to recognize that the
between groups of patients. abzence of any association between a drug and
The epidemiologic literature provides several any given adverse event has to be judged in the
sets of criteria for helping to decide whether an context of the limited amount of patient exposure
empirical association is likely to be causal. The in pre-approval clinical trials. This topic is
best known criteria are those proposed in 1965 reviewed in ICH Guide- line E1A, which
by Bradford-Hill to help evaluate evidence linking addresses the extent of population exposure
cigarette smoking with lung cancer16 (see also needed to assess clinical safety for drugs
Chapter 2). The nine Bradford-Hill criteria are intended for long term treatment of non-life-
discussed briefly below as they relate to threatening conditions.7, 8 The guideline notes
evaluation of adverse experiences in that
premarketing clinical trials.
It is expected that short-term event rates (cumu-
lative 3-month incidence of about 1%) will be
Strength of Association well characterized. Events where the rate of
occurrence changes over a longer period of time
This is commonly quantified in terms of a suitably may need to be characterized depending on
their severity and importance to the risk−benefit
adjusted hazard function ratio or risk ratio assessment of the drug. The safety evaluation
(relative risk), rather than a p-value. In general, during clinical drug development is not expected
the farther the ratio is from unity, the less likely it to characterize rare adverse events, for
could be entirely attributable to imbalances in risk example, those occurring in less than 1 in 1000
patients.
factors between groups. An exception to this
occurs when the ratio is based on very small
numbers or highly influenced by only a few cases. Consistency
In addition, it is worth noting that with the large
number of different kinds of adverse events often The original wording was, ‘‘Has [the association]
seen in large clinical trials, it is quite likely that been repeatedly observed by different persons,
some risk ratios far from unity will occur by in different places, circumstances and times.’’ This
chance alone. Multiple comparisons not only is a useful criterion for assessing adverse
distort p-values, but can also bias risk ratios and experiences in a program of several clinical trials.
their confidence intervals away from unity. This Results that show a consistently elevated
happens because screening many different adverse incidence on drug across studies are generally
event terms and selecting those with very low p- more convincing than those in which the
values inherently selects for relative risks that are elevated risk is largely due to one study. It has
biased away from unity.17 Since the confidence been noted by FDA, however, that an apparent
interval is centered about the observed relative lack of consistency among trials may simply
risk, it will also be biased away from unity. This reflect differences in trial design, making the
type of selection bias, a form of regression to the event less likely in some trials than others. For
mean, is common to all programs of screening for example, in a combined analysis of several
unusually large or small values and is well known studies in an NDA, a reassuringly low incidence
in the statistics literature. One example where this of phototoxicity was seen. Examination of
appears to have occurred is in some individual studies revealed that in one study the
observational studies of vasectomy and rate of phototoxicity was substantial— a finding
prostate cancer.17, 18 Because pre-approval that had been obscured by combining the data.
reviews of drug safety use the same set of data This study was the only outpatient study and
for identifying potentially drug related events and thus these were the only patients who had an
for providing preliminary estimates of the risk, the opportunity to have the event by virtue of having
relative risk estimates can sun exposure.14
be biased away from unity.
TemporaFity
In both epidemiologic and clinical
assessments of causality, it is important to
distinguish between
events having onset before a drug was started lytic anemia, serum-sickness), altered
and those having onset after drug therapy was metabolism, or drug interactions.15 A classic
started. Especially in the evaluation of adverse example where timing provided highly convincing
experiences from studies without a comparison evidence of causality was in the evaluation of
group, it sometimes occurs that early symptoms of Guillain-Barre´ syndrome in association with the
a disease which is present but not yet ‘‘swine-flu’’ vaccine20 (see Figure 28.1).
recognized lead a patient to be prescribed a
drug, which then appears to be the cause of the
disease when it is eventually diagnosed. This Dose– Response
has been called ‘‘protopathic bias’’ and is a Adverse effects caused by exaggerated pharmaco-
special case of the broader concept of logical actions of a drug are often dose
‘‘confounding by indication’’ (see also Chapter dependent. Examples include hypotension
34). A classic example of this form of bias was resulting from the use of antihypertensive drugs
seen in uncontrolled postmarketing surveillance and gastrointestinal hemorrhage from
studies of cimetidine, where a higher than nonsteroidal anti-inflammatory drugs. For such
expected incidence of gastric carcinoma was adverse events it is especially important to
found among users than among nonusers. It is characterize how the incidence of the event
likely that many of the cancers were present but varies with dose in different patient popula-
undiagnosed at the time the cimetidine was tions. This is not only important in assessing
started. Subsequent studies showed that causality and quantifying incidence, but also in
elevations in gastric cancer risk diminished with understanding the mechanism and providing
duration of followup, returning to baseline with guidance to clinicians. The case of renal failure
long term use.19 in patients with congestive heart failure treated
Another way in which the concept of with the angiotensin converting enzyme (ACE)
temporality plays a role in the evaluation of inhibi- tor enalapril provides an excellent
adverse experiences is whether the timing of the example of where a life saving drug was initially
reaction in relation to duration of exposure is thought by some commentators to be inherently
consistent with the proposed mechanism. Thus, nephrotoxic, when the actualproblem was that
an elevated incidence of cancer in patients who the starting dose in patients with severe
had been taking a drug for several years would congestive heart failure was too high. Uhile ACE
be of more concern than would an elevated inhibitors improve survival in patients with
incidence in the first year of therapy. Timing congestive heart failure, too high a dose can
plays a major role in the evaluation of adverse shut off renal function, because angio- tensin-II
events that are thought to be due to immune is part of the compensatory process for
mechanisms (e.g., anaphylaxis, angioedema,
hemo-
Figwre 28.1. Onset of SwiFFain-Barre´ syndrome in reFation to day of vaccination.16 Reprodwced with permission.
maintaining adequate renal perfusion pressure may strengthen a causalinterpretation, but lack
in the face of low cardiac output. This example of it cannot be used to nullify one.
illustrates the importance of understanding both
pathophysiology and dose−response relationships
in evaluating drug safety. AnaFogy
Reasoning by analogy often serves as a basis
ExperimentaF Evidence for having a lower threshold for judging an
adverse event to be causally related to one drug
Evidence from animal experiments or from pre- in a class when the same effect is considered to
vious human clinical trials can be especially be causally related to another drug in the class.
important in helping to interpret adverse experi- Such reasoning is even more problematic than
ences. Uell designed animal studies can be reasoning based on biological plausibility and
especially helpful in determining the extent to this criterion has been criticized as being
which animal results are applicable to humans. potentially misleading. It is worth noting,
On the other hand, poorly designed animal however, that certain types of adverse event are
studies may produce misleading results, which commonly enough associated with drugs to be
can require carefully designed further worth anticipating as topics which regulatory
experiments to correct. agencies are likely to ask to be specifically
addressed. The FDA medical reviewer guidance
BioFogicaF PFawsibiFity document14 states that
an adverse event, but also to describe its clinical approval. This requires effort to build and main-
features and spectrum of severity. tain a repository of disease incidence and
natural history data from governmental surveys,
public- use data tapes of governmental studies,
THE FUTURE publica- tions, and other sources.42 It also
requires under- taking epidemiologic studies far
Premarketing and postmarketing applications of enough in advance of clinical trials to be able to
pharmacoepidemiology differ in the speed of contribute to clinical trial planning and analysis.
response required and in the fact that the Finally, it requires that the epidemiologic group
threshold for halting human exposure to a drug maintain an awareness of pertinent research
is lower before market approval than after findings in areas relevant to the diseases under
market ap- proval. In addition, premarketing study and to the types of measurement and
pharmacoepide- miology uses cohort study analytic technique likely to be needed.
designs almost exclusively, while postmarketing
pharmacoepide- miology often uses
case−control study designs as well. Safety
monitoring in clinical trials can be improved by REFERENCES
anticipating potential questions and organizing
data resources in advance to answer them as
1. Lawson DH. Pharmacoepidemiology„ a new disci-
rapidly as possible. Having a responsive data pline. Bv Wed J 1984; 289„ 940 − 1.
management system is as essential to pre- 2. Tilson HH. Pharmacoepidemiology„ the future.
marketing applications of Dvug Intell Clin Phavm 1988; 22„ 416 − 21.
pharmacoepidemiology as it is to the entire 3. Guess HA. Adverse and beneficial effects in
process of drug development. Epidemiologic pharmacoepidemiology. In„ Armitage P, Colton T,
eds, Encyclopedia of Bioztatizticz. New York„ Uiley,
planning should be regarded as one of the first 1998; 3332 − 8.
steps in drug development, rather than as
something to be considered at or after market
4. Mattison N. The FDA’s treatment IND„ current DL, Bryan JA„ Guillain-Barre´ syndrome following
controversies. Phavmaceut Wed 1988; 3„ 159 − 71. vaccination in the NationalInfluenza Immunization
5. Calandra GB, Garelik JP, Hohler PT, Brown HB. Program, United States, 1976 − 1977. Am J
Problems and benefits of an antibiotic compassio-
Epide- miol 1979; 11Є 105 − 23.
nate therapy program. Rer Infect Diz 1987; 9„ 1095 21. Ueed DL, Hursting SD„ Biologic plausibility in
− 101. causal inference„ current method and practice.
6. Yedeval Regiztev. Expedited Safety Reporting
Am J Epidemiol 1998; 147„ 415 − 25.
Requirements for New Drugs and Biologics. 7
22. Lane DA. The Bayesian approach to causality
October, 1997, 62 (194)„ 52 237 − 53. Available
assessment„ an introduction. Dvug Inf J 1986; 2Є
at
455 − 61.
http„]]www.access.gpo.gov]sudocs]aces]aces140.html.
23. Naranjo CA, Lanctot HL, Lane DA. The Bayesian
7. International Conference on Harmonisation of differentialdiagnosis of neutropenia associated
Technical Requirements for Registration of Phar- with antiarrhythmic agents. J Clin Phavmacol 1990;
maceuticals for Human Use (ICH). Available at„ 3Є 1120 − 7.
http„]]www.ifpma.org/ich1.html. Accessed 30 Janu- 24. Guess HA. Behavior of the exposure odds ratio in
ary, 1999. a case control study when the hazard function is
8. ICH Topics and Guidelines. Available at„ http„]] not constant over time. J Clin Epidemiol 1989; 42„
www.ifpma.org]ich5.html. Accessed 30 January, 1179 − 84.
1999. 25. Suissa S, Blais L, Spitzer UO, Cusson J, Lewis M,
9. Council for International Organizations of Medical Heinemann L. First-time use of newer oral contra-
Sciences (CIOMS). Available at ceptives and the risk of thromboembolism. Contva-
http„]]www.who.int] dsa]cat98]zcioms8.htm. ception 1997; 56„ 141 − 6.
Accessed 30 January, 1999. 26. Yola M, Lucien A„ Evidence of the depletion of
10. CIOMS Uorking Groups I−IV. Available at„ http„]] susceptibles effect in non-experimental pharmacoe-
www.who.int]dsa]cat98]zcioms8.htm. Accessed 30 pidemiologic research. J Clin Epidemiol 1994; 47„
January, 1999. 731 − 7.
11. National Cancer Institute. Suvreillance, Epidemiol- 27. Frome EL, Checkoway H. Epidemiologic
ogy and End Rezultz. Available at„ http„]]www- programs for computers and calculators. Use of
seer.ims.nci.nih.gov]. Accessed 25 November, 1998. Poisson regression models in estimating
12. Rotterdam Cancer Registry. Cancev Regiztviez on the incidence rates and ratios. Am J Epidemiol 1985;
Eeb. Available at„ http„]]www.ikr.nl]can- regs.htm. 121„ 309 − 23.
Accessed 25 November, 1998. 28. Greenland S, Rothman HJ. Introduction to
13. International Agency for Research on Cancer. stratified analysis. In„ Rothman HJ, Greenland S,
Cancev in the Euvopean Jnion. Available at„ http„]] eds, Wodevn Epidemiology, 2nd edn. Philadelphia,
www-dep.iarc.fr]dataava]eucandat.htm. Accessed PA„ Lippincott−Raven, 1998; Ch. 15, 253 − 79.
25 November, 1998. 29. Guess HA, Thomas JE. A rapidly converging
14. US Food and Drug Administration, Center for Drug algorithm for exact binomial confidence intervals
Evaluation and Research. Conducting a Safety Reriew about the relative risk in follow-up studies with
of a New Pvoduct Application and Pvepaving a Repovt on the stratified incidence density data. Epidemiology 1990;
Reriew. Available at„ http„]]www.fda.gov]cder] 1„ 75 − 7.
guidance]index.htm. Accessed 22 November, 30. Breslow NE, Day NE. Statiztical Wethodz in Cancev
1998. Rezeavch, Volume II„ the Dezign and Analyziz of Cohovt
15. Food and Drug Administration Staff College. Studiez, International Agency for Research on
Clinical Zhevapeuticz and the Recognition of Dvug- Induced Cancer Scientific Publication No. 82. Lyon„ Uorld
Dizeaze. MedUatch Continuing Education Article, Health Organization, 1987.
June 1995. Available at„ http„]]www.fda.- 31. Dean AG, Imrey HH, Dusich H, Hall UN.
gov]medwatch]articles]dig]rcontent.htm. Accessed Adjusting morbidity ratios in two communities
24 November, 1998. using risk factor prevalence in cases. Am J Epidemiol
16. Hill AB. The environment and disease„ 1988; 127„ 654 − 62.
association or causation. Pvoc R Soc Wed 1965; 32. Hleinbaum DG, Hupper LL, Morgenstern H. Epide-
58„ 295 − 300. miologic Rezeavch„ Pvinciplez and Quantitatire Weth- odz.
17. Guess HA. Invited commentary„ vasectomy and Belmont, CA„ Lifetime Learning, 1982; 163.
prostate cancer. Am J Epidemiol 1990; 132„ 1062 − 5. 33. Bruzzi P, Green SB, Byar DP, Brinton LA, Schairer
18. Rosenberg L, Palmer JR, Zauber AG, Uarshauer C. Estimating the population attributable risk for
ME, Strom BL, Harlap S, Shapiro S. The relation multiple risk factors using case−control data. Am
of vasectomy to the risk of cancer. Am J Epidemiol J Epidemiol 1985; 122„ 904 − 14.
1994; 14Є 431 − 8. 34. Guess HA. Uorldwide drug safety information
19. Ualker AM. Confounding by indication [editorial]. processing„ what’s ahead. Dvug Inf J 1989; 23„
Epidemiology 1996; 7„ 335 − 6. 169−70.
20. Schonberger LB, Bregman DJ, Sullivan-Bolyai JZ,
Heenlyside RA, Ziegler DU, Retailliau HF, Eddins
35. Calandra G, Lydick E, Carrigan J, Ueiss L, Guess adolescents„ a population-based study. Pediatvicz
H. Factors predisposing to seizures in seriously ill 1985; 76„ 512 − 7.
infected patients receiving antibiotics„ experience 39. Plotkin SA. Hell’s fire and varicella−vaccine safety
with Imipenem/Cilastatin. Am J Wed 1988;
[editorial]. New Engl J Wed 1988; 318„ 573 − 5.
84„ 911 − 8. 40. Guess HA, Stephenson UP, Sacks ST, Gardner
36. Guess HA, Resseguie LJ, Melton LJ III, JS. Beyond pharmacoepidemiology„ the larger
Hurland LT, Lydick EG, Uilson UR. Factors role of epidemiology in drug development. J Clin
predictive of seizures among intensive care Epidemiol 1988; 41„ 995 − 6.
unit patients with gram-negative infections. 41. Slater EE, Merrill DD, Guess HA, Roylance PJ,
Epilepzia 1990; 31„ 567 − 73. Cooper UD, Inman UHU, Ewan PU. Clinical
37. Shaw FE, Graham DJ, Guess HA, Milstien JB, profile of angioedema associated with angiotensin
Johnson JM, Schatz GC, Hadler SC, Huritsky JN, converting-enzyme inhibition. J Am Wed Azzoc
Hiner EE, Bregman DJ, Maynard JE. Postmarket- 1988; 26Є 967 − 70.
ing surveillance for neurological adverse events 42. Guess HA. The role of epidemiology in pharmaceu-
reported after hepatitis B vaccination„ experience tical research, development, and marketing.
of the first three years. Am J Epidemiol 1988; 127„ In„ Hartzema AG, Porta MS, Tilson HH, eds,
337 − 52. Phavma- coepidemiology„ an Intvoduction, 2nd edn.
38. Guess HA, Broughton DD, Melton LJ III, Cincinnati, OH„ Uhitney, 1991; 321− 9.
Hurland LT. Epidemiology of herpes zoster in
children and