results are restricted to statements regarding the
specific patient populations studied and the sample
sizes available (see above).
Post-marketing surveillance studies attempt to
study toxicity under conditions of actual use. These
studies differ from early phase investigations in
several ways (Wardellet al., 1979). Larger sample
size, lower cost nonrandom assignment, lack of
control over subgroups, long-term open-ended stu-
dies and no formal regulation may all be exploited.
Longitudinal studies investigate nonrandomized
groups(s) using a specific drug, and follow cohorts
of patients through time to see if a specific event
occurs. Case–control studies investigate nonrando-
mized groups of subjects with and without an
adverse event, reviewed retrospectively to deter-
minewhich drugs the subjects took; in this case, the
two or more patient groups are matched for inci-
dental features such as age or race.
39.10 The need for better
communication to the
prescribers and patientsThe most important responsibility of the pharma-
ceutical industry is to ensure that safety messages
are communicated clearly and effectively to prescri-
bers, and sometimes also to patients. Adding to the
core safety information is pointless when it is not
known whether such messages reach the target
audience. This is particularly relevant to contraindi-
cations, precautions and warnings. It is also presum-
ably the responsibility of the regulatory authorities
to identify and counsel any prescriber who they
identify may have misprescribed a drug to the detri-
ment of a patient. These mistakes may not be delib-
erate,butinviewofthevolumeofliteraturereceived
by busy physicians, it is essential that important
information concerning the administration and
safety of drugs is read and understood.
Modern technology should help. For example,
pharmacists are developing databases that help
them to identify drug interactions. In the future,
the medical history of a patient could be added to
a card which couldbe used bya pharmacist toensure
thatthe patient’s prescribedmedicine isappropriate.
It would also be possible to input safety data on
drugstocomputersystems already usedby prescrib-
ing physicians to store their patients’ records. The
physician would then be alerted to any contraindi-
cations, warnings or precautions that may be rele-
vant to individual patients if prescribed the drug.39.11 Summary
In this chapter, we have outlined the principal
motives and methods that pertain to good pharma-
covigilance. We have also tried to show how, in
particular, risk–benefit analysis must always be on
a case-by-case basis, and how it relies, ultimately,
on the judgment of those experienced in this field
rather than some automatically applicable arith-
metic algorithm. Large-scale patient exposures
will always trump clinical trials databases for
rare types of adverse event, and this continues to
be demonstrated by cases such as thalidomide,
terfenadine and rofecoxib.References and Resources
Bruppacher R, Gelzer J. 1991. Identifying, evaluating,
and quantifying adverse drug reactions in children:
opportunities and obstacles for the manufacturer.
Bratisl Lek Listy 92 : 549–553.
Castle WM, Cook SF. 2001. Pharmaceutical packaging
to prescribing of drugs. InEncyclopedia of Phar-
maceutical Technology, vol. 12, Swarbrick J, Boylan
JC (eds). Marcel Dekker: New York; 327.
CIOMS Working Group I. 1990.International Report-
ing of Adverse Drug Reactions.Final Report of
CIOMS Working Group. CIOMS: Geneva.
CIOMS Working Group II. 1992.International Report-
ing of Periodic Drug-safety Update Summaries.
CIOMS: Geneva.
CIOMS Working Group III. 1995.Guidelines for Pre-
paring Core Clinical-safety Information on Drugs.
CIOMS: Geneva.
CIOMS Working Group IV. 1999.Benefit-Risk Balance
for Marketed Drugs. Evaluating Safety Signals.
CIOMS: Geneva.
CIOMS Working Group V. 2001.Current Challenges
in Pharmacovigilance: Pragmatic Approaches.
CIOMS: Geneva.542 CH39 DRUG SURVEILLANCE