registered before their outcome is known: the
diagnosed birth anomaly can cause bias in report-
ing frequency, and converts a prospective
approach into a retrospective one. The choice of
comparison population is highly complex. In
general, the assumption of such registry studies
is that the appropriate comparison is the general
population, effectively a prospective cohort-
controlled approach. Thus, the objective of such
studies is to detect increases of specific defects
over the expected rates in the general population.
However, the detected occurrence of defects is a
function of follow-up method and definition, and
may be influenced by many (undocumented or
even unknown) factors. Therefore, the conduct of
registries and development of such comparisons
must be done with great caution. Guidance on
these matters has recently been proferred by the
US FDA (2002).
24.9 Balancing benefits
against risksCritical to understanding the evolving picture of
risk with a marketed product is to recognize that
such risks emerge with progressively greater
experience in increasingly broad and diverse popu-
lations over time...experiences which include
patients who differ in age, underlying disease
state and stage or even indication from those who
were studied in the development process. These
uses will have varying impacts...and may materi-
ally change the benefits profile from that in the
approved product information. Often these issues
become particularly important when a newly iden-
tified risk may appear to change the balance
of benefits against risks to an unacceptable
extent...precipitating consideration of possible
regulatory withdrawal.
Thus, a program of ongoing research may well
be needed which calls upon the skills of the phar-
macoepidemiologist, observational studies of
drug utilization patterns, often tied to medical
outcomes.
The professional epidemiologist is the first to
caution against cavalier approaches to such effec-
tiveness research. Without the protections of
randomization and blinding, many of the biases
which have plagued clinical research since its
inception can creep in to such studies and render
them uninterpretable, or, perhaps worse, lead to
wrong interpretations. Thus, the research team
contemplating the use of observational follow-up
studies to understand effectiveness, particularly as
part of a revised benefit to risk assessment, needs a
pharmacoepidemiologist on board.24.10 Training to be a
pharmacoepidemiologistNo parent has ever heard the statement: ‘Mummy
(or Daddy), I want to be a pharmaceutical phy-
sician pharmacoepidemiologist when I grow up’!
Few physicians and very few pharmaceutical
physicians choose this route. But what is along
that route? The ‘high road’, in the United States
at least, is graduation from medical school,
obtaining at least one year of intense clinical
postgraduate training (often leading to internal
medicine or other primary care boards, the
equivalent of MRCP in the United Kingdom),
and then to undertake a further formal residency
training program that leads to certification by the
American Board of Preventive Medicine. Resi-
dencies are inspected and approved by the Board,
and may be in one of four areas: public health,
general preventive medicine, occupational med-
icine or aerospace medicine. Various concentra-
tions in medical management are also becoming
recognized. In the United Kingdom, registrar
positions in most of these specialties are adver-
tised, and the Diplomas in Occupational or Pub-
lic Health, membership of the Faculties of
Occupational or Public Health and the Diploma
in Aviation Medicine would provide equivalent
experience and certification.
In the United States, the academic equivalent
of these, which can often be pursued in parallel,
is to obtain the additional degree of Master of
Public Health (MPH). Board certification
through a preventive medicine residency requires
such formal academic training and an MPH310 CH24 PHARMACOEPIDEMIOLOGY AND THE PHARMACEUTICAL PHYSICIAN