regulatory action can influence the rate of reporting
as well as the types of events reported. In addition,
this database can contain duplicate reports because
of multiple potential reporters, and the same
case may come from different reporters. Also, a
single case report may contain more than one
preferred term.
Data interpretation caveats
Lincoln Technologies, the company that prepares
the data, prepares it for WebVDME without
changes to the reports.
These signaling data are generated by computer-
assisted algorithms based on relative ratio – the
ratio of the observed event for the specific drug as
compared with the expected occurrence based on
other drugs within the database. A potential signal
is generated when an AE–drug combination has a
disproportionately high occurrence compared with
the background. It is important to stress that an
elevated ratio is a measure of statistical association
and not a clinical association between an AE and
a drug.
These identified AE–drug combinations are
hypotheses for further testing; follow-up may be
necessary to determine whether they represent a
potential drug safety issue. Further evaluation may
include a query and review of the company’s safety
database, review of the scientific literature and
preclinical data and consultation/discussion with
internal and external experts. If a product safety
issue was identified, next steps would include the
development of a risk management and risk com-
munication plan.
It is important to note that the numerator reflects
only reports in the database. The actual number of
patients exposed to the drugs is generally not
known, so it is not possible to calculate the true
incidence of an AE from this database. Thus, com-
paring EB05 scores across drugs should not be
done as a surrogate calculation for incidence. Inci-
dence data are not available from this database;
there are too many factors that could influence
the EB05 in an asymmetric manner, for exampleFigure 40.5 The cumulative annual disproportionality score (EB05) of cases of rhabdomyolysis in the AERS database
associated with the statins
552 CH40 DATA MINING