Some of the limitations of the FDA’s AERS
database are that the lag time can be several
months, there is underreporting and there can be
increased reports as a result of stimulated report-
ing. Additionally, there can be biased reporting
due to a number of factors, including publicity,
regulatory letters and so on, and because of the
differential interpretation of the reporting regula-
tions, reports may differ by country and company.
Additionally, duplication, coding errors, variable
historical data, poor quality of information and
changes over time are all limitations to information
recorded in the AERS database.
The data mining output is similar to the safety
surveillance output in that hypotheses are gener-
ated and these may need to be evaluated with
additional quantitative analyses as appropriate,
using the company database, stimulated reporting,
enhanced surveillance or they may require the
conduct of epidemiological studies.
40.5 Case study
Rhabdomyolysis and statins
Introduction
In this case example, the FDA’s SRSþAERS
database, through the end of the second quarter
of 2005, was data mined to determine the lower
95% confidence interval limit of the EBGM scores
(denoted as EB05), a measure of disproportional-
ity, for rhabdomyolysis associated with the use of
statins. The drugs of interest were atorvastatin,
cerivastatin, fluvastatin, lovastatin, pravastatin,
rosuvastatin and simvastatin. The event of interest
was rhabdomyolysis.
EB05 guideline
A guideline that has been used for identifying a
signal score for pairwise combinations as higher-
than-expected is an EB052. This criterion
ensures with a high degree of confidence that,
regardless of count size, the particular drug–event
combination is being reported at least twice as
often as it would be if there was no association
between the drug and the event (Szarfmanet al.,
2002).Data sourceThis report contains the most currently available
cumulative data from the FDA’s SRSþAERS data-
base, through the end of the second quarter of 2005.
This database contains approximately 2.7million
patient records. It includes branded and generic
prescription products that are marketed in the
United States. The database contains both US
reports (including consumer reports) and a subset
of non-US reports (AEs that are both serious and
unexpected, which are not contained in the US
package insert).
All data were retrieved utilizing Lincoln Tech-
nologies WebVDME 5.2, which is a data mining
application used in post-marketing safety surveil-
lance to support product risk management. Unless
specified, individual case reports were not specifi-
cally checked for duplicate reporting. However, the
vendor does implement an algorithm to screen the
database for duplicates as part of standard data
cleansing. Searches were conducted based on
‘drug mentions within a report’. This means that
all case reports where the selected drug is classified
as either a concomitant or suspect drug are
included.Data outputFigures 40.2–40.5 show the frequency and EB05
scores, both total and cumulative by year, of rhab-
domyolysis associated with the use of the statins.
AEs in the FDA database are codified using the
MedDRA dictionary. It is important to note that a
single case report may contain more than one pre-
ferred term.
The color of the bar represents a measure of
disproportionality, that is ‘how disproportionate’
is the observed report frequency of the AE–drug
combination compared to what might be expected,
if all AE–drug combinations in the database were
independent. The color scale ranges from a light40.5 CASE STUDY 549