Case ascertainment is a further strength of our study. Allmembers of
the study cohortwhowere admitted to hospitalwith rhabdomyolysis or
myopathy during the study period are likely to have been identified
using the NationalMinimum Dataset, since public and private hospitals
are required to report all inpatient and day patient discharges and record
diagnoses according to strict ICD coding standards [30]. Similarly,
cause of death coding in theMortality Collection is subject to ICD coding
conventions and the quality of such coding was recently found to be
very good [31]. Our reviewof the spontaneous reports of rhabdomyolysis
and myopathy in statin users which were received by CARM also
provided reassurance about the likely completeness of case ascertainment.
We validated the diagnosis of rhabdomyolysis and myopathy by
reference to hospital discharge letters and death records using standard
criteria. For a minority of patients, these sources did not provide CK
concentrations. However, for all five rhabdomyolysis cases (one fatal)
without recorded CK levels a clinical diagnosis of rhabdomyolysis had
been made and there was documented evidence of acute renal failure.
Referral bias is unlikely to have played a role in our study, as the serious
nature of rhabdomyolysis is such that patients are likely to have been
referred to hospital regardless of the dose of simvastatin they were, or
were not, taking.
We took the date of diagnosis as the index date because the precise
date of onset of muscle symptoms was not known for all cases. Potentially
this could have led to amisclassification of exposure status, and biased
estimates of relative risk, if a gradual onset of symptoms had led to
an early discontinuation of simvastatin or a reduction in dose. However,
this does not appear to have been a substantive issue; of the rhabdomyolysis
cases who were classified as current users, only three were dispensed
a different dose of simvastatin in the 90 day period before the
index date (a reduction in dose from 80 mg to 40 mg [32 days before
index date], a short-term reduction from 40 mg to 20 mg [57 days before
index date, resumed 40 mg 20 days later], and an increase from
20 to 40 mg [51 days before index date]). Moreover, the three rhabdomyolysis
cases whowere past users (including the current user of atorvastatin)
stopped simvastatin well before the index date (155 days,
16 months, 24 months).