Confounding is also unlikely to have had a substantive impact on the
results of the case–control analysis; controls were matched to cases by
age and sex, most of the relevant risk factors for rhabdomyolysis were
identifiable using hospital discharge and dispensing data, and we explored
potential confounding by concomitant medicine use and other
risk factors for rhabdomyolysis in a matched analysis. As would be expected
on the basis of clinical guidelines, cases and controls who had a
history of hospital admission with coronary artery disease at cohort
entryweremore likely to have been started on higher doses of simvastatin
than patientswithout such histories; a similar, but lessmarked, pattern
was seen for both renal impairment or failure and diabetes (Supplemental
Table 2). Caseswere also more likely than controls to have been admitted
with these conditions at any time before the index date (Table 1).
However, adjusting for a history of coronary artery disease and diabetes
did not change the odds ratios, and we adjusted for a history of renal impairment
or failure.Hence, confounding by these co-morbidities is unlikely
to explain the strong associations we observed.