Estimating the extent of reporting to FDA: a case study of statin-associated rhabdomyolysis

Purpose To estimate the extent of reporting to FDA through statin-associated rhabdomyolysis data. Methods Data included incidence rates (IRs) of hospitalized rhabdomyolysis among statin users from a population-based study, and comparable reported AERS cases and national estimates of statin use from an AERS analysis. Using IRs, national estimates of statin use and average days supply per prescription, we estimated the number of US statin-associated cases of hospitalized rhabdomyolysis. We compared this estimate to the observed number of cases reported to FDA to evaluate the extent of reporting. We repeated this method for atorvastatin, cerivastatin, pravastatin, and simvastatin and statin combinations. We performed sensitivity analyses to check for biases such as misclassification of statin use and cohort selection bias. We evaluated potential time-dependent cerivastatin reporting by a Dear Health Care Provider (DHCP) letter. Results The estimated extent of reporting to FDA varied by statin (atorvastatin, 5.0%; cerivastatin, 31.2%; simvastatin, 14.2%; all four combined, 17.7%; and non-cerivastatin statins combined, 9.9%). No pravastatin-associated cohort cases occurred. Across a reasonable value range, sensitivity analyses did not significantly alter the results; overall the cohort was similar to national statin-users. There was a large increase in AERS reports after the cerivastatin DHCP letter and the estimated extent of reporting increased from 14.8 to 35.0%. Conclusions The extent of reporting of adverse events to FDA varied by statin and may be influenced by publicity. For statins-associated rhabdomyolysis, the estimated extent of reporting appears to range from 5 to 30% but in the absence of stimulated reporting appears to be 5-15%. Copyright (c) 2008 John Wiley & Sons, Ltd.