Statistical analyses

We used summary tables to document the intervals between launch year and the year of first report of an adverse drug reaction, the interval between launch year and the year of first withdrawal, and the interval between the first report of an adverse drug reaction and the year of first withdrawal. Because these intervals were skewed, we used medians and interquartile ranges (IQRs). We also compared intervals to reports of adverse reactions and withdrawals between products based on their mechanistic actions. Fisher’s exact test was used to test for differences in the proportions of withdrawals based on the pharmacological mechanisms of action; a P value of <0.05 was considered statistically significant.