Appendix for Online

In this section we report statin usage rates and compare trends in observables over the entire population covered in our claims database who fit at least one of the criteria defined as risk factors for statin usage: above age 45 (if male) or 55 (if female), hypertension, diabetes, or cardiac disease. If physicians responded to Zocor’s patent expiration by prescribing statins more broadly, to patients who might be more price-sensitive, or in less need of a statin, then our estimates of the average copay effect on drug choice will be biased. Although we control for observable patient characteristics such as age and chronic conditions, significant changes in the distribution of these characteristics over time could be indicative of changes in unobservable characteristics. When we compare the observable characteristics in the period before and after the patent expirations, however, the differences are small (see Table 2). In fact, most of these differences are either smaller than or equivalent to the changes over the same time period among the entire population of covered patients with a statin risk factor in our data, which are reported in Appendix Table A2, Panel B. To address the acute concern that more price-sensitive patients might be more likely to appear after the patent expiration, Table 2 reports the averages of each patient’s generic share of unique drugs purchased in the past year, which is a proxy measure of price sensitivity or willingness to try generic drugs. Table 2 shows that the mean generic share of past drugs increased significantly in our sample, from 60 to 65%, from the preto post-expiration period. This increase is close in size, however, to the increase seen among people with at least one risk factor for statin use who did not use any statin during the entirety of the prior year (see Appendix Table A2, Panel B, and Figure A1.). Therefore, it reflects the market growth of generics over this time period rather than a change in the type of patient initiating statin Given the variance in generic drug prominence across different drug classes, this a rather crude measure, but it is clearly correlated with generic statin use: In the period before the patent expiration, the average value of this variable is 60% among those who started on a brand-name statin and 67% among those who started on generic Mevacor. After the patent expiration, the average value is 63% among those who start on a brand-name statin and 68% among those who start on an offpatent statin (Zocor, Pravachol, or Mevacor).