LAUSR

Thank you for sharply spotting and kindly bringing to our attention an error in the units of event rates of the LEADER trial in Table 1 of our article (1). As you correctly surmise (2), we failed to notice the different units in this trial from the other trials, also due to the fact that we added trials as the manuscript went through rounds of revision. However, the composite end point “HHF or CV death” cannot always be calculated as the sum of hospitalization for heart failure (HHF) and cardiovascular (CV) death because these two outcomes may overlap significantly. In the LEADER trial, the authors do not explicitly provide a figure for this composite end point, so rather than adding up its components we chose to use the “expanded major adverse cardiovascular events” for our exemplification. We have corrected these numbers—and doublechecked all the other values—in the new Table 1 and replotted them in the new Fig. 2, which replace the old table and figure in the online version of the article. In contrast to your calculation, however, we find that the new regression function (Fig. 2A) has much the same slope and statistical significance (r = 0.69, P = 1.9e-9) as the old one (should you be interested, we could share the data file with you). More importantly, none of the arguments or conclusions in the article are affected by the correction. With regard to this, please consider that the plots in Fig. 2 were used merely as examples of the different relation of absolute and relative risk to background risk rate, with no pretense to establish quantitative norms or formal between-trial comparisons. Thank you again for your interest in our work.