The interventions aimed at patients with unprotected left main coronary disease are supported by a vast amount of data, the interpretation of which is increasingly difficult. For this purpose, studies using traditional meta-analysis and/or metaregression analysis are helpful, because they effectively synthesize the available information.
In these patients, the analysis of outcomes other than mortality can be subjected to different end point definitions and/or different interpretations. Therefore, we undertook a new examination of the same data and focused our analysis only on the end point of all-cause mortality. The study described, designed as a temporal trend analysis based on metaregression, was intended to be the completion of our previous metaregression that focused only on nonmortality end points. The present metaregression examined all observational studies included in 3 previously published meta-analyses. All-cause mortality at 1 to 2 years was the end point of our analysis. Each observational study was assigned to a specific calendar year that represented the midpoint of the respective enrollment interval. Our analysis used standard techniques of metaregression and specific software (OMA program, Open Meta-Analyst, version 4.16.12, Tufts University, available at: http://tuftscaes.org/open_meta/ ). In proportion meta-analysis, the rates were subjected to arcsine transformation and then back-transformed to avoid negative values in the 95% confidence intervals.
A total of 18 observational studies were included in our meta-regression, the results of which are summarized in Figure 1 , showing the temporal trend of mortality in patients receiving coronary stenting and for patients who underwent coronary artery bypass grafting. Interestingly enough, in patients who underwent coronary stenting, the increase in 1-year mortality with time was highly significant ( Figure 1 ). However, this end point was very stable with time in the group treated with coronary artery bypass grafting ( Figure 1 ). In the comparison of stenting versus coronary artery bypass grafting ( Figure 1 ), the increase in the risk ratio with time (at limits of statistical significance) was a direct consequence of the end point findings. Additional details on these metaregression findings are provided in Appendix A .
