**4.3 Meta-analysis of MVD and DM**

In this meta-analysis of the individual database of patients, where they analyzed 11 trials of patients with MVD followed in the long term, who were randomized to PCI or CABG, in the subgroup of patients with DM, it was observed that mortality was significantly higher in patients with PCI 15.7% than with CABG, which was 10.7% (p = 0.0001), while no differences were found among non-DM patients, 8.4% for CABG and 8.7% in the PCI group (p = 0.81) [43].

## **4.4 Our experience**

*The Current Perspectives on Coronary Artery Bypass Grafting*

and randomized. However, for the 5-year outcome rates, the denominator was 752 for PCI and 781 for CABG. These numbers are not the group totals but rather the number of patients remaining at risk at the end of the study. The number of events and the number remaining at risk are independent of each other. A basic occupant of a rate is that the subjects in the numerator are included in the denominator. The percentages the authors report are not rates; they are ratios and are very misleading. Calculating the events among the number randomized in each group results in a relative difference of 26% that is less significant than reported (**Figure 3B**). We would have more confidence in these recalculated rates if the study included all subjects in the denominator and accounted for outcomes on all subjects. They do not include the 214 (11.3%) patients lost to follow-up for whom we have no outcome data. This study also experienced a significant differential in attrition by group. The CABG group had twice the patients lost to follow-up (14.9%) as the PCI group did (7.7%). Revising the comparison by adding in the lost patients as events and calculating it with an intention-to-treat analysis (attributing events to the group of original assignment), we get a very different picture for the 5-year outcome (**Figure 3C**). The relative 5% difference is not significant (p = 0.42). This finding is in line with the 2-year composite outcomes in which the study authors observed no difference in outcome rates (13.0% vs. 11.9%, p = 0.51). The 5-year finding is significantly biased by the differential FREEDOM trial results comparing ITT analyses [72]. Other points of FREEDOM, which used first-generation stents that are currently discontinued, we remember presented a high rate of thrombosis stent [73]. Also in the trial a great geographical disparity was observed, since this difference marked by the study only was able to observe in the United States and the other centers in the randomization, and there were no significant differences outside of North American centers [68]. VA CARDS trial [74] is a study of veteran hospitals in the USA, in 22 centers, and included diabetic patients with MVD and 198 patients to be revascularized to PCI with DES or CABG with a 2-year follow-up. The primary end point of the study was the combined death events of all causes and nonfatal MI. The study was stopped early due to very slow recruitment by enrolling a quarter of the pre-established patients, which did not produce the power necessary for the evaluation of events. Within the study, it was observed that mortality in the 2-year PCI group reached a very high number up to 21% vs. 5% for CABG, while mortality was very high in the CABG group up to 15% compared with 6.2% for the PCI. This study

*A.B.C. FREEDOM trial results comparing ITT analyses. CABG, coronary arterial bypass graft surgery; ITT, intention-to-treat analysis; PCI, percutaneous coronary intervention. From "Critical appraisal of cardiology* 

*guidelines on revascularization: clinical practice" (David R Dobies and Kimberly R Barber).*

**20**

**Figure 3.**

In the ERACI III registry [76] which included 3 cohorts of 225 patients in each group with multiple MVD and PCI with DES, PCI with BMS, and patients with CABG, we analyzed the results of the subgroup of diabetic patients in each group at 3 years of follow-up. The incidence of MACCE at 3 years was significantly higher in diabetics than nondiabetics (RR, 0.81 [0.66–0.99]; p = 0.018). Higher rates of death and nonfatal AMI and a trend toward increased TVR, among others, were the principal determinants of increased MACCE. When stratified by treatment modality, MACCE rates among diabetics at 3 years were 36.2% in the DES arm, 43.6% in the BMS arm, and 30.8% in the CABG group (p = 0.49). There was a nonsignificant trend toward more death and nonfatal MI among diabetics in the ERACI III-DES cohort (19.1%) than in the BMS (12.8%) or CABG (15.4%) arms of ERACI II. Just as in the FREEDOM trial, the only stents used were the first-generation stents. Another limitation is that it was not a randomized trial, but they were two wellfollowed cohorts.
