Journal of Invasive Cardiology

The definition of clinical endpoints, such as “myocardial infarction,” was predetermined by the authors of original studies. For creatinine levels, the highest value in the first 72 hours after surgery was used. Myocardial infarction was reported as rate of events and summarized as a risk ratio. Heterogeneity was evaluated using the I2 test. If results were considered homogenous, a fixed effects model was used for data synthesis; if not, a random effects model was used. Publication bias was evaluated by visual analysis of the funnel plot. Alpha was set at P<.05. Data analysis was done using Review Manager 5 Software (2010, The Cochrane Collaboration).

The following subgroup analyses were planned: type of surgery, type of ischemic stimulus, use of beta-blockers, diabetes, and use of sevoflurane or desflurane. Sensitivity analysis was conducted in all comparisons with significant heterogeneity, by systematic elimination of each study from the analysis, until results were as homogenous as possible.

Results

Our search strategy identified 23 eligible studies (Figure 1). Three articles were excluded from final analysis as they did not report relevant data on cardiac biomarkers, incidence of myocardial infarction, or creatinine.^50-52 Two meeting abstracts were eliminated due to incomplete data.^53,54 The final meta-analysis included 1371 patients, with 689 randomized to RIPC and 682 to placebo.

Table 1 summarizes the characteristics of included studies. The most common settings were CABG for chronic coronary artery disease^42,43,55-60 and open AAA repair.^61-63 CABG studies induced RIPC using a pressure cuff around the arm, which was inflated for 5 minutes, than deflated for another 5 minutes, 2-4 times. Aortic aneurysm repair studies used a protocol of 10-minute clamping and release of each iliac artery.

Patients receiving RIPC had lower levels of markers of myocardial injury in the first few days after surgery (standardized mean difference [SMD], 0.54; 95% confidence interval [CI], -1.01 to -0.08; P=.01) with highly heterogeneous results (I2 = 93%) (Figure 2) and no serious signs of publication bias (Figure 3). They also had a lower incidence of perioperative myocardial infarction (7.9% RIPC vs 13.9% placebo; relative risk [RR], 0.56; 95% CI, 0.37-0.84; P=.005; I2 = 0%) (Figure 4). The effect of RIPC on markers of myocardial injury seemed constant for all subgroups evaluated (Figure 2), especially patients undergoing cardiac surgery (Figure 5). The main source of heterogeneity identified was the study by Rahman et al.⁴² If we exclude it from the subgroup on-pump CABG, the effect size of this group rises to SMD of -0.65 (95% CI, -0.9 to -0.4), the heterogeneity though falls to I2 = 0%, and the results reach statistical significance (P<.001).