Risk of bias in meta-analysis on erythropoietin-stimulating agents in heart failure