To the Editor: Although the concept of a prolonged QRS duration having prognostic value, as raised in the study by Dr Wang and colleagues,1 is both logical and appealing, there are a number of features of this study that require it to be interpreted with caution.
First, the authors chose to retain covariates in the multivariate model if they met P < .05 for all-cause mortality. This is not an optimal approach,2 particularly in observational studies in which it is critical to include all covariates that are clinically important. The presentation is unclear regarding which covariates were actually used and whether the same method for covariate selection was applied to all analyses.
QRS Duration in Patients Hospitalized for Worsening Heart Failure. JAMA. 2008;300(16):1879-1881. doi:10.1001/jama.2008.503