632 followers
E-values are really for causal models and confounding. Given the above study focuses on prognosis/prediction using stepwise model selection rather than a confounder selected causal model (e.g. www.atsjournals.org/doi/10.1513/...), would not consider e-v…