Comments on Right Ventricular Dysfunction Score in the Prognostic Evaluation of Patients with Sepsis and Septic Shock

Ventricular Dysfunction Score in Septic Shock: Points Warranting Further Discussion" by Kasinathan et al., evaluating a five-parameter right ventricular dysfunction (RVD) score and the tricuspid annular plane systolic excursion (TAPSE)/pulmonary artery systolic pressure (PASP) ratio as predictors of 28-day mortality in septic shock, reporting excellent discrimination for both [Area under the receiver operating characteristic curve (AUROC): 0.857 and 0.883, respectively]. 1 The authors are to be commended for extending a comprehensive RV assessment tool into a critically ill sepsis cohort.A few methodological points, however, merit further discussion.First, both the RVD score and the reference-standard echocardiographic measurements were acquired and interpreted by a single unblinded investigator.1 Given that RV assessment, particularly TAPSE, PASP, and inferior vena cava (IVC) collapsibility, is known to exhibit substantial inter-observer variability, the absence of a second blinded reader introduces the possibility of assessment bias, especially since the same investigator's clinical impression of illness severity could plausibly influence the measurement.Second, the numerically higher AUROC for TAPSE/PASP (0.883) than for the RVD score (0.857) was interpreted as equivalent performance, based on a non-significant DeLong test (p = 0.093). 1 With only 189 patients and 75 events, this comparison may be underpowered to detect a real difference.We would caution against concluding "no difference" from a non-significant result in a modestly sized cohort, a caveat borne out in larger meta-analyses of RV function as a mortality predictor, in which pooled effect sizes vary appreciably across component measures.2 These concerns do not detract from the value of a standardized, multiparametric approach to RV assessment in septic shock, an area where consensus on effect size and definitions remains limited.3 External, multicenter validation with blinded, dual-reader assessment would considerably strengthen confidence in these promising cut-offs.