The authors reply.

We thank Prof. Dandel (1) for his many insightful comments with regard to our article (2). We agree that the prevalence and severity of tricuspid regurgitation (TR) are integral to interpreting right ventricular dysfunction (RVD) in acute respiratory distress syndrome (ARDS). This is included in Supplementary Table 2 in (2): approximately one-third of patients had none, 60% had mild, and 6% had moderate TR. There were no patients with severe TR. There was a trend toward an increased prevalence of moderate TR in nonsurvivors compared with survivors (10% vs 4%), although this was not statistically significant. We acknowledge that the presence of TR can lead to an overestimation of RV fractional area change and create false negatives in diagnosing RV systolic impairment. We were unable to provide evidence of this in the current study.

We also agree that it is important to interpret RV systolic function relative to pulmonary artery systolic pressure. The severity of pulmonary arterial hypertension (PASP) can vary between patients due to varying degrees of hypoxic pulmonary vasoconstriction, hypercarbia, acidosis, microangiopathy, pulmonary emboli, and ventilator settings. With the Anrep and Frank Starling effect, an acute increase in PASP is coupled to an increase in RV contractility. When this process is uncoupled, RV forward flow is reduced. A marker of RV:PA uncoupling was independently associated with mortality in patients with COVID-19 ARDS (3), and this warrants prospective evaluation in larger critically ill cohorts. Given that PASP varies among patients, measuring RV systolic function relative to PASP is likely to provide additional useful information relating to RV forward flow and prognostication. Unfortunately, we were unable to measure PASP reliably in all patients and, therefore, unable to measure the degree of RV:PA coupling.

We wholeheartedly agree that assessment of RV function is important in critically ill patients including those with COVID-19 ARDS. However, it is not possible to elucidate whether RVD directly contributes to mortality or is merely associated with death due to its correlation with disease severity/comorbid illness (4). As such, it is unknown whether efforts to improve RVD will translate to patient benefit. It is also unknown which therapies should be investigated to improve RVD: should these focus on reducing RV afterload with the use of pulmonary vasodilators, optimizing RV preload and reducing systemic congestion with the use of diuretics, and improving RV contractility with the use of inotropic agents, heart rate optimization, or mechanical circulatory support/devices (5). The use of prone positioning and RV protective ventilator settings on improving RV function also warrants investigation. Although these therapies have physiologic equipoise, we are unaware of any randomized trials demonstrating benefit of these treatments in patients with RVD in ARDS. With so many unknowns, defining and treating RVD in critically ill patients is an important area for further research. We hope to aid these efforts by providing a common and prognostically enriched RV phenotype (RV dilation with systolic impairment) that, if validated in other populations (6), may be used as a target in randomized trials of interventions.