Prevalence of Left Ventricular Diastolic Dysfunction and Prognostic Value in Septic Shock Patients According to Different Guidelines: Comparison Between Contemporary Cardiological Algorithms and Simplified Approaches
Filipe A. Gonzalez, Érico Costa, Deborah Rondinella, Filippo SanfilippoABSTRACT
Background
Diagnosis of left ventricular diastolic dysfunction (LVDD) in the context of septic shock is challenging. We aimed to compare different algorithms applied to septic patients.
Methods
Secondary analysis of a prospective, single‐center study of 98 consecutive adult septic shock patients undergoing comprehensive speckle‐tracking echocardiography within 24 h of ICU admission. We studied the prevalence and the prognostic value of five LVDD diagnostic algorithms:2016‐American Society of Echocardiography (ASE)/European Association CardioVascular Imaging(EACVI), 2024‐British Society of Echocardiography(BSE), 2025‐ASE, Lanspa‐simplified criteria, and a further Lanspa‐modified version hereby proposed (adding LVDD diagnosis when LV ejection fraction< 50%, as in 2016‐ASE‐EACVI guidelines). Logistic regression and Kaplan‐Meier curves were used to explore the association between algorithms and 30‐day mortality.
Results
The prevalence of LVDD on day 1 varied widely, ranging from 32.7%(Lanspa) to 72.4%(2024‐BSE). Concordance between the three cardiology guidelines ranged from slight to moderate: the lowest agreement among them was observed between the 2016‐ASE/EACVI and the 2025‐ASE criteria(55.1%, Kappa 0.10), whereas the highest agreement across all five algorithms was seen between the 2016‐ASE/EACVI and the Lanspa‐modified criteria(88.8%, Kappa 0.77). When correcting for severity score, need for invasive ventilation, and norepinephrine dose, independent prognostic values for 30‐day mortality were observed only for the 2025‐ASE, the Lanspa and the Lanspa‐modified algorithms, but all algorithms had similar predictive performances.
Conclusions
The prevalence of LVDD in septic shock is highly dependent on the diagnostic algorithm utilized. Although only the 2025‐ASE, Lanspa, and Lanspa‐modified algorithms demonstrated independent prognostic value after multivariable adjustment, prognostic discrimination for 30‐day mortality was moderate and comparable across all approaches.