Abstract
Non-invasive, real-time monitorable indicators for early assessment of sepsis-associated myocardial injury (SMI) are still lacking. We aimed to develop non-invasive, real-time indicators for early assessment of SMI using bedside heart rate (HR) and diastolic arterial pressure (DAP) monitors. In this multi-center cohort study, piece-wise exponential additive mixed models were used to estimate the exposure window and time fraction of the hazardous exposure proportion, and secondarily to analyze the exposure characterization on this basis to identify high-risk exposure pattern. In total, 20,043 patients were finally included; we found that SMI patients had the highest survival rate when HR was <90 bpm or DAP was between 50 and 70 mmHg. Further investigation revealed that the SMI high-risk exposure pattern was the H(1)D(-1) (HR ≥ 90 bpm and DAP ≤ 50 mmHg, exposure proportion > 0.3 and 0.2, respectively, and exposure window on admission day 1). H(1)D(-1) exposure pattern using glucocorticoids significantly increased the risk of mortality in H(1)D(-1). Validation against various methodologies and data sources demonstrated acceptable consistency.