Despite the increasing use of high-flow nasal cannulas (HFNCs) to treat critically ill patients, data on their effectiveness for sepsis patients remains very limited. We studied a prospective cohort of sepsis patients from the Korean Sepsis Registry (18 intensive care units (ICUs)). Patients started on HFNC therapy for hypoxemia within the first three ICU days were enrolled. HFNC failure was defined as intubation or ICU death, and the primary outcome was early HFNC failure occurring within 72 h of HFNC initiation. Of 901 patients with sepsis admitted to the ICU, 206 who received HFNC therapy were finally included (117 with pneumonia vs. 89 with non-pneumonia sepsis; median age, 71.0 (63.0-78.0) years; PaO2/FiO2 ratio, 160.2 (107.9-228.2) mm Hg; septic shock, n = 81 (39.3%)). During HFNC therapy, 72 (35.0%) patients were intubated and 51 (24.8%) died. HFNC failure developed in 95 (46.1%) patients, and among them, early failure rate was 85.3% (81/95). On multivariate analysis, an immunocompromised state (odds ratio (OR) = 2.730), use of a combination of antibiotics (OR = 0.219), and the PaO2/FiO2 ratio (OR = 0.308) were significantly associated with early HFNC failure in pneumonia sepsis patients. However, in non-pneumonia sepsis patients, lactate levels (OR = 1.532) were significantly associated with early HFNC failure. In conclusion, a high proportion of sepsis patients experience HFNC failure, usually within 72 h after therapy initiation, which emphasizes the importance of close monitoring. Furthermore, unlike in pneumonia sepsis, organ failure (i.e., lactate) might serve as a prognostic marker in non-pneumonia sepsis (i.e., type IV respiratory failure).
Keywords: high flow nasal cannula; intubation; outcomes; sepsis.