Université de Lille Predictors of ventilatory supports failure in patients with interstitial lung diseases admitted to intensive care unit Interstitial lung disease intensive care unit acute respiratory failure ventilatory supports failure prognostic bedside indices 610 Pneumopathie infiltrante diffuse Complications (médecine) Insuffisance respiratoire aigüe Oxygénothérapie nasale à haut débit Respiration artificielle Échec thérapeutique (médecine) Soins intensifs en pneumologie Pronostic (médecine) Pneumopathies interstitielles complications Insuffisance respiratoire Oxygénothérapie Ventilation artificielle Échec thérapeutique Unité de soins intensifs respiratoires Pronostic Background: Acute respiratory failure (ARF) is a frequent and severe complication in patients with interstitial lung diseases (ILD) admitted to intensive care unit (ICU), and is associated with high short- and long-term mortality. High-flow nasal oxygen (HFNO) and invasive mechanical ventilation (MV) are commonly used and reliable bedside prognostic markers to predict ventilatory supports failure are lacking. Objectives: To evaluate the prognostic value of four readily available respiratory indices—the relationship of Respiratory Rate-Oxygenation (ROX index), the ratio of pulse oximetry oxygen saturation to the fraction of inspired oxygen (SpO?/FiO?), the ventilatory ratio (VR), and the mechanical power (MP)—to predict ventilatory supports failure in ICU patients with ILD. Methods: We conducted a retrospective, single-center observational study including all adult ILD patients admitted to ICU for ARF between January 2013 and November 2024. HFNO failure was defined as a composite endpoint of 30-day mortality or need for invasive MV. MV failure was defined as 30-day mortality among intubated patients. ROX index and SpO?/FiO? ratio were assessed at 0, 24, and 48 hours during HFNO therapy, while VR and MP were evaluated at the same time points during invasive MV. Discriminative performance was assessed using receiver operating characteristic curves with area under the curve (AUC) values, Youden’s index, and survival analyses. Multivariable analyses were adjusted for decisions regarding limitations of life-sustaining therapies given sample size constraints. Results: A total of 100 ILD patients were included, 71 received HFNO (25 subsequently undergoing tracheal intubation) and 54 required invasive MV. The overall 30-day mortality rate was 41%. Among HFNO patients, 57% experienced failure of HFNO therapy. ROX index ? 7.07 at 48 hours was associated with a significantly lower risk of HFNO failure (AUC = 0.81). SpO?/FiO? ratio ? 181.9 at 48 hours was significantly associated with a decreased risk of HFNO failure (AUC = 0.79). After adjustment for treatment-limitation decisions, both indices retained independent prognostic value. Among mechanically ventilated patients, VR and MP were significantly higher at 24 hours in the MV failure group. VR at 24 hours showed fair discriminative performance (AUC = 0.72), with a threshold of 1.89 associated with an approximately fourfold increased risk of MV failure. MP at 24 hours demonstrated acceptable discrimination (AUC = 0.73), with a threshold of 31.38 J/min associated with a fourfold increased risk of MV failure. These associations remained robust after adjustment for limitations of life-sustaining therapies. Conclusions: In critically ill patients with ILD, simple bedside respiratory indices provided clinically relevant prognostic information. ROX index and SpO?/FiO? ratio at 48 hours could early facilitate and accurate prediction of HFNO therapy failure. In mechanically ventilated patients, VR and MP measurements at 24 hours were significantly associated with 30-day mortality. The study results should be validated in larger prospective studies to assess their clinical uses. Electronic Thesis or Dissertation Text en application/pdf 1 : 0 Ko https://pepite-depot.univ-lille.fr/LIBRE/Th_Medecine/2026/2026ULILM014.pdf Lacombe Charlotte 1995-11-14 FR 295101571 2026ULILM014 2026-01-23 Médecine. Anesthésie Réanimation Université de Lille (2022-...) thesis.degree.grantor_1 259265152 Doctorat de médecine non oui Robriquet Laurent intervenant_1 060740140 ddc:610 Robriquet Laurent Université de Lille (2022-...) Université de Lille Unicode PDF 50