Epidemiology, ventilation, and outcomes of acute respiratory failure in immunocompromised patients from 103 intensive care units in 26 countries: a retrospective observational study

Background Acute hypoxaemic respiratory failure (ARF) is the leading cause of intensive care unit (ICU) admission among immunocompromised patients. However, contemporary data regarding the epidemiology, management, and outcomes of ARF in this population remain scarce. We aimed to identify predictors of mortality and intubation in immunocompromised patients admitted to the ICU with ARF. Methods This retrospective observational study was conducted in 103 ICUs in 26 countries. Adults (≥18 years) with ARF and immunodeficiency were eligible for inclusion. Patient data, including information on the nature of underlying immunosuppression, the cause of ARF, and the oxygenation strategy, were obtained from electronic medical records or medical charts. The primary outcome was to report 30-day mortality and identify associated factors in patients with complete data for all variables. Cox proportional hazards models were used to identify variables associated with mortality, and differences between groups were compared with χ 2 tests or two-sided Wilcoxon rank-sum tests, with p values of less than 0·05 considered significant. Findings 9854 immunocompromised patients with ARF admitted to participating ICUs between Jan 1, 2017, and Dec 31, 2023, were included in the study. The median age was 64 years (IQR 54–71); 3941 (40·0%) patients were female and 5913 (60·0%) were male. The main causes of immunodeficiency were a haematological malignancy (4759 [48·3%] of 9854 patients) or solid malignancy (3818 [38·7%] patients). Infection was the leading cause of ARF (6610 [62·0%] of 9854 patients); 5288 (53·7%) patients had more than one contributing cause of ARF, and no cause was identified in 1490 (15·1%) patients. The median partial pressure of oxygen in arterial blood (PaO 2 )/fractional concentration of oxygen in inspired air (FiO 2 ) ratio was 198 [IQR 141–208]. The 30-day mortality rate was 47·3% (4662 patients). Predictors of higher mortality were older age (hazard ratio 1·01 [IQR 1·00–1·02]), higher Charlson Comorbidity Index score (1·04 [1·01–1·07]), higher Frailty Index score (1·22 [1·16–1·28]), longer time from hospital to ICU admission (1·02 [1·01–1·03]), higher respiratory rate (1·02 [1·02–1·03]), coma at ICU admission (2·04 [1·72–2·43]), invasive fungal infection as cause of ARF (1·82 [1·45–2·28]), disease-specific infiltrates (1·73 [1·32–2·26]), unidentified cause of ARF (2·16 [1·74–2·68]), and use of vasoactive drugs (2·45 [2·10–2·86]) or renal replacement therapy (2·07 [1·74–2·48]). Protective factors included receipt of a solid organ transplant (0·62 [0·49–0·79]), systemic vasculitis or connective tissue disease (0·61 (0·47–0·78]), higher PaO 2 /FiO 2 ratio (0·78 [0·72–0·84]), receipt of high-flow nasal oxygen therapy (0·78 [0·64–0·95]), and cardiogenic pulmonary oedema (0·67 [0·51–0·89]). Interpretation In this large international cohort of immunocompromised patients with ARF, we identified key risk and protective factors for mortality and intubation. These findings could improve outcomes by informing timely clinical decisions, goals-of-care discussions, and management in this vulnerable population. Funding Kirsten and Freddy Johansen Foundation and Groupe de Recherche en Réanimation Onco-Hématologique.