Statistical analysis

Normoxemic patients with AHF were divided into the oxygen therapy group and the ambient-air group. If the measurement data were normally distributed and the variance was homogeneous, data were characterized as mean ± standardized differences (SD) and compared between groups using a Student’s t test. If the requirements were not satisfied, data were represented by median [interquartile range (IQR)], and the Kruskal Wallis rank test was used for comparisons between groups. Numeration data were expressed as absolute values and percentages; the Pearson’s χ² test or Fisher’s exact test was chosen for statistical analyses as appropriate.

Given the observational nature of the current study, propensity score matching (PSM) was used to minimize the effect of potential confounders. A logistic regression model was constructed to calculate and assign each patient a propensity score, which was defined as the likelihood of being exposed to an intervention. Next, 1:1 matching (the oxygen therapy group vs. the ambient-air group) without replacement was performed using a nearest neighbor matching algorithm, with a fixed caliper width of 0.05. The standardized mean difference (SMD) was calculated to evaluate the efficiency of PSM in reducing the differences between the two groups.

In the pre-PSM and post-PSM cohorts, logistic regression models were employed to investigate associations between oxygen therapy and clinical outcomes adjusting for confounding variables selected based on P <  0.05 in the univariate analysis, in which the Akaike information criterion was applied as the selection criteria of the optimal model. Linear regression was used to assess the correlation of oxygen therapy with length of stay, and the odds ratios (ORs) were presented using the formula OR = e^βi. A series of subgroup analyses were performed to further assess the association between oxygen therapy and all-cause in-hospital mortality, including duration of oxygen therapy, median SpO₂ during hospitalization, age, De Nova AHF, history of atrial fibrillation, history of myocardial infarction, history of stroke, history of hypertension, history of chronic kidney disease, and renal replacement treatment. In the subgroup analyses of patients’ duration of oxygen therapy, the reference group was defined as all the patients of the ambient-air group. In the other subgroup analyses, the reference group was defined as the patients of the ambient-air group in the corresponding subgroup.

A two-tailed P value of less than 0.050 was considered to be statistical significance. All statistical analyses were performed using SPSS software (version 22.0; IBM Corporation, St. Louis, Missouri, USA) and R software (version.3.6.1,The R Project for Statistical Computing, TX, USA; http://www.r-project.org).