Objective:To compare the clinical characteristics and analyze the prognostic factors between human immunodeficiency virus (HIV)-infected patients and non-HIV-infected immunocompromised patients with pneumocystis pneumonia (PCP) complicated with acute respiratory failure (ARF) in intensive care unit (ICU).Methods:The clinical data of patients with PCP complicated with ARF admitted in ICU of The First Affiliated Hospital of Zhengzhou University and The Sixth People′s Hospital of Zhengzhou City between May 2018 and October 2020 were retrospectively reviewed. All subjects were divided into HIV-infected group and non-HIV-infected immunocompromised group. General characteristics and underlying diseases of patients in the two groups were analyzed. Laboratory parameters, treatment and outcomes between two groups were compared. Independent sample t test, Mann-Whitney U test and chi-square test were used for statistical analysis, and univariate and multivariate logistic regression models were used to identify the risk factors for the clinical outcome. Results:A total of 129 PCP complicated with ARF patients were enrolled, including 75 HIV-infected patients and 54 non-HIV-infected immunocompromised patients. Only 10.7%(8/75) patients of HIV-infected group received anti-retroviral therapy (ART), but none of the patients in either groups had previously received trimethoprim-sulfamethoxazole (TMP-SMX) for PCP prophylaxis. Acute physiology and chronic health evaluation (APACHE) Ⅱ score of HIV-infected group was 18.7±6.0, which was higher than that in non-HIV-infected immunocompromised group (13.1±4.4) when admitted in ICU ( t=-5.45, P<0.001). Hypoproteinemia was common in both groups. Ninety-six percent (72/75) of HIV-infected patients had CD4 + T lymphocyte counts lower than 200/μL and 84.0%(63/75) of patients had CD4 + T lymphocyte counts even lower than 50/μL, while 5.74%(31/54) of patients in non-HIV-infected immunocompromised group had CD4 + T lymphocyte counts lower than 200/μL. The CD4 + /CD8 + T lymphocyte counts ratio was 0.05(0.02, 0.12) in HIV-infected group, which was lower than that in non-HIV-infected immunocompromised group (0.96(0.64, 1.44)), and the difference was statistically significant ( Z=-9.16, P<0.001). The length of ICU stay and hospital stay of non-HIV-infected immunocompromised patients were 10.0(7.0, 14.0) days and 18.0(11.8, 32.5) days, respectively, which were both longer than those in HIV-infected patients (7.0(4.0, 9.0) days and 13.0(7.0, 23.0) days, respectively), and the differences were both statistically significant ( Z=-3.58 and -2.73, respectively, both P<0.050). The hospital mortality of HIV-infected patients was 57.3%(43/75), which was significantly higher than that in non-HIV-infected immunocompromised patients (38.9%, 21/54) ( χ2=4.27, P=0.039). Multivariable logistic regression identified that lactic dehydrogenase (LDH), C-reactive protein (CRP) and APACHE Ⅱ score were the risk factors for the clinical outcome of HIV-infected patients (odds ratio ( OR)= 1.006, 1.015 and 1.736, respectively, all P<0.050). The partial pressure of oxygen in arterial blood/fractional concentration of inspiratory oxygen (PaO 2/FiO 2), LDH and CD4 + T lymphocyte counts were the risk factors for the clinical outcome of non-HIV infected immunocompromised patients ( OR=0.970, 1.008 and 0.989, respectively, all P<0.050). Conclusions:PCP patients with ARF are critically ill with high mortality rate. LDH, CRP and APACHEⅡscore are predictors for prognosis of HIV-infected patients with PCP, while PaO 2/FiO 2, LDH and CD4 + T lymphocyte counts are predictors for prognosis of non-HIV infected immunocompromised patients with PCP.