Objective  To explore the etiological characteristics of hospital-acquired pneumonia (HAP) in intensive care unit (ICU), analyze risk factors and construct the prediction model.

  Methods  Fifty HAP patients in ICU were included in infection group, while another 70 ICU patients without HAP during the same period were included in non-infection group. The etiological characteristics of infection in the infection group were analyzed. The risk factors of HAP were screened by Logistic regression analysis. The risk prediction model was constructed based on the regression coefficients of risk factors. The discrimination of the prediction model was assessed by receiver operating characteristics (ROC) curves.

  Results  There were 73 positive strains isolated from the 50 patients in the infection group, including 16 strains of Gram-positive bacteria (21.92%), 51 strains of Gram-negative bacteria (69.86%) and 6 strains of fungi (8.22%). The detection rate of Gram-negative bacteria was higher than that of the other types (P < 0.05). Logisitic regression analysis showed that age of 60 years and above, diabetic history, consciousness disorder, mechanical ventilation, history of chronic obstructive pulmonary disease and stay time in ICU over 7 d were independent risk factors of HAP (P < 0.05). The area under the ROC curve (AUC) predicted by the risk model for HAP in ICU patients was 0.855.

  Conclusion  The main pathogens of HAP infection are Gram-negative bacteria in ICU patients. The constructed risk model has better discrimination, which can be applied to identify high-risk groups susceptible to HAP.