Analysis Clinical Characteristics,development And Validation Of Clinical Models For Predicting The Prognosis Of Non-HIV Pneumocystis Pneumonia Patients

ObjectivesIn those non-HIV(Human Immunodeficiency Virus)patients with Pneumocystis jirovecii Pneumonia(PJP):(1)describe the clinical characteristics of patients and analyze the risk factors related to ICU admission and death;(2)develop two clinical model to predict patients’mortality risk and ICU admission risk;(3)validation of models.MethodsThe study retrospectively collected the clinical data of all PJP patients diagnosed in Peking Union Medical College Hospital between January 2012 to December 2021.The clinical characteristics of the included population were described by univariate analysis.The Least absolute shrinkage and selection operator and multivariate logistic regression were used to screen the risk factors related to patients’death and ICU admission.Based on these factors,two clinical models were developed to predict the risk of ICU admission and death.Those models were transferring into the nomogram.Also,internal and external validation were conducted.The prediction accuracy of the two models was evaluated by calculating the area under the receiver characteristic curve(AUC)and drawing the calibration curves.ResultsIn total,508 patients were included.All of PJP patients had underlying diseases.There were 423 non-HIV PJP patients including in the development cohort,and the number of patients with rheumatic diseases was the most(n=198,47%).When analyzing the risk-factor of ICU admission,the results revealed that dyspnea,rale in the lung,respiratory rate,pleural effusion,BUN,C-reactive protein to albumin ratio,and LDH were the independent risk factors for ICU admission.Those seven variables were included in the clinical model to predict the risk of ICU admission.The under area of ROC curve of the model was 0.880(95%CI:0.806-0.955)and the model was visualized as a nomogram.There were 8 variables in the clinical model of predicting the risk of death,including age,chronic lung disease,respiratory rate,LDH,BUN,CMV infection,shock and invasive mechanical ventilation.The model was steady and the area under the ROC curve of was 0.915(95%CI:0.8310.997).The model was visualized as a nomogram.ConclusionAll non-HIV PJP patients have underlying disease.The risk of ICU admission and death were high in the non-HIV PJP patients.Based on the information from those models,the doctors could choose different treatment and monitoring levels for those PJP patients with different risks,and finally reduce the mortality of PJP patient.