Clinical characteristics and risk factor analysis of Pneumocystis jirovecii pneumonia in patients with CKD: a machine learning-based approach.

Abstract

Patients with chronic kidney disease (CKD) who are being treated with immunosuppressive medications are at risk for developing Pneumocystis jirovecii pneumonia (PCP). We attempted to characterize the clinical aspects of PCP in CKD patients in order to alert high-risk patients with bad prognosis. A retrospective study of CKD patients was conducted from June 2018 to June 2022. Based on PCP diagnostic criteria, these patients were divided into PCP and non-PCP groups. Using univariate and multivariate logistic regression analysis, risk indicators were evaluated, and nomogram and decision tree were developed. Of the CKD patients screened for Pneumocystis carinii nucleic acid, 1512 were included. Two-hundred forty four (16.14%) were diagnosed with PCP. Of the PCP, 88.5% was receiving glucocorticoid (GC) therapy, of which 66.3% received more than 0.5 mg/kg GC. Multivariate analysis showed that membranous nephropathy (OR 2.35, 95% CI 1.45-3.80), immunosuppressive therapy (OR 1.94, 95% CI 1.06-3.69), and ground glass opacity of CT scanning (OR 1.71, 95% CI 1.10-2.65) were associated with increased risk of Pneumocystis carinii infection. The AUC of nomogram based on logistics regression was 0.78 (0.75-0.81). The mortality in patients with PCP was 32.40%. Univariate analysis and decision tree showed that pulmonary insufficiency (PO2: OR 0.98, 95% CI 0.96-1.00), elevated APTT (OR 1.07, 95% CI 1.04-1.11), and reduced hemoglobin (OR 0.97, 95% CI 0.96-0.98) were associated with poor prognosis. PCP is not rare in CKD patients, particularly in those treated with immunosuppressive therapy. Considering the high mortality of the cases, further studies on the prevention and management of these patients are needed.