Prediction Model for Severe Community-acquired Pneumonia Development among Patients with Diabetes Mellitus
Introduction: Diabetes is an independent risk factor for the development of severe community-acquired pneumonia (CAP) and associated with pneumonia-related hospitalization as well as mortality. Here, we assessed several selected biomarkers to determine their predictive value for progression to severe CAP among diabetic patients. Research design and methods: A retrospective cohort study of diabetic patients with CAP was conducted at a tertiary teaching hospital. The prediction model group (N=100) comprised patients registered between January 2015 to December 2016. Multivariate analysis was performed to identify predictive biomarkers from this cohort. The validation group (N=108) comprised the patients between January 2017 to February 2019. Predictive performance was assessed in the validation group. Results: A total of 208 diabetic inpatients with CAP were recruited. Further multivariate analysis showed that C-reactive protein (CRP), absolute lymphocyte count (ALC), immunoglobulin (IgM), and HbA1C at admission were independently associated with progression to severe CAP in diabetic patients during hospitalization. The prediction model =0.0179555*CRP+1.975918* HbA1C-2.879364* ALC -0.026255* IgM -8.220555. The area under ROC (AUROC) curve in the validation group was 0.851 (0.781-0.921) with statistical significance (P<.05). Conclusions: In diabetic patients with CAP, a combination of CRP, ALC, IgM, and HbA1C at admission could be used to predict the progression to severe CAP.