A new score to predict Clostridioides difficile infection in medical patients: a sub-analysis of the FADOI-PRACTICE study

Medical divisions are at high risk of Clostridioides difficile infection (CDI) due to patients' frailty and complexity. This sub-analysis of the FADOI-PRACTICE study included patients presenting with diarrhea either at admission or during hospitalization. CDI diagnosis was confirmed when both enzyme immunoassay and A and B toxin detection were found positive. The aim of this sub-analysis was the identification of a new score to predict CDI in hospitalized, medical patients. Five hundred and seventy-two patients with diarrhea were considered. More than half of patients was female, 40% on antibiotics in the previous 4 weeks and 60% on proton pump inhibitors (PPIs). CDI diagnosis occurred in 103 patients (18%). Patients diagnosed with CDI were older, more frequently of female sex, recently hospitalized and bed-ridden, and treated with antibiotics and PPIs. Through a backward stepwise logistic regression model, age > 65 years, female sex, recent hospitalization, recent antibiotic therapy, active cancer, prolonged hospital stay (> 12 days), hypoalbuminemia (albumin < 3 g/dL), and leukocytosis (white blood cells > 9 x 10<^>9/L) were found to independently predict CDI occurrence. These variables contributed to building a clinical prognostic score with a good sensitivity and a modest specificity for a value > 3 ( 79% and 58%, respectively; AUC 0.75, 95% CI 0.71-0.79, p < 0.001), that identified low-risk (score <= 3; 42.5%) and high-risk (score > 3; 57.5%) patients. Although some classical risk factors were confirmed to increase CDI occurrence, the changing landscape of CDI epidemiology suggests a reappraisal of common risk factors and the development of novel risk scores based on local epidemiology.