Development and validation of a clinical prediction model for Clostridioides difficile associated diarrhea

Abstract

Introduction: The aim of this study was to develop and validate a clinical prediction model for Clostridioides difficile associated diarrhea (CDAD) based on routine laboratory tests.

Methodology: Data from 121 CDAD patients and 123 patients with non-CDAD who presented at the First Affiliated Hospital of Nanjing Medical University between May 2017 and January 2022 were used to create a nomogram based on logistic regression. In addition, 109 stool samples from diarrhea patients in Jurong People's Hospital were collected to detect Clostridioides difficile toxin genes. The performance of the prediction model was assessed by the area under the curve (AUC), Hosmer-Lemeshow goodness of fit, and decision curve analysis (DCA).

Results: The following variables were included in the new multivariate regression model: white blood cell (WBC), lymphocyte (LY), hemoglobin (HGB), mean corpuscular volume (MCV), activated partial thromboplastin time (APTT), D-dimer, urea, creatinine (Cr), and uric acid (UA). The AUC of the prediction model was 0.793 (95% CI = 0.737–0.849) for the derivation sets and 0.708 (95% CI = 0.506–0.910) for the validation set. The calibrated values were 0.874 and 0.543, respectively. The nomogram showed better net benefit when prediction probability values were above 0.1 in the DCA curve.

Conclusions: A new diagnostic prediction model for CDAD was established. Clinicians can use the nomogram to initially assess the likelihood of CDAD when the patient suffers diarrhea, to ensure timely specific laboratory tests, and appropriate diagnostic and treatment measures.