1. External validation of two prediction tools for patients at risk for recurrent Clostridioides difficile infection
- Author
-
Yvette H. van Beurden, Christina M. J. E. Vandenbroucke-Grauls, Olaf M. Dekkers, Laura J. van Dijk, Tessel M. van Rossen, Martijn W. Heymans, Medical Microbiology and Infection Prevention, AII - Infectious diseases, Epidemiology and Data Science, APH - Methodology, APH - Personalized Medicine, Amsterdam Gastroenterology Endocrinology Metabolism, and Gastroenterology and hepatology
- Subjects
0301 basic medicine ,medicine.medical_specialty ,recurrence ,business.industry ,Clostridioides difficile ,030106 microbiology ,Gastroenterology ,External validation ,prediction models ,prognostic factors ,Clostridium difficile ,03 medical and health sciences ,0302 clinical medicine ,Internal medicine ,medicine ,risk factors ,030212 general & internal medicine ,business ,Clostridioides - Abstract
Background: One in four patients with primary Clostridioides difficile infection (CDI) develops recurrent CDI (rCDI). With every recurrence, the chance of a subsequent CDI episode increases. Early identification of patients at risk for rCDI might help doctors to guide treatment. The aim of this study was to externally validate published clinical prediction tools for rCDI. Methods: The validation cohort consisted of 129 patients, diagnosed with CDI between 2018 and 2020. rCDI risk scores were calculated for each individual patient in the validation cohort using the scoring tools described in the derivation studies. Per score value, we compared the average predicted risk of rCDI with the observed number of rCDI cases. Discrimination was assessed by calculating the area under the receiver operating characteristic curve (AUC). Results: Two prediction tools were selected for validation (Cobo 2018 and Larrainzar-Coghen 2016). The two derivation studies used different definitions for rCDI. Using Cobo’s definition, rCDI occurred in 34 patients (26%) of the validation cohort: using the definition of Larrainzar-Coghen, we observed 19 recurrences (15%). The performance of both prediction tools was poor when applied to our validation cohort. The estimated AUC was 0.43 [95% confidence interval (CI); 0.32–0.54] for Cobo’s tool and 0.42 (95% CI; 0.28–0.56) for Larrainzar-Coghen’s tool. Conclusion: Performance of both prediction tools was disappointing in the external validation cohort. Currently identified clinical risk factors may not be sufficient for accurate prediction of rCDI.
- Published
- 2021