Waterlow score to predict patients at risk of developing Clostridium difficile-associated disease.

This study describes the development and testing of an assessment tool to predict the risk of patients developing Clostridium difficile-associated disease (CDAD). The three phases of the study include the development of the tool, prospective testing of the validity of the tool using 1468 patients in a medical assessment unit and external retrospective testing using data from 29 425 patients. In the first phase of the study, receiver operating characteristic (ROC) analysis identified the Waterlow assessment score as having the ability to predict CDAD (area under the curve: 0.827). The Waterlow tool was then tested prospectively with 1468 patients admitted to a medical assessment unit. A total of 1385 patients (94%) had a Waterlow score <20 and 83 patients (6%) had a Waterlow score of > or = 20. After a three-month follow-up, six patients in the low Waterlow score group developed CDAD (0.4%) and 14 patients in the high score group developed CDAD (17%). The sensitivity and specificity of the Waterlow score to predict the risk of developing CDAD were 70% and 95%, respectively. Similar results were obtained when the tool was tested retrospectively on a large external patient data set. The Waterlow score appears to predict patients' risk of developing CDAD and although it did not identify all cases, it highlighted a small group of patients who had a disproportionately large number of CDAD cases. The Waterlow score can be used to target patients most at risk of developing CDAD.