A PREDICTIVE MODEL TO IDENTIFY CLOSTRIDIUM-DIFFICILE TOXIN IN HOSPITALIZED-PATIENTS WITH DIARRHEA

Objectives: Although Clostridium difficile is a common pathogen in hos pitalized patients with diarrhea, no study has attempted to develop a predictive model to estimate the likelihood of C. difficile positivity . Methods: We conducted a cross-sectional study at a single hospital o f 271 patients with diarrhea for whom stool was tested for C. difficil e toxin. The sample was randomly divided into a subset to derive the m odel (n = 180) and another to validate it (n = 91), and independent pr edictors of toxin positivity were identified using logistical regressi on analysis, Results: C. difficile toxin was present in 81 patients an d absent in 190. In the derivation set, a positive toxin was independe ntly predicted (p < 0.0005) by readmission within 2 wk of prior hospit alization, by a white blood cell count greater than or equal to 10,000 /mm(3), and by presence of fecal leukocytes. In the validation set, C. difficile toxin was present in 24, 29, and 77% of patients with 0, 1, and greater than or equal to 2 risk factors, respectively. Conclusion : If validated prospectively and/or in other centers, the model could identify patients who should be considered for empirical management wh ile awaiting results of toxin assays.