Performance of severity of illness classification for Clostridium difficile infection to predict need-for-colectomy...

In current state of practice, disease severity assessment for Clostridium difficile infection (CDI) lacks consensus between different organizations. In the latest guidelines for management of CDI, authors have proposed a new disease severity classification. This classification has been derived from expert opinion and includes previously reported patient related factors that predict unfavorable outcome in CDI.

To evaluate the performance of new disease severity classification to predict CDI-related colectomy or in-patient deaths.

This observational study was performed at a 300-bed community hospital catering to a suburban population. All the adult patients discharged during October 2005 through September 2012 and diagnosed with CDI were included in the study. Cases of CDI were identified using the billing database. Demographic and clinical characteristics of CDI cases were extracted from medical chart reviews performed by two physician researchers. Cases were classified as mild-to-moderate, severe, or severe and complicated CDI. Major outcomes measured were in-patient deaths and colectomy attributed to CDI. For risk stratification, each variable of severe and complicated CDI was counted.

In total, 59,897 patients were discharged from our hospital during the study period; 894 of them were diagnosed with CDI. Mean age of CDI cases was 74 years (standard deviation 15 years), 345 (39%) were male, and median length of hospital stay was 7 days (interquartile range 4-13). One hundred ninety-one patients (21%) were classified as mild-to-moderate, 93 patients (10%) as severe, and 610 patients (68%) as severe and complicated. In total, 14 patients underwent colectomy for CDI and 53 patients expired. In 35 of these patients, the cause of death was thought to be CDI. The combined outcome (CDI-related colectomy and/or death in hospital) occurred in 47 patients. None of the patients in mild-to-moderate disease or severe disease had combined outcome. On severe and complicated cases, as the score increased, rate of combined outcome increased significantly (z-12.7, p<0.0001). On plotting receiver-operating curve, the simple variable count had area under the curve of 0.91.

The newly proposed disease severity classification for CDI categorized more than two-thirds of patients as severe and complicated. Increase in number of severe and complicated classifying variables increases the chance of adverse outcomes significantly. Patients meeting only one variable could be potentially treated as a severe case instead of severe and complicated. This data needs to validated prospectively before could be used in clinical practice.