RESUMO
Treatment guidelines for Clostridium difficile infection (CDI) are limited by a lack of widely accepted clinical prediction tools (CPTs). Two published CPTs, the Velazquez-Gomez Severity Score Index (VGSSI) and ATLAS scores, were evaluated, and variables showing the greatest correlation with mortality in patients with CDI were identified to further develop an objective, mortality-based CPT. A retrospective review of the charts of 271 hospitalized patients with CDI was performed. VGSSI and ATLAS scores were assigned. Means and correlations of these scores with mortality were evaluated. Multivariate logistic regression analysis was performed on 32 known potential mortality predictor variables. Mortality was overall strongly associated with VGSSI and ATLAS scores with poor correlation within the intermediate ranges. Mean scores for nonsurvivors indicated poor calibration. The variables most associated with mortality were Age, vasopressors, steroids, creatinine level, and albumin. Although both CPTs revealed the ability to discriminate patients at greater risk for mortality, precision and overall calibration were lacking. Five variables were identified which had the greatest correlation with mortality. Utilization of these variables to enhance or modify the existing CPTs is suggested as the next step in the development of a useful and accurate mortality-based CPT for the treatment of CDI.
