Optimizing a Clostridium difficile Screen for Intensive Care Unit Admissions

Background: Clostridium difficile infection is one of the most common healthcare-associated infections. Careful screening upon hospital admission enables clinicians to mitigate the spread and sequelae of this illness through early intervention and isolation but this screening must be accurate and highly sensitive to maximize benefit.

Local problem: The site for this project, an ICU in Tennessee, currently employs a highly sensitive but nonspecific tool to detect CDI. This has resulted in low accuracy, low staff compliance, low patient morale, and unnecessarily high personal protective equipment (PPE) costs.

Methods: Levin’s evidence-based practice improvement model was chosen to guide this project from evidence gathering through dissemination. An evidence-based screening tool was developed using risk factors and exclusion criteria gathered from major hospitals nationwide. This was distributed to and administered by bedside nurses following CDI project education.

Interventions: Charts of patients who flagged the existing screening tool were analyzed for the selected contextual factors. The secondary screening tool could be interpreted two ways: 1) patients with more than 1 risk factor flag the intervention tool, and 2) patients with no risk factors and one or more confounding factors do not flag the intervention tool. Aggregated data were analyzed to assess changes in specificity (measured with accuracy) and sensitivity compared with the baseline tool.

Results: During the implementation period, the baseline sensitivity was assigned to the current screening tool, which also exhibited 18% accuracy. Flagging all patients with at least two risk factors increased accuracy to 59% but dropped sensitivity to 60% relative to the current tool. Excluding patients who have no risk factors but exhibit at least one confounding factor maintained the baseline sensitivity while improving accuracy to 31%.

Conclusions: These data demonstrate that giving context to diarrhea has the potential to decrease unnecessary CDI isolation without sacrificing sensitivity, thus improving the screening process.