Background:Clostridioides difficile remains a pervasive issue throughout healthcare facilities in the United States. Currently, no national guidelines exist for healthcare facilities to notify public health about suspected C. difficile transmission. Identification of a threshold for public health notification is needed to improve efforts to target prevention in facilities and to contain the spread of C. difficile.Methods: We analyzed C. difficile data reported by acute-care hospitals (ACHs) during October 2017–September 2018 via the CDC NHSN in Colorado and Tennessee. Threshold levels of ≥2, ≥3, and ≥4 C. difficile infections per calendar month per unit were assessed to identify ACH units that would trigger facility reporting to public health. Values meeting thresholds were defined as “alerts.” Facilities were further stratified by size and medical teaching status. Recurrent alerts were defined as meeting the threshold at least twice within 12 months. Presence and recurrence of facility alerts were compared to facility-specific standardized infection ratios (SIRs) and cumulative attributable differences (CADs). Results: Of 105 ACHs in Tennessee and 50 in Colorado, 46 in Tennessee (44%) and 28 in Colorado (56%) had alerts with a threshold of ≥2 cases per calendar month per unit; 20 in Tennessee (19%) and 19 in Colorado (38%) had ≥3 cases per calendar month per unit; and 7 in Tennessee (7%) and 10 in Colorado (20%) had ≥4 cases per calendar month per unit. Most alerts with each threshold were in facilities with ≥400 beds and in major teaching hospitals. Using a threshold of ≥2, 64% of Tennessee and 79% of Colorado alerts were associated with recurrent alerting units. Using an alert threshold of ≥3, 85% of Tennessee facilities (17 of 20) and 75% of Colorado facilities (15 of 20) with the highest CAD values had at least 1 alert. Using state-based CAD values, 79% of the CAD value for Tennessee (356 of 449) and 91% of the CAD value for Colorado (309 of 340) were attributable to facilities with at least 1 alert. Facilities above a threshold of ≥3 had a pooled SIR of 0.92 in Tennessee (range, 0.46–7.94) and 1.07 in Colorado (range, 0.74–1.74). Conclusions: Using alert threshold levels identified ACHs with higher levels of C. difficile. Recurrent alerts account for a substantial proportion of the total alerts in ACHs, even as thresholds increased. Alerts were strongly correlated with high CAD values. Because NHSN C. difficile data are not available to public health departments until several months after cases are identified, public health departments should consider working with ACHs to implement a threshold model for public health notification, enabling earlier intervention than those prompted by SIR and CAD calculations.

Disclosures: None

Funding: None