BACKGROUND: State law requires New York State (NYS) hospitals to report all laboratory-identified carbapenem-resistant Enterobacteriaceae (CRE) clinical cultures to the National Healthcare Safety Network (NHSN). State regulation also requires that regulated facilities report outbreaks of communicable diseases to the NYS Nosocomial Outbreak Reporting Application (NORA). An outbreak is defined as an increased incidence above the baseline. NYS encourages reporting of outbreaks so that timely information on emerging infections can be disseminated and so that health department staff can offer epidemiologic assistance and laboratory support.
METHODS: We used the SaTScan temporal scan statistic to identify the most likely clusters of hospital onset (HO) CRE in the 2015 NHSN data. We performed the scan test by species (Klebsiella, Enterobacter, E. coli) at three resolutions (facility-wide, NHSN location type, and subgroups of location if available) within each hospital. The maximum cluster size was set to 60 days. We asked hospital infection preventionists (IPs) to retrospectively evaluate whether the newly identified SaTScan clusters represented possible outbreaks.
RESULTS: In 2015, 109 (62%) NYS hospitals reported 1,303 incident cases of HO CRE in NHSN. SaTScan identified 14 clusters with average duration of 10 days. Hospitals evaluated 12 of the clusters: 3 were subsequently reported as outbreaks to NORA, 2 were the result of previous under-reporting of positive cases to NHSN and should not have been identified as incident clusters, and 7 were not considered outbreaks by the IPs. In many cases the IPs suspected that the patients previously carried CRE, but no previous hospital lab results were available. Two separate outbreaks were reported to NORA, but these were not considered statistically significant by SaTScan.
CONCLUSIONS: This activity heightened IPs’ attention to CRE surveillance and outbreak reporting, and may complement the IP’s routine process by using a high temporal resolution (daily vs. monthly). Limitations to this method include lack of information on patient movement within the facility, and lack of control for prevalent cases. In addition, more timely identification of the clusters would have improved the ability of the IPs to evaluate CRE transmission. Further work is needed to better define outbreaks so that outbreak reporting is more consistent across the state.