BACKGROUND: Carbapenem-resistant Enterobacteriaceae (CRE) cause approximately 9,300 healthcare-associated infections annually and are resistant to nearly all antibiotics. Mortality rates up to 50% have been reported. During March 1, 2015 to September 1, 2016, the North Carolina Division of Public Health (NC DPH) implemented CRE sentinel site surveillance. Public health epidemiologists (PHEs) at the state’s seven largest hospital systems conducted CRE surveillance by reviewing laboratory results, submitting case report forms (CRFs), and coordinating the submission of isolates for molecular phenotyping. We evaluated this surveillance system to assess performance and inform CRE policy decisions.
METHODS: Referencing CDC guidelines, we conducted semi-structured interviews with PHEs and laboratory personnel to assess simplicity, data quality, acceptability, stability, and overall usefulness of the system. We assessed data completeness, timeliness, and predictive value positive (PVP) by analyzing surveillance database variables using SAS. PVP was calculated as the proportion of reported cases that met the case definition. We compared facility-level variables reported to the National Healthcare Safety Network (NHSN) in 2015 between the seven surveillance sites and all other NC acute care hospitals (n=98) to assess representativeness.
RESULTS: Simplicity was moderate; surveillance involved 16 key individuals across nine organizations. Flexibility was high; all system components (e.g. electronic CRF, Access database) were developed, maintained, and readily adaptable by DPH. Data completeness was high; four of 37 (10.8%) key variables had missing values in more than 10% of records. Acceptability was high with six of seven (85.7%) PHEs and four of six (66.7%) clinical laboratorians willing to continue voluntary surveillance. Stability was moderate in terms of personnel and programs; four of seven (57.1%) PHEs and six of six (100%) clinical laboratorians indicated there was another individual at their facility capable of performing CRE surveillance responsibilities. Timeliness was low (median 34.5 days from specimen collection to report to DPH and median 61 days from specimen collection to arrival at the State Laboratory of Public Health). PVP was 94.4%. Representativeness was moderate; sentinel sites represented 6,043 of 21,603 (28.0%) acute care hospital beds and 1,731,622 of 5,291,220 (32.7%) acute care hospital patient days reported in NHSN.
CONCLUSIONS: The system has been useful in describing the burden of CRE in NC; its flexibility will permit improvements to simplicity, acceptability, data quality, and timeliness. Feedback from key stakeholders emphasized the need for a standardized CRE case definition that incorporates clinical and laboratory components. The future of CRE surveillance in NC is under discussion.