Implementation of an adaptable semi-automated nonventilator hospital-acquired pneumonia (nvHAP) surveillance system in swiss acute care hospitals: a feasibility study

OBJECTIVE: nvHAP (nonventilator hospital-acquired pneumonia) can affect all non-intubated patients, and semi-automated systems enable incidence surveillance. This feasibility study evaluated the performance and implementation of a semi-automated nvHAP surveillance in Swiss acute care hospitals. DESIGN: Multicenter feasibility study. SETTING: Seven Swiss acute care pilot hospitals representing different sizes and language regions. PATIENTS: Inpatients of the participating hospitals. METHODS: Hospitals implemented an adaptable nvHAP selection algorithm including one to five indicators (radiology procedure, radiology report, leukocytes, body temperature, intubation). Five hospitals performed manual review on the preselected patients following standardized definitions. Performance characteristics of the algorithms (i.e., sensitivity and percentage records to manually review) and time investment to implement the semi-automated surveillance were evaluated. Barriers and facilitators for implementation were identified through interviews. RESULTS: Hospitals implemented algorithms including one, two, four and five indicators. Sensitivity ranged from 98% to 100% in larger hospitals. Percentage of records to manually review ranged from 1% to 6% in hospitals that surveyed the total patient population and was 13% in one hospital that focused on two high-risk departments. Time for technical implementation varied from 55 to 437 hours. Mean time for manual review per preselected patient was 14 minutes and decreased with experience. Implementation facilitators included preprocessed data, team experience in similar projects, and external support. CONCLUSIONS: Semi-automated nvHAP surveillance was feasible and sufficiently sensitive regardless of the algorithm. It required effort for setup and manual review. Algorithm adaptability enabled the implementation in hospitals with limited electronically available data or IT resources.