Surveillance of hospital-acquired infections: a model for settings with resource constraints

Abstract

Background: Surveillance activities have been considered of paramount importance for effective infection control programs in health care organizations.

Objectives: Our objective was to design a capture system able to assure surveillance of hospital-acquired infections (HAI) in acute hospitals with few resources devoted to infection control.

Methods: We performed 4 biweekly repeated prevalence studies to identify major HAI (urinary tract infections, surgical site infections, lower respiratory tract infection, bloodstream infections) as defined by the Centers for Disease Control and Prevention (CDC) criteria in 3 large hospitals in northeastern Italy (6 internal medicine departments, 5 general surgery departments, 3 intensive care units, and 1 bone marrow transplant unit).

Results: One thousand five hundred fifty-four patients were screened (63.9% in medical wards, 27.5% in surgical wards, and 8.5% in intensive care units and bone transplant unit). The overall prevalence of infection was 4.9% (77/1,554); 4.5% (70/1,554) of patients were infected. A capture system based on the presence of fever >or=38 degrees C, antibiotic use, and presence of devices guarantees 100% sensitivity in detecting HAI but requires an assessment of 62% of the population. Using the presence of fever and devices as criteria guarantees a sensitivity of 98%, requiring an assessment of 41.4% of patients, whereas presence of fever and antibiotic use has the same sensitivity but requires an assessment of 50% of patients. Using nursing records, physician records, and direct patient examination as sources of documentation guarantees that all necessary data are collected while requiring a mean of 4 minutes and 42 seconds per patient (standard deviation, 1 minute and 30 seconds).

Conclusion: A capture system based on biweekly repeated prevalence studies that select patients for the presence of fever, antibiotics, and medical devices ensures the detection of all HAI in a resource-limited environment.