Feasibility of Whole-hospital CLABSI Surveillance in a Large Community-Based Health Care Center

Background: Central line-associated bloodstream infections (CLABSI) are often monitored only in intensive care units (ICUs), but many non-ICU patients have central lines as well.  Availability of unit-level data to feed back to staff is critical to CLABSI prevention efforts, but limited resources often serve as a barrier to conducting surveillance in non-ICU settings.  

Objective: To determine CLABSI rates in all non-ICU wards at our institution, and to determine feasibility of such surveillance.

Methods: Our 1100-bed, 2-hospital community-based academic health care center has reported ICU CLABSI to NHSN since 1995.  In Oct 2009, we began CLABSI surveillance in 3 stepdown units, and in Feb 2010, we began a 3-month pilot of CLABSI surveillance in all other non-ICU wards (n = 20), excluding maternity, pediatric and psychiatric floors.  Each unit’s clerical staff were required to submit daily logs of patients with central lines.  Infection Prevention (IP) staff (n = 6) reviewed all positive blood cultures after identification using Safety Surveillor (Premier), cross-referenced them with the device logs, and applied standard NHSN CLABSI definitions.  We calculated CLABSI rates as number of infections per 1000 catheter-days, and calculated device utilization ratios (DUR) by dividing catheter-days by patient-days.  We excluded units that failed to submit device logs at least 75% of the time from analysis. 

Results: The three stepdown units under surveillance from Oct 09-Sept 10 reported 32 CLABSI, 11,030 catheter-days and 34,446 patient-days (mean CLABSI rate 2.9/1000, DUR 0.32).  Of the 20 non-ICU wards (10 medical, 7 surgical and 3 additional stepdown units) included in the pilot surveillance program, 3 were excluded from analysis due to >25% missing denominator data.  The remaining wards reported 23 CLABSI, 8893 catheter-days and 53,711 patient days (overall DUR 0.17), but averaged 10.2% missing denominator data (range, 0.7 – 21.1%).  The mean catheter-days per month was 555 (median, 473; range, 19 – 2326).  The mean CLABSI rate was 2.6 per 1000 catheter-days (median 0.9), with higher mean rates among medical (3.6/1000) than surgical (0.5/1000) units. Nine units had zero CLABSI identified throughout the pilot period.  IP staff spent approximately 10 hours weekly to conduct this surveillance.

Conclusions: While central lines were utilized less frequently, CLABSI rates for medical floors were similar to those for stepdown units. Microbiology surveillance software decreased the time required of IP staff to identify potential CLABSI, but significant resources were still required.  Obtaining accurate catheter-days was the major barrier to conducting non-ICU CLABSI surveillance.  Lack of complete denominator data resulted in overestimation of CLABSI rates and underestimation of DUR.  Automated acquisition of device days is necessary to obtain accurate whole-hospital CLABSI rates.