LB 9 Evaluation of sampling denominator data to estimate central line days: Validation of methods to reduce data collection burden of the National Healthcare Safety Network

Sunday, April 3, 2011: 2:15 PM
Coronado BCD (Hilton Anatole)
Nicola D. Thompson, PhD , Centers for Disease Control and Prevention, Atlanta, GA
Deborah Godine, RN, CIC , California EIP, Oakland, CA
Julie Duran, MPH , Colorado Department of Public Health and Environment, Denver, CO
Deborah Thompson, MD, MSPH, FACPM , New Mexico Department of Health, Santa Fe, NM
Patricia Lawson, RN, MS, MPH, CIC , Maryland Department of Health & Mental Hygiene, Baltimore, MD
Zintars Beldavs, MS , Oregon Health Department, Portland, OR
Ghinwa Dumyati, MD , University of Rochester, Rochester, NY
Matthew B. Crist, MD, MPH , Tennessee Department of Health, Nashville, TN
Jonathan Edwards, MStat , Centers for Disease Control and Prevention, Atlanta, GA
Shelley Magill, MD, PhD , Centers for Disease Control and Prevention, Atlanta, GA

Background: Financial incentives to report central line associated bloodstream infection (CLABSI) data to the National Healthcare Safety Network (NHSN) began in 2011; increasing demands for data and reporting require considerable staff time. Valid methods that reduce data collection burden are needed. Reducing the burden of collecting CLABSI denominator data (DD) (patient days [PD] and central line days [CLD]) by sampling one day/week has shown promise, but has not been subject to large scale evaluation.

Objective: To evaluate the accuracy of sampling DD one day/week to estimate CLD (eCLD) and calculate CLABSI rates.

Methods: Eligible facilities were acute care hospitals performing NHSN CLABSI surveillance in Emerging Infections Program (EIP) locations. Facilities volunteering to participate submitted ≥ 6 consecutive months of 2009/10 CLABSI DD showing daily totals of PD and CLD. eCLD were calculated for each weekday for all units (sample device utilization ratio (DUR) obtained from CLD/PD corresponding to each weekday X total PD). The difference and percent difference between eCLD and actual CLD (CLD diff, CLD %), and estimated and actual CLABSI rate (Rate diff) were calculated for each weekday sample. Summary statistics comparing the range in accuracy of estimated to actual data for each weekday were calculated.

Results: Between September and December 2010, 65 ICUs (55% Med/Surg) in 45 facilities submitted a total of 688 months of CLABSI DD (mean 10.6 months/unit), yielding 205,140 PD and 105,338 CLD (median DUR 0.49, range/unit 0.10 - 0.94). The difference in eCLD and actual CLD varied by weekday, and was greatest (absolute difference >20%) for Sat, Sun, and Fri (Table) and lower when sampling DD Mon through Fri. The CLABSI rate difference for sampling DD was smallest for Tue (narrowest 5th 95th percentile interval); for 90% of units the estimated CLABSI rate was no more than -0.06 to +0.14 CLABSI per 1,000 CLD from the actual CLABSI rate (Table).

Conclusions: These preliminary results from a variety of ICU types suggest once weekly DD sampling is a viable alternative to manual daily collection. Evaluating the feasibility of implementing DD sampling and identification of factors associated with poor precision of eCLD are next steps to develop parameters for guiding the implementation of these methods into NHSN. Compared to daily DD collection (365 counts), adoption of once weekly (52 counts) sampling would substantially reduce the DD data collection burden and increase the efficiency of CLABSI surveillance.