Objective: To determine whether incidence and characteristics of CLABSI are different in non-ICU compared to ICU locations.
Methods: We calculated pooled incidence of CLABSI/1000 central line days by location type for 2006-2008. We determined incidence density ratios (IDRs) and device utilization ratios (DURs) for medical, medical-surgical, and surgical wards compared to respective ICUs. The analysis accounted for type of hospital (teaching and non-teaching) and ICU size (small (<=15 beds) and large (>15 beds)). The chi-square test was used to evaluate pathogen distributions.
Results: The number of CLABSIs reported from non-ICU locations increased from 361 reported by 24 facilities in 2006 to 1625 reported by 197 facilities in 2008. During 2006-2008, the majority of CLABSIs were reported from medical-surgical wards (1070; 44.6%), followed by medical wards (562; 23.4%) and surgical wards (246, 10%). Overall, non-ICUs had lower CLABSI incidence compared to ICU locations: IDR was 0.6 (95%CI 0.5-0.7) for surgical wards vs. surgical ICUs; 0.6 (95%CI 0.5-0.7) for medical wards vs. medical ICUs in major teaching hospitals and 0.9 (95%CI 0.8-1.0) for medical wards vs. medical ICUs in non-teaching hospitals. In major teaching hospitals, the IDR was 0.6 (95%CI 0.5-0.7) for medical-surgical wards vs. medical-surgical ICUs; in non-teaching hospitals, the IDR for medical-surgical wards was 0.9 (95%CI 0.8-1) compared to small and 0.8 (95%CI 0.7-0.9) compared to large medical-surgical ICUs. Differences in commonly reported pathogens included significantly higher proportions of Staphylococcus aureus (18.2% vs. 9.1%, p<0.01) from wards CLABSIs compared to ICUs, while less Enterococcus spp. (14.1% vs. 20.3%, p<0.01) and Candida spp. (12.5% vs. 16%, p<0.01) from wards compared to ICUs. Conclusions: Incidence of CLABSIs was 10%-60% lower in surgical and medical non-ICU locations compared to their ICU counterparts and very similar in medical-surgical wards compared to medical-surgical ICUs. Translating these rates to total disease burden will help to quantify potential impact of prevention efforts in these settings.