Primary bloodstream infections are a frequent cause of morbidity and mortality in critically ill patients. According to the report from the National Healthcare Safety Network (NHSN) of the U.S. Centers for Disease Control and Prevention (CDC), catheter-associated bloodstream infections (CA-BSI) occur at a mean rate of 2.9 per 1000 catheter-days for medical intensive care units (ICUs) in American hospitals [1]. Mortality attributable to CA-BSI has been estimated to be as high as 35%, and the length of hospital stay is increased among infected patients [2-6]. In Korea, primary bloodstream infection was the third common nosocomial infection (24%) and CA-BSI occurred at a rate of 2.58 per 1000 catheter-days in 2006 [7]. However, CA-BSI rate varied among institutions: 10 percentile 0, 90 percentile 8.38/1000 catheter days in hospitals with more than 900 beds [7]. Therefore, interventions aimed to decrease the infection rate are necessary to reduce serious public health consequences of this nosocomial infection.

Prior studies from a single hospital have shown reductions in the rates of CA-BSI by various interventional efforts, including education and surveillance [8-14]. Recent multicenter studies also showed successful sustained reductions in the rates of infection [15, 16]. However, CA-BSI rate is still high in developing countries, and their efforts to reduce CA-BSI rate have scarcely been reported [17]. In Korea, there are some reports on the interventions that was employed to reduce CA-BSI [14, 18, 19] but all have been done in a single center setting and no report on multicenter trial exist.

Therefore, we initiated a prospective multicenter study to implement a multifaceted, education-based intervention in ICUs at 3 academic medical centers in Korea. The primary purpose of this study was to assess the effect of intervention and to examine the sustainability of its effect on lowering the incidence of bloodstream infection which was associated with central venous catheters during and after the intervention.

This study is the first multicenter study in Korea which showed that an active multidisciplinary intervention could reduce the incidence of catheter-related bloodstream infection. The maximal reduction of CA-BSI rates in our study occurred during the 3 months of intervention and was sustained after an active intervention. However, the effect declined shortly after finishing intervention. The result of infection rate surveillance in the post-intervention period showed slight increase of CA-BSI rate. Three months of intervention might be too short to achieve a maximum decrease of infection rate. According to a multicenter intervention by Warren et al., it took approximately 7-12 months to change practice pattern of nurses and physicians, thereby getting a maximum reduction of infection rate [15]. In another multicenter study including more than 100 ICUs which had low infection rate (2.7 per 1000 catheter-days) before the intervention [16], the median rate of infection reached 0 within 3 months after implementation of intervention to decrease CA-BSIs, and continued throughout 15 months of follow-up. In the baseline period of this study, the CA-BSI rates of each hospital were higher than 75 percentile of CA-BSI rates from medical or surgical ICUs in the National Healthcare Safety Network of the USA (4.2 or 4.4 case per 1,000 catheter days) [1], but lower than mean rate obtained in 8 developing countries (12.5 case per 1,000 catheter days) [17], It is, therefore, warranted to compare the efficacies of 3-month intervention and that of longer-period intervention in order to determine whether longer period of intervention is needed to achieve a maximal reduction of infection and sustain this reduction in hospitals with high CA-BSI rate.

Industrialized countries such as the USA and UK have established standards for institutional surveillance and control of healthcare-associated infection [1, 22]. Most studies related to healthcare-associated infection have been conducted in developed countries and demonstrated the efficacy of surveillance and its significant contribution to minimizing morbidity, mortality, and medical cost [2, 3, 8-10, 12, 13, 15, 16, 23]. On the other hand, only a few studies on healthcare-associated infection using standardized definitions and intervention are available in developing countries, especially in Asia [24, 25]. As suggested earlier, studies from developing countries showed higher rates of device-associated infection and this could be due to the lack of guidlines mandating healthcare-associated infection control programs, variable hand hygiene standards, scarce funds and resource - especially lower nurse-to-patient staffing ratios in ICUs - for infection control, and limited use of outdated technology for prevention [17]. A higher baseline infection rate together with these limitations seemed to work against maintaining sustained improvement after the active intervention in the current study. Surveillance of healthcare-associated infections appears to be truly the first step toward reducing the risk of infection in vulnerable hospitalized patients. Our experiences, however, indicate that continued efforts of active intervention are needed to sustain the reduced infection rate after the surveillance.

In this study, 3 university hospitals harboring similar number of ICU beds were enrolled, but the infection rate and outcome of intervention were different in each hospital. The reason can be partly because of several differences in catheter-practices and clinical characteristics of the patients in each hospital (data not shown). For example, subclavian vein insertion rate was practiced in 28.6%, 67% or 76% in each hospital patients, respectively (P<0.001), and the places where catheter was inserted were significantly different (ICU, ER or OR): 33%, 23% or 30% vs. 79%, 7% or 8% vs. 28%, 34% or 29% in each hospital, respectively (P<0.001). Mean age of the patients, severity score of baseline diseases, and duration of admission were also different. Therefore, to decrease the CA-BSI, careful observation and understanding the problems in catheter-practice in each institution are necessary.

It is noted in this study that infection by gram-negative bacteria was dramatically decreased during and after the intervention. Bacteremia due to these organisms emerges usually from non-catheter-related sources, such as nosocomial urinary tract infection or nosocomial pneumonia. During the intervention period, one-time use of sterile 3-way cap of infusion hub and cleaning of injection port with 70% alcohol before accessing the infusion system were recommended to reduce hub-originated catheter infection in this study. In one report, surveillance hub culture predicted the pathogen of CA-BSI especially when gram-negative organisms were identified [26] and in another report, hub contamination control reduced CA-BSI caused by gram-negative organisms, although statistical significance was not found [27]. Therefore we thought that those interventions might reduce bacteremia due to gram-negative bacteria in this study. When CVC is no longer necessary, immediate removal would also contribute to the reduction of infection.

This study has several limitations. The study design, a preintervention-postintervention comparison, diminished the ability to make a causal relation between the intervention and reduction in CA-BSI rates. However, since randomized controlled assignment of the intervention in an ICU setting was not feasible, several intervention studies adopted pre-postintervention design [15, 16]. Second, CA-BSI rates varied among three hospitals involved, suggesting substantial differences in severity of illness, efficiency of surveillance, and availability of institutional resources for prevention between the hospitals. This variation is likely to have occurred, because bloodstream infections are rare and the presence of small number of cases or outbreaks substantially affects the infection rate. Third, because of the same reason as mentioned above, each center did not show significant reduction in CA-BSI rates both during and after the intervention, in spite of the reduction in overall CA-BSI rate. Last, we included only three medical centers that voluntarily participated in the intervention, thus limiting the ability to generalize the findings.

Both in the developing and the developed countries, catheter-associated bloodstream infections are a huge threat to the safety of patients. This study demonstrated that a multi-strategy approach to reduce CA-BSI can be implemented in a diverse setting of medical and surgical units in Korea and can decrease CA-BSI rates during the intervention. Future interventional effort is urgently needed to prolong the effect of intervention which includes diverse strategy for a longer period of time, depending on each center's demand.