Catheter-associated bloodstream infections: Looking outside of the ICU

CA-BSI in non-ICU Settings 

The objective of our previous study12 was to determine the rate of CA-BSIs in non-ICU medical patients by developing a prospective surveillance program in a major tertiary care hospital. The study was conducted at Barnes-Jewish Hospital, a 1250-bed teaching hospital in St. Louis, Missouri. Four general medical wards were selected for surveillance. Each ward had separate nursing staff; 2 wards shared a medical director, interns, residents, and a nurse manager. Three wards were 26-bed units, and 1 ward was a 27-bed unit (a total of 105 beds).

All positive blood cultures reported by an electronic detection tool (GermWatcher, St. Louis, MO)13 were reviewed. The surveillance period was April 1, 2002, to April 30, 2003. A single infection control specialist collected data by chart review (catheter start and stop dates, ward admission and discharge dates, microbiology data). Any adult patient admitted to 1 of the 4 wards was included in the study as soon as a CVC was inserted.

The following definitions were used: “Catheter utilization ratio” = total number of days with a CVC/total number of patient-days; “Catheter-associated BSI” = defined by CDC criteria14, eg, a patient had to have a catheter at least 48 hours before detection of infection; “CA-BSI rate” = CA-BSIs/1000 catheter-days.

Results 

The 13-month study included 7337 catheter-days and 33,174 patient-days. Thirty-five patients (83%) had a nontunneled catheter, 3 (7%) had totally implanted ports, 2 (5%) had tunneled catheters, and 2 (5%) had more than 1 catheter in place. The overall catheter-utilization ratio was 0.22 (range, 0.19-0.25). Forty-two CA-BSIs were identified. Gram-positive organisms were recovered in 24 (57%) cases; gram-negative bacteria in 7 (17%); and Candida spp in 6 (14%). The CA-BSI rate was 5.7 (95% confidence interval: 3.4-8.0) and varied from 4.3 to 8.0 between the 4 wards. There was no significant difference in CA-BSI rates between the wards (χ2 for linear trend, 0.42; P = .52). The overall rate of CA-BSIs decreased steadily during the study period, from 7.8 during the first 6 months to 3.9 during the following 7 months, representing a rate ratio of 0.5 (95% confidence interval: 0.27-0.93). This may be partly due to an intervention that was started during the study period (a mandatory educational module for nurses on 2 of the 4 wards studied). Surveillance itself (and the feeding back of rates to wards) has also been shown to decrease infection rates.15

The catheter utilization ratio of 0.22 is similar to data reported by Climo et al6 for patients in the non-ICU setting, lower than NNIS ICU data (range, 0.51-0.69)3 but higher than what Vonberg and Behnke reported in German hospitals (0.046).10 It is unclear why US device utilization ratios should be higher, and this raises the question whether the United States has catheter policies different from those in other countries. At least for the ICU setting, the results of an international study on CA-BSIs argue against significant differences among countries. The multicenter prospective Evaluation of Processes and Indicators in Infection Control study compared CA-BSI rates in ICUs in the United States to rates in 14 other countries (3.82 vs 5.02, respectively; P = .27), and no significant difference was found.16 There may still be differences in the much more diverse non-ICU wards of hospitals. Vonberg and Behnke10 included community hospitals and documented longer hospital lengths of stay than we did, both indicating lower acuity of illness and possibly resulting in lower rates of catheter use and therefore nosocomial infections.

Although catheter utilization was lower than for the ICUs in the NNIS system, the CA-BSI rate of 5.7 was similar to the 5.0 rate for NNIS ICUs and similar to the CA-BSI rate in ICUs at Barnes-Jewish Hospital during the same time period (5.2). This discrepancy is surprising, and we can only hypothesize what the reasons may have been. As discussed earlier, Trick et al pointed to an increased number of idle catheters outside the ICU and to possible deficits in optimal care.9 This suggests that patients in non-ICU wards might be at a higher rates of unnecessary catheters and/or suboptimal care and therefore at a high risk of catheter-associated infections.

Both Trick et al9 and Climo et al6 showed differences in the types of catheters used in different settings within the hospital. Therefore, patients in non-ICU wards might have catheters that are more prone to infections.

In a facility with shortage of ICU beds, Mnatzaganian et al examined ward patients who qualified as critically ill (by criteria derived from the Society of Critical Care Medicine) and reported lower rates of BSIs in critically ill patients admitted to regular wards compared with the ICU.17 This argues against severity of illness as the primary risk factor for BSI. Other factors may influence the pathogenesis of BSIs and lead to high rates outside the ICUs as encountered in our study.

The Institute for Healthcare Improvement leads a campaign called “5 Million Lives,” referring to the number of lives that could be saved if infection control was optimized.18 The campaign Web site mentions CA-BSIs but only under the heading “intensive care.” On the other hand, the NNIS has recognized the need for surveillance data in other areas of the health care system. In 2004, Clinical Infectious Diseases published an article describing surveillance and surveillance systems in the United States.19 It was stated that the transformed NNIS system would also “expand surveillance beyond intensive care units.”

The NHSN report for 2006 is the first article to provide nationwide surveillance data for non-ICU units.20 Unfortunately, compared with the large number of ICUs described, only 18 inpatient medical wards were included. Catheter utilization was similar to what we found (0.24 vs 0.22, respectively), but CA-BSI rates were much lower (2.1 vs 5.7, respectively). However, no range was given because of the low number of wards studied, and there is no information readily available to determine whether these were wards in community or teaching hospitals.

Conclusions 

Benchmark data for hospital infections in the non-ICU setting are starting to become available. If maintenance of insertion-site dressings is less stringent outside ICUs, efforts to improve care may have greater impact here than in the ICU. When a patient is transferred out of the ICU, it should be determined whether the catheter can be removed. Educational measures targeted at non-ICU wards are needed.

Computer-assisted surveillance has been proposed as a means to facilitate surveillance of larger number of patients because manual chart review—the gold standard—to determine CA-BSIs is time-consuming.21 First results of computer-assisted surveillance are promising.22

Mandatory public reporting of hospital-acquired infections is becoming more common as more states in the United States adopt such legislation.23 The Healthcare Infection Control Practices Advisory Committee recommends that CA-BSIs be reported but, again, reporting limited to ICUs.24 Soon, reportage of CA-BSIs in non-ICU patients could become mandatory.