Venous catheters are among the most widely used medical devices and are indispensable for medical care. However, they can also lead to serious complications by causing bloodstream infections (BSIs). Therefore, it is crucial for patient safety to prevent catheter-related BSIs (CRBSIs) in hospital settings because they are not only associated with an increased hospital stay and extra attendant costs but also with increased mortality [1,2,3].

Most CRBSIs are related to the presence of central venous catheters (CVCs) rather than peripheral venous catheters (PVCs) [4]. The incidence rate of PVC-BSIs is estimated to be 0.05–0.5 per 1000 device days, which is far less than the incidence rate of CVC-BSIs [5,6,7]. However, PVCs are more frequently used than CVCs in clinical settings [8, 9], and, therefore, the absolute number of PVC-BSI cases is not negligible.

CVC-BSIs have been well studied for their prevalence, pathogen species distribution, and major complications because their incidence rate is higher than that of PVC-BSIs. Only a few studies have focused specifically on the pathogen species distribution and prognosis in PVC-BSIs, despite the more frequent PVC use compared to CVC use and potential for serious infections [6, 9, 10]. Considering that PVC-BSIs may be quite different from CVC-BSIs, some studies have focused on PVC-associated phlebitis or exit-site infections [11,12,13,14,15]. However, PVC-BSIs and phlebitis are not synonymous. In this study, we have compared the clinical characteristics and prognoses of CVC-BSIs and PVC-BSIs to further our understanding of their relative clinical impacts.


Study Setting

This study was performed at the University of Tokyo Hospital [a 1210-bed tertiary care teaching hospital including 34 intensive care unit (ICU) beds for adults] in Tokyo, Japan, from April 2011 through March 2013. The infection control teams consisted of full-time infection control and infectious disease physicians and nurses.

Study Design

This was a retrospective observational cohort study. The enrolled cases were selected from those that involved patients with positive blood culture results. Patients with PVC-BSIs and CVC-BSIs were enrolled based on the criteria given below.

Inclusion Criteria

The bloodstream infection occurred more than 2 days after admission. CRBSI was determined when (1) a recognized pathogen was cultured from one or more blood cultures, and the organism was not related to infection at another site, or (2) for common commensal bacteria, such as coagulase-negative staphylococci (CNS), at least two consecutive positive blood cultures were obtained, with clear resolution of clinical symptoms after catheter withdrawal. CVC-BSI was defined according to the surveillance definition given by the Centers for Disease Control and Prevention National Healthcare Safety Network [16]. A patient with PVC who met the above criteria for bloodstream infection with a recognized pathogen or with commensal bacteria was defined as having a PVC-BSI. If more than two episodes occurred after the treatment of BSI was completed, they were enrolled as separate cases of CRBSI.

Exclusion Criteria

Patients who (1) were < 18 years of age at the time of detection of positive blood cultures, (2) had both CVC and PVC, (3) had CVC-BSI associated with intravenous hyperalimentation therapy at home, and (4) had other venous catheters (such as peripherally inserted central catheter, hemodialysis catheter, central venous access port, Hickman-type catheters, or arterial catheters) along with a PVC or CVC were excluded from the study.

Data Collection and Definitions

Data of the following clinical characteristics were obtained from medical records: age, sex, comorbidities, length of catheter indwelling, length of hospitalization, ward type on the onset day (medicine, surgery, emergency department, or ICU), receipt of mechanical ventilation, and severity of illness at BSI onset which was calculated using the Sequential Organ Failure Assessment (SOFA) score [17]). To calculate the age-adjusted Charlson score for mortality risk due to comorbid conditions [18, 19], the following comorbidities were documented: myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic obstructive pulmonary diseases, connective tissue disease, peptic ulcer disease, diabetes mellitus, chronic kidney disease, hemiplegia, leukemia, malignant lymphoma, solid tumor, liver disease, and acquired immune deficiency syndrome. The day of catheter removal, day of antibiotic regimen initiation, and the day of initiation and duration of appropriate antimicrobial therapy were all confirmed from the medical records. The first day was defined as the initial day that blood cultures were taken from the patient. Appropriate antimicrobial therapy was defined as systemic administration of at least one antimicrobial agent to which the causative pathogen showed susceptibility in vitro. The 30-day all-cause mortality was measured as the clinical outcome. Estimates of incidence rates of CRBSIs were calculated as episodes of BSI per 1000 patient-days.

Microbiological Analysis

Blood culture specimens were inoculated into BACTEC standard culture bottles in the BACTEC 9000 system (Becton Dickinson, Franklin Lakes, NJ, USA). The identification and susceptibility tests were performed using a WalkAway system (Siemens, Berlin, Germany). The broth microdilution method was used for susceptibility testing, and the results were interpreted according to the Clinical Laboratory Standards Institute guidelines (M100-S19).

Statistical Analysis

For continuous variables, mean values were compared using the two-sample Student’s t test for independent samples. The Mann–Whitney U test was used when variables were skewed. Differences in proportion were compared using the Chi square test or Fisher’s exact test, as appropriate. The Kaplan–Meier survival analysis and log-rank test were used to estimate the probability of survival after the onset of BSI. All p values for these tests were two-tailed, and values of p < 0.05 were considered statistically significant. All statistical analyses were performed using GraphPad Prism (v.5.0; GraphPad Software, San Diego, CA, USA). The study was approved by the ethics committee of Graduate School of Medicine and Faculty of Medicine, The University of Tokyo (#3538). The requirement for written informed consent was waived due to the observational retrospective nature of the study.


Case Characteristics

During the study period, a total of 1,236 bacteremia cases were identified, of which 414 were associated with CRBSIs. Of these, 166 were PVC-BSIs, 183 were CVC-BSIs, and 65 were caused due to other intravascular devices. Finally, 124 cases met the criteria for PVC-BSIs and 110 for CVC-BSIs (Fig. 1). The 47 cases of BSIs (30 CVC-BSIs and 27 PVC-BSIs) involved 21 patients, each with two or three episodes of BSI. The estimations of incidence rates of PVC-BSIs and CVC-BSIs were 0.17 and 0.15 per 1000 patient-days, respectively. The median age, age-adjusted Charlson scores, SOFA scores, sex distribution, and ward types were similar for both the groups. The median length of hospitalization and catheter indwelling times were significantly shorter in the PVC-BSI group than in the CVC-BSI group (Table 1).

Fig. 1
figure 1

Patient selection. The flow chart shows the selection of patients with peripheral venous catheter-associated and central venous catheter-associated bloodstream infections

Table 1 Characteristics of patients with peripheral venous catheter-associated and central venous catheter-associated bloodstream infections

Pathogens Responsible for CVC-BSIs and PVC-BSIs

The distribution of the causative pathogens differed between the CVC-BSI and PVC-BSI groups. Staphylococcus aureus and Gram-negative bacilli infections were more frequent, and coagulase-negative staphylococci (CNS) and Candida spp. infections were less frequent in the PVC-BSI group than in the CVC-BSI group. The prevalence of oxacillin resistance of both S. aureus and CNS did not differ between the two groups (Table 2).

Table 2 Distribution of pathogens in patients with peripheral venous catheter-associated and central venous catheter-associated bloodstream infections

Therapeutic Approach

Catheter removal after BSI onset occurred earlier in the PVC-BSI group than in the CVC-BSI group (Table 3). While empirical treatment regimens and their appropriateness were comparable between the two groups, definitive treatment regimens differed significantly between the two groups because of the differences in the distribution of causative pathogens. In total, 33 patients in the PVC-BSI group and 15 in the CVC-BSI group received no antibiotics at the time of blood culture. No patient without antibiotic treatment died within 30 days of blood culture, even though 4 and 3 patients from the PVC-BSI and CVC-BSI groups, respectively, showed positive blood culture results.

Table 3 Treatment of peripheral venous catheter-associated and central venous catheter-associated bloodstream infections

Clinical Outcomes

The all-cause mortality at 30 days was 12.0% (14 patients) for the PVC-BSI group and 12.8% (14 patients) for the CVC-BSI group (p = 0.883, log-rank test; Fig. 2). Survival rates did not differ significantly between the two groups.

Fig. 2
figure 2

Survival in patients with catheter-related bloodstream infections, showing the curves for the 30-day all-cause mortality among patients with peripheral venous catheter-associated and central venous catheter-associated bloodstream infections. The all-cause mortality at 30 days was 12.0% (14 patients) for the PVC-BSI group and 12.8% (14 patients) for the CVC-BSI group PVC-BSI peripheral venous catheter-associated blood stream infection, CVC-BSI central venous catheter-associated blood stream infection


In general, the incidence rates of PVC-BSIs were far less than CVC-BSIs [5,6,7]. However, PVCs were much more frequently used than CVCs [8, 9], and, therefore, the absolute number of PVC-BSIs surpassed that of CVC-BSIs. This study showed that the overall clinical outcomes did not differ between patients with CVC-BSIs and PVC-BSIs. However, the distribution of the causative pathogens differed between the two groups, indicating that PVCs are not safer than CVCs with regards to BSIs.

The length of catheter indwelling required to develop a PVC-BSI was far shorter than that required to develop a CVC-BSI. The Infectious Diseases Society of America guidelines for the prevention of intravascular catheter-related infections [20] state “there is no need to replace peripheral catheters more frequently than every 72–96 h to reduce the risk of infection and phlebitis in adults”. Based on this guideline, PVCs were replaced every 72–96 h in our hospital, and, therefore, the median length of PVC indwelling was 3 days. As many as 68 cases (55%) of PVC-BSI developed within 3 days of indwelling, increasing to 123 cases (99%) within 10 days of indwelling. In contrast, only 35 cases (32%) of CVC-BSI developed within 10 days of indwelling. Colonization and biofilm formation occurs within 3 days of indwelling [21]. Molecular methods have shown frequent extraluminal and intraluminal bacterial colonization in PVCs, even though culture methods detected no bacteria [22]. Biofilms tend to form on the external surface and not in the internal lumen of the catheter for up to 10 days [23, 24]. Based on these results and reports, it is believed that the main mechanism of PVC-BSI development involves contamination outside the catheter, mainly from the skin during the insertion procedure. The Catalonian Nosocomial Infection Surveillance (VINCat) program reported the incidence of PVC-BSIs and CVC-BSIs to be 0.05 and 0.10–0.14 per 1000 patient-days, respectively, in the non-ICU wards. The incidence of PVC-BSIs in our study was slightly higher than that in the VINCat program. Therefore, cleaning of the skin using aseptic techniques before insertion of PVC is critical for reducing the incidence of PVC-BSIs; however, replacing the catheters more frequently than every 72–96 h does not help.

Our findings demonstrated a difference in the distribution of pathogen species between the two groups. In line with previous reports [25,26,27], we found that 49 (45%) CVC-BSI cases were caused by coagulase-negative staphylococci. Contrastingly, S. aureus was the most common causative pathogen in PVC-BSI cases, and Gram-negative bacilli were not observed less frequently than coagulase-negative staphylococci. Our findings on causative pathogens in PVC-BSI cases were consistent with those reported previously [6, 10, 28].The duration of S. aureus-associated bacteremia is reported to be significantly longer in the PVC-BSI group than in the non-PVC-BSI group [29], highlighting the fact that PVC-BSIs are a major source of S. aureus-associated bacteremia in clinical settings. To treat S. aureus infections, glycopeptides are important agents used for empirical therapy until the blood culture results are available [30]. β-lactams may also be important for the treatment of PVC-BSIs because of the dominance of Gram-negative bacilli. It has recently been reported that Gram-negative rods were more frequently identified in PVC-BSIs than in CVC-BSIs [31]. However, it is necessary to be careful while interpreting these results due to differences in the antibiograms for each setting, which decide the appropriate initial empirical therapy.

The third point is the prognosis of PVC-BSIs. Consistent with a previous study that showed comparable overall mortality between the CVC and PVC groups [8, 32], we also found that the 30-day all-cause mortality was comparable between the two groups. While some studies have reported a deterioration in prognoses of critically ill patients in ICUs due to CVC-BSIs [1, 3, 33], another study concluded that CVC-BSIs had no influence on the prognosis [34]. These studies may indicate that the impact of CRBSIs on mortality depends on the severity of the patient’s condition. Thus, the prognoses and the distribution of pathogen species between CVC-BSIs and PVC-BSIs were consistent with other reports.

Our study has some limitations. First, there is a possibility of over or under diagnosis of CRBSIs, since catheter-tip culture information was not used. Second, as the lengths of hospitalization and catheter indwelling were shorter in the PVC-BSI group, there could be a slight bias in patient selection due to the retrospective nature of the study. Third, pathogen species distribution may vary depending on the facility. Finally, this study focused on cases from 2011 to 2013; the findings might be different today. Further studies will be needed to determine the clinical impact of PVC-BSIs on patient safety so that PVC-BSIs can be managed optimally.


Our findings suggest that while pathogen species distribution varies between PVC-BSIs and CVC-BSIs, the all-cause mortality rate did not differ significantly between the two groups. Because PVCs are more frequently used in hospital settings, their absolute contribution to nosocomial infections could be comparable to that of CVCs. Therefore, PVC-BSIs should be given the same attention that is given to CVC-BSIs in order to prevent their incidence.