Mario Fernández‐Ruiz, MD

Catheter‐related bloodstream infections (CRBSIs) are among the most common forms of hospital‐acquired infection and increase both length of stay and cost of hospitalization.[1, 2] Notable efforts are being devoted to reduce the rate of CRBSI, usually by implementing a bundle of measures.[3, 4, 5] One measure focuses on reducing to a minimum the exposure to vascular catheters.[3] In addition, various studies have shown that multilumen central venous catheters (CVCs) are associated with a higher risk of CRBSI than are single‐lumen catheters.[6, 7, 8, 9, 10] Accordingly, the Healthcare Infection Control Practices Advisory Committee (HICPAC) guideline recommends that clinicians use a CVC with the minimum number of ports or lumens essential for the management of the patient.[3]

Despite the fact that most CRBSIs occur in conventional wards,[1, 11] only a few studies have been focused on the potential magnitude of reducing the number of unnecessary vascular catheters and catheter lumens in the noncritical‐care setting.[12, 13, 14, 15, 16] The adequacy of vascular‐catheter use has been predominantly assessed for nontunneled CVCs.[14, 16, 17, 18] An institutional program aimed at reducing the overall rate of CRBSI should also include other sources, such as conventional peripheral venous catheters (PVCs), peripherally inserted central catheters (PICCs), and arterial catheters (ACs).[3] The need to extend surveillance to other types of catheters has been identified by the Infectious Diseases Society of America (IDSA) as an unresolved issue.[19]

We sought to investigate the rate and appropriateness of use of vascular catheters in the entire population of inpatients at a tertiary‐care center on a single day, as well as the adequacy of the number of catheter lumens harbored by each patient, by using a set of preestablished objective criteria.

METHODS

RESULTS

Out of 1134 reviewed inpatient beds, 834 (73.5%) were occupied on the day of the survey. The mean age of the included patients was 64.518.8 years, and 415 (49.8%) were male. Of these patients, 575 (68.9%) had 1 vascular catheter in place. The proportion of patients with a vascular catheter was significantly higher in ICUs compared with conventional wards (100% vs 66.7%, P<0.0001; Table 2). The overall numbers of vascular catheters and catheter lumens analyzed were 703 and 1448, respectively. Regarding the type of device, 567 (80.6%) were PVCs, 111 (15.8%) were CVCs (including 65 nontunneled CVCs, 16 dialysis catheters, 15 PICCs, 7 skin‐tunneled CVCs, 5 totally implantable CVCs, and 3 Swan‐Ganz catheters), and 25 (3.5%) were ACs. The distribution according to hospital ward and anatomic site of insertion is detailed in Table 2. The use of CVCs and ACs was higher in ICUs (42.0% and 28.4% of all catheters in place, respectively) compared with conventional wards (12.0% and 0.0%, P<0.0001). The use of the subclavian vein insertion site was more common in medical wards (65.7% of all CVCs, excluding PICCs) than in surgical wards or ICUs (26.9%, P=0.0002). Most of the catheters had been inserted by nursing staff members (75.2%), followed by anesthesia physicians (13.4%) and critical‐care medicine physicians (4.8%). An opaque gauze or transparent polyurethane insertion‐site dressing was present in 378 (53.8%) and 319 (45.3%) catheters, respectively, with no significant differences according to the type of device or hospital ward.

After excluding ACs, we found an overall mean number of 2.060.82 lumens per catheter (1.860.45 per PVC and 3.091.39 per CVC), with significant differences between ICUs and conventional wards (P<0.0001; Table 2). There was a mean of 0.860.57 3‐way stopcocks per catheter. The mean number of concurrent IV medications per patient was 2.82.7 (ranging from 2.32.1 in those with a single catheter to 10.52.6 in those with 4 catheters). The most commonly administered medications were antimicrobials (46.6% of patients with a vascular catheter), fluid therapy (33.4%), chemotherapy (2.3%), and vasoactive and inotropic drugs (1.0%). According to the administration regimen, 455 (79.1%), 30 (5.2%), and 182 (31.6%) patients were receiving medications by rapid infusion, infusion over 124 hours, or continuous infusion over a 24‐hour period, respectively. In 57 patients (9.9%), the catheter was used only as preemptive vascular access. No apparent indication for the use of a vascular catheter was found in 63 patients (10.9%).

Based on our criteria, 126 out of 834 inpatients (15.1%, 95% CI: 12.817.7) had 1 inappropriate catheter, with significant differences between conventional wards and ICUs (13.2% vs 26.3%, P=0.014). This prevalence rate increases to 21.9% (95% CI: 18.725.5) when only patients with 1 vascular catheter in place were analyzed.

Focusing on the number of catheter lumens, 631 out of 1448 (43.6%, 95% CI: 41.046.1) were considered unnecessary. There was a nonsignificant trend toward a higher rate of unnecessary lumens in conventional wards compared with ICUs (44.8% vs 39.4%, P=0.086; Table 3). Because some centers have policies requiring all inpatients to harbor 1 PVC in place throughout the entire hospitalization period, we performed a first sensitivity analysis in which we assumed that having a single functional vascular lumen was appropriate in all cases, regardless of any other criteria. Under this assumption, only 248 lumens (17.1%, 95% CI: 15.319.2) could be regarded as unnecessary. We conducted a second sensitivity analysis by including in the rate denominator only those catheter lumens potentially removable (eg, PVCs, nontunneled CVCs, and noncoring needles inserted in Port‐A‐Cath catheters). By applying this method, 48.6% of lumens (631 out of 1298, 95% CI: 45.951.3) could be considered inappropriate.

DISCUSSION

In this cross‐sectional survey, we found that 1 out of every 5 (20%) adult inpatients with a vascular catheter in place in our tertiary‐care center had an inappropriate number of catheters. This figure increased to 43.6% when the number of catheter lumens was analyzed (or 17.1% if we assumed that all patients should have at least 1 vascular access during their hospitalization period solely on the basis of preemptive reasons). Such rates of unnecessary catheter use throughout an entire institution offer an opportunity for improvement in clinical practice and, eventually, for reducing catheter‐related morbidity.

Other authors have also assessed the adequacy of CVC use in either ICU[16, 17] or non‐ICU settings.[14, 15, 16, 18, 21] A recent hospital‐wide survey found that 4.8% of catheter‐days were unnecessary, with a higher proportion in conventional wards than in the ICU,[16] mirroring the results from previous studies.[14] Another prospective study, limited to conventional wards, reported that almost half of the patients had 1 day with inappropriate vascular‐device use; age, total number of catheters used, and duration of catheterization were significantly associated with this event.[21] On the contrary, compliance with the criteria drawn up by the HICPAC and the Infusion Nurses Society for PICC use was found to be high overall in a medium‐sized community hospital.[22] Interestingly, we found a differential pattern in the adequacy of catheter use between hospital areas in function of the variable analyzed: number of inappropriate vascular catheters (higher rate in ICUs) or number of unnecessary vascular lumens (higher rate in conventional wards). Although our assessment criteria may partially account for such differences (ie, drug infusions for >1 hour justified the use of an exclusive catheter lumen), this finding raises the question of whether future interventions should be aimed at modifying specific catheter practices according to the type of hospital ward.

In contrast to the amount of literature on CVC, there is a scarcity of studies evaluating the appropriateness of PVC use in clinical practice. Lederle et al. reported that 17% of patients admitted in conventional wards of a university hospital had an idle PVC, with 20% of patient‐days of catheter exposure considered unnecessary.[12] The same authors subsequently demonstrated a significant decrease in these figures by implementing a multidisciplinary quality‐improvement intervention.[13] A previous cross‐sectional survey in our center revealed a PVC use rate as high as 46.2% among non‐critically ill adult inpatients.[23] Phlebitis is a common complication of PVC use, occurring in about 7% of inpatients and usually leading to catheter removal and replacement.[24] Although at a much lower incidence, peripheral catheterization also represents a non‐negligible source of CRBSI.[25] In our institution, in which a recently implemented specific bundle has resulted in a clear improvement in CVC care,[4, 20] about 60% of CRBSI occurring during the first 3 months of 2013 were due to PVCs (unpublished data). Therefore, this type of device should be routinely included in future surveys seeking to investigate the local epidemiology of catheter use at each institution. In that sense, it should be noted that a recent clinical trial showed no benefit of routine third‐day replacement vs clinically indicated replacement for phlebitis or CRBSI.[24]

Our study suggests that the daily review of the need for maintaining the vascular catheter should take into account the number of vascular lumens, as >40% of them were deemed unnecessary. To our knowledge, this area for potential intervention has not been addressed in previous surveys. Numerous studies have long demonstrated that the use of double‐lumen or triple‐lumen CVCs is associated with a higher rate of CRBSI than single‐lumen devices.[6, 7, 8] Even though a meta‐analysis concluded that this relationship diminishes when only high‐quality studies were included,[9] a more recent prospective study reported a hazard ratio for infection of 4.4 for each additional lumen.[10] However, it might be argued against our decision to count each inflow port in multiway stopcocks as a separate vascular lumen. Because the present survey was ultimately aimed at identifying opportunities to reduce the risk of CRBSI by decreasing catheter exposure, such an approach was chosen to properly capture and quantify every single potential source of infection in catheterized patients. We hypothesize that the use of 3‐lumen stopcocks could involve an increased number of manipulations, thus jeopardizing the integrity of the insertion‐site dressing and subsequently favoring the intraluminal bacterial colonization of the common catheter. Although the current guidelines do not provide specific recommendations regarding the number of lumens in devices other than CVCs,[3, 19] the potential benefit of reducing this figure to the minimum in PVCs and ACs should also be assumed. In our opinion, specific efforts have to be focused on improving the use of 3‐way stopcocks, as we found a mean of 1.86 lumens per PVC and >3 lumens per CVC in our study. Maybe the need for 3‐way stopcocks should be reassessed on a daily basis in a similar way as that recommended for temporary CVCs.[3, 19] By eliminating unnecessary vascular lumens, the risk of CRBSI could be diminished without compromising the availability of vascular access for preemptive purposes.

The present surveillance also provides an accurate insight into the real‐life vascular catheter practices in a hospital‐wide setting, in contrast with most of the previous studies, which have been conducted in specific wards or units.[13, 15, 17] One relevant finding was the relatively low use of the subclavian vein site for central venous access (only 40.8% of all CVCs inserted), with significant differences between medical wards and the remaining hospital areas. Various studies have shown that the subclavian site is associated with a lower risk of infectious and thrombotic complications.[4, 26, 27, 28] Therefore, the HICPAC and IDSA guidelines strongly recommend using a subclavian site, rather than a jugular or a femoral site, to minimize infection risk for nontunneled CVCs.[3, 19] Nevertheless, recent studies have suggested that internal jugular and femoral sites could be acceptable when a subclavian approach is not feasible, particularly if chlorhexidine‐impregnated dressings are used and catheters are left in place for <4 days.[29]

The current study has a number of inherent limitations; the most significant is its cross‐sectional design, which precludes direct comparison of the rates of catheter use with other prospective cohort surveys.[14, 16, 21] In addition, we were not able to assess the changing dynamics of catheter use over time. In other words, the lack of use of a given device on the day of the survey does not necessarily imply inappropriateness. The criteria used to determine appropriateness of vascular catheterization were consensus opinion and not evidence‐based, a weakness shared by previous studies,[21] as current guidelines only address the indications for certain devices (ie, HICPAC and Infusion Nurses Society recommendations for PICC use).[3, 30] Although we have attempted to be liberal in accepting indications for catheter use (ie, preemptive access in patients deemed at risk of hemodynamic instability), some misclassification bias cannot be ruled out. In evaluating the adequacy of catheter lumens, we did not take into account the simultaneous delivery of incompatible infusateswhich must be infused through separate linesor other relevant variables (eg, nursing availability). Because the aim of our study was simply to determine whether a patient had an appropriate number of vascular catheters and vascular lumens in overall terms, all vascular lumens in each subject were individually counted and added regardless of the nature of the device, and therefore we were not able to disaggregate the adequacy rate by specific catheter types. Finally, the generalizability of the results may be hampered by their single‐center nature, and this limitation applies particularly to institutions with different policies than ours regarding preemptive vascular catheterization (ie, those requiring that all inpatients have at least 1 vascular lumen at any time during hospitalization).

On the other hand, some strengths of this survey merit consideration, namely its comprehensive design, capturing the entire population of adult inpatients in different hospital areas and every type of vascular catheter. Moreover, we addressed the adequacy of maintaining catheterization not only in terms of idle catheters in place, but also in terms of unnecessary lumens. In conclusion, there remains room for improvement in daily practice regarding the prompt removal of vascular catheters and vascular lumens that are no longer medically necessary. Further educational efforts among physicians and nursing staff should be targeted toward achieving this simple but effective measure to reduce the incidence of CRBSI.

Catheter‐related bloodstream infections (CRBSIs) are among the most common forms of hospital‐acquired infection and increase both length of stay and cost of hospitalization.[1, 2] Notable efforts are being devoted to reduce the rate of CRBSI, usually by implementing a bundle of measures.[3, 4, 5] One measure focuses on reducing to a minimum the exposure to vascular catheters.[3] In addition, various studies have shown that multilumen central venous catheters (CVCs) are associated with a higher risk of CRBSI than are single‐lumen catheters.[6, 7, 8, 9, 10] Accordingly, the Healthcare Infection Control Practices Advisory Committee (HICPAC) guideline recommends that clinicians use a CVC with the minimum number of ports or lumens essential for the management of the patient.[3]

Despite the fact that most CRBSIs occur in conventional wards,[1, 11] only a few studies have been focused on the potential magnitude of reducing the number of unnecessary vascular catheters and catheter lumens in the noncritical‐care setting.[12, 13, 14, 15, 16] The adequacy of vascular‐catheter use has been predominantly assessed for nontunneled CVCs.[14, 16, 17, 18] An institutional program aimed at reducing the overall rate of CRBSI should also include other sources, such as conventional peripheral venous catheters (PVCs), peripherally inserted central catheters (PICCs), and arterial catheters (ACs).[3] The need to extend surveillance to other types of catheters has been identified by the Infectious Diseases Society of America (IDSA) as an unresolved issue.[19]

We sought to investigate the rate and appropriateness of use of vascular catheters in the entire population of inpatients at a tertiary‐care center on a single day, as well as the adequacy of the number of catheter lumens harbored by each patient, by using a set of preestablished objective criteria.

METHODS

RESULTS

Out of 1134 reviewed inpatient beds, 834 (73.5%) were occupied on the day of the survey. The mean age of the included patients was 64.518.8 years, and 415 (49.8%) were male. Of these patients, 575 (68.9%) had 1 vascular catheter in place. The proportion of patients with a vascular catheter was significantly higher in ICUs compared with conventional wards (100% vs 66.7%, P<0.0001; Table 2). The overall numbers of vascular catheters and catheter lumens analyzed were 703 and 1448, respectively. Regarding the type of device, 567 (80.6%) were PVCs, 111 (15.8%) were CVCs (including 65 nontunneled CVCs, 16 dialysis catheters, 15 PICCs, 7 skin‐tunneled CVCs, 5 totally implantable CVCs, and 3 Swan‐Ganz catheters), and 25 (3.5%) were ACs. The distribution according to hospital ward and anatomic site of insertion is detailed in Table 2. The use of CVCs and ACs was higher in ICUs (42.0% and 28.4% of all catheters in place, respectively) compared with conventional wards (12.0% and 0.0%, P<0.0001). The use of the subclavian vein insertion site was more common in medical wards (65.7% of all CVCs, excluding PICCs) than in surgical wards or ICUs (26.9%, P=0.0002). Most of the catheters had been inserted by nursing staff members (75.2%), followed by anesthesia physicians (13.4%) and critical‐care medicine physicians (4.8%). An opaque gauze or transparent polyurethane insertion‐site dressing was present in 378 (53.8%) and 319 (45.3%) catheters, respectively, with no significant differences according to the type of device or hospital ward.

After excluding ACs, we found an overall mean number of 2.060.82 lumens per catheter (1.860.45 per PVC and 3.091.39 per CVC), with significant differences between ICUs and conventional wards (P<0.0001; Table 2). There was a mean of 0.860.57 3‐way stopcocks per catheter. The mean number of concurrent IV medications per patient was 2.82.7 (ranging from 2.32.1 in those with a single catheter to 10.52.6 in those with 4 catheters). The most commonly administered medications were antimicrobials (46.6% of patients with a vascular catheter), fluid therapy (33.4%), chemotherapy (2.3%), and vasoactive and inotropic drugs (1.0%). According to the administration regimen, 455 (79.1%), 30 (5.2%), and 182 (31.6%) patients were receiving medications by rapid infusion, infusion over 124 hours, or continuous infusion over a 24‐hour period, respectively. In 57 patients (9.9%), the catheter was used only as preemptive vascular access. No apparent indication for the use of a vascular catheter was found in 63 patients (10.9%).

Based on our criteria, 126 out of 834 inpatients (15.1%, 95% CI: 12.817.7) had 1 inappropriate catheter, with significant differences between conventional wards and ICUs (13.2% vs 26.3%, P=0.014). This prevalence rate increases to 21.9% (95% CI: 18.725.5) when only patients with 1 vascular catheter in place were analyzed.

Focusing on the number of catheter lumens, 631 out of 1448 (43.6%, 95% CI: 41.046.1) were considered unnecessary. There was a nonsignificant trend toward a higher rate of unnecessary lumens in conventional wards compared with ICUs (44.8% vs 39.4%, P=0.086; Table 3). Because some centers have policies requiring all inpatients to harbor 1 PVC in place throughout the entire hospitalization period, we performed a first sensitivity analysis in which we assumed that having a single functional vascular lumen was appropriate in all cases, regardless of any other criteria. Under this assumption, only 248 lumens (17.1%, 95% CI: 15.319.2) could be regarded as unnecessary. We conducted a second sensitivity analysis by including in the rate denominator only those catheter lumens potentially removable (eg, PVCs, nontunneled CVCs, and noncoring needles inserted in Port‐A‐Cath catheters). By applying this method, 48.6% of lumens (631 out of 1298, 95% CI: 45.951.3) could be considered inappropriate.

DISCUSSION

In this cross‐sectional survey, we found that 1 out of every 5 (20%) adult inpatients with a vascular catheter in place in our tertiary‐care center had an inappropriate number of catheters. This figure increased to 43.6% when the number of catheter lumens was analyzed (or 17.1% if we assumed that all patients should have at least 1 vascular access during their hospitalization period solely on the basis of preemptive reasons). Such rates of unnecessary catheter use throughout an entire institution offer an opportunity for improvement in clinical practice and, eventually, for reducing catheter‐related morbidity.

Other authors have also assessed the adequacy of CVC use in either ICU[16, 17] or non‐ICU settings.[14, 15, 16, 18, 21] A recent hospital‐wide survey found that 4.8% of catheter‐days were unnecessary, with a higher proportion in conventional wards than in the ICU,[16] mirroring the results from previous studies.[14] Another prospective study, limited to conventional wards, reported that almost half of the patients had 1 day with inappropriate vascular‐device use; age, total number of catheters used, and duration of catheterization were significantly associated with this event.[21] On the contrary, compliance with the criteria drawn up by the HICPAC and the Infusion Nurses Society for PICC use was found to be high overall in a medium‐sized community hospital.[22] Interestingly, we found a differential pattern in the adequacy of catheter use between hospital areas in function of the variable analyzed: number of inappropriate vascular catheters (higher rate in ICUs) or number of unnecessary vascular lumens (higher rate in conventional wards). Although our assessment criteria may partially account for such differences (ie, drug infusions for >1 hour justified the use of an exclusive catheter lumen), this finding raises the question of whether future interventions should be aimed at modifying specific catheter practices according to the type of hospital ward.

In contrast to the amount of literature on CVC, there is a scarcity of studies evaluating the appropriateness of PVC use in clinical practice. Lederle et al. reported that 17% of patients admitted in conventional wards of a university hospital had an idle PVC, with 20% of patient‐days of catheter exposure considered unnecessary.[12] The same authors subsequently demonstrated a significant decrease in these figures by implementing a multidisciplinary quality‐improvement intervention.[13] A previous cross‐sectional survey in our center revealed a PVC use rate as high as 46.2% among non‐critically ill adult inpatients.[23] Phlebitis is a common complication of PVC use, occurring in about 7% of inpatients and usually leading to catheter removal and replacement.[24] Although at a much lower incidence, peripheral catheterization also represents a non‐negligible source of CRBSI.[25] In our institution, in which a recently implemented specific bundle has resulted in a clear improvement in CVC care,[4, 20] about 60% of CRBSI occurring during the first 3 months of 2013 were due to PVCs (unpublished data). Therefore, this type of device should be routinely included in future surveys seeking to investigate the local epidemiology of catheter use at each institution. In that sense, it should be noted that a recent clinical trial showed no benefit of routine third‐day replacement vs clinically indicated replacement for phlebitis or CRBSI.[24]

Our study suggests that the daily review of the need for maintaining the vascular catheter should take into account the number of vascular lumens, as >40% of them were deemed unnecessary. To our knowledge, this area for potential intervention has not been addressed in previous surveys. Numerous studies have long demonstrated that the use of double‐lumen or triple‐lumen CVCs is associated with a higher rate of CRBSI than single‐lumen devices.[6, 7, 8] Even though a meta‐analysis concluded that this relationship diminishes when only high‐quality studies were included,[9] a more recent prospective study reported a hazard ratio for infection of 4.4 for each additional lumen.[10] However, it might be argued against our decision to count each inflow port in multiway stopcocks as a separate vascular lumen. Because the present survey was ultimately aimed at identifying opportunities to reduce the risk of CRBSI by decreasing catheter exposure, such an approach was chosen to properly capture and quantify every single potential source of infection in catheterized patients. We hypothesize that the use of 3‐lumen stopcocks could involve an increased number of manipulations, thus jeopardizing the integrity of the insertion‐site dressing and subsequently favoring the intraluminal bacterial colonization of the common catheter. Although the current guidelines do not provide specific recommendations regarding the number of lumens in devices other than CVCs,[3, 19] the potential benefit of reducing this figure to the minimum in PVCs and ACs should also be assumed. In our opinion, specific efforts have to be focused on improving the use of 3‐way stopcocks, as we found a mean of 1.86 lumens per PVC and >3 lumens per CVC in our study. Maybe the need for 3‐way stopcocks should be reassessed on a daily basis in a similar way as that recommended for temporary CVCs.[3, 19] By eliminating unnecessary vascular lumens, the risk of CRBSI could be diminished without compromising the availability of vascular access for preemptive purposes.

The present surveillance also provides an accurate insight into the real‐life vascular catheter practices in a hospital‐wide setting, in contrast with most of the previous studies, which have been conducted in specific wards or units.[13, 15, 17] One relevant finding was the relatively low use of the subclavian vein site for central venous access (only 40.8% of all CVCs inserted), with significant differences between medical wards and the remaining hospital areas. Various studies have shown that the subclavian site is associated with a lower risk of infectious and thrombotic complications.[4, 26, 27, 28] Therefore, the HICPAC and IDSA guidelines strongly recommend using a subclavian site, rather than a jugular or a femoral site, to minimize infection risk for nontunneled CVCs.[3, 19] Nevertheless, recent studies have suggested that internal jugular and femoral sites could be acceptable when a subclavian approach is not feasible, particularly if chlorhexidine‐impregnated dressings are used and catheters are left in place for <4 days.[29]

The current study has a number of inherent limitations; the most significant is its cross‐sectional design, which precludes direct comparison of the rates of catheter use with other prospective cohort surveys.[14, 16, 21] In addition, we were not able to assess the changing dynamics of catheter use over time. In other words, the lack of use of a given device on the day of the survey does not necessarily imply inappropriateness. The criteria used to determine appropriateness of vascular catheterization were consensus opinion and not evidence‐based, a weakness shared by previous studies,[21] as current guidelines only address the indications for certain devices (ie, HICPAC and Infusion Nurses Society recommendations for PICC use).[3, 30] Although we have attempted to be liberal in accepting indications for catheter use (ie, preemptive access in patients deemed at risk of hemodynamic instability), some misclassification bias cannot be ruled out. In evaluating the adequacy of catheter lumens, we did not take into account the simultaneous delivery of incompatible infusateswhich must be infused through separate linesor other relevant variables (eg, nursing availability). Because the aim of our study was simply to determine whether a patient had an appropriate number of vascular catheters and vascular lumens in overall terms, all vascular lumens in each subject were individually counted and added regardless of the nature of the device, and therefore we were not able to disaggregate the adequacy rate by specific catheter types. Finally, the generalizability of the results may be hampered by their single‐center nature, and this limitation applies particularly to institutions with different policies than ours regarding preemptive vascular catheterization (ie, those requiring that all inpatients have at least 1 vascular lumen at any time during hospitalization).

On the other hand, some strengths of this survey merit consideration, namely its comprehensive design, capturing the entire population of adult inpatients in different hospital areas and every type of vascular catheter. Moreover, we addressed the adequacy of maintaining catheterization not only in terms of idle catheters in place, but also in terms of unnecessary lumens. In conclusion, there remains room for improvement in daily practice regarding the prompt removal of vascular catheters and vascular lumens that are no longer medically necessary. Further educational efforts among physicians and nursing staff should be targeted toward achieving this simple but effective measure to reduce the incidence of CRBSI.

Catheter‐related bloodstream infections (CRBSIs) are among the most common forms of hospital‐acquired infection and increase both length of stay and cost of hospitalization.[1, 2] Notable efforts are being devoted to reduce the rate of CRBSI, usually by implementing a bundle of measures.[3, 4, 5] One measure focuses on reducing to a minimum the exposure to vascular catheters.[3] In addition, various studies have shown that multilumen central venous catheters (CVCs) are associated with a higher risk of CRBSI than are single‐lumen catheters.[6, 7, 8, 9, 10] Accordingly, the Healthcare Infection Control Practices Advisory Committee (HICPAC) guideline recommends that clinicians use a CVC with the minimum number of ports or lumens essential for the management of the patient.[3]

Despite the fact that most CRBSIs occur in conventional wards,[1, 11] only a few studies have been focused on the potential magnitude of reducing the number of unnecessary vascular catheters and catheter lumens in the noncritical‐care setting.[12, 13, 14, 15, 16] The adequacy of vascular‐catheter use has been predominantly assessed for nontunneled CVCs.[14, 16, 17, 18] An institutional program aimed at reducing the overall rate of CRBSI should also include other sources, such as conventional peripheral venous catheters (PVCs), peripherally inserted central catheters (PICCs), and arterial catheters (ACs).[3] The need to extend surveillance to other types of catheters has been identified by the Infectious Diseases Society of America (IDSA) as an unresolved issue.[19]

We sought to investigate the rate and appropriateness of use of vascular catheters in the entire population of inpatients at a tertiary‐care center on a single day, as well as the adequacy of the number of catheter lumens harbored by each patient, by using a set of preestablished objective criteria.

METHODS

RESULTS

Out of 1134 reviewed inpatient beds, 834 (73.5%) were occupied on the day of the survey. The mean age of the included patients was 64.518.8 years, and 415 (49.8%) were male. Of these patients, 575 (68.9%) had 1 vascular catheter in place. The proportion of patients with a vascular catheter was significantly higher in ICUs compared with conventional wards (100% vs 66.7%, P<0.0001; Table 2). The overall numbers of vascular catheters and catheter lumens analyzed were 703 and 1448, respectively. Regarding the type of device, 567 (80.6%) were PVCs, 111 (15.8%) were CVCs (including 65 nontunneled CVCs, 16 dialysis catheters, 15 PICCs, 7 skin‐tunneled CVCs, 5 totally implantable CVCs, and 3 Swan‐Ganz catheters), and 25 (3.5%) were ACs. The distribution according to hospital ward and anatomic site of insertion is detailed in Table 2. The use of CVCs and ACs was higher in ICUs (42.0% and 28.4% of all catheters in place, respectively) compared with conventional wards (12.0% and 0.0%, P<0.0001). The use of the subclavian vein insertion site was more common in medical wards (65.7% of all CVCs, excluding PICCs) than in surgical wards or ICUs (26.9%, P=0.0002). Most of the catheters had been inserted by nursing staff members (75.2%), followed by anesthesia physicians (13.4%) and critical‐care medicine physicians (4.8%). An opaque gauze or transparent polyurethane insertion‐site dressing was present in 378 (53.8%) and 319 (45.3%) catheters, respectively, with no significant differences according to the type of device or hospital ward.

After excluding ACs, we found an overall mean number of 2.060.82 lumens per catheter (1.860.45 per PVC and 3.091.39 per CVC), with significant differences between ICUs and conventional wards (P<0.0001; Table 2). There was a mean of 0.860.57 3‐way stopcocks per catheter. The mean number of concurrent IV medications per patient was 2.82.7 (ranging from 2.32.1 in those with a single catheter to 10.52.6 in those with 4 catheters). The most commonly administered medications were antimicrobials (46.6% of patients with a vascular catheter), fluid therapy (33.4%), chemotherapy (2.3%), and vasoactive and inotropic drugs (1.0%). According to the administration regimen, 455 (79.1%), 30 (5.2%), and 182 (31.6%) patients were receiving medications by rapid infusion, infusion over 124 hours, or continuous infusion over a 24‐hour period, respectively. In 57 patients (9.9%), the catheter was used only as preemptive vascular access. No apparent indication for the use of a vascular catheter was found in 63 patients (10.9%).

Based on our criteria, 126 out of 834 inpatients (15.1%, 95% CI: 12.817.7) had 1 inappropriate catheter, with significant differences between conventional wards and ICUs (13.2% vs 26.3%, P=0.014). This prevalence rate increases to 21.9% (95% CI: 18.725.5) when only patients with 1 vascular catheter in place were analyzed.

Focusing on the number of catheter lumens, 631 out of 1448 (43.6%, 95% CI: 41.046.1) were considered unnecessary. There was a nonsignificant trend toward a higher rate of unnecessary lumens in conventional wards compared with ICUs (44.8% vs 39.4%, P=0.086; Table 3). Because some centers have policies requiring all inpatients to harbor 1 PVC in place throughout the entire hospitalization period, we performed a first sensitivity analysis in which we assumed that having a single functional vascular lumen was appropriate in all cases, regardless of any other criteria. Under this assumption, only 248 lumens (17.1%, 95% CI: 15.319.2) could be regarded as unnecessary. We conducted a second sensitivity analysis by including in the rate denominator only those catheter lumens potentially removable (eg, PVCs, nontunneled CVCs, and noncoring needles inserted in Port‐A‐Cath catheters). By applying this method, 48.6% of lumens (631 out of 1298, 95% CI: 45.951.3) could be considered inappropriate.

DISCUSSION

In this cross‐sectional survey, we found that 1 out of every 5 (20%) adult inpatients with a vascular catheter in place in our tertiary‐care center had an inappropriate number of catheters. This figure increased to 43.6% when the number of catheter lumens was analyzed (or 17.1% if we assumed that all patients should have at least 1 vascular access during their hospitalization period solely on the basis of preemptive reasons). Such rates of unnecessary catheter use throughout an entire institution offer an opportunity for improvement in clinical practice and, eventually, for reducing catheter‐related morbidity.

Other authors have also assessed the adequacy of CVC use in either ICU[16, 17] or non‐ICU settings.[14, 15, 16, 18, 21] A recent hospital‐wide survey found that 4.8% of catheter‐days were unnecessary, with a higher proportion in conventional wards than in the ICU,[16] mirroring the results from previous studies.[14] Another prospective study, limited to conventional wards, reported that almost half of the patients had 1 day with inappropriate vascular‐device use; age, total number of catheters used, and duration of catheterization were significantly associated with this event.[21] On the contrary, compliance with the criteria drawn up by the HICPAC and the Infusion Nurses Society for PICC use was found to be high overall in a medium‐sized community hospital.[22] Interestingly, we found a differential pattern in the adequacy of catheter use between hospital areas in function of the variable analyzed: number of inappropriate vascular catheters (higher rate in ICUs) or number of unnecessary vascular lumens (higher rate in conventional wards). Although our assessment criteria may partially account for such differences (ie, drug infusions for >1 hour justified the use of an exclusive catheter lumen), this finding raises the question of whether future interventions should be aimed at modifying specific catheter practices according to the type of hospital ward.

In contrast to the amount of literature on CVC, there is a scarcity of studies evaluating the appropriateness of PVC use in clinical practice. Lederle et al. reported that 17% of patients admitted in conventional wards of a university hospital had an idle PVC, with 20% of patient‐days of catheter exposure considered unnecessary.[12] The same authors subsequently demonstrated a significant decrease in these figures by implementing a multidisciplinary quality‐improvement intervention.[13] A previous cross‐sectional survey in our center revealed a PVC use rate as high as 46.2% among non‐critically ill adult inpatients.[23] Phlebitis is a common complication of PVC use, occurring in about 7% of inpatients and usually leading to catheter removal and replacement.[24] Although at a much lower incidence, peripheral catheterization also represents a non‐negligible source of CRBSI.[25] In our institution, in which a recently implemented specific bundle has resulted in a clear improvement in CVC care,[4, 20] about 60% of CRBSI occurring during the first 3 months of 2013 were due to PVCs (unpublished data). Therefore, this type of device should be routinely included in future surveys seeking to investigate the local epidemiology of catheter use at each institution. In that sense, it should be noted that a recent clinical trial showed no benefit of routine third‐day replacement vs clinically indicated replacement for phlebitis or CRBSI.[24]

Our study suggests that the daily review of the need for maintaining the vascular catheter should take into account the number of vascular lumens, as >40% of them were deemed unnecessary. To our knowledge, this area for potential intervention has not been addressed in previous surveys. Numerous studies have long demonstrated that the use of double‐lumen or triple‐lumen CVCs is associated with a higher rate of CRBSI than single‐lumen devices.[6, 7, 8] Even though a meta‐analysis concluded that this relationship diminishes when only high‐quality studies were included,[9] a more recent prospective study reported a hazard ratio for infection of 4.4 for each additional lumen.[10] However, it might be argued against our decision to count each inflow port in multiway stopcocks as a separate vascular lumen. Because the present survey was ultimately aimed at identifying opportunities to reduce the risk of CRBSI by decreasing catheter exposure, such an approach was chosen to properly capture and quantify every single potential source of infection in catheterized patients. We hypothesize that the use of 3‐lumen stopcocks could involve an increased number of manipulations, thus jeopardizing the integrity of the insertion‐site dressing and subsequently favoring the intraluminal bacterial colonization of the common catheter. Although the current guidelines do not provide specific recommendations regarding the number of lumens in devices other than CVCs,[3, 19] the potential benefit of reducing this figure to the minimum in PVCs and ACs should also be assumed. In our opinion, specific efforts have to be focused on improving the use of 3‐way stopcocks, as we found a mean of 1.86 lumens per PVC and >3 lumens per CVC in our study. Maybe the need for 3‐way stopcocks should be reassessed on a daily basis in a similar way as that recommended for temporary CVCs.[3, 19] By eliminating unnecessary vascular lumens, the risk of CRBSI could be diminished without compromising the availability of vascular access for preemptive purposes.

The present surveillance also provides an accurate insight into the real‐life vascular catheter practices in a hospital‐wide setting, in contrast with most of the previous studies, which have been conducted in specific wards or units.[13, 15, 17] One relevant finding was the relatively low use of the subclavian vein site for central venous access (only 40.8% of all CVCs inserted), with significant differences between medical wards and the remaining hospital areas. Various studies have shown that the subclavian site is associated with a lower risk of infectious and thrombotic complications.[4, 26, 27, 28] Therefore, the HICPAC and IDSA guidelines strongly recommend using a subclavian site, rather than a jugular or a femoral site, to minimize infection risk for nontunneled CVCs.[3, 19] Nevertheless, recent studies have suggested that internal jugular and femoral sites could be acceptable when a subclavian approach is not feasible, particularly if chlorhexidine‐impregnated dressings are used and catheters are left in place for <4 days.[29]

The current study has a number of inherent limitations; the most significant is its cross‐sectional design, which precludes direct comparison of the rates of catheter use with other prospective cohort surveys.[14, 16, 21] In addition, we were not able to assess the changing dynamics of catheter use over time. In other words, the lack of use of a given device on the day of the survey does not necessarily imply inappropriateness. The criteria used to determine appropriateness of vascular catheterization were consensus opinion and not evidence‐based, a weakness shared by previous studies,[21] as current guidelines only address the indications for certain devices (ie, HICPAC and Infusion Nurses Society recommendations for PICC use).[3, 30] Although we have attempted to be liberal in accepting indications for catheter use (ie, preemptive access in patients deemed at risk of hemodynamic instability), some misclassification bias cannot be ruled out. In evaluating the adequacy of catheter lumens, we did not take into account the simultaneous delivery of incompatible infusateswhich must be infused through separate linesor other relevant variables (eg, nursing availability). Because the aim of our study was simply to determine whether a patient had an appropriate number of vascular catheters and vascular lumens in overall terms, all vascular lumens in each subject were individually counted and added regardless of the nature of the device, and therefore we were not able to disaggregate the adequacy rate by specific catheter types. Finally, the generalizability of the results may be hampered by their single‐center nature, and this limitation applies particularly to institutions with different policies than ours regarding preemptive vascular catheterization (ie, those requiring that all inpatients have at least 1 vascular lumen at any time during hospitalization).

On the other hand, some strengths of this survey merit consideration, namely its comprehensive design, capturing the entire population of adult inpatients in different hospital areas and every type of vascular catheter. Moreover, we addressed the adequacy of maintaining catheterization not only in terms of idle catheters in place, but also in terms of unnecessary lumens. In conclusion, there remains room for improvement in daily practice regarding the prompt removal of vascular catheters and vascular lumens that are no longer medically necessary. Further educational efforts among physicians and nursing staff should be targeted toward achieving this simple but effective measure to reduce the incidence of CRBSI.