Impact of a multimodal intervention to reduce bloodstream infections related to vascular catheters in non-ICU wards: a multicentre study

Authors


Corresponding author: N. Freixas, RN, Infectious Disease Service, Hospital Universitari MútuaTerrassa, Plaça Dr. Robert 5, 08221 Terrassa, Spain
E-mail: nfreixas@mutuaterrassa.es

Abstract

Clin Microbiol Infect

Abstract

To determine the impact of a multimodal intervention designed to reduce the incidence of catheter-related bloodstream infections (CRBSIs) outside the ICU, we conducted a prospective, quasi-experimental, before-after intervention study in 11 hospitals participating in the VINCat programme in Catalonia, Spain. The intervention consists of: (i) an evidence-based bundle of practices relating to catheter insertion and maintenance; (ii) a training programme for healthcare workers; (iii) four point–prevalence surveys to track the status of the catheters; and (iv) feedback reports to the staff involved. The study included both central (CVC) and peripheral venous catheters (PVCs). Rates of CRBSI per 1000 patient-days were prospectively measured in 2009 (pre-intervention period) and 2010 (post-intervention period). The analysis included 1 191 843 patient-days in 2009 and 1 173 672 patient-days in 2010. The overall incidence of CRBSI decreased from 0.19 to 0.15 (p 0.04) and the incidence of CRBSI associated with a CVC decreased from 0.14 to 0.10 (p 0.004) after the intervention. The incidence in PVCs remained unchanged. There was a statistically significant improvement in the adequate maintenance of both CVCs and PVCs. Among the CRBSIs originating in PVCs, 61.8% appeared more than 72 h every insertion. There was a lower infection rate in the hospitals with a higher adherence to the recommendation to replace PVCs after 72 h. Our findings suggest that the implementation of intervention programmes similar to ours could have a major impact on patient safety by reducing the incidence of CRBSIs, and that routine replacement of PVCs might additionally prevent a significant number of bloodstream infections.

Introduction

Vascular catheter-related bloodstream infection (CRBSI) is one of the main forms of nosocomial bacteraemia [1,2]. It is associated with significant morbidity and mortality and has a high economic impact [1–5]. Most CRBSIs in hospitals occur in conventional wards [6].

Interventions aimed at reducing CRBSI in the ICU have shown that the application of a bundle of measures can result in a significant reduction in the incidence of CRBSI [7–9]. Recently, the need for strategies for preventing CRBSI outside the ICU has been emphasized [6,10,11]. However, few interventions have focused on conventional hospital wards. Measures to reduce CRBSI in this setting need to modify the practices of a large number of professionals regarding the management of both central venous catheters (CVCs) and peripheral venous catheters (PVCs).

The aim of this study was to evaluate the effectiveness of a multimodal intervention programme for preventing CRBSI in non-ICU wards. The primary endpoint was the reduction of CRBSI rates related to both CVCs and PVCs; the secondary endpoint was the improvement of adherence to preventive measures. The study was designed by the VINCat programme (http://vincat.gencat.cat), a standardized surveillance system providing risk-adjusted, procedure-specific rates of nosocomial infections in Catalonia, Spain.

Methods

Study design, study setting and patient population

This multicentre, prospective, quasi-experimental, before-after intervention study compared the baseline phase (2009 calendar year) and the intervention phase (2010 calendar year).

Eleven hospitals affiliated to VINCat took part in the study: four university hospitals with over 500 beds, and seven general hospitals with under 500 beds. All the adult patients hospitalized in acute care wards outside the ICU were included.

Intervention

Between January and December 2010 we carried out a multimodal intervention programme designed by a coordinating team appointed by the VINCat Technical Committee. Participating hospitals designated at least one physician and one nurse to take charge of the project at their centre. The intervention was divided into four sections.

(i) Proposing evidence-based recommendations [12–16] for the insertion and maintenance of CVCs and PVCs. Recommendations for CVCs included hand hygiene, preference for a subclavian or jugular insertion site rather than a femoral site, full barrier precautions (sterile full body drape, sterile gown, sterile gloves, hair cover, and mask with eye protection) for insertion, use of 2% chlorhexidine alcohol solution for skin antisepsis, disinfection of the connector before access, proper maintenance of the dressing and daily review of the need for catheterization. Antibiotic-coated catheters were not used in the participating hospitals. Recommendations for PVCs included hand hygiene, use of 2% alcoholic chlorhexidine for skin antisepsis, disinfection of the connector prior to access, proper maintenance of the dressing, daily review of the need for catheterization and replacement of short PVCs within 72 h of insertion (or 48 h for those inserted in the emergency department) or in the case of signs of inflammation or extravasation. For both types of catheter, a transparent semipermeable dressing was used to cover the catheter exit site during the study period. The programme was publicized through pocket leaflets and posters placed visibly in all the hospital wards, and through training sessions for the staff. In all, 1000 posters were distributed at the participating centres. (ii) Staff training. During the first quarter of 2010 several training sessions were held at each of the participating hospitals for medical and nursing staff. The sessions comprised a standard PowerPoint presentation highlighting the risks associated with CRBSI, CRBSI rates in Catalonia, the pathogenesis of infection and recommendations for prevention. Previously, training had been provided for the trainers via a meeting with the researchers at the different centres. Training sessions were backed up with a prior self-assessment questionnaire containing multiple choice questions on the epidemiology and prevention of CRBSI. Each hospital informed the coordinating centre of the sessions held, the number of participants, and the number of self-assessment questionnaires. (iii) Follow-up and evaluation of the process. The level of adherence to the recommendations was evaluated by the infection control teams at each centre by means of point-prevalence surveys, which reviewed all venous catheters inserted in non-ICU patients, using a structured grid for data collection, including the number of patients with a venous catheter, the number and type of catheters, and compliance with the recommended measures. Table 1 shows the items evaluated and the criteria used. Surveys were conducted at baseline, before the dissemination of the recommendations, and then every 2 months during the study. At the end of each of these controls, feedback was given to healthcare workers on the ward in the form of a report with the results on the rate of adherence to the recommendations and the problems identified. The results obtained at each hospital were sent to the coordinating centre for pooled analysis. (iv) Giving information to the health staff regarding all cases of CRBSI. The team responsible for patient care received written information about each case, and possible areas for improvement were discussed.

Table 1.   Criteria used to evaluate the catheter maintenance status at each point-prevalence study
Items evaluatedCriteria
  1. PVC, peripheral venous catheter.

Catheter was neededThe catheter had been used in the last 24 h, or procedures requiring its presence were scheduled.
PVC dwell time was correctThe catheter was in situ <72 h or <48 h if it was inserted in the emergency department.
Dressing was intactThe dressing was clean and adhering to the skin
The infusion set was correctNo blood was visible and a protective cap was in place if the equipment was disconnected.
No phlebitisAbsence of induration, pain or signs of inflammation at the insertion point or in the catheter trajectory.
Catheter use was recordedInformation in the clinical records included the date of insertion and catheter size.

Detection of cases of CRBSI and definitions

Surveillance of CRBSI episodes was conducted by a multidisciplinary team at each centre, which performed daily evaluation of hospitalized patients with positive blood cultures. Table 2 shows the criteria used for the diagnosis of CRBSI. All hospital-acquired bloodstream infections related to CVC (including peripherally inserted central venous catheters) or PVC (both short-line and mid-line) were included. CRBSI acquired outside the hospital and those originating in implantable vascular ports were excluded.

Table 2.   Definition of venous catheter-related bloodstream infection
  1. CFU, colony forming units.

Bacteraemia or fungaemia in a patient with a venous catheter with at least one set of positive blood cultures obtained from a peripheral vein (or two sets in the case of habitual skin-colonizing microorganisms).
and
 Clinical manifestations of infections (i.e. fever, chills, and or hypotension) in the absence of an apparent source of the bloodstream infection other than the catheter.
and one of the following:
 Positive semi-quantitative culture (>15 CFU/catheter segment) or quantitative (>10CFU/catheter segment) with the same microorganism as in the blood cultures obtained from the peripheral blood.
 Simultaneous quantitative blood cultures with a ≥5:1 ratio catheter vs. peripheral.
 Difference in time to positivity of the blood cultures of above 2 h between the cultures obtained from a peripheral vein and those obtained from the lumen of a catheter.
 A culture of the secretion in the insertion point shows growth of the same microorganism detected in the blood cultures.
 Resolution of clinical signs and symptoms after the withdrawal of the catheter with or without appropriate antibiotic treatment (this will be accepted as a condition if it was not possible to perform the above procedures).
 For the clinical diagnosis of peripheral venous catheter-related bloodstream infection, the presence of signs of phlebitis is required.

Statistical analysis

Incidence density of CRBSI was calculated as episodes of bloodstream infection per 1000 patient-days, according to the VINCat Program criteria [17]. Patient-days were obtained from administrative data. In addition, an estimation of CRBSI rate per 1000 catheter-days was carried out, as reported by Trinh et al. [18]. To estimate catheter-days, we obtained the total adult patient-days at each centre during the study period and then multiplied it by the mean prevalence rate of catheter use at each centre in the five point-prevalence studies conducted during the study.

Descriptive analysis was performed using means and standard deviation or median and interquartile range for continuous variables and percentages with 95% confidence intervals for categorical variables. To assess the impact of the intervention, we compared the incidence density of CRBSI obtained at all participating hospitals during 2010 with that obtained in 2009 using the incidence rates comparison test through the Z statistic. To assess the impact of the intervention on the implementation of preventive measures, we compared the adherence rates observed in the baseline point-prevalence study with those observed at the four subsequent pooled time-points using the proportion comparison test for independent samples through the Z statistic. In addition, we compared the median duration of catheter insertion time before bacteraemia for each type of catheter by year (2010 compared with that obtained in 2009) using the Mann–Whitney U-test. A p value of 0.05 or less was considered statistically significant. Statistical analyses were performed using spss V12.0 (SPSS Inc., Chicago, IL, USA) and EPIDAT V 3.1.

Results

The participating centres included 4304 beds, with a total of 1 191 843 patient-days in 2009 and 1 173 672 patient-days in 2010.

In total, 2176 healthcare workers participated in the training sessions held at the participating hospitals. In the surveys conducted before the training sessions, the mean score obtained was 58 out of a maximum of 100. The mean level of knowledge was higher for nurses (62 ± 19) than for physicians (57 ± 17) or nursing assistants (50 ± 23).

In the period after the multimodal intervention a statistically significant reduction was recorded in the overall incidence of CRBSI (Table 3). This decrease was due to a reduction in CRBSI originating in CVCs, particularly in hospitals with over 500 beds (Table S1 in supporting information), while the incidence of cases due to PVCs remained unchanged. When the incidence of CRBSI was calculated as episodes per 1000 estimated catheter-days, a similar significant reduction was observed in the overall incidence of CRBSI and in CRBSI originating in PVCs (results are shown in Table S2 in supporting information).

Table 3.   Catheter-related bloodstream infections in the pre-intervention (2009) and post-intervention (2010) periods
Catheter type20092010p
No. of CRBSINo. of patient-daysCRBSI per 1000 patient-days95% CINo. of CRBSINo. of patient-daysCRBSI per 1000 patient-days95% CI
  1. CRBSI, catheter-related bloodstream infections; CVC, central venous catheter (also including peripherally inserted central venous catheters); PVC, peripheral venous catheter.

CVC1661 191 8430.140.12–0.161151 173 6720.100.08–0.110.004
PVC541 191 8430.050.03–0.6611 173 6720.050.04–0.070.52
Overall2201 191 8430.190.16–0.211761 173 6720.150.13–0.180.04

The median duration of CVC before bloodstream infection was 13 days (interquartile range, 8–23 days) in 2009 and 11 days (interquartile range, 6–19 days) in 2010 (p 0.20). In PVCs, the corresponding figures were 4 days (interquartile range, 3–6) for both 2009 and 2010. In the case of short peripheral venous catheters, in which replacement was recommended after 72 h of insertion, 35 infections (63.3%) occurred after the first 72 h of insertion in 2009, while 34 infections (61.8%) appeared after this interval in 2010.

Table 4 shows the microorganisms responsible for CRBSI in the two periods. In the post-intervention period, the incidence of Gram-positive cocci CRBSI decreased significantly from 0.16 to 0.11 episodes per 1000 patient-days (p 0.003), mainly due to a reduction in CVC-BSI, whereas the incidence of Gram-negative bacilli CRBSI increased slightly from 0.02 to 0.03 episodes per 1000 patient-days (p 0.44). The incidence of Staphylococcus aureus CRBSI decreased from 0.05 to 0.04 episodes per 1000 patient-days, although the difference was not statistically significant (p 0.18). S. aureus was the main organism responsible for PVC-BSI, accounting for 68.4% in the pre-intervention period and 52.3% in the post-intervention period (p 0.07).

Table 4.   CRBSI causative microorganisms in pre-intervention and post-intervention periods
Microorganism20092010p*
Overall N (%)CVC N (%)PVC N (%)Overall N (%)CVC N (%)PVC N (%)
  1. *Overall BSI-CV 2009 vs. overall BSI-CV 2010.

  2. CVC, central venous catheter (also including peripherally inserted central venous catheters); PVC, peripheral venous catheter; CoNS, coagulase-negative staphylococci; MSSA, methicillin-sensitive Staphylococcus aureus; MRSA, methicillin-resistant Staphylococcus aureus.

Gram-positive microorganisms195 (82.6)145 (81.0)50 (87.7)138 (72.6)88 (69.3)50 (79.4)0.01
 CoNS113 (47.9)103 (57.5)10 (17.5)67 (35.3)55 (43.3)12 (19.0)0.01
 Staphylococcus aureus 71 (30.0)32 (17.9)39 (68.4)54 (28.4)21 (16.5)33 (52.3)0.78
  MSSA63 (26.7)29 (16.2)34 (59.6)49 (25.8)21 (16.5)28 (44.4)0.92
  MRSA8 (3.4)3 (1.7)5 (8.8)5 (2.6)5 (7.9)0.86
 Enterococcus spp.9 (3.8)8 (4.5)1 (1.8)13 (6.8)10 (7.9)3 (4.8)0.23
 Miscellaneous2 (0.8)2 (1.1)4 (2.1)2 (1.6)2 (3.2)0.49
Gram-negative microorganisms33 (14.0)26 (14.5)7 (12.3)40 (21.1)28 (22.0)12 (19.0)0.07
 Escherichia coli 3 (1.3)3 (1.7)4 (2.1)4 (3.1)0.77
 Klebsiella spp.12 (5.1)11 (6.1)1 (1.8)12 (6.3)6 (4.7)6 (9.5)0.73
 Proteus spp.2 (0.8)2 (1.1)1 (0.5)1 (0.8)0.85
 Pseudomonas aeruginosa 6 (2.5)6 (3.4)3 (1.6)3 (2.4)0.72
 Enterobacter species 7 (3.0)3 (1.7)4 (7.0)13 (6.8)10 (7.9)3 (4.8)0.09
 Miscellaneous3 (1.3)1 (0.6)2 (3.5)7 (3.7)4 (3.1)3 (4.8)0.18
Candida spp.8 (3.4)8 (4.5)12 (6.3)11 (8.7)1 (1.6)0.23
All microorganisms236 (100)179 (100)57 (100)190 (100)127 (100)63 (100) 
Polymicrobial bloodstream infection1613314122 

Table 5 shows the results regarding adherence to preventive measures observed in the point-prevalence studies. There was a statistically significant improvement in the implementation of the measures in the post-intervention period for both central and peripheral catheters. The measure with the lowest level of adherence was the recommendation to replace short PVCs at 72 h after insertion: in spite of this specific recommendation, 25.8% of PVCs remained in place beyond this time-point. The three participating hospitals with a lower proportion (<15%) of PVCs in place for more than 72 h, according to the follow-up point-prevalence studies, obtained a lower rate of CRBSI originating in PVCs, as compared with the eight hospitals with a higher proportion of PVCs in place for more than 72 h (0.02 vs 0.06 per 1000 patient-days; p 0.05).

Table 5.   Adherence to the recommended precautionary measures evaluated in the point-prevalence studies*
Item evaluatedCentral venous cathetersa Peripheral venous cathetersp
Baseline observation (266 catheters)Follow-up observationsb (877 catheters)Baseline observation (1809 catheters)Follow-up observationsb (5654 catheters)
No.% (95% CI)No.% (95% CI)pNo.% (95% CI)No.% (95% CI)
  1. *Results are expressed as percentage of catheters fulfilling the recommendation.

  2. aAlso including peripherally inserted central venous catheters.

  3. bThe four point-prevalence studies performed after the dissemination of the recommendations are pooled together.

  4. NA, not applicable.

Catheter was needed25495.5 (92.3–97.7)83194.8 (93.1–96.1)0.75151283.6 (81.9–85.3)483285.5 (84.5–86.4)0.05
PVC dwell time was correctNANANANA115663.9 (61.7–66.1)419374.2 (73.0–75.3)<0.001
Dressing was intact21279.7 (74.9–84.5)79190.2 (88.0–92.1)<0.0001140678.7 (75.8–79.6)490486.7 85.9–87.6)<0.001
The infusion set was correct23186.8 (82.8–90.9)77288.0 (85.9–90.2)0.68149182.4 (80.7–84.2)451479.8 (78.8–80.9)0.02
No phlebitis266100 (98.6–100)86798.9 (97.9–99.5)0.16176497.5 (96.7–98.2)548797.0 (96.6–97.5)0.33
Catheter use was recorded23186.8 (82.8–90.9)78289.2 (87.1–91.2)0.34144679.9 (78.1–81.8)484985.8 (84.9–86.7)<0.001
All the items evaluated were correct15959.8 (53.9–65.7)63172.0 (69.0–74.9)<0.00178543.4 (41.1–45.7)308054.5 (53.2–55.8)<0.001

Discussion

The implementation of a programme designed to decrease the rate of CRBSI originating in conventional hospital wards achieved a significant reduction over the period of 1 year. This decrease was due to the 30% fall in infections originating in central venous catheters, while the incidence of PVC-BSI did not change significantly. The programme also achieved a significant improvement in the adequate maintenance of intravenous catheters, evaluated in four point-prevalence studies after the start of the programme, especially regarding the state of preservation of the dressings and the replacement of PVCs. Although the improvement was observed for both central and peripheral catheters, the rate of catheters fulfilling all the items evaluated during the point-prevalence studies was lower in the PVC group compared with CVC group (54.5% vs. 72%), probably due to the high number of healthcare workers involved in the insertion and maintenance of PVCs and the generally low perception of the risk of complications associated with their use.

There are few published data on CRBSI rates outside the ICU. For CVCs, studies from different countries and settings showed a rate from 1.2 to 7.7 cases per 1000 catheter-days [19–22]. Few previous studies have shown reductions in the rate of CRBSI after introducing specific measures for catheter insertion and maintenance outside the ICU [23–25]. These studies are based on educational and feedback interventions. The widespread implementation of intervention programmes similar to ours could have a major impact on patient safety, with significant reductions in morbidity, mortality and cost. With regard to PVCs, the information is very limited. A systematic review reported an average incidence of 0.5 cases per 1000 catheter-days for short plastic catheters and 8.6 for steel needles [26]. In our study, as in other previous studies [18,27], a significant proportion of cases were caused by S. aureus, showing a potential for serious complications and high mortality [28–31]. Significantly, over 60% of CRBSI originating in short PVCs appeared more than 72 h after insertion of the catheter. Moreover, participating hospitals with a higher adherence to removing PVCs when recommended, obtained a lower rate of CRBSI originating in PVCs. Both facts suggest that strict adherence to the recommendation of routine catheter replacement could have prevented a significant number of cases, but often healthcare workers appear reluctant to follow this recommendation [32,33]. Although a recent meta-analysis indicated that routine replacement of peripheral catheters does not reduce the risk of infection [34], the study may have been underpowered to address the issue of CRBSI because the data analysed included only 8779 catheter-days, a figure too low to assess a risk of between 0.1 and 0.5 cases per 1000 days. Interestingly, the application of a bundle in a Irish hospital including the replacement of PVC within 72 h reduced cases of CRBSI due to S. aureus [35]. More recently, a controlled study of education and feedback in Detroit (USA), incorporating a recommendation to replace PVCs within 96 h, obtained a significant reduction in PVC-associated bloodstream infection [36].

Our study has several limitations. First, it is not a randomized study, with the resulting risk of bias due to confounding factors. However, we believe that the prospective multicentre nature of the study and the proximity in time of the baseline and post-intervention periods reduce this risk. Second, the CRBSI incidence is expressed per 1000 patient-days instead of catheter-days. However, continuous monitoring of all patients with venous catheters outside the ICU would have required a great deal of resources and entailed excessive costs to the participating hospitals. This prevents comparing our rates with other studies expressing the rates per catheter-days, and also may prevent identification of the dwell-time of CVC that benefited most from the intervention. Third, the methodology used to study adherence to the recommendations focused on the state of the catheters observed through the four prevalence studies, but did not include observational studies on adherence to the measures recommended at the time of catheter insertion.

In conclusion, this multimodal intervention significantly reduced the overall rate of CRBSI outside the ICU and the rate of CRBSI originating in CVCs. However, the intervention was unable to decrease the rate of CRBSI originating in PVCs. Our findings suggest that the implementation of intervention programmes similar to ours could have a major impact on patient safety and that routine replacement of peripheral catheters may additionally prevent a significant number of bloodstream infections.

Participating researchers and hospitals

Fundació Hospital Asil de Granollers (Jordi Cuquet, Mª Dolors Navarro). Consorci Sanitari de Vic (Mª Jesús Martínez, Josefa Rifa, Josep Vilaró). Hospital de Barcelona (Montserrat Sierra, Montserrat Vaqué, Marta Calsina). Fundació Privada Hospital de Mollet (Eva Redón, Rosa Mª Vidal, Elisabeth Mauri). Fundació Sanitària d’Igualada (Anna Marron, Montserrat Brugués). Hospital Universitari Mútua de Terrassa (Esther Calbo, Montserrat Riera, Carme Nicolàs). Consorci Sanitari de Terrassa (Núria Boada, Elena Espejo, Lurdes Pagespetit). Hospital Universitari de Bellvitge (Ana Hornero, Olga Arch). Hospital de la Santa Creu i Sant Pau (Virginia Pomar, Àngels Cotura, Joaquín López-Contreras). Corporació Sanitaria del Parc Taulí de Sabadell (Marta Fernández, Marta Piriz, Inmaculada Fernández). Hospital Universitari Joan XXIII de Tarragona (Montserrat Olona, Graciano García, Rosa Antúnez).

Acknowledgements

We thank Antonio Carrillo (Fellowship Universitat de Barcelona) for his contribution to data collecting. We are also grateful to J. Clapés and J. Casas (Information Systems, Catalan Health Service).

Funding

This study was supported in part by a grant from the National Ministry of Health (RD 829/2010).

Transparency Declarations

All authors have no conflicts of interest to declare with regard to this article.

Ancillary