Interventions aiming at reducing the incidence of central line-associated blood stream infections (CLABSI) have been well described in the literature, as strategies to reduce morbidity, mortality and associated cost with the provision of care (Bell et al., 2012; O'Grady et al., 2011; Gorski et al., 2016). Bundles of interventions targeting insertion and maintenance have demonstrated effectiveness (Bell et al., 2012). Maintenance intervention measures include assessment of insertion site with prompt dressing change when soiled, damp or non-adherent, reduction of unnecessary catheter access, appropriate disinfection of the connector with every access, removal of the catheter when no longer needed, and proper flushing of the device (Bell et al., 2012; O'Grady et al., 2011; Gorski et al., 2016; Ferroni et al., 2014).
CLABSI pathogenesis in centrally inserted central catheters (CICCs) is slightly different than in peripherally inserted central catheters (PICCs) where contamination occurs more often in the inner lumen of the catheter. Studies have demonstrated that, for CICCs in particular, external contamination at the insertion site of the vascular access most likely originates in the skin (Raad et al., 1993; Timsit et al., 2012). Literature suggests that transparent semipermeable dressings are preferred by both nurses and patients, as they require less-frequent replacement provided they remain clean (Treston-Aurand et al., 1997; de Barros et al., 2009). According to the Center for Disease Control and the Infusion Nurses Society (INS) guidelines, both gauze and transparent polyurethane dressings are recommended for protection of central venous access devices (CVADs) (O'Grady et al., 2011; Gorski et al., 2016).
Current guidelines recommend that dry dressings be changed every 48 hours, whereas transparent semi-permeable dressings can remain in situ for up to seven days, as long as they are dry and clean (O'Grady et al., 2011; Gorski et al., 2016). In our ICU, dressing care required usage of a transparent semipermeable dressing to be replaced when soiled, damp or loose, as well as every seven days, whichever came first.
This project took place at McGill University Health Centre, a university teaching hospital, that has a major regional trauma centre with a 22-bed ICU. Trauma patients account for 40 percent of admissions and the remaining admissions are medical or surgical. In this ICU, we have approximately 1,500 admissions per year. Since 2007, surveillance of CLABSI is performed year-round, following the National Health & Safety Network definition. All positive blood cultures are reviewed by the Infection Control Preventionist (ICP) and the Director of Infection Prevention and Control. Yearly rates of infection varied from 0.93 to 3.34/1,000 catheter days between 2007 and 2017, excluding chronic dialysis cases.
In 2016–2017, we noted an increase in the CLABSI rate at 2.36 per 1,000 catheter days with a total of 12 incidences of CLABSI in a total of 5,084 catheter days (Figure 1).
Figure 1. Central line-associated blood stream infection rates pre-implementation
To further understand the possible sources of the infection, an investigation consisting of audits and reviews of the CLABSI cases was undertaken.
We performed 51 audits (over a period of 3.5 months) on dressing integrity, presence of blood in the tubing or needlefree connector, need for an intravenous (IV) access, and vein accessed for the insertion of the CVAD. Despite a protocol that required dressings to be changed when soiled, results indicated that only 16% of the dressings were intact (clean and holding well). We also noticed there was no visible blood in 59% of the tubings and that 92% of the patients had a prescription for an IV infusion.
The review of CLABSI cases revealed that lines inserted in the trauma room, as emergencies, were not systematically reinserted once the patient's condition was stable and that most CICCs were inserted in the internal jugular vein. Insertions performed in the emergency department have been reported to influence the incidence of CLABSI (Gowardman, Robertson, Parkes, & Rickard, 2008). Table 1 describes the epidemiology of the cases. Furthermore, studies indicate an increased risk of colonization with CICCs inserted in the internal jugular vein compared with catheters inserted in the subclavian vein (O'Grady et al., 2011; Timsit et al., 2012; Arvaniti et al., 2017).
Table 1. Epidemiology of cases
Case number | ICU service (trauma, medicine, surgery) | Insertion site | Time to infection in days | Micro-organism category | Inserted in emergency department (Yes/No) |
---|---|---|---|---|---|
1 | Trauma | Subclavian | 7 | Enterococcus | Yes |
2 | Trauma | Subclavian | 9 | Multiple organism | No |
3 | Trauma | PICC | 11 | Skin contaminant | Yes |
4 | Trauma | PICC | 32 | Serratia | No |
5 | Surgery | IJ | 4 | Pseudomonas | No |
6 | Medicine | IJ | 4 | Candida | Yes |
7 | Medicine | IJ | 7 | Enterococcus | Yes |
8 | Medicine | IJ | 8 | Skin contaminant | No |
9 | Medicine | IJ | 9 | Skin contaminant | No |
10 | Medicine | PICC | 10 | Skin contaminant | No |
11 | Medicine | PICC | 15 | Pseudomonas | No |
12 | Medicine | IJ | 94 | Multiple organisms | No |
IJ=internal jugular; PICC= peripherally inserted central catheters
In our ICU, CICCs are inserted predominantly in the internal jugular vein (55%) compared with 35% in the subclavian and 10% in the femoral veins. However, clinically, those inserted in the jugular tend to be harder to protect adequately because of pistoning of the device into the insertion site with patient movement and gravity pulling on the dressing when the patient is sitting. In addition, the use of cervical collars promotes an increase in heat and humidity build-up under the dressing, and makes assessment difficult. Last, the presence of facial hair in males prevents or hinders adherence of the dressing.
Based on these findings, we conducted an additional literature review specifically on dressings for CVADs in ICUs, but we could not find results for this specific population. However, we found a review by Webster, Gillies, O'Riodan, Sherriff and Rickard (2015) that concluded a four-fold increase in the risk of CLABSI when polyurethane dressings were used to secure CVADs. The authors indicated the high uncertainty of this estimate and recommended more quality research in this topic. A discussion took place among the infection control preventionists, the vascular access nurse, physicians, bed-side nurses and nurse educators, and a change in practice regarding the type of dressing used for protection of all CVADs was proposed.
Methods
In March 2017, a short education session, lasting approximately 15 minutes, was given to ICU nurses. This in-service reiterated interventions that were already known by the nurses such as proper disinfection of the needle-free connector for 15 seconds, preferred use of phlebotomy for blood sampling when possible, and if the CICC had been placed in the emergency room, its replacement within 48 hours of the patient's transfer to the ICU. Additionally, specific instructions regarding dressing changes were given to replace transparent dressings (previously changed every seven days and when soiled or loose) with sterile gauze and non-woven tape dressings to be changed every 48 hours. No other intervention occurred during this period, for the type of catheter or its insertion site. All CICCs were chlorhexidine coated, and PICCs were non-coated. All CICCs were secured with sutures, whereas PICCs were stabilized with an adhesive securement device.
As described earlier, 51 audits were performed weekly by the ICP prior to the education sessions and twice a month thereafter. Audits included cleanliness of the insertion site, adherence of the dressing, and assessment of the tubing and connectors (free of visible blood). Statistical analysis was performed using a Mann-Whitney U test (http://www.socscistatistics.com/tests/mannwhitney/Default2.aspx). P-values less than 0.05 were considered statistically significant. The audit on dressing compliance is presented as a percentage of the catheters audited and the CLABSI rate is reported per 1,000 catheter days. The research ethics board review was deemed unnecessary as this was a quality improvement project, and the proposed intervention was based on the available literature and current guidelines and was recommended by the hospital infection control committee for prevention of CLABSI.
Results
Twelve in-services were given at various times to cover all shifts in the ICU. A total of 84 nurses (93%) received the in-service. The remaining nurses were all seen individually by a nurse educator.
A total of 242 audits were performed (51 pre-;191 post-). In the pre-intervention period, 16% of the dressings were dry and clean, compared to 85% in the post-intervention group (p<0.00001) (Table 2). Dressing adherence also improved from 63% to 91% (P <0.00001). There was no change in practice for the vein accessed.
Table 2. Central venous access device audits
Element audited | Pre (n=51) | Post (n=191) | P value |
---|---|---|---|
Dressing is clean (absence of blood or serous fluid under dressing) | 19.61% | 85.34% | <0.00001 |
Dressing is holding well (insertion site completely covered and dressing not peeling off) | 62.75% | 90.58% | <0.00001 |
No blood in tubing/connector | 58.82% | 54.97% | 0.67 |
From April 1, 2017, to March 31, 2018, zero cases of CLABSI occurred (total of 5,060 catheter days) compared to 12 events (rate of 2.36/1,000 catheter days) in the year prior to the practice change. This information was shared with the nurses during the weekly audits along with the results of the bimonthly audits. Additionally, after three months and again after one year, rewards were offered to the team to recognize the improvement in practice.
Discussion
Results of the initial audits indicate that dressings were soiled 84% of the time despite a protocol that required change of the dressing when needed. This suggests that it was not possible to maintain a transparent dressing clean and intact or that dressings were not changed when needed.
Challenges of maintaining a clean transparent dressing in the ICU
As mentioned previously, most CICCs in our intensive care units are inserted into the internal jugular vein. This access increases risk of pistoning, especially if the puncture is performed high in the neck. In a patient population often anticoagulated or with a bleeding disorder, movement of the catheter may contribute to an increased risk of bleeding at the insertion site, thus preventing the dressing from remaining clean and intact. A second challenge arises from the frequent use of cervical collars required by our patient population. The cervical collar does not enable rapid visualization of the dressing and promotes humidity build-up under the dressing. This can contribute to lifting of the dressing and preventing adequate adherence.
Challenges with ‘PRN’ dressing change
Changing of the dressing “when needed” requires that the planning of the ICU nurse be adjusted based on the assessment performed. In an environment where the patient's condition changes rapidly, it is possible that performing an activity when needed is more challenging than a planned activity, as per a calendar day. Three months after the practice changes, we asked the nurses if they wanted to go back to transparent dressings for the insertion sites that were dry and clean, as staff usually favours this type of dressing (Treston-Aurand, Olmsted, Allen-Bridson, & Craig, 1997; Barros et al., 2009). The majority of nurses mentioned they preferred dry dressings with routine changes. Nurses stated they felt dry dressings were holding better and were more appropriate for their population.
Previously published literature indicates unplanned dressing disruption is associated with an increased risk of CLABSI (Timsit et al., 2012). In our study, planned dressing changes, although more frequent, were associated with a significant reduction of CLABSI. Our study shows that dry dressings might be superior to transparent dressings for a trauma and medical/surgical ICU population, because of frequent oozing, bleeding disorders, use of jugular vein for insertion and frequent use of cervical collars. Our study also suggests that dry dressings may be more appropriate for this type of population as they reduced the risk of keeping a soiled, damp or unstuck dressing in situ.
Limitations
These results were obtained on a trauma-medical-surgical ICU population and may not be transferable to a different population. Additionally, the authors recognize that audit, feedback and reward may have contributed to the reduction of CLABSI, as these interventions have been associated with small, but potentially important practice change improvements (Damschroder et al., 2009; Ivers et al., 2012).
Conclusion
In this pre-post study, a simple change in dressing type was implemented, resulting in a significant reduction in the CLABSI rate. More studies are needed on the impact that routine versus “when needed” dressing changes have on nurses' adherence to protocols.