Determinants and outcomes of access-related blood-stream infections among Irish haemodialysis patients; a cohort study

Patients on haemodialysis (HD) endure infection rates that are more than 26 times higher than that of the general population [1], and more than 100 to 200-fold higher for specific organisms [2]. They are the second leading cause of hospitalisation and mortality in the dialysis population [3‐5]. National and international guidelines along with national policy initiatives [6‐9] recommend the use of arteriovenous fistula (AVF) whenever possible, as the risk of infections and other complications is highest among patients using central venous catheters (CVCs) [3, 10, 11]. Despite the dangers associated with CVC use, these devices remain the principal type of access in the majority of HD patients in Ireland [12, 13] and internationally [14].

The alarmingly high rates of access-related bloodstream infections (AR-BSI) in patients undergoing dialysis with a CVC has forced changes in clinical practices that include better anti-infective protocols, increasing adoption of catheter lock solutions, and better anti-microbial surveillance protocols in order to reduce CVC-related infection rates [15‐18]. It is unclear, however, to what extent these changes have curbed the high rates of AR-BSI in the context of an increasing elderly HD phenotype with a high burden of complex health problems. It is also uncertain whether any benefit derived from these measures extends to very high-risk groups especially the elderly, patients with diabetes and those dialysed with a femoral CVC. While the formation of a functioning AVF is the preferred vascular access, this is not easily attainable in all individuals, especially elderly patients on HD [19]. Furthermore it remains controversial whether CVCs are superior to AVFs among elderly patients undergoing dialysis with a recent study finding lower rates of catheter-related bacteraemia in elderly patients compared to younger patients [18, 20‐22].

Within the Irish health system, data is lacking on the on the frequency and impact of AR-BSI in HD. The availability of such data along with clinical outcomes will help inform healthcare providers and policy-decision makers on access type, and will drive quality improvement initiatives to improve patient outcomes. We determined the rates of AR-BSI in a contemporary cohort of HD patients dialysed with a CVC or AVF and explored the relative contribution of demographic and clinical factors to overall rates of AR-BSI.

In this large centre-based study, we emphasise the significant risk of bloodstream infections associated with use of tunnelled dialysis catheters. Compared to patients who were using an AVF, patients with a CVC experienced a 20-fold higher risk of access-related bacteraemia. The risk associated with CVC use was independent of age and comorbid disease measured at baseline. Subgroup analysis confirmed that the pattern of risk from CVC was present in younger and in older patients, men and women, and in patients with and without diabetes. These results would suggest that despite advances in anti-infective protocols, innovative catheter designs, and the implementation of national guidelines, CVCs remain a major source of serious morbidity in HD patients.

The adverse impact of CVC over AVF on catheter-related bacteraemia rates was overwhelmingly apparent from this analysis. Our observed rates of AR-BSI events were 2.22 and 0.11 events per 100 patient-months for CVC and AVF respectively, a 20-fold difference. Our findings are concordant with reports from other parts of the world. AR-BSI rates in patients with CVC and AVF were 3.1 and 0.6, and 3.5 and 1.7 per 100 patient-months from Greece and Brazil respectively [4, 18]. A study from the National Healthcare Safety Network (NHSN) in the US reported pooled rates of 2.55 and 0.23 events per 100 patient-months for CVC and AVF respectively from 2007 and 2011 [27], while a more recent report suggested improvements with estimates of 1.83 and 0.16 for CVCs and AVF respectively [24]. The patient-months distribution in this last report was 19% for CVC and 63% for AVF, reflecting a much lower dependence on tunnelled catheters than in our Irish cohort. The rates of AR-BSI in our cohort compare favourably with those from the CDC report in the US [24] in that rates were at the 50th percentile for CVC-related BSIs, and below the 25th for AVF-related infections. Despite these reassuring statistics, there is emerging evidence that further improvements are possible. Hymes et al, reported significant reductions in AR-BSI to 0.67 per 100 patient-months with the introduction of antimicrobial barrier caps [28]. A further study by the CDC Dialysis BSI Prevention Collaborative, demonstrated a sustained reduction in CVC-related BSI’s from 2.26 to 1.08 events per 100 patient-months using a bundle of BSI-preventative interventions [29]. These encouraging findings suggest that there is further scope to reduce infection rates associated with CVC use and emphasise the need for sustained quality improvement initiatives.

Controversy currently exists as to whether tunnelled dialysis catheters should be considered a satisfactory access type for dialysis in older patients [30]. A lower rate of complications in older patients would support this approach. In support of this hypothesis, Murea et al found lower rates of catheter-related bacteraemia in patients above 75 years versus younger patients (1.67 versus 5.99 events per 100 patient-months respectively, HR 0.33 (95% CI 0.20–0.55) [17], citing lower rates of nasal colonisation, less sweating, and less mechanical stress on the catheter as potential reasons. Wang et al showed similar results [16]. However, several studies have found no association between age and AR-BSIs [18, 20‐22]. Furthermore, mortality risks (infection-related, cardiovascular-related, and all-cause) are higher in patients on dialysis by CVC even in elderly patients [31, 32]. The findings from our study are in direct contrast with those of Murea et al in that elderly patients experienced risks that were similar to those of younger patients.

The most significant factor associated with increased catheter-related BSIs in our population was site of the tunnelled dialysis catheter. In univariate and multivariable analysis, femoral access was associated with a fivefold increase in the rate of AR-BSI when compared to non-femoral access. This observation may relate to greater levels of skin contamination at the femoral area, relatively more difficult access for cleaning and observation, or may relate to some patient characteristics such as vintage or poor health. Femoral access is known to have higher rates of complications overall, including infection and malfunction [16, 17]. Quality improvement programs need to focus on this high-risk group of patients. We did not observe a difference in AR-BSI rates between patients with and without diabetes in univariate or multivariable analysis. Diabetes was found to be a risk factor for bacteraemia in some [16, 18] but not all studies [17]. Similarly, gender did not have an effect on infection rates, and this is consistent with multiple prior studies [16‐18].

Gram positive organisms were the predominant microbes from positive blood cultures in our cohort, with only a smaller proportion of AR-BSIs attributable to Gram negative (GN) organisms. Whereas studies from the US, Brazil, Greece and Singapore reported GN bacterial growth in 15 to 26% of positive cultures [17, 18, 24, 27, 33, 34], GN bacteria were identified in less than 5% of specimens in our study. CoNS were identified in more than 60% of cases, which is a higher rate than compared to published literature. CoNS, despite being common constituents of the normal flora of the skin, can be major nosocomial pathogens and cause significant morbidity in patients with CVCs [35]. Their spread is facilitated by poor hand hygiene and inadequate disinfection or sterilisation of instruments or surfaces [35]. It is difficult however to compare frequencies between different studies because of different study inclusion and exclusion criteria, and different definitions used. Our study also highlights the high burden of these events on both the patient and the health system. Two bacteraemia-related fatalities were identified. The majority of patients with AR-BSIs required hospitalisation, and catheter replacement was required in more than 50% of patients.

A few limitations are worth mentioning. The study was retrospective in design and thus not all known risk factors were measured at baseline. Patients receiving dialysis by CVC may be inherently different to those with AVF. Our data did not enable us to characterise patients beyond the variables used in our models. We did not capture exit-site and tunnel infections in our study. However, it should be noted that there is subjectivity in the definition of these events and these may or may not be associated with bacteraemia. The study reflects a single centre experience, which may limit generalisability. In addition, we did not differentiate between incident or prevalent HD patients in our study. Therefore, we must acknowledge we were unable to assess whether dialysis/ catheter vintage modifies the relationship with infection risk in those with CVC’s. Finally, our unit’s policy on management of access-related bacteraemia did not have a standardised protocol to check for clearance of bacteraemia prior to or shortly after discontinuation of the antimicrobial agent. This precluded conducting a reliable comparison of clearance duration. These limitations, however, were counterbalanced by several strengths. First, our study included all chronic HD patients who received dialysis at a large centre, and none of the patients had missing data. Second, all microbiology test results were available from a single central laboratory ensuring consistency and reliability of reporting. This was of concern in data reported from dialysis facilities in the US, as blood cultures were analysed at several different laboratories, particularly for samples obtained after hospitalisation [24, 36]. We reported on BSI, a measure that is based on an objective test, using standard definitions. Finally, the period of follow-up was long relative to other published studies in the literature.

AR-BSI remain a significant complication particularly among contemporary cohorts of patients undergoing haemodialysis by tunnelled catheters. The risk is present for all subgroups including the elderly. Access-related bloodstream infections impose a huge burden on patients and on health systems. Active surveillance of BSI linked to quality improvement initiatives should remain an integral part of all dialysis programmes to reduce catheter- associated infections and improve patient outcomes.