Blood stream Infections in chronic hemodialysis patients - characteristics and outcomes

The aim of the present study was to investigate characteristics of BSI in HD patients – the incidence rates of infection, susceptible patients’ characteristics, causative pathogens and clinical outcomes. The main finding was a significant incidence of BSI events amongst HD patients with nearly half of these events caused by gram-negative bacteria. Increased age and diabetes were associated with an increased risk for mortality whereas Kt/V was associated with a decreased risk of mortality. In our cohort, 17.5% of patients developed BSI during follow up. This is a lower incidence than that reported by other studies. Rteil et al. reported an incidence rate of 32.7% BSI in their HD patients and Sahil et al. reported an incidence of 22.4% BSI in their HD patients [9]. This lower reported incidence of BSI was probably related by successful implementation evidence-based preventive infection control strategies which were intensified in HD units worldwide in recent years [15], such as strict adherence to hand hygiene protocols, screening and isolation of patients with MDR pathogens and Clostridium difficile, and sterile attendance to catheters. Despite these measures and similar to other reports almost a third of the BSI were due to resistant organisms [16, 17]. This is an important finding because resistant pathogens are associated with an increased risk of mortality, which may be due to the pathogenicity of the bacteria itself and/or inadequate initial empiric antibiotic therapy [18]. We observed a high percentage of gram-negative bacteria causing 46.3% of all BSI events in our cohort. The most frequently isolated pathogen was K. pneumoniae, found in 18.5%, while E. coli and P. aeruginosa were responsible for 14.8% and 9.26% BSIs, respectively. There is an increasing body of evidence demonstrating worldwide changes in the pattern of pathogens causing BSI in HD patients [19‐21]. The high prevalence of gram-negative bacteremia in our study is unusual as BSI in HD patients are mostly caused by gram-positive bacteria [16]. Gram positive bacteria commonly reside on the colonizing areas around the catheter site insertion commonly used to deliver HD. The colonization leads to biofilm formation and a nidus for virulent bacteria [22, 23]. The increase in gram-negative pathogens is probably multifactorial and could be a result of an increase in preventative measures mainly targeting gram-positive organisms. Current strategies, such as intranasal or dialysis catheter exit site application of mupirocin, have been shown to be very effective in preventing gram-positive bacterial infections, especially S. aureus infection [24, 25]. It is also possible that some of the gram-negative infections reported in our study were related to the spread of bacteria within the hospital (nosocomial) and inadequate infection control practices. Unfortunately, preventive strategies for gram-negative BSI are less well defined as the mechanism of acquiring these infections still warrants further study. In a cohort study on gram negative bacteria BSI amongst HD patients, the source of infection was more likely to be urinary or abdominal [20]. However, attempts of selective decontamination of the gastrointestinal tract and use of chlorhexidine washing were not consistently effective [25]. Nevertheless, some treatment strategies for central line associated BSI (CLABSI) proved to be more effective for gram-negative than for gram-positive bacteria. Indeed, the clinical success of an antibiotic lock protocol in eradicating CLABSI without removal of the catheter was as high as 87% for gram-negative infections compared to 40% and 75% for S. aureus and S. epidermidis, respectively [26, 27]. The finding of increased prevalence of gram-negative bacteria and a high percentage of drug resistant organisms may also have implications in the choice of empirical antibiotic treatment for BSI. The results support adding gram-negative coverage for empiric antibiotic therapy among HD patients with suspected BSI. The decisions regarding empirical antibiotic coverage should be taken according to local epidemiology to prevent over and underuse of antibiotics.

We identified bacteremia, older age and diabetes as independent risk factors for mortality amongst HD patients. These findings are concurrent with other studies that have previously described these risk factors in HD patients with BSI [28‐30]. We also identified risk factors for bacteremia. Of particular note a central line catheter. This risk factor is well known from other studies and emphasizes the importance of administrating HD through a fistula with or without a graft [29, 30]. Our findings of an increased risk with patients with a history of a cerebrovascular accident may be attributed to the fact that these patients are less mobile and have associated vasculopathy. There is limited data regarding the association between weight and risk of BSI in HD patients. Our findings that patients’ weight was mildly protective are congruent with other studies showing that lower body mass index has a significant association with adverse outcomes including infection-related death in similar populations such as those with end stage kidney disease and peritoneal dialysis [31, 32]. Our study has several limitations. This is a single center study so our results may not be highly generalizable. This study was not designed to investigate or compare different preventive strategies for CLABSI. However, our findings did change our local empiric antibiotic regimen for HD patients. We only assessed BSI and no other types of infections in HD patients. Patient included had a primary bacteremia. Patients underwent the standard evaluations related to focus of infections. However, some patients may have been included were the primary infection was from a secondary focus of infection. In this study we did not follow the clinical course of patients regarding metastasis of the primary bacteremia to other foci. The baseline vascular access was reported as per initial vascular access for HD treatment and therefore these is a predominance of catheter-based vascular access. However, the vascular access was changed to fistula or graft vascular access in many patients during the duration of HD treatment.

We also did not assess the source of infection, source control measures, or appropriateness of antibiotic therapy. Another limitation is that, due to missing data, we could not adjust for potential confounding factors such as antibiotic regimen, catheter indwelling time, and duration of hemodialysis in the logistic regression analysis. The data for cause of death was not available for many patients and was not included in this study. These limitations may be counterbalanced by several strengths. Due to fully computerized patients’ files in this center, there was no missing data. All results of microbiology tests were performed in a single laboratory highly certificated for reliability and consistency of results. Follow-up period was longer relative to other studies published on this topic [12, 33]. In this cohort there was a high percentage of patients with tunneled catheters and this may have increased the overall rate of infections and in particular the rate of gram negative infections. This may also affect the generality of the results.

In conclusion, this study demonstrated a high percentage of gram-negative bacteria causing 46.3% of all BSI events in our HD patients. This result reflects an increasing prevalence of gram-negative pathogens in the etiology of BSI in HD patients. Our findings highlight the need to identify local epidemiology in order to implement appropriate preventative measures and tailor empiric antibiotic therapy in HD patients.