Abstract
Background and aims
Infective endocarditis is an important cause of morbidity and mortality in patients receiving haemodialysis for established renal failure. We carried out a prospective audit of patients developing infective endocarditis in a single renal unit.
Methods and results
From 1 January 2000 to 31 December 2013, we collected data on all cases of endocarditis occurring in patients receiving haemodialysis at Monklands Hospital, Airdrie. Twenty-nine patients developed endocarditis during our audit period. Twenty-three (79.3%) of the patients had pre-existing cardiac valve abnormalities such as regurgitation or calcification. Staphylococcus aureus was the most common microorganism cultured from the blood of 22 patients (75.9%). MRSA bacteraemia was identified in eight of these patients and all eight patients died during that first presentation. Different strategies were introduced within the unit during the audit period aiming to reduce the rate of bacteraemia. Since 2011, a successful strategy has been introduced under the auspices of the Scottish Patient Safety Programme. This led to our Staph aureus bacteraemia rate related to non-tunnelled venous catheters going from an average of 15 days between episodes to having had no episodes between 2 December 2011 and the end of the study period (760 days). This also appears to have had a positive impact on reducing the rate of endocarditis.
Conclusion
Infective endocarditis remains a devastating consequence of bacteraemia in patients receiving haemodialysis. An effective strategy aimed at reducing the rate of bacteraemia appears to have a similar effect on the rate of endocarditis.
Introduction
Infective endocarditis (IE) is an important cause of mortality and morbidity in haemodialysis (HD) patients.1,2 The median age of patients starting renal replacement therapy in Scotland is 65 years. 3 This group of patients is more likely to have abnormal cardiac valves as result of degenerative changes and calcification 4 which makes them more susceptible to development of IE.5,6 The type of vascular access used for dialysis contributes to the risk with a much higher risk of IE in patients receiving HD via catheters compared to arterio-venous fistulae (AVF). 6 We report an observational prospective cohort series of HD patients from a single centre who developed IE over a period of 13 years.
Materials and methods
From 1 January 2000, each patient with established renal failure (ERF) in our dialysis unit who developed IE was followed up from the date of diagnosis to the end of 2013. A combination of demographic and clinical data was collected. Demographic data included age, sex, cause of renal failure, duration of HD before the diagnosis of IE, vascular access, and the presence of any pre-existing cardiac valve lesions as reported by previous echocardiography. Clinical data included laboratory investigations, microbiological results, echocardiography findings and whether surgical valve repair was required. Data were analysed using Microsoft Excel for Windows 2010.
Diagnosis of IE was confirmed in all patients with positive blood cultures and echocardiography examination and therefore met at least two of the Duke University major criteria for the diagnosis of IE. 7 It was routine for most dialysis patients to undergo echocardiography if they developed Staph aureus bacteraemia (SAB) particularly if they received dialysis via venous catheter or had clinical signs of IE. Since 2010, a more formal policy was adopted and all patients with SAB now have echocardiography. Trans-oesophageal echocardiography was used to confirm the diagnosis when transthoracic echocardiography was equivocal. We provide the only inpatient dialysis for our region and therefore most of our dialysis patients who become seriously unwell are treated in our unit. Our relatively small medical team knew the study and alerted the authors to any new cases. As such we have a high degree of certainty that we identified every case of endocarditis during the study period.
Several changes took place within the unit during the study period, which are noteworthy when interpreting our data. These include changes to the number of prevalent patients receiving dialysis, changes in cannulation technique, and several initiatives aimed at reducing rates of bacteraemia. Our prevalent HD population almost doubled during the study period. Since 2007, several strategies were introduced for our patients with tunnelled catheters with the aim of reducing rates of bacteraemia and in turn, deep-seated infections such as endocarditis. These measures included improved exit site care with use of stronger 2% chlorhexidine solution and less frequent dressing changes (weekly) coupled with the use of chlorhexidine patch over the exit site. In October 2009, we started using citrate 46.7% ‘locks’ within each lumen of the catheter after dialysis as an anti-microbial agent. This led to a reduction in bacteraemia rates but was not sustained as the use of citrate dropped due to unacceptable side-effects.
Since August 2011, we modified our approach by adopting a series of improvements under the auspices of the Scottish Patient Safety Programme (SPSP – http://www.scottishpatientsafetyprogramme.scot.nhs.uk/). This national initiative was launched in 2008 with the aim of improving the safety and reliability of healthcare across different areas. The initial focus was on the practices in intensive care including those related to use of central venous catheters. 8 The lessons learned were felt to be applicable to our patients when we provide HD. The adoption of this approach including changing our ‘lock’ solution to taurolidine led to a dramatic and sustained drop in line-related bacteraemia episodes. During the study period, the last bacteraemia associated with a non-tunnelled line was on 2 December 2011. During 2012–2013, there was a period of over 400 days without any SAB related to a tunnelled dialysis line.
We also changed our practice regards AVF care. We moved to the use of 2% chlorhexidine to cleanse the skin but also moved to using the ‘buttonhole’ technique for inserting needles instead of the ‘rope-ladder’ technique in late 2009. The buttonhole method uses the exact same tract for each session whereas the rope-ladder technique rotates the area needled. The buttonhole technique is considered to lead to fewer complications 9 but is associated with potential increase in bacteraemia particularly if adequate care is not taken to prepare the needling site each time. 10 An audit of our buttonhole technique demonstrated a modest increase in bacteraemia following the move to buttonhole but the rate was still very low (0.095 episodes per 1000 HD days). Also, throughout this study, our main source of bacteraemia (and endocarditis) was related to the use of tunnelled cuffed venous catheters (TCVCs).
Results
Demographic data of patients developing infective endocarditis (n = 29).
SD: standard deviation; HD: haemodialysis; ERF: established renal failure; AVF: arterio-venous fistulae.
Twelve patients had more than one abnormal valve.
Two patients had no previous echocardiographic examination prior to this illness.
Microorganisms cultured from each patients and cardiac valves involved with infective endocarditis.
MSSA: methicillin-sensitive Staphylococcus aureus; MRSA: methicillin-resistant Staphylococcus aureus.
This patient was on immunosuppressive treatment for cardiac transplant.
All patients had positive blood cultures. The causative microorganism in each case and patient outcome are shown in Table 2. Staphylococcus aureus was the most common microorganism in 22 patients (75.9%). MRSA bacteraemia was identified in 8 patients (27.6%), and 14 patients (48.3%) had methicillin-sensitive Staphylococcus aureus (MSSA) bacteraemia. Coagulase-negative staphylococcus was cultured in five patients (17.2%). Streptococcus viridans was isolated from one patient (3.4%). Candida guilliermondii was isolated in one patient (3.4%) who was receiving immunosuppression for cardiac transplant.
Figure 1 demonstrates the annual rate of IE during the study period as well as causative organism and prevalent number of HD patients at the end of each year. During the study period, the number of patients receiving HD has almost doubled. The three-year period 2004–2006 appeared to have the highest annual rate of IE.
Annual number of infective endocarditis episodes and causative organism between 2000 and 2013.
Five of the patients (17.2%) required surgical intervention for valve repair. Antibiotic used for treatment was based on microbiology advice and the result of blood culture sensitivities. Historically, vancomycin and gentamicin were most commonly advised although some patients required rifampicin to be added as a second agent. Since early 2012, on microbiology advice, we adopted a more formal policy of treating MSSA bacteraemia with intravenous flucloxacillin for two weeks.
Eleven patients (37.9%) died during initial hospital admission, and a further 10 (34.5%) died within a year of hospital discharge. All patients with MRSA IE died during the first admission. Eight patients (27.6%) survived more than one year after the episode and six (21%) were still alive up to the end of last follow-up with the longest survival being just under five years.
Five patients (17.2%) required surgical intervention due to the severity of damage to the cardiac valves. In one case, the valve was a metallic prosthesis which did not respond to antibiotic therapy alone. The group of patients who did not have surgical intervention either responded to antibiotic treatment or were considered too high a risk for surgical treatment.
Discussion
This observational study provides long-term follow-up of patients experiencing IE from one of the largest single renal units in Scotland. The study confirms the high incidence of IE as well as the devastating consequences. It also highlights the positive impact of a programme aimed at reducing bacteraemia episodes.
Most patients (79.3%) in this cohort who developed IE had documented pre-existing cardiac abnormalities; a further two (6.8%) had a prosthetic valve. These abnormalities are likely to have increased the risk of developing IE as reported in other centres. 5 The median age of patients starting dialysis in our renal unit over the last decade is 63.2 (IQR 48.5, 72.3) which is consistent with national data. 3 Older patients are likely to have a higher prevalence of degenerative cardiac changes including cardiac valvular calcification which is known to increase the risk of IE. 11
IE is often preceded by an episode of bacteraemia. Twenty patients (69.0%) in our series were receiving HD via a TCVC at the time of development of IE. Although HD via non-tunnelled venous catheters poses the greatest risk of bacteraemia,12,13 the risk associated with TCVCs has been reported as being up to 7.6 times greater when compared with receiving dialysis via AVF. 14 The prevalence of use of TCVC in our institution has generally been very high. From 2007 onwards, we collected annual census data on vascular access use for the Scottish Renal Registry and our prevalent use of TCVC for dialysis access has varied between 35 and 55%. 3
Our patients with IE had a higher proportion of diabetic nephropathy and multi-system disease as their PRD than the general dialysis population in Scotland and fewer patients had interstitial nephritis. 3 This is perhaps not surprising as patients with diabetes and multi-system diseases tend to have more co-morbidity and as such a larger proportion of this group may not be fit for surgery or have limited options for AVF creation.
Our unit is one of the largest in Scotland serving a population of 550,000 over three hospital sites. We must consider the fact that some cases may have been missed. However, the fact that we work in a single inpatient unit and strive to admit our HD patients to our unit coupled with the prospective nature of our data collection means we are confident that we have identified all cases of endocarditis within our population. We have not carried out any comparison with the rest of our HD population as a control group and have instead provided a descriptive narrative of the predominant features of our patients who experienced endocarditis. We contend that this is legitimate since it is clear that the main focus on reducing endocarditis should be on reducing bacteraemia in all HD patients and that comparison by way of a control group would not add anything.
Staph aureus was the main cause of endocarditis in our cohort. It is known that IE with this organism can occur in normal cardiac valves in the general population although the frequency has varied considerably with quoted ranges of 3% to 25%. 15 It is also clear that bacteraemia with Staph aureus is associated with metastatic infection such as endocarditis in up to a third of patients. 15 That Staphylococcus aureus was the most frequently cultured microorganism (76%) is not surprising as we know that this is the most common microorganism causing bacteraemia in HD patients. 4 However, in this series, there was also a significant number patients with endocarditis associated with methicillin-resistant Staphylococcus aureus (MRSA).
The patients in our series had high within-hospital mortality (37.9%) and a further 34.5% died within a year of discharge. Only six patients (20.7%) had survived to the last follow-up. All patients with endocarditis associated with MRSA bacteraemia died during the initial hospital admission. This is similar to the general population where the mortality rate of MRSA endocarditis has been reported as 80% in one series. 16 The prognosis of IE in the general population has improved with a mortality rate of 30–40% in the 1960s reducing to 20–25% in recent years. 17 However, the prognosis of IE in HD patients compared to general population has remained poor. This is due to a combination of factors including cardiac abnormalities, often extensive co-morbidity as well as frequent potential exposure to organisms due to the dialysis process either via TCVC or AVF cannulation.
We are not aware of any recent series similar to ours from a UK-cohort but similar studies from Tunisia and China have demonstrated comparable results. Rekik et al. 18 showed that in 16 patients with IE, Staphylococcal species predominated (68.7%) with mitral involvement also being very common (56.2%) and in-hospital mortality rate of 43.7%. Chang et al. 19 reported on 18 HD patients and found 67% had Staph aureus and that the overall mortality rate was 61.1% but 100% for those with MRSA. With similar rates of causative organism and mortality, we consider that our data are generalisable to the wider HD population.
Within renal units, measures aimed at reducing bacteraemia have focused on promotion of arterio-venous access over TCVC as well as specific measures to reduce exposure to bacteria for all patients receiving dialysis regardless of the method of access. In patients where TCVC use is necessary, a rigorous approach to exit site care coupled with anti-microbial lock solutions should perhaps be viewed as the optimal approach.
Within Monklands renal unit since 2007, several different strategies have been introduced for our patients with tunnelled catheters. These appear to have been successful at reducing rates of endocarditis at a time when our dialysis population was increasing. It is difficult to know with any certainty which aspect has been the most useful but it is clear that it was not until we adopted the SPSP approach in 2011 that we achieved a sustained improvement in bacteraemia rates. During the study period, the last bacteraemia associated with a non-tunnelled line was on 2 December 2011, and during the two-year period 2012–2013, there was a period of over 400 days without any SAB related to a tunnelled dialysis line. It is pleasing to note that there is a corresponding drop in the number of cases of endocarditis during this same time period.
Conclusions
This prospective cohort study highlights the poor mortality associated with endocarditis in HD patients and in turn highlights the dangers of undergoing dialysis via a venous catheter. Despite improvements in the prognosis of IE in the general population the mortality rate is still much higher in the HD population. This study also demonstrates that a successful strategy to reduce overall bacteraemia rates can lead to a reduction in severe secondary infections such as IE.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.