Investigation of risk factors for tunneled hemodialysis catheters dysfunction: competing risk analysis of a tertiary center data

Suitable vascular access is the main concern in chronic hemodialysis patients. Tunneled central venous catheters (CVC) is a flexible tube with prolonged vascular access providing for the management of intravenous medication treatments, fluids, or total parenteral nutrition, repeated blood sampling, and hemodialysis (HD) [1]. CVCs are used as temporary access to the vascular until permanent access to the vascular (arteriovenous fistula [AV F] or arteriovenous graft [AVG]) can be placed [1]. Approximately 80% of newly hemodialysis patients require a CVC because they do not have suitable AFV to use or they have not had permanent access placed before dialysis initiation.

Identification and prevention of CVCs complications in hemodialysis patients (Common complications such as catheter infection and thrombosis) are serious to improving patient care. CVC’s main complications include infection, catheter dysfunction, and central vein obstruction (CVO). Infection associated with hemodialysis catheters is the most serious complication that may cause significant morbidity and mortality, systemic complications, hospitalizations, and considerable costs to the healthcare system [2]. Catheter dysfunction is defined as the lack of ability of a central venous catheter to: (1) doing a single dialysis period without causing recurrent pressure alarms or (2) generate hand over a mean dialysis blood flow of > 300 ml/min on two sequential dialysis periods or deliver a Kt/V of 1.2 in 4 hours or less [3]. Catheter dysfunction may be caused due thrombosis, fibrin sheath obstruction and mechanical complications. Central vein obstruction (CVO) is identified in patients with > 70% stenosis of a central vein by contrast venography, or upper limb edema on the same side, and having a history of a central venous catheter [3]. The incidence rate of CVCs complications is very wide because it can be determined by the definition of complications, patient population, units of measurement, duration of catheterization and follow-up, catheter location, and diagnostic methods [4]. The US Renal Data System delivers guidelines for coding complications of the catheter in hemodialysis patients [5].

Previous research about the variables influencing the complications of the tunneled central venous catheters occurred in hemodialysis patients is shown that the Localization of the catheter into the right internal jugular vein [1], duration of catheterization and diabetes [6] increase the risk of infection and also younger hemodialysis patients have a higher risk of infection [7]. However, the use of anticoagulant agents decreases the risk of thrombosis but is not significant [1]. Since there is a little knowledge about the variables that effect on the time until common complications of CVCs occurs, this study aims to find the effect of several risk factors on the time until thrombosis and infections of the catheter in hemodialysis patients. A better understanding of these factors may lead to having better survival and lower complication in the CVCs.

Throughout 1048 central venous catheterization cases in the study period, 466 patients were completely eligible for analysis (Fig. 1). Within these cases, 242(52%) patients experienced thrombosis, 123(23.4%) dysfunction happened due to infection, and the remaining catheter continued working into the end of follow-up period. The median (range) time for proper catheter working is 243(36–1131) days in total. This median (range) time were 217(36–1075), 247(37–1060), 486(36–1131) in thrombosed, infected and working catheter respectively.

There was no case of death before catheter dysfunction. Jugular catheterization was performed in 95.5% of patients whereas sub-clavian and femoral catheter was fixed in 2.4 and 2.1% cases respectively. Patients’ care level was 6.9, 50.5, 34.7 and 7.9% from level 1 to 4 respectively. Patients’ clinical characteristics are described in Table 1.

Single variable analysis showed that older patients’ are at higher risk of thrombosis and lower risk of infection (Table 2), although these differences were not observed when age is adjusted with other variables (Table 3). The female gender showed a considerably higher risk of thrombosis both in single and adjusted analysis (Odds ratio (OR) = 2.66, 95% Confidence interval (CI):1.77–4.00), whereas there was no relationship between gender and infection (Tables 2 and 3).

Multivariate regression for thrombosis, which competed by infection, demonstrated that female gender and hypertension predispose patients to thrombosis while DM is associated with a reduced risk of it. On the other hand, higher age and lower diastolic pressure are associated with a lower chance of infection in competing risk of thrombosis since diabetes makes patients more susceptible to infection (Table 4).

This single-center study tried to investigate many predisposing risk factors for catheter dysfunction with about 116,000 days follow-up through two types of a regression model. Findings proposed that female and hypertensive patients besides more obese individuals, are considerably higher at risk of catheter thrombosis. In contrast, diabetes is the most critical risk factor for infectious catheter dysfunction. The result of the competing risk model is more reliable due to the best adjustment.

Previous studies reported different rates of thrombosis. A similar percentage were observed in 46.7% of cases with hemodialysis catheter in the Sahli study [6]. In a study in Croatia, this rate was 16.8 in 1 year following up of patients with the central catheter for dialysis [9]. Develter et al. reported a 12% rate for thrombosis as a catheter dysfunction cause which is considerably lower than our results. No comorbidity was found as a predictor [1]. The difference in observed thrombosis rate might be affected by the length of patient follow-up and also the use of antithrombotic/thrombolytic protocols since Hemmelgarn et al. and also ward et al. reported a significant difference between recombinant tissue-plasminogen activator and heparin lock solution in thrombosis rate [10, 11].

The infection rate reported widely from 6 to 29% in the various surgical settings and also patients population [6, 12‐14] which is compatible with our findings of 26.4%. Sometimes the infection is the main reason for catheter removal, which differs from our results. This variety might be due to patients’ care level distribution and exposure to infectious organisms. In our study, care level 1 is directly associated with a lower risk of infection.

Multivariate analysis to assess the risk factors of thrombosis and also infection in patients with tunneled central catheters is performed frequently using Cox or logistic regression, but most studies used these methods for each endpoint separately [4, 6, 11, 15]. A cohort study compared the risk factors of death rate in tunneled hemodialysis catheter and aterio-venous fistula, which is different in final outcomes [2]. These risk calculation methods assume that thrombosis and infection are independent, and there is no mutual risk factor that confounds or interacts between these two endpoints outcomes. This assumption may be a risk of bias because it seems thrombosis or infection compete with each other in catheter dysfunction, and this competition should be considered in any factor analysis [16]. Therefore competing risk analysis is the better selection for adjusting the risk of different variables in the multivariate analysis approach.

On the base of the author’s knowledge, the competing risk regression model has not been used for risk factor analysis in hemodialysis tunneled catheters but it was used recently for peritoneal dialysis-associated factors [17].

Patients’ age in our study is lower compared with most previous studies especially compared in developed countries [1, 11, 12, 18] but this might be due to earlier incidence of end-stage renal disease in developing country which observed similarly in some reported articles [6, 13, 15]. Competing risk analyses have shown a higher risk of infection in younger patients, which is compatible with the results of Wang et al. study [15] but against Sahli et al. results which rejected the role of age in this outcome. The patients’ care level and outdoor activity should be considered to infer this association. On the other hand, age is associated with a higher chance of thrombosis adjusted with other risk factors, which was reported previously by Ward et al. [11].

A higher rate of thrombosis in females was reported considerably by Ward in the direction of our results. However, Pasara et al. reported a higher mortality rate in men due to catheter dysfunction [9]. Our results or previous studies did not find any association between infection and gender.

Our results show hypertensive patients have a considerably higher risk of thrombosis SHR = 1.4. Equivalent results were reported in the Ward study, where higher mean blood pressure at the time of dialysis is reported as a risk factor for thrombosis. Unfortunately, this variable is not available in our database, although to quantify the effect of hypertension on thrombosis, we include diastolic blood pressure at the time of catheter insertion in a multivariate competing regression model, but the results were not statistically significant. Pasara reported a lower chance of death in hemodialysis patients with previous hypertension history.

Higher BMI and obesity are frequently mentioned as risk factors for catheter thrombosis, and this is entirely compatible with the current study’s result [19].

The current study introduces new methods of risk factor analysis for hemodialysis catheter dysfunction whit competing for risk calculation which is recommended to be considered for future studies. Single-center study population made the intervention protocol more homogenous, although this might reduce its external validity. The study’s main limitation is its retrospective nature, leading to missing data, loss of patient follow-up, and a lack of good quality laboratory data registry.