Starting hemodialysis with catheter and mortality risk: persistent association in a competing risk analysis

Introduction

Despite improvements achieved in both the healthcare of chronic kidney disease (CKD) patients and hemodialysis (HD) technology in recent years, the clinical outcomes of this treatment are still influenced by the vascular access (VA) type in use (1). It has been suggested that the VA type could be involved in the variability of current mortality observed between countries (2) or between dialysis techniques (peritoneal dialysis vs. HD) (3). In this regard, bearing in mind the existing biases (4), initiating HD treatment using a central venous catheter (CVC) has been associated with higher risk of death over time compared with a mature fístula (AVF) (5-6-7-8-9-10-11-12-13-13).

The highest all-cause mortality rate of incident HD patients occurs during the first year of treatment (14) and, more specifically, within the first 120 days (early period) compared to 121-365 days (late period) after starting HD (15-16-17). It has been reported that CVC use at the HD inception is one of the factors independently associated with the increased mortality risk during the first year of HD treatment (15, 16). However, is this mortality risk constant throughout the whole first year of follow-up (early and late periods) for both types of CVC? It’s necessary to identify the temporal profile of mortality associated with tunneled (TCC) and untunneled (UCC) central catheters and also clarify its specific weight on the mortality risk.

Using the data from the Catalan Registry, we performed a retrospective observational study to assess the survival of chronic HD patients over time according to the initial VA and to analyze the VA profile of incident patients who died during the first year of follow-up.

Materials and methods

The Catalan Renal Registry (RMRC)

The data source used for this study is the Registry of Catalonia, an autonomous community of Spain with a population of 7,553.197 million residents in 2011. This Registry collects data from 46 centers about HD initiation, changes of center and/or treatment modality, deaths and patient exclusion, as well as annual update of all active cases on the 31^st of December of each year.

Patients and vascular accesses

We analyzed the VA of the incident HD patients from the 1^st of January 2000 until December 31, 2011. We have included all patients resident in Catalonia aged 18 years or older with end-stage renal disease (ESRD) starting chronic HD program as their initial form of renal replacement treatment (RRT).

The following baseline variables were analyzed: gender, age, primary kidney disease, associated comorbidities (coronary artery disease, cardiomyopathy, cardiac conduction disorders, cerebrovascular disease, peripheral vascular disease, chronic respiratory disease and diabetes not considered as primary kidney disease) grouped according to the International Classification of Diseases (ICD-9) (18). The duration of nephrology care before HD start was also considered, as well as ESRD presentation form (acute presentation of CKD not previously known or acute-on-chronic renal disease or steady progression of known CKD) and period of HD initiation (2000-2005 or 2006-2011).

The causes of death were codified according to the European Renal Association-European Dialysis and Transplant Association (ERA-EDTA) (19) codes and we have classified them into seven categories: unknown, cardiovascular, infection, liver or gastrointestinal disease, cancer, social (suicide, dialysis withdrawal) and miscellaneous.

The VA used at HD inception was classified as one of four types: AVF, arteriovenous graft (AVG), TCC and UCC. To assess the changes from the initial VA, we used the information collected in the two first annual follow-up reports at 31 December, in order to have at least one year follow-up in all cases. We determined the direction of changes comparing the initial VA with the VA used in the first and second follow-up at December 31.

Statistical analyses

Continuous and categorical variables were described as mean ± standard deviation and as percentages, respectively. We evaluated the differences between each type of VA and the variables studied using the analysis of variance and the chi-square test.

The risk of death for all-causes over time was obtained by means of a multivariate competing risk model where kidney transplantation, switch to peritoneal dialysis or renal function recovery acts as competing risk (20). This model was adjusted taking into account all potential confounding aforementioned baseline variables.

Considering the incident patients who died during the first year of follow-up, we describe the initial VA and the causes of death overall, during the first 120 days (early period) and during the period of 120-365 days (late period). For this purpose, although the patients considered for the study were those included until December 31, 2011, the observation period ended on December 31, 2012. The odds ratios (OR) for all-cause, cardiovascular- and infectious-related mortality during these periods, according to the initial VA type, were estimated using logistic regression models taking into account all the potential confounding variables previously outlined. We also analyzed the change in the adjusted all-cause mortality risk associated with the switch of the initial VA during the first year of follow-up. Patients starting HD with AVG and who died during the first year of follow-up were excluded from the final analysis of mortality because of the low number (n = 14).

Data analysis was performed using STATA software version 11.

Results

Patients included

During 2000-2011, 9956 patients met the inclusion criteria. Table I summarizes the characteristics of these patients by initial VA type at baseline. Considering globally the 12-year period, the VA distribution for initiating HD was the following: AVF 47.9% (4473), AVG 1.2% (116), TCC 15.9% (1582) and UCC 35.0% (3485). Their follow-up was as following: changing the RRT to peritoneal dialysis (1.8%) or kidney transplantation (23.7%), death (47.2%), recovery of renal function (1.4%), exclusion (0.1%) and remained on HD at December 31, 2011 (25.7%).

TABLE I -

Characteristics of the incident hemodialysis patients according to the initial vascular access during the study period

% Missing (n = 9956)			% UCC (3485)	% TCC (1582)	% AVF (4773)	% AVG (116)	p value*
AVF = arteriovenous fistula; AVG = arteriovenous graft; ESRD = end-stage renal disease; TCC = tunneled central vein catheter; UCC = untunneled central vein catheter.
*p<0.05.
		18-44	40.3	14.5	44.7	0.5	p<0.001
0	Age group (years)	45-64	32.6	13.8	52.3	1.3
		65-74	35.7	15.8	47.2	1.3
		>74	34.9	17.8	46.2	1.1
0	Gender	Male	34.4	14.8	50.1	0.7	p<0.001
		Female	36.1	17.7	44.1	2.1
	Primary Renal Disease (PRD)	Glomerular disease	36.7	13	49.2	1.1
		Polycystic kidney disease	16.2	7.3	74.7	1.8
		Interstitial disease	32.6	14.9	50.8	1.7
0		Vascular disease	35.9	15.6	47.4	1.1	p<0.001
		Diabetic nephropathy	35.8	17.2	45.8	1.2
		Others	47.2	22.3	30	0.5
		Unknown	34.9	17.1	47	1
3.2	Duration of predialysis nephrology care (years)	<1	59.1	22.1	18.5	0.3	p<0.001
		1-2	30.5	16.3	52.3	0.9
		>2	23.7	12	62.5	1.8
0.9	Coronary artery disease	No	34.6	14.8	49.5	1.1	p<0.001
		Yes	37.1	19.5	41.8	1.6
0.9	Cardiomyophathy	No	32.8	14.6	51.4	1.2	p<0.001
		Yes	40.7	19.2	39	1.1
0.9	Cardiac conduction disorders	No	33.8	15.3	49.7	1.2	p<0.001
		Yes	41.1	19	38.8	1.1
0.9	Cerebrovascular disease	No	34.8	15.4	48.7	1.1	p<0.001
		Yes	37.2	18.9	42.6	1.3
1	Peripheral vascular disease	No	34.2	15.4	49.2	1.2	p<0.001
		Yes	38.2	17.6	43	1.2
0.9	Chronic respiratory disease	No	33.9	15.4	49.5	1.2	p<0.001
		Yes	40.3	17.9	40.6	1.2
0.8	Diabetes (not PRD)	No	34.8	15.2	48.7	1.3	p<0.001
		Yes	39.1	22.9	37.6	0.4
2.1	ESRD presentation	Unknown chronic renal disease	70.4	22.3	7.2	0.1	p<0.001
		Acute-on-chronic renal disease	53.6	20.4	25.4	0.6
		Gradual progression	20.3	12.6	65.5	1.6
0	Period of time	2000-2005	40.1	11.1	47.6	1.2	p<0.001
		2006-2011	29.6	20.9	48.4	1.1

Changes in the VA during the first year of follow-up

We were able to study a possible VA change in 6705 (67.3%) incident patients of whom there was available information about the VA in two consecutive follow-up reports at 31 December. Overall, 66.6% of these patients changed the initial VA during the first year of follow-up. This percentage was 90.1% and 81.7% in patients with UCC and TCC as first VA, respectively.

Regarding the direction of these VA changes, more than half of incident patients who started HD with UCC and TCC changed to an AVF during the first year of follow-up (67.2 and 59.6%, respectively). Despite having wide variability, the period of time necessary for AVF placement was significantly higher for patients who initiated HD by TCC (107.5 ± 212.9 days) in relation to that of UCC (85.8 ± 194.4 days) (p = 0.015).

Survival of patients according to the initial VA

We found significant differences in patient survival over time according to the VA used in the first HD performed. Regarding initiating HD with a mature AVF (reference), the adjusted hazard ratio (HR) of death was 1.55 [95% CI: 1.42-1.69, p<0.001], 1.43 (95% CI: 1.33-1.54, p<0.001) and 1.08 (95% CI: 0.84-1.38, p = 0.56) for the patients who started HD through TCC, UCC and AVG, respectively (Fig. 1).

Fig. 1 -

Characteristics of the incident HD patients who died during the first year of follow-up

Of the 4700 patients who died during the study period, 1481 (31.5%) patients died during the first year of follow-up reaching a crude mortality rate of 14.6 deaths/100 patient-years (95% CI: 14.2-15.9). When comparing these patients with the survivors after the first year of follow-up (n = 8475), we have observed significant differences in all variables analyzed excepting gender (Tab. II). Half of the patients who died were aged over 74 years, but only one-third of the patients who survived were included in this age group. The percentage of each comorbidity considered was always significantly higher in the patients who died compared to the survivors.

TABLE II -

Cumulative incidence of dying over time according to the vascular ­access used in the first hemodialysis (HD) session. Multivariate competing risk regression model. Meaning of the curves from top to bottom: tunnelled central catheter (TCC), untunnelled central catheter (UCC), arteriovenous graft (AVG), arteriovenous fistula (AVF).

Characteristics of the patients who died and who survived during the first year of follow-up

		Death 1-365 days (n = 1481)		Survived >365 days (n = 8475)		p-value1
		n	%	n	%
1p<0.05.
2 UCC = untunneled central vein catheter; TCC = tunneled central vein catheter; AVF = arteriovenous fistula; AVG = arteriovenous graft.
Age group (years)	18-44	31	2.1	884	10.4	<0.001
	45-64	201	13.6	2,498	29.5
	65-74	456	30.8	2,261	26.7
	>74	793	53.5	2,832	33.4
Gender	Male	958	64.7	5,389	63.6	0.417
Primary renal disease (PRD)	Glomerular	118	8.0	1,087	12.8	<0.001
	Polycystic	33	2.2	670	7.9
	Interstitial	115	7.8	814	9.6
	Vascular	303	20.5	1,489	17.6
	Diabetes	355	24.0	1,820	21.5
	Others	204	13.8	647	7.6
	Unknown	353	23.8	1,948	23.0
Duration of pre-dialysis nephrology care (years)	<1	527	36.6	2,168	26.5	<0.001
	1-2	398	27.6	2,025	24.7
	>2	515	35.8	4,003	48.8
Initial comorbidity	Coronary artery disease	474	32.2	1,715	20.4	<0.001
	Cardiomyophathy	651	44.3	2,245	26.8	<0.001
	Cardiac conduction disorders	453	30.9	1,321	15.7	<0.001
	Cerebrovascular disease	315	21.4	1,071	12.8	<0.001
	Peripheral vascular disease	500	34.1	1,757	20.9	<0.001
	Chronic respiratory disease	413	28.1	1,487	17.7	<0.001
	Diabetes (not PRD)	183	12.4	644	7.7	<0.001
End-stage renal disease presentation	Unknown chronic renal disease	230	15.7	790	9.5	<0.001
	Acute-on-chronic renal disease	543	37.1	2,328	28.1
	Steady progression	691	47.2	5,169	62.4
Period of hemodialysis inception	2000-2005	839	56.7	4,301	50.7	<0.001
	2006-2011	642	43.3	4,174	49.3
First vascular access2	UCC	746	50.4	2,739	32.3	<0.001
	TCC	309	20.9	1,273	15.0
	AVF	412	27.8	4,361	51.5
	AVG	14	0.9	102	1.2

VA profile of incident HD patients who died during the first year of follow-up

Patients who died during the first year of follow-up showed a higher percentage of UCC (50.4% vs. 32.3%) and lower percentage of AVF (27.8% vs. 51.5%) compared with the survivors (for both comparisons, p<0.001) (Tab. II). In each of the four age groups considered, UCC and AVF were the most frequent VA used to start HD in the patients who died and those survived, respectively (Tab. III).

TABLE III -

Classification of the patients who died and who survived during the first year of follow-up according to age group and type of initial vascular access

Age group (years)	Type of initial vascular access	Death 1-365 days N (%)	Survived >365 days N (%)
UCC = untunneled central vein catheter; TCC = tunneled central vein catheter; AVF = arteriovenous fistula; AVG = arteriovenous graft.
18-44	UCC	20 (64.5)	349 (39.5)
	TCC	8 (25.8)	125 (14.1)
	AVF	3 (9.7)	405 (45.8)
	AVG	0 (0.0)	5 (0.6)
45-64	UCC	112 (55.7)	769 (30.8)
	TCC	40 (19.9)	333 (13.3)
	AVF	44 (21.9)	1,365 (54.6)
	AVG	5 (2.5)	31 (1.3)
65-74	UCC	234 (51.3)	735 (32.5)
	TCC	93 (20.4)	337 (14.9)
	AVF	126 (27.6)	1,157 (51.2)
	AVG	3 (0.7)	32 (1.4)
>74	UCC	380 (47.9)	886 (31.3)
	TCC	168 (2.12)	478 (16.9)
	AVF	239 (30.1)	1,434 (50.6)
	AVG	6 (0.8)	34 (1.2)

The crude all-cause mortality curve associated with any type of CVC as the first VA used was always above that for AVF at any time during the first year of follow-up (Fig. 2). Regarding initiating HD with a mature AVF (reference), the adjusted OR for the patients who started HD through UCC and TCC for all-causes of death was 2.18 (95% CI: 1.83-2.59) and 1.94 (95% CI: 1.59-2.37), respectively. For cardiovascular cause of death it was 2.05 (95% CI: 1.59-2.64) and 1.89 (95% CI: 1.41-2.53) and for infection-related cause of death it was 2.17 (95% CI: 1.45-3.23) and 1.56 (95% CI: 0.97-2.51), respectively.

Fig. 2 -

The all-cause mortality risk of patients switching from a CVC to an AVF during the first year (n = 1932) decreases by half (OR: 0.46, 95% CI: 0.36-0.58) compared with patients dialyzed for as long through a CVC (using a single CVC or changing from one CVC to another) (n = 873). The all-cause mortality risk of patients switching from an AVF to a CVC during the first year (n = 226) is 6-fold (OR: 5.88, 95% CI: 4.00-9.09) compared with patients dialyzed for as long through an AVF (using a single AVF or changing from one AVF to another) (n = 2945).

Characteristics of the patients who died within 120 days (early period) and from 121 to 365 days (late period) after initiating HD

Of the 1481 patients who died during the first year of follow-up, 577 (39%) died within 120 days (early period) and 904 (61%) from 121 to 365 days (late period) after initiating HD. No significant differences between either group of patients in age or in the distribution of virtually all comorbidities were found (Tab. IV).

TABLE IV -

Crude all-cause mortality rates by month and initial vascular access type.

Characteristics of the patients who died within 120 days and from 121 to 365 days after initiating hemodialysis

		Death ≤120 days (n = 577)		Death 121-365 days (n = 904)		p-value1
		n	%	n	%
1p<0.05.
2UCC = untunneled central vein catheter; TCC = tunneled central vein catheter; AVF = arteriovenous fistula; AVG = arteriovenous graft.
Age group (years)	18-44	16	2.8	15	1.7	0.378
	45-64	79	13.7	122	13.5
	65-74	168	29.1	288	31.9
	>74	314	54.4	479	53.0
Gender	Male	353	61.2	605	66.9	0.024
Primary renal disease (PRD)	Glomerular	51	8.8	67	7.4	0.022
	Polycystic	6	1.0	27	3.0
	Interstitial	36	6.2	79	8.7
	Vascular	121	21.0	182	20.1
	Diabetes	126	21.8	229	25.3
	Others	89	15.4	115	12.7
	Unknown	148	25.6	205	22.7
Duration of pre-dialysis nephrology care (years)	<1	236	42.4	291	32.9	<0.001
	1-2	152	27.3	246	27.8
	>2	168	30.2	347	39.3
Initial comorbidity	Coronary artery disease	180	31.4	294	32.8	0.571
	Cardiomyophathy	263	45.9	388	43.3	0.320
	Cardiac conduction disorders	192	33.6	261	29.1	0.073
	Cerebrovascular disease	116	20.2	199	22.2	0.373
	Peripheral vascular disease	191	33.3	309	34.6	0.601
	Chronic respiratory disease	163	28.4	250	27.8	0.816
	Diabetes (not PRD)	87	15.2	96	10.7	0.011
End-stage renal disease presentation	Unknown chronic renal disease	105	18.4	125	14.0	<0.001
	Acute-on-chronic renal disease	232	40.7	311	34.8
	Steady progression	233	40.9	458	51.2
Period of hemodialysis inception	2000-2005	330	57.2	509	56.3	0.737
	2006-2011	247	42.8	395	43.7
First vascular access2	UCC	338	58.6	408	45.1	<0.001
	TCC	119	20.6	190	21.0
	AVF	113	19.6	299	33.1
	AVG	7	1.2	7	0.8

The crude all-cause mortality rate (deaths/100 patient-years) was significantly higher during the early period (18.3, 95% CI: 16.8-19.8) compared to the late period (15.4, 95% CI: 14.5-16.5). This rate peaked at 90 days, reaching a crude mortality rate of 21.6 deaths/100 patient-years and, then, dropped until remaining stable around a rate of 15 deaths/100 patient-years from 121 to 365 days after HD inception (Fig. 2).

Cause-specific mortality rates were significantly higher in the early than in the late period for virtually all-causes (Fig. 3). Cardiovascular was the leading cause of death in both periods, followed by infection, and showed the greatest gap between early and late periods (7.5 vs. 4.2 deaths/100 patient-years).

Fig. 3 -

VA profile of incident patients who died within (the first) 120 days (early period) and from 121 to 365 days (late period) after initiating HD

The distribution of the first VA type was significantly different between the early and late mortality periods (Tab. III). Patients with early mortality showed a significantly higher percentage of UCC (58.6% vs. 45.1%) and lower percentage of AVF (19.6% vs. 33.1%) compared to the patients with late mortality (for both comparisons, p<0.001).

The crude all-cause mortality rate (deaths/100 patient-years) for UCC was significantly higher during the early mortality period (31.2, 95% CI: 28.0-34.6) compared to the late period (21.3, 95% CI: 19.3-23.6). The highest crude all-cause mortality rate (deaths/100 patient-years) associated with the CVC was recorded during this period: 37.4 at 60 days for UCC and 29.9 at 90 days for TCC.

Regarding starting HD with a mature AVF (reference), the adjusted multivariate OR of all-cause mortality risk for the patients using UCC or TCC was 3.66 (95% CI: 2.80-4.81) and 2.97 (95% CI: 2.17-4.06) during the early period, and 1.87 (95% CI: 1.56-2.25) and 1.98 (95% CI: 1.66-2.44) during the late period. Regarding cardiovascular mortality, 2.76 (95% CI: 1.90-4.01) and 1.84 (95% CI: 1.17-2.89) during the early period, and 1.79 (95% CI: 1.36-2.37) and 1.98 (95% CI: 1.45-2.71) during the late period. Regarding infection-related mortality, 6.62 (95% CI: 3.11-14.05) and 4.58 (95% CI: 2.00-10.52) during the early period, and 1.80 (95% CI: 1.17-2.75) and 1.54 (95% CI: 0.93-2.55) during the late period.

Discussion

Those patients who initiated HD by any type of CVC in Catalonia had a higher associated all-cause mortality risk over time compared to those patients with a mature AVF at HD inception, as has been previously reported (5-6-7-8-9-10-11-12-13). Gruss et al showed a risk-adjusted all-cause mortality rate 1.86 times higher for patients who started HD with a TCC than for those patients using an AVF, and demonstrated that patients who needed a TCC at some point during follow-up had a risk-adjusted mortality rate 1.68 times higher than those patients who had always used an AVF (10). From the Scottish Renal Registry (2009-2011), Bray et al reported that patients using only TCC during follow-up had higher adjusted risk of all-cause, cardiovascular- and infection-related cause of death compared with patients dialyzing only with an arteriovenous access (21).

Using a TCC at the time of HD initiation in Catalonia is independently associated with the highest adjusted risk of death over time related to AVF, even above the UCC. Why don’t TCC and UCC share the same associated all-cause mortality risk? The answer may be related to the duration of CVC exposure, significantly higher for TCC than for UCC. Regarding this “time factor”, Gruss et al reported that the mortality risk is graded according to the length of TCC exposure: the risk-adjusted all-cause mortality rate was 7.66 times higher for patients who were dialyzed with a TCC for over 52% of the follow-up time in relation to patients with a TCC for less than 18% of the follow-up time (10).

Confirming other studies (15, 16), the use of a CVC at first HD session is one factor independently associated with the increased risk of adjusted all-cause mortality during the first year of HD treatment in Catalonia. Therefore, the VA type at HD initiation should be included in predictive models of patient’ survival such as those developed and validated by Mauri et al using data from RMRC (22, 23).

The crude all-cause mortality rate was the highest during the early period of follow-up (first 120 days) and peaked in month 3 after HD inception in Catalonia. According to the United States Renal Data System (USRDS), all-cause mortality rate peaked in month 2 following HD initiation (24). This temporal profile is very similar with the crude all-cause mortality rate associated with any type of CVC in Catalonia because it peaked in month 2 and 3 after starting HD through UCC and TCC, respectively. Lukowsky et al showed that the highest all-cause mortality rate among 18,707 incident HD patients occurred during the first 6 months, including an 80% higher death risk within the first 2 months, and using CVC as the first VA explained 34% of all deaths after HD inception (25).

Why this excess of mortality risk associated with the CVC as the initial VA? Catheter-related bacteremia can explain some cases of death in incident patients using a CVC (infection-related mortality) (26, 27). In fact, in Catalonia both UCC and TCC were associated with a significantly higher risk of infection-related death during the first 120 days after HD inception - that is within the highest mortality risk period.

However, sepsis secondary to CVC cannot fully explain the increase in the mortality rate related to CVC. Confirming previous findings (11), we have demonstrated a significantly higher cardiovascular mortality risk associated with starting HD treatment by CVC during the first year of follow-up. This risk remains significantly higher when considering UCC and TCC separately or also both early and late periods of mortality. The reason for this increased one-year cardiovascular mortality risk is unclear. It has been postulated that the use of CVC itself, through the foreign body effect and/or the presence of bacterial biofilm without overt CVC infection, may amplify the baseline inflammatory status which, in turn, has been linked with an increased cardiovascular risk (28-29-30-31-32-33). It appears that the elapsing time from CVC insertion until developing a fatal cardiovascular disease is limited and would be against this hypothesis. However, according to the CHOICE study, CVC insertion is associated with a heightened state of inflammation that persists even after 60 days from placement (34).

In a recent adjusted multivariate logistic analysis, both age and comorbidity are risk factors independently associated with the likelihood to use a mature AVF as the first VA in the incident patient (35). It has been suggested that some patient characteristics, such as older age and the presence of cardiovascular comorbidities, are associated, on the one hand, with a lower probability to start chronic HD through a mature AVF (and an higher likelihood to use a CVC) and, on the other hand, with a higher cardiovascular mortality risk during the first year of HD treatment (36). In this regard, the clinical profile between patients who died and who survived during the first year of follow-up in the present study was significantly different suggesting the existence of selection bias (1, 4). In addition to age and comorbidity, other unmeasured confounding factors could play a role in the mortality rate often attributed to CVC (37). In this regard, Sachdeva et al have pointed out that CKD patients initiating HD with TCC or AVG had a higher inflammatory status before VA placement compared to patients using a mature AVF, which could be related to the mortality excess at HD inception (33). Grubbs et al analyzed 117,277 incident HD patients and reported that, after adjustment for health status, the mortality risk associated with CVC after HD start was partially attenuated compared to AVF (37). They suggested that some clinical variables related to incident patients could explain a part of the increased mortality risk previously attributed to CVC itself (37). However, two findings of our study suggest that CVC is somehow involved in the all-cause mortality rate after HD inception: a) in all age groups considered, even in young patients, the UCC was the VA most used to start HD in the patients who died during the first year of HD treatment, and b) as in previous studies (7, 9, 38), the change in the adjusted mortality risk was related to the change of the VA type during the first year of follow-up.

This study has a number of weaknesses inherent to any study performed completely using patients’ data from a population register database, in which the variables utilized are restricted in number and can have low clinical specificity. The adjusted competitive all-cause mortality risk over time related to VA was calculated according to the VA used only in the first HD, without accounting for further changes in VA type during the follow-up. Another limitation of this study is the small proportion of incident patients that initiated HD through AVG and, therefore, conclusions about the mortality associated with this type of VA are not possible in Catalonia.

Strengths of the present study include its large sample size (almost 10,000 incident HD patients) and the prolonged duration of the study period (12 years), sufficient numbers and sufficient time to make valuable conclusions about the mortality risk associated with VA in Catalonia. Another strength of our study is that, in contrast with previous epidemiological studies where UCCs and TCCs were combined into one “CVC group” (15, 16), we have considered UCC and TCC separately to analyze possible differences between both types of CVC. In the present study, we included the mortality data from day 1 of HD treatment unlike other studies that only consider patients who died after day 90 of HD inception. That is, they excluded all incident HD patients dying within the first 3 months after HD initiation (11). According to USRDS, death rates from day 1 are 17-20% higher than those followed from day 90 (24).

In conclusion, the present registry-based study over a period of 12 years shows that around 50% of ESRD patients started HD without a mature AVF in Catalonia. Initiating HD treatment by any type of CVC was independently associated with a higher all-cause mortality risk over time than with a mature AVF. The period of the highest crude all-cause mortality rate in Catalonia was the first 120 days after HD inception. During this period, the adjusted OR of death for all-cause, cardiovascular- and infection-related mortality was significantly higher for the incident patients who started HD through any type of CVC regarding initiating HD with a mature AVF. Therefore, about half incident HD patients in Catalonia are exposed annually to an excessive mortality risk related to CVC, which is a potentially modifiable risk factor by means of early AVF placement.

Cause-specific mortality rates and 95 confidence intervals for the ≤120 days and 121 to 365 days periods.

• 1. Ravani P Palmer SC Oliver MJ et al. Associations between hemodialysis access type and clinical outcomes: a systematic review. J Am Soc Nephrol 2013 24 3 465 473
• 2. Pisoni RL Arrington CJ Albert JM et al. Facility hemodialysis vascular access use and mortality in countries participating in DOPPS: an instrumental variable analysis. Am J Kidney Dis 2009 53 3 475 491
• 3. Perl J Wald R McFarlane P et al. Hemodialysis vascular access modifies the association between dialysis modality and survival. J Am Soc Nephrol 2011 22 6 1113 1121
• 4. Quinn RR Ravani P Fistula-first and catheter-last: fading certainties and growing doubts. Nephrol Dial Transplant 2014 29 4 727 730
• 5. Polkinghorne KR McDonald SP Atkins RC Kerr PG Vascular access and all-cause mortality: a propensity score analysis. J Am Soc Nephrol 2004 15 2 477 486
• 6. Astor BC Eustace JA Powe NR Klag MJ Fink NE Coresh J CHOICE Study. Type of vascular access and survival among incident hemodialysis patients: the Choices for Healthy Outcomes in Caring for ESRD (CHOICE) Study. J Am Soc Nephrol 2005 16 5 1449 1455
• 7. Allon M Daugirdas J Depner TA Greene T Ornt D Schwab SJ HEMO study group. Effect of change in vascular access on patient mortality in hemodialysis patients. Am J Kidney Dis 2006 47 3 469 477
• 8. Moist LM Trpeski L Na Y Lok CE Increased hemodialysis catheter use in Canada and associated mortality risk: data from the Canadian Organ Replacement Registry 2001-2004. Clin J Am Soc Nephrol 2008 3 6 1726 1732
• 9. Lacson E Jr Wang W Lazarus JM Hakim RM Change in vascular access and mortality in maintenance hemodialysis patients. Am J Kidney Dis 2009 54 5 912 921
• 10. Gruss E Portolés J Tato A et al. Clinical and economic repercussions of the use of tunneled haemodialysis catheters in a health area. Nefrologia 2009 29 2 123 129
• 11. Ocak G Halbesma N le Cessie S et al. Haemodialysis catheters increase mortality as compared to arteriovenous accesses especially in elderly patients. Nephrol Dial Transplant 2011 26 8 2611 2617
• 12. Antón-Pérez G Pérez-Borges P Alonso-Almán F Vega-Díaz N Vascular accesses in haemodialysis: a challenge to be met. Nefrologia 2012 32 1 103 107
• 13. DeSilva RN Patibandla BK Vin Y et al. Fistula first is not always the best strategy for the elderly. J Am Soc Nephrol 2013 24 8 1297 1304
• 14. Nordio M Tessitore N Feriani M et al. Scientific Committee of Veneto Dialysis and Transplantation Registry. Mortality in the Veneto population on renal replacement therapy. J Nephrol 2013 26 Suppl 20 S23 S33
• 15. Bradbury BD Fissell RB Albert JM et al. Predictors of early mortality among incident US hemodialysis patients in the Dialysis Outcomes and Practice Patterns Study (DOPPS). Clin J Am Soc Nephrol 2007 2 1 89 99
• 16. Foley RN Chen SC Solid CA Gilbertson DT Collins AJ Early mortality in patients starting dialysis appears to go unregistered. Kidney Int 2014 86 2 392 398
• 17. Robinson BM Zhang J Morgenstern H et al. Worldwide, mortality risk is high soon after initiation of hemodialysis. Kidney Int 2014 85 1 158 165
• 18. The International Classification of Diseases. 9^th Revision, Clinical Modification 1978. Comission on Professional and Hospital Activities, Ann Harbor, Michigan 48705, USA
• 19. van Dijk PC Jager KJ de Charro F et al. ERA-EDTA registry. ­Renal replacement therapy in Europe: the results of a collaborative effort by the ERA-EDTA registry and six national or regional registries. Nephrol Dial Transplant 2001 16 6 1120 1129
• 20. Fine J Gray R A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999 94 446 496 509
• 21. Bray BD Boyd J Daly C et al. Scottish Renal Registry. Vascular access type and risk of mortality in a national prospective cohort of hemodialysis patients. Q J Med 2012 105 11 1097 1103
• 22. Mauri JM Clèries M Vela E Registry CR Catalan Renal Registry. Design and validation of a model to predict early mortality in haemodialysis patients. Nephrol Dial Transplant 2008 23 5 1690 1696
• 23. Catalan Society of Nephrology. Mortality risk calculator: predictive model of mortality within one year after starting HD. Available online at: http://www.socane.cat Accessed December 24, 2014.
• 24. U.S. Renal Data System. USRDS 2013 Annual Data Report: Atlas of Chronic Kidney Disease and End-Stage Renal Disease in the United States. National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases. Bethesda, MD 2013
• 25. Lukowsky LR Kheifets L Arah OA Nissenson AR Kalantar-Zadeh K Patterns and predictors of early mortality in incident hemodialysis patients: new insights. Am J Nephrol 2012 35 6 548 558
• 26. Lok CE Mokrzycki MH Prevention and management of catheter-related infection in hemodialysis patients. Kidney Int 2011 79 6 587 598
• 27. Xue H Ix JH Wang W et al. Hemodialysis access usage patterns in the incident dialysis year and associated catheter-related complications. Am J Kidney Dis 2013 61 1 123 130
• 28. Canaud B Sénécal L Leray-Moragués H et al. L’accès vasculaire, une cause d’inflammation sous-estimée chez l’hémodialysé. Nephrologie 2003 24 7 353 358
• 29. Cappelli G Tetta C Canaud B Is biofilm a cause of silent chronic inflammation in haemodialysis patients? A fascinating working hypothesis. Nephrol Dial Transplant 2005 20 2 266 270
• 30. Movilli E Brunori G Camerini C et al. The kind of vascular access influences the baseline inflammatory status and epoetin response in chronic hemodialysis patients. Blood Purif 2006 24 4 387 393
• 31. Hung[J12] AM Alp Ikizler TA Hemodialysis central venous catheters as a source of inflammation and its implications. Semin Dial 2008 21 5 401 404
• 32. Goldstein SL Ikizler TA Zappitelli M Silverstein DM Ayus JC Non-infected hemodialysis catheters are associated with increased inflammation compared to arteriovenous fistulas. Kidney Int 2009 76 10 1063 1069
• 33. Sachdeva M Hung A Kovalchuk O Bitzer M Mokrzycki MH The initial vascular access type contributes to inflammation in incident hemodialysis patients. Int J Nephrol 2012 2012 917465
• 34. Banerjee T Kim SJ Astor B Shafi T Coresh J Powe NR Vascular access type, inflammatory markers, and mortality in incident hemodialysis patients: the Choices for Healthy Outcomes in Caring for End-Stage Renal Disease (CHOICE) Study. Am J Kidney Dis 2014 64 6 954 961
• 35. Roca-Tey R Arcos E Comas J Cao H Tort J Committee CR Vascular access for incident hemodialysis patients in Catalonia: analysis of data from the Catalan Renal Registry (2000-2011). J Vasc Access 2015 doi: 10.5301/jva.5000410 [Epub ahead of print].
• 36. Roca Tey R Vascular access for haemodialysis: an unresolved issue. Nefrologia 2010 30 3 280 287
• 37. Grubbs V Wasse H Vittinghoff E Grimes BA Johansen KL Health status as a potential mediator of the association between hemodialysis vascular access and mortality. Nephrol Dial Transplant 2014 29 4 892 898
• 38. Bradbury BD Chen F Furniss A et al. Conversion of vascular access type among incident hemodialysis patients: description and association with mortality. Am J Kidney Dis 2009 53 5 804 814