Association between post-dialysis hemoglobin level and the survival of vascular access

Abstract

Introduction:

Although a few dialysis facilities conduct a complete blood cell count for some patients at post-dialysis, including hemoglobin, clinical findings supporting the interpretation of results are scarce. The aim of this study was to investigate the association between post-dialysis hemoglobin level and vascular access failure with clinical data.

Methods:

Study design: Case crossover design. Setting: Japanese dialysis facilities, which routinely take post-dialysis blood samples, including complete blood cell counts at least once a month. Participants: Hemodialysis patients who experienced vascular access failure in January 2010 until December 2014. Exposure: Post-dialysis hemoglobin level. Main outcome: Vascular access failure treated with endovascular treatment or operation. Statistical analysis: Self-matched odds ratios and 95% confidence intervals were estimated by comparing post-dialysis hemoglobin just before events (“case”) with levels at 6 and 12 months before events (“control”) using conditional logistic regression, and presented with restricted cubic spline.

Results:

Two hundred and thirty hemodialysis patients with vascular access failure were identified. Mean post-dialysis hemoglobin level before the failure was 11.8 g/dL (standard deviation 1.7). The spline curve showed that higher post-dialysis hemoglobin levels above 11.8 g/dL had a greater odds ratio for vascular access failure. Post-dialysis hemoglobin levels and odds ratios (95% confidence interval) for vascular access failure relative to the reference value (Hb 11.8 g/dL) were Hb 12.0 g/dL, 1.1 (1.0-1.1); Hb 14.0 g/dL, 1.4 (1.0-2.0); and Hb 16.0 g/dL, 2.1 (1.1-4.3).

Conclusions:

A higher post-dialysis hemoglobin level was associated with vascular access failure. Higher post-dialysis Hb could be a factor that triggers vascular access failure.

Keywords

Hemodialysis Post-dialysis hemoglobin Vascular access

Introduction

Pre-dialysis hemoglobin level (Hb) is a widely used indicator in the management of renal anemia in patients with hemodialysis, and nearly all clinical research and guidelines cite this variable (1, 2).

In addition to biochemical measurement both pre- and post-dialysis, most centers also collect blood samples for complete blood count (CBC) at pre-dialysis to assess dialysis adequacy, anemia and electrolyte levels. However, other centers check CBC both pre- and post-dialysis. Some researchers consider that real Hb values are closer to the post-hemodialysis (HD) than pre-HD level, based on pre- and post-dialysis Hb findings that Hb significantly increased following an HD session, particularly in the first 24 hours, then gradually decreased over the rest of the inter-dialysis period (3 -9) (Supplementary Tab. I, available online as Supplementary material at http://www.vascular-access.info), and a more recent finding that Hb at 4, 24, and 48 hours after an HD session remained elevated compared to the pre-HD Hb level, with no significant difference from the immediate post-HD Hb concentration (6). Despite this evidence that post-dialysis Hb might be an indicator for real Hb concentration in HD patients, the outcome-based clinical research data required to interpret post-dialysis Hb are scarce. This lack of outcome-based research might suggest that interpretations of post-dialysis Hb to date have been experience-based.

Vascular access failure is an important clinical issue and a major burden on HD patients. Well-known risk factors for vascular access failure include gender, surgical site of vascular access, type of vascular access, diabetes mellitus and secondary hyperparathyroidism. Protective factors include antihypertensive drugs, such as angiotensin-converting enzyme inhibitors and angiotensin-receptor II antagonists, and aspirin (10 -13). One study of the association between the patency of a newly created vascular access and pre-dialysis Hb (13) reported an association between access failure and the lowest hemoglobin group (Hb <10g/dL).

Although findings that post-HD Hb might represent actual Hb values in HD patients, no study has yet directly examined the association between post-HD Hb and vascular access failure. Here, we investigated the association between post-HD Hb and vascular access failure.

Methods

Design, setting and patients

The study was conducted under a case-crossover design. The setting was nine facilities across Japan (five secondary care centers, one tertiary medical care center, and three private clinics), which routinely assessed post-dialysis Hb levels at least once a month and consented to participate.

The study subjects were maintenance hemodialysis patients who experienced vascular access failure from April 2010 to December 2014. Inclusion criteria included at least 12 months since the initiation of HD or the creation of their virgin vascular access (whichever occurred later), three maintenance HD sessions a week, and the passage of more than 12 months since their vascular access failure. Patients receiving peritoneal dialysis or with a permanent catheter for dialysis were excluded.

Main outcomes and measures

The primary end-point was complete or incomplete vascular access failure. We defined the primary outcome as vascular access failure requiring surgery, such as surgical reconstruction and replacement of the arteriovenous graft; or an endovascular procedure, such as percutaneous transluminal angioplasty, thrombectomy and use of a thrombolytic agent. When a patient experienced several vascular access failures during the study period, we defined the outcome occurrence as the first vascular access failure.

Exposure and definition of cases and controls

We defined the exposure as post-dialysis Hb. Because Hb levels closest to the vascular access failure were likely the most heavily influenced by the failure, cases were defined as the post-dialysis Hb collected routinely at the beginning of the month in which the vascular access failure occurred. Cases were also defined as the first outcome during the observational period. Considering the carry-over effects of Hb, we used data at two timepoints for comparison, namely 6 months and 12 months before the case (Fig. 1). In cases when the post-dialysis Hb sample was collected routinely two or more times a month, we used the first sample in the month.

Fig. 1

A case-crossover design for post-dialysis hemoglobin and vascular access failure.

Confounding factors and other variables

Data at respective timepoints for the following confounding factors were used, selected based on previous studies (10 -13): pre-dialysis serum albumin; C-reactive protein (CRP); amount of fluid removed; pre-dialysis systolic blood pressure; medications such as antiplatelet agents, anticoagulant agents, angiotensin-converting enzyme inhibitors (ACEIs) and angiotensin-receptor II antagonists; anticoagulant in the hemodialysis circuit; amount of erythropoiesis stimulant agents (ESA); and dialysis hypotension. For the amount of ESA, when the patient used darbepoetin, we converted one unit of darbepoetin into 250 units of epoetin. When the patient used epoetin beta pegol, we converted one unit of epoetin beta pegol into 62.5 units of epoetin (14, 15). For parathyroid hormone (PTH), when whole PTH was measured, we converted one unit of whole PTH into 1.7 units of intact PTH (iPTH). Dialysis hypotension was primarily defined using pre- and post-dialysis blood pressure as a difference in systolic blood pressure ≥20 mmHg or a difference in mean arterial pressure by ≥10 mmHg between pre- and post-dialysis blood pressure.

Statistical analysis

Continuous variables with normal distribution were expressed as the mean (standard deviation [SD]), and continuous variables without normal distribution as the median (interquartile range). Categorical variables were expressed as number and proportion, as appropriate. For the association between post-dialysis Hb and vascular access failure, we estimated the adjusted odds ratio (OR) using a conditional logistic model with vascular access failure as the dependent variable and post-dialysis Hb as the independent variable with adjustment for the variables mentioned above. We did not use factors that showed little change over time, such as gender and cause of end-stage renal disease (ESRD). As secondary analysis, we similarly estimated the adjusted OR using the difference between pre- and post-dialysis Hb. Based on a previous study with pre-dialysis Hb (13), we expected that the association between post-dialysis Hb and the outcome would be non-linear. Accordingly, we showed the association using the restricted cubic spline method, using the three knots of the 25^th, 50^th and 75^th percentile points of each Hb level. The association was presented as linear for less than 25 and more than 75 percentile points, and as a restricted cubic spline from the 25 to 75 percentiles. Reference values were set at the median of post- and pre-dialysis Hb, namely 11.8 g/dL for post-dialysis Hb and 10.8 g/dL for pre-dialysis Hb. With regard to confounding factors, CRP was transformed logarithmically and used in the model. We performed a complete case analysis in the present study. All analyses were conducted using STATA 13.1 (StataCorp, College Station, TX, USA), with 2-sided significance set at 0.05.

Ethical issues

The study complied with the Declaration of Helsinki and was approved by the Ethics Committee of Kyoto University (Approval Numbers R0387) and Fukushima Medical University (Approval Numbers 2315).

Results

During the study period, outcomes occurred in 230 hemodialysis patients (Fig. 2). Mean age of these patients was 69.7 years, median duration of hemodialysis treatment was 3 years, and the proportion with arteriovenous fistula was 86.5% (Tab. I). Post-hemodialysis Hb just before the vascular access failure and at 6 and 12 months before failure was 11.7 (SD 1.7), 11.6 (SD 1.6) and 11.6 (SD 1.6), respectively (Tab. II and Fig. 3).

Fig. 2

Patient flow.

Table I

Baseline characteristics

	n = 230
Age, years (SD)	69.7	(10.9)
Gender, female, %	118	(51.3%)
Duration of dialysis, years (quartile)	3	(1, 8)
Cause of ESRD, n (%)
Diabetic nephropathy	85	(37.0%)
Glomerulonephritis	48	(20.9%)
Others	97	(42.1%)
Vascular access
AVF	199	(86.5%)
AVG	31	(13.5%)
Comorbidities
Coronary heart disease, %	91	(39.6%)
Diabetes, %	76	(33.0%)
Duration of vascular access patency, years (quartile)	4	(1, 14)

SD = standard deviation; ESRD = end-stage renal disease; AVF = arteriovenous fistula; AVG = arteriovenous graft.

Table II

Laboratory data and other data at each time point

	Right before obstruction				6 months before				12 months before
	Pre-dialysis		Post-dialysis		Pre-dialysis		Post-dialysis		Pre-dialysis		Post-dialysis
Hb, g/dL (SD)	10.8	(1.4)	11.7	(1.7)	10.5	(1.3)	11.6	(1.6)	10.7	(2.3)	11.6	(1.6)
Hct, % (SD)	33.1	(4.3)	36.1	(5.6)	32.7	(4.2)	35.8	(5.0)	32.5	(4.7)	35.7	(5.0)
%⊿BW, % (SD)	4.4		(1.8)		4.5		(1.8)		4.5		(1.7)
Oral antiplatelet agents, n (%)	106		(46.1%)		88		(38.2%)		88.0		(38.2%)
Oral anticoagulant agents, n (%)	19		(8.3%)		17		(7.4%)		17.0		(7.4%)
ACEI/ARB, n (%)	114		(49.6%)		121		(52.6%)		121.0		(52.6%)
Amount of ESA/week, Units (quartile)	5,500		(3,750, 10,000)		5,000		(3,000, 10,000)		5,000		(3,000, 15,000)
Usage of intravenous iron agents, n (%)	59		(25.7%)		64		(27.8%)		64		(27.8%)
Anticoagulant in hemodialysis circuit, n (%)
UFH	192		(83.5%)		192		(83.5%)		192		(83.5%)
LMWH	26		(11.3%)		28		(12.2%)		28		(12.2%)
NM	11		(4.8%)		13		(3.8%)		13		(3.8%)
Others	1		(0.5%)		1		(0.5%)		1		(0.5%)
Systolic blood pressure, mmHg (SD)	144	(24)	142	(25)	145	(22)	144	(26)	146	(22)	143	(24)
Diastolic blood pressure, mmHg (SD)	75	(15)	75	(14)	76	(14)	76	(12)	76	(13)	76	(14)
Intradialytic hypotension, n (%)	38		(16.5%)	38		(16.5%)		30			(13.0%)
WBC, /μL (SD)	5829	(1906)	NA		5763	(1950)	NA		5331	(2235)	NA
PLT, 10⁴/μL (SD)	18.9	(6.9)	NA		18.7	(7.4)	NA		18.6	(7.0)	NA
TP, g/dL (SD)	6.3	(0.6)	7.1	(0.9)	6.4	(0.6)	7.2	(0.9)	6.6	(2.6)	7.3	(0.9)
Alb, g/dL (SD)	3.6	(0.4)	4.1	(0.6)	3.6	(0.4)	4.1	(0.7)	3.7	(0.4)	4.2	(0.6)
BUN, mg/dL (SD)	61	(16)	19	(8)	61	(16)	19	(8)	61	(15.2)	19	(7)
Cre, mg/dL (SD)	10.0	(2.0)	3.9	(1.5)	9.7	(2.8)	3.7	(1.3)	10.7	(6.0)	3.9	(1.4)
CRP, mg/dL (SD)	0.17	(0.05, 0.68)	NA		0.13	(0.05, 0.42)	NA		0.14	(0.04, 0.48)	NA
iPTH, pg/mL (SD)	139	(99)	NA		179	(239)	NA		165	(202)	NA

ESA = erythropoiesis stimulating agent; UFH = unfractionated heparin; LMWH = low-molecular-weight heparin; NM = nafamostat mesylate;

WBC = white blood cell; PLT = platelet; Alb = albumin; BUN = blood urea nitrogen; Cre = creatinine; CRP = C-reactive protein; iPTH = intact parathyroid hormone; NA = not available; %⊿BW = (post-dialysis BW - pre-dialysis BW)/dry weight.

Fig. 3a

Hemoglobin (Hb) levels and odds ratio (OR) of vascular access failure and histogram. The solid lines present the adjusted odds ratio of (A) post-dialysis Hb and (B) pre-dialysis Hb derived from restricted cubic spline models. Reference values are 11.8 g/dL for post-dialysis Hb and 10.8 g/dL for pre-dialysis Hb. The dashed line shows the 95% confidence intervals (CI). Each histogram presents the frequency of (A) post- and (B) pre-dialysis Hb, respectively.

Fig. 3B

Hemoglobin (Hb) levels and odds ratio (OR) of vascular access failure and histogram. The solid lines present the adjusted odds ratio of (A) post- and (B) pre-dialysis Hb levels derived from restricted cubic spline models. Reference values are 11.8 and 10.8 g/dL for post- and pre-dialysis Hb, respectively. The dashed line shows the 95% confidence intervals (CI). Each histogram presents the frequency of (A) postand (B) pre-dialysis Hb levels, respectively.

Table IIIA and Figure 3A show the results of conditional logistic models with post-dialysis and pre-dialysis Hb as exposure. Multivariable analysis showed that greater post-dialysis Hb levels above 11.8 g/dL were associated with greater ORs for vascular access failure. Further, multivariable analysis revealed that the association between pre-dialysis Hb and outcome was not statistically significant at any pre-dialysis Hb level.

Table IIIA

Association between pre- or post-hemoglobin and the outcome adjusted dialysis hypotension

Hb (g/dL)	Post-dialysis Hb		Pre-dialysis Hb
	OR	(95%CI)	OR	(95%CI)
6	0.4	(0.1-2.9)	0.2	(0.0-1.8)
8	0.6	(0.3-1.1)	0.5	(0.2-1.3)
10	0.8	(0.5-1.0)	0.8	(0.7-1.1)
10.8			Reference
11.8	Reference
12	1.1	(1.0-1.1)	2.0	(1.0-3.7)
14	1.4	(1.0-2.0)	4.1	(0.9-19.6)
16	2.1	(1.1-4.3)	8.4	(0.6-110.1)
18	3.0	(1.2-9.1)	17.0	(0.5-617.5)

Hb = hemoglobin; OR = odds ratio; CI = confidence interval.

We adjusted pre-dialysis serum albumin; C-reactive protein (CRP); amount of fluid removed using bodyweight difference between pre- and post-dialysis; pre-dialysis systolic blood pressure; medications such as antiplatelet agents, anticoagulant agents, angiotensin-converting enzyme inhibitors and angiotensin-receptor II antagonists; anticoagulant in the hemodialysis circuit; amount of erythropoiesis stimulant agents (ESA); and dialysis hypotension in each analysis.

Discussion

In this study of the association between post-dialysis Hb and vascular access failure, we found that greater post-dialysis Hb was associated with vascular access failure. No other pre-dialysis Hb was statistically associated with this outcome. These findings may suggest that measurement of post-dialysis Hb is valuable in improving patient outcomes.

A meta-analysis of randomized control trials (RCTs) for ESA in patients with chronic kidney disease showed that higher Hb groups were at higher risk for thrombosis of the vascular access (relative risk 1.33, 95% confidence interval [CI] 1.16-1.53) than the lower Hb groups (16). Five of the eight RCTs in the meta-analysis dealt with HD patients and three dealt with patients without dialysis. Although participants’ backgrounds differed from the present study, the meta-analysis finding that higher Hb is associated with more frequent vascular access failure is consistent with our study. Our study revealed that a higher post-dialysis Hb above 11.8 g/dL showed a large effect size, with an OR greater than 10. We considered three possible reasons for this. First, we have possibly observed recirculation of the vascular access. In other words, stenosis from the central vascular access induced recirculation. The blood concentrated in the recirculation route, could have resulted in the rapid increase in Hb concentration post-dialysis. Briefly, there may be a reversal of cause and effect. Second, we considered the effects of outliers. However, the same trend was seen in a sensitivity analysis after the exclusion of patients with the top 1% of post-dialysis Hb values. Finally, because of the small sample size in the higher post-dialysis Hb group, the estimation of OR might be unstable. Nevertheless, these studies revealed that higher pre-dialysis Hb level was associated with vascular access failure; however, our study showed that pre-dialysis Hb level was not related to vascular access failure. This difference could be caused by the following conditions. First, we used a case-crossover design in this study. Hence, we assessed the effect of post-dialysis Hb for the trigger factor, whereas the previous studies used pre-dialysis Hb at baseline data or maintained higher pre-dialysis Hb for intervention or exposure. Consequently, the difference of the study design might change the results. Second, although it was not statistically significant, we showed that higher pre-dialysis Hb levels were likely higher point estimates for vascular access failure in our study, which might be caused by lack of sample size.

One previous study of this association showed that a pre-dialysis Hb less than 10 g/dL was more likely to be associated with vascular access failure than a pre-dialysis Hb level of 10.0 to 12.0 (hazard ratio 1.80 [95% CI 1.13-2.86]), and that a level of over 12.0 g/dL was not associated with access failure (13). In that study, we analyzed not only post- but also pre-dialysis Hb, and found that pre-dialysis Hb was not related to outcome at any level. The results of our previous and the present study are therefore clearly inconsistent, regardless of the same exposure. This difference might be due to the following: (i) the subjects in the previous study were patients with a newly created vascular access; (ii) the present study period was considerably shorter than that of the previous study, which measured occurrence at up to 30 months after exposure; or (iii) the previous study’s adjustment for confounding factors of age, dialysis vintage, diabetes mellitus, cardiovascular disease, use of ESA, angiotensin II receptor blockers (ARBs) and ACEI, and parathyroid hormone, but not CRP, serum albumin or amount of fluid removed. In particular, factors that were unadjusted in the previous study were important in determining whether the outcome was affected by malnutrition and generalized weakness.

Hypotension and shock are strong prognostic factors for vascular access failure (17). We adjusted dialysis hypotension in this study. However, when Hb is an independent variable, and vascular access failure is a dependent variable, hypotension could be an intermediate factor. Thus, this result could be attenuated and biased, and we also analyzed excepting dialysis hypotension (Tab. IIIB and Fig. 3B).

Table IIIB

Association between pre- or post-hemoglobin and the outcome not adjusted dialysis hypotension

Hb (g/dL)	Post-dialysis Hb		Pre-dialysis Hb
	OR	(95%CI)	OR	(95%CI)
6	5.8	(0.40-95.5)	0.4	(0.0-4.9)
8	2.7	(0.5-15.0)	0.6	(0.2-2.2)
10	1.3	(0.7-2.4)	0.8	(0.7-1.1)
10.8			Reference
11.8	Reference
12	1.1	(1.0-1.1)	1.4	(0.8-2.5)
14	3.3	(1.3-8.3)	2.6	(0.2-16.6)
16	12.7	(1.8-89.4)	4.9	(0.2-109.9)
18	49.1	(2.5-967.4)	9.2	(0.1-728.7)

Hb = hemoglobin; OR = odds ratio; CI = confidence interval.

We adjusted pre-dialysis serum albumin; C-reactive protein (CRP); amount of fluid removed using bodyweight difference between pre- and post-dialysis; pre-dialysis systolic blood pressure; medications such as antiplatelet agents, anticoagulant agents, angiotensin-converting enzyme inhibitors and angiotensin-receptor II antagonists; anticoagulant in the hemodialysis circuit; and amount of erythropoiesis stimulant agents (ESA).

This study had several strengths. First, the use of a case-crossover design prevented confounding due to major effects on outcome such as past vascular access reoperation. Second, to our knowledge, this study is the first study to test the hypothesis that the association between post-dialysis outcome and vascular access failure.

Several limitations also warrant mention. First, because the inclusion criteria restricted facilities to those that routinely conduct post-dialysis CBC sampling, they might have tended toward more frequent laboratory testing and detailed examination; if so, a degree of selection bias might be present. Second, it was difficult to assess vascular access stenosis: while acknowledging the importance of vascular access stenosis in vascular access failure, our wish to avoid detection bias meant we were unable to define vascular access stenosis. Third, many vascular access failures might have occurred gradually. Because our study used a case-crossover design, we revealed only that higher post-dialysis Hb might be a trigger factor for vascular access failure. Longitudinal studies with repeated measurement are needed to prove the long-term effect of post-dialysis Hb. Fourth, as we mentioned, high post-dialysis Hb level could be related to recirculation. However, we did not measure the presence of recirculation. Finally, we were unable to adjust for all of the potential confounding factors.

In conclusion, this study identified an association between post-dialysis Hb and vascular access failure. Our study revealed that the post-dialysis Hb value might be more closely associated with vascular failure than the differential value or pre-dialysis Hb. We consider that higher post-dialysis Hb might be a trigger for vascular access failure. Our findings might aid in clinically interpreting the value of post-dialysis Hb. Use of post-dialysis Hb as an indicator might help improve patient outcomes through the optimization of fluid removal quantity and dry weight, calibration of dosages of ESA and iron drugs, and evaluation of vascular access to predict access failure. Validation of this potential indicator in larger studies is warranted.

Footnotes

Acknowledgements

Study Investigators

In addition to the authors, the following investigators participated in the study: M. Saito and A. Kiyan (Shirakawa Kosei General Hospital), M. Nakano (Hanawa Kosei Hospital), Y. Odagaki (Bange Kosei General Hospital), S. Nakanishi (Chibune Hospital), F. Ebihara (Ebihara Clinic), H. Shimizu (Chubu Rosai Hospital), T. Saito and K. Ueno (Kawashima Internal Medicine Clinic), K. Okada (Yokohama Minami Kyousai Hospital) and Y. Ishibashi (Japan Red Cross Medical Center).

We would like to thank Guy Harris and DMC Corp. (www.dmed.co.jp) and editage () for English editing.

Disclosures

Financial support: This work was supported by grants from the Institute for Health Outcomes & Process Evaluation research. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Conflict of interest: The authors have declared that no competing interests exist.

References

Kliger

Foley

Goldfarb

et al . KDOQI US commentary on the 2012 KDIGO Clinical Practice Guideline for Anemia in CKD. Am J Kidney Dis. 2013;62(5):849–859.

Tsubakihara

Nishi

Akiba

et al . 2008 Japanese Society for Dialysis Therapy: guidelines for renal anemia in chronic kidney disease. Ther Apher Dial. 2010;14(3):240–275.

Bellizzi

Minutolo

Terracciano

et al . Influence of the cyclic variation of hydration status on hemoglobin levels in hemodialysis patients. Am J Kidney Dis. 2002;40(3):549–555.

Movilli

Pertica

Camerini

et al . Predialysis versus postdialysis hematocrit evaluation during erythropoietin therapy. Am J Kidney Dis. 2002;39(4):850–853.

Korzets

Ori

Zevin

Gafter

. Posthemodialysis hemoglobin levels: overdue or not? Am J Kidney Dis. 2002;40(4):873–874.

Minutolo

De Nicola

Bellizzi

et al . Intra- and post-dialytic changes of haemoglobin concentrations in non-anaemic haemodialysis patients. Nephrol Dial Transplant. 2003;18(12):2606–2612.

Geller

DeVita

Marku-Podvorica

Rosenstock

Panagopoulos

Michelis

. Increase in post-dialysis hemoglobin can be out of proportion and unrelated to ultrafiltration. Dial Transplant. 2010;39(2):57–62.

Castillo

García-García

Rivero

et al . Should we adjust erythropoiesis-stimulating agent dosage to post-dialysis hemoglobin levels? A pilot study. BMC Nephrol. 2012;13(1):60.

Sagheb

Fallahzadeh

Moaref

Fallahzadeh

Dormanesh

. Comparison of hemoglobin levels before and after hemodialysis and their effects on erythropoietin dosing and cost. Nephrourol Mon. 2016;8(4):e38495.

10.

Grandaliano

Teutonico

Allegretti

et al . The role of hyperparathyroidism, erythropoietin therapy, and CMV infection in the failure of arteriovenous fistula in hemodialysis. Kidney Int. 2003;64(2):715–719.

11.

Saran

Dykstra

Wolfe

Gillespie

Held

Young

. Dialysis Outcomes and Practice Patterns Study. Association between vascular access failure and the use of specific drugs: the Dialysis Outcomes and Practice Patterns Study (DOPPS). Am J Kidney Dis. 2002;40(6):1255–1263.

12.

Hasegawa

Elder

Bragg-Gresham

et al . Consistent aspirin use associated with improved arteriovenous fistula survival among incident hemodialysis patients in the dialysis outcomes and practice patterns study. Clin J Am Soc Nephrol. 2008;3(5):1373–1378.

13.

Garrancho

Kirchgessner

Arranz

et al . Haemoglobin level and vascular access survival in haemodialysis patients. Nephrol Dial Transplant. 2005;20(11):2453–2457.

14.

Raymond

Wazny

Vercaigne

Lesperance

Skwarchuk

Bernstein

. Conversion from epoetin alfa to darbepoetin alfa within the Manitoba Renal Program: evaluation of dose ratios. CANNT J. 2008;18(1):39–43.

15.

Kuwahara

Mandai

Kasagi

et al . Responsiveness to erythropoiesis-stimulating agents and renal survival in patients with chronic kidney disease. Clin Exp Nephrol. 2015;19(4):598–605.

16.

Palmer

Navaneethan

Craig

et al . Meta-analysis: erythropoiesis-stimulating agents in patients with chronic kidney disease. Ann Intern Med. 2010;153(1):23–33.

17.

Chang

Paik

Greene

et al . Intradialytic hypotension and vascular access thrombosis. J Am Soc Nephrol. 2011;22(8):1526–1533.