Prehemodialysis arteriovenous access creation is associated with better cardiovascular outcomes in patients receiving hemodialysis: a population-based cohort study

Introduction

End-stage renal disease (ESRD) has become a major public health issue because of its prevalence in more than two million people worldwide and increasing incidence. Compared with general cohorts, patients with ESRD have a higher relative risk of 5-year mortality (Robinson et al., 2016). Among the major causes of mortality, cardiovascular (CV) disease contributes nearly half of the events in this population (Collins et al., 2010). Therefore, identifying management of CV complications is essential.

Prehemodialysis (pre-HD) care has been proved to ameliorate the outcomes of ESRD patients on HD maintenance (Baek et al., 2015; Bradbury et al., 2007). Timely creation of arteriovenous access (AVA), such as native fistula or artificial graft, is one of the crucial methods of care planning. It prevents not only the complications from delayed dialysis but also catheter-related infectious events (Oliver et al., 2004). However, CV outcomes following pre-HD AVA surgery are currently divergent. Once pre-HD AVA is created, cardiac output increases and leads to functional and structural changes of the heart, lungs, and vasculature (Guyton & Sagawa, 1961; Munclinger et al., 1987). London et al. (1993) reported that the arteriovenous shunt might result in chronic flow overload and cause cardiac hypertrophy. Nakhoul et al. (2005) observed that nitric oxygen production was decreased in patients with arteriovenous fistula and contributed to pulmonary hypertension (HTN). Korsheed, Burton & McIntyre (2009) and Korsheed et al. (2011) reported improved arterial stiffness, better left ventricular ejection fraction, and lesser heart damage after native fistula creation. Variation in laboratory and imaging parameters makes it difficult to predict the clinical outcomes. Several small-scale studies have reported negative clinical CV results following fistula creation (MacRae et al., 2004; Reddy et al., 2017; Vizinho et al., 2014), while a national study from the United States Renal Data System showed that using pre-HD fistula was strongly associated with lower CV mortality (Wasse, Speckman & McClellan, 2008). At present, only a few large-scale studies have explored the association between AVA creation and CV-related hospitalization.

Although ESRD is reported to have the highest prevalence in Taiwan compared with other countries, the 5-year survival rate of Taiwanese patients with ESRD seems better (Robinson et al., 2016). The Taiwan pre-ESRD pay-for-performance program, involving education and promotion of pre-HD AVA establishment, might have contributed to the higher survival rate (Lin et al., 2018). In Taiwan, more than half of the patients undergoing dialysis received access surgery before their first dialysis session, and access creation was completed in more than 80% of them before their chronic dialysis sessions (Hsu et al., 2018). Our study investigated the association between timing of AVA creation and CV outcomes in patients who underwent HD. We hypothesized that pre-HD AVA creation improves the CV outcomes of patients undergoing HD.

Materials and Methods

Results

Figure 1 illustrates the research design and sampling strategy. Overall, this study analyzed 3,147 patients—837 patients (26.6%) receiving pre-HD AVA creation and 2,310 patients (73.4%) not receiving pre-HD AVA creation.

The baseline characteristics of the recruited patients are presented in Table 1. The median age of patients with pre-HD AVA creation was lower than that of those without (66 years vs. 71 years, ASD: 29.9%). No statistically significant difference was observed in terms of sex between the two groups. Considering comorbidities, the patients in the pre-HD AVA group had lower proportions of CVA, dysrhythmia, COPD, liver disease, and dementia than those in the non-pre-HD AVA group. The groups exhibited similar proportions of patients with IHD, CHF, PVD, and DM. The mean Taiwan index of pre-HD AVA group was significantly lower than that of the non-pre-HD AVA group (5.55 ± 4.0 vs. 6.55 ± 4.4, ASD: 23%). Additionally, more patients with pre-HD AVA creation received ESAs than did those without pre-HD AVA creation (81.72% vs. 34.81%, ASD: 108%), whereas less patients with pre-HD AVA received ACEIs and ARBs than did those without pre-HD AVA (60.57% vs. 68.35%, ASD: 16.3%). The prescription of antiplatelet agent, anticoagulant agent, or statin was similar among the groups.

Table 2 shows the primary and secondary outcomes of the pre-HD AVA and non-pre-HD AVA groups. The patients with pre-HD AVA creation had a lower rate of MACEs during the follow-up period than did those without pre-HD AVA creation (crude HR 0.73, 95% CI [0.6–0.89]), but the effect became nonsignificant after matching for age, sex, year of HD, comorbidities, and medicines (adjusted HR 0.89, 95% CI [0.71–1.11], p = 0.29, Fig. 2). We further analyzed the MACEs separately and observed that patients with pre-HD AVA creation had a 35% lower CHF hospitalization rate after matching (adjusted HR 0.65, 95% CI [0.48–0.88], p < 0.01, Fig. 3). We examined CHF hospitalization rate in the propensity-score-matched groups, which showed a similar trend (adjusted HR 0.63, 95% CI [0.46–0.87], p < 0.01).

Regarding secondary outcomes, we revealed that patients in the pre-HD AVA group had a 52% lower rate of all-cause mortality (adjusted HR 0.48, 95% CI [0.39–0.59], p < 0.001, Fig. 4), a marginally lower rate of MACE-related mortality (adjusted HR 0.7, 95% CI [0.45–1.08], p = 0.1, Fig. 5), and a 68% lower rate of BSI-related mortality (adjusted HR 0.32, 95% CI [0.22–0.46], p < 0.001, Fig. S1) than those in the non-pre-HD AVA group. We next examined the outcomes in the propensity-score-matched groups. The patients with pre-HD AVA creation exhibited a 54% lower rate of all-cause mortality (adjusted HR 0.46, 95% CI [0.36–0.59], p < 0.001), a 40% lower rate of MACE-related mortality (adjusted HR 0.60, 95% CI [0.38–0.96] , p < 0.05), and a 71% lower rate of BSI-related mortality (adjusted HR 0.29, 95% CI [0.19–0.44], p < 0.01) than those without pre-HD AVA creation.

Sensitivity analyses of the outcomes are presented in Table 3. They revealed a consistently lower CHF hospitalization rates, ranging from 29% to 43%, of patients receiving pre-HD AVA creation. In addition, they exhibited a consistently lower BSI-related mortality rates of patients in the pre-HD AVA group.

Discussion

In our nationally representative cohort, we observed that patients with pre-HD AVA creation had a 35% lower CHF hospitalization rate and 52% lower all-cause mortality rate than those without during the first year of HD—significant differences. Additionally, we disclosed a marginally lower rate of MACEs and MACE-related mortality during the same follow-up period. Pre-HD AVA creation might be associated with better CV outcomes within the first year of HD.

In this study, the patients with pre-HD AVA creation were younger, had a lesser burden of comorbidities, had a higher percentage of ESAs but a lower percentage of ACEIs and ARBs usage (Table 1). Age is a well-established factor affecting postsurgery prognosis. The significantly lower Taiwan index of the pre-HD AVA group indicated the lower disease burden of the patients receiving pre-HD AVA compared with those not receiving pre-HD AVA. We also observed that the patients without pre-HD AVA creation were significantly more likely to have CVA, dysrhythmia, COPD, or liver disease than those with pre-HD AVA creation, and those diseases might reflect higher neurologic, respiratory, and coagulatory risks during the operation. In one previous study, patients with dementia had a greater risk of early death and fatal complications postoperatively (Kassahun, 2018). Thus, patients who were young or had fewer comorbidities were willing to undergo pre-HD AVA surgery. Additionally, a different proportion of patients receiving ESAs and renin–angiotensin–aldosterone system blockers might imply more recruitment of pre-HD care, which promotes the possibility of dialysis access creation by education (Ishani et al., 2014).

Our study revealed a significantly lower CHF hospitalization rate within the first year of HD among the patients receiving pre-HD AVA creation (Fig. 3). Consistent results were obtained for the other matched models (Table 2) and sensitivity analyses (Table 3). Patients who undergo AVA surgery before HD might avoid delayed HD, thus preventing exacerbated fluid overload and increased CHF risk. The increased cardiac preload after AVA surgery is compensated by a corresponding decrease in peripheral vascular resistance following surgery (Ori et al., 1996) and consequent fluid removal during HD sessions. The fluid status of most patients undergoing HD has been proved to achieve a new balance shortly (Alkhouli et al., 2015; Dal Canton et al., 1981). In the Dialysis Outcomes and Practice Patterns Study, Rayner et al. (2003) observed a low flow rate of the fistula in Japanese patients, which might be related to low-caliber vessels in the Asian population. The degree of CV damage due to blood volume following AVA creation might differ according to vascular characteristics. Several studies have supported our findings of CV benefits following pre-HD AVA creation: Ori et al. (1996) conducted an echocardiographic study to observe cardiac performance before and after AVA creation. A gentle volume overload developed postoperatively but was offset by decreased vascular resistance. The shortening and ejection fractions of the left ventricle improved 2 weeks after the AVA operation. Sandhu et al. (2004) observed that none of 17 patients receiving native fistula before HD developed CHF during the 6-week-period following surgery. They concluded that the postoperative changes in cardiac index, stroke volume, and vascular resistance were physically minimal and without extra loading of patients’ hemodynamics. Thus, CHF might not occur or worsen after AVA creation. Further investigation is warranted to clarify the causality of AVA in CV outcomes.

Some studies have obtained contrasting results from ours: MacRae et al. (2004) noted that a patient undergoing HD developed cardiac failure under a high-flow arteriovenous fistula and concluded that the high fistula flow caused myocardium decompensation with a decline in the ejection fraction. Other studies have adopted an opposite viewpoint on ejection fraction alteration following AVA creation (Iwashima et al., 2002; Korsheed et al., 2011; Ori et al., 1996). Vizinho et al. (2014) reported that pre-HD AVA creation was associated with a decrease in the subendocardial viability ratio, which predicted a poorer outcome regarding CV hospitalization. Nevertheless, a small sample size and lack of CV comorbidity adjustments limit the relevance of their speculation. Reddy et al. (2017) traced CV changes of patients following native shunt creation for 2.6 years and observed that remodeling and dysfunction of the right ventricle developing after shunt operation and dialysis initiation caused increased risks of CHF and death. However, the absence of controls and uncertainty in the effect of AVA and dialysis on cardiac dysfunction make the supposition inconclusive. More large-scale and close-matching studies should be planned to confirm the relationship.

Considering secondary outcomes, we evaluated the effect of pre-HD AVA creation on overall mortality and disclosed a 52% lower rate of all-cause mortality in the pre-HD AVA group (Table 2; Fig. 4). In addition to CV disease, catheter-associated infectious disease, mainly those transmitted through the bloodstream, is another major cause of mortality in patients undergoing dialysis. We assumed that the lower all-cause mortality might have been related to the 68% reduction in the rate of BSI-related mortality in the pre-HD AVA group (Table 2; Fig. S1), which was due to lesser usage of HD catheters. Furthermore, we also observed that the patients with pre-HD AVA creation had an insignificantly lower rate of MACE-related mortality compared with those without after propensity score matching (Table 2; Fig. 5). We believed that this type of mortality was mainly affected by the underlying diseases of patients rather than AVA surgery. The literature has suggested that HTN, IHD, CHF, and DM influence the CV mortality rate for patients undergoing HD (Banerjee et al., 2007; Lee et al., 2016; Zoccali, Tripepi & Mallamaci, 2005). Because our groups had similar distributions of these diseases, our finding was in fair agreement with the literature.

This was a country-based study including all pre-HD patients underlined CKD matched for age, sex, income, year of HD, comorbidities, and associated medicines. In Taiwan, the NHI Bureau has launched a pay-for-performance program focusing on patients of glomerulus filtration rate <45 ml/min/1.73 m² from 2006. The incentive payment from the bureau is paid to medical institutions if their recruited patients achieved the targets on blood pressure, glycated hemoglobin, nursing education, nutrition consult, and so on. Patients with ESRD have received comprehensive access evaluation by qualified nephrologists at outpatient clinics or during admissions. Most studies exploring the effects of pre-HD AVA creation on CV outcomes compared associated parameters before and after surgery (Dal Canton et al., 1981; Dundon et al., 2014; Iwashima et al., 2002; Korsheed et al., 2011; Munclinger et al., 1987; Ori et al., 2002, 1996; Reddy et al., 2017; Sandhu et al., 2004; Savage et al., 2002; Utescu et al., 2009; Vizinho et al., 2014). In addition, some compared the effects before dialysis initiation to exclude the impact of dialysis on CV performance (Dundon et al., 2014; Iwashima et al., 2002; Korsheed et al., 2011; Ori et al., 1996; Savage et al., 2002). However, selection bias would have been unavoidable in these studies because patients with pre-HD AVA creation tend to be compliant in medical practice, which would affect their overall outcomes. Furthermore, excluding the dialysis effect appears impractical considering the goal of AVA preparation. Moreover, dialysis is a well-known risk factor of cardiac injury, and its vintage is positively associated with the degree of injury (McIntyre, 2009). Once the AVA was used for dialysis, evaluating the CV prognoses in combination with HD was difficult. We followed up for 1 year after HD initiation because we assumed that the effect of pre-HD AVA creation would be offset by a longer period of HD. Our study provides another perspective regarding evaluation of the benefits and hazards of pre-HD AVA surgery.

This study had several limitations. First, the present study was an observational study, which non-observed confounders might restrict the inference. Second, the NHIRD is an administrative database in which the identification of comorbidities is based solely on ICD-9-CM codes rather than clinical criteria; misclassifications might thus have occurred, leading to residual confounding. Additionally, we recruited our patients until the end of 2012 since the data was valid until December 31, 2013 in LHID 2000. Third, patients could choose the preferred medical providers for AVA creation or HD freely owing to high medical accessibility in Taiwan. It is difficult to determine the relationship between patients and medical providers, which influenced the timing of AVA creation. Fourth, the indications of CHF hospitalization are varied among patients and medical facilities. However, the NHIRD does not provide objective parameters of cardiac alteration, such as the level of natriuretic peptide, ejection fraction of ventricles, or pulse wave velocity of vessels, which could support our findings. Fifth, our study did not consider medicines such as calcium channel blockers, beta blockers, or diuretics, which have been shown to influence CV outcomes in patients undergoing HD (Georgianos & Agarwal, 2016; Karaboyas et al., 2018). Lastly, it is difficult to distinguish the absolute effect of pre-HD AVA on CV outcomes in combination with personal compliance and dialysis factors affecting the CV system. Integrated trials comprising data and imaging should be further conducted to corroborate our results.

Conclusions

In this population-based cohort study, patients with pre-HD AVA creation had a 35% lower CHF hospitalization rate and a 52% lower all-cause mortality rate than those without pre-HD AVA creation within the first year of HD. Marginal benefits were also observed in terms of MACEs and MACE-related mortality during the same follow-up period. Pre-HD AVA creation might be associated with better CV outcomes in the first year of HD and should be promoted in the pre-HD care focusing on patients with late-stage CKD.

Supplemental Information