Frequent use of percutaneous transluminal angioplasty is not associated with reduced vascular access recreation in a longitudinal ecology study

Abstract

The frequency of percutaneous transluminal angioplasty (PTA) procedures has substantially increased, but its effect on vascular access recreation (VAR) remains inconclusive. We conducted a population-based retrospective analysis of Taiwan hemodialysis (HD) centers from 2004 to 2012. Data was accumulated into center-level characteristics, including patients' demographics, baseline characteristics, PTA procedures, and VAR. Center-level characteristics were summarized annually using appropriate measures. A mixed model assessed the association between PTA frequency and VAR rates, considering within-center correlation and adjusting for potential confounders. A total of 82,005 patients (mean age 62.7 ± 13.9 years, 50.5% male, 48.5% with diabetes mellitus) from 820 HD centers were analyzed. From 2004 to 2012, PTA frequency significantly increased from 1.24 to 3.23 per 1000 HD sessions, while VAR rates did not decline as expected (0.5–0.8 per 1000 HD sessions). Compared with the HD centers of infrequent use of PTA (annual lowest quartile, range 0.39–1.20 per 1000 HD sessions), the ones of frequent use (annual highest quartile, range 2.52–5.10 per 1000 HD sessions) didn't have lower VAR (range 0.54–0.99 vs. 0.50–0.91 per 1000 HD sessions, respectively). After controlling the potential confounders, the HD centers' PTA rates were not significantly associated with lower VAR rates (− 2.6, 95% confidence interval: − 30.3; 25.0, p = 0.85). Frequent use of PTA does not seem to improve VA patency at the center level, with no significant association identified with lower VAR. The indication of PTA in daily practice should be re-evaluated in terms of its efficiency in lowering VAR.

Introduction

Hemodialysis (HD) is the worldwide primary kidney replacement therapy modality for end-stage kidney disease (ESKD)¹. Keeping one functional vascular access (VA) is substantial in delivering hemodialysis efficiently. The 3-year functional patency of upper arm arteriovenous fistula (AVF) and arteriovenous graft (AVG) are reported 74% ± 2.0% and 69% ± 5%, respectively², which means many of the HD patients would need more than one time of VA creation³. Being the primary determining factor of VA failure, VA stenosis has long been the therapeutic target in order to prolong VA patency.

Percutaneous transluminal angioplasty (PTA) has been the most commonly used intervention over the past three decades to relieve the stenosis of VA, either fistula or graft, in HD patients^3,4. In addition to reducing the need for VA recreation, PTA has other advantages, such as preventing total occlusion of the VA to maintain dialysis and avoiding the need for temporary central catheterization. However, given the following reasons, whether PTA can prolong the accumulative patency of VA is inconclusive^5,6. Firstly, Angiography is not routinely employed for assessing VA in hemodialysis. This omission risks overlooking undiagnosed stenosis, potentially leading to misattributed patency outcomes following PTA. Studies focusing on PTA efficiency may inadvertently attribute VA patency to PTA, obscuring the true efficacy of the procedure in cases where underlying stenosis was not properly identified. Secondly, stenosis severity, commonly assessed by diameter reduction, significantly influences VA patency. However, published studies have inadequately adjusted for this critical parameters before PTA, potentially confounding results⁷. The lack of standardization in PTA procedures for HD VA, both during intervention and post-angioplasty vascular flow recovery, results in varied treatment outcomes and hinders uniform clinical approaches.

Addressing the aforementioned limitations poses a formidable challenge given the wide-ranging considerations, including ethical implications, various clinical practices, patients' preferences, and the technical complexities of measuring and alleviating VA stenosis severity. Despite the advancements in modern VA surveillance, state-of-the-art PTA, and sophisticated machine learning models, these concerns persist and remain difficult to overcome. Using a large national dataset and analyzing the result by counting each HD centers as an analytic unit, assuming these potential confounders randomly distributed among HD centers, we conducted this retrospective study to test the hypothesis that frequent PTA could prolong HD patients' cumulative patency of VA.

Methods and material

Study design and population

We conducted a retrospective population-based study, treating each dialysis care center as the analytical unit. All information in the study is extracted from the National Health Insurance Research Database, which is authorized by the Health and Welfare Data Science Centre, Ministry of Health and Welfare (NO: H106015). The prevalent HD population from 2004 to 2012 is determined using a series of reimbursed codes as described in our previous reports^8,9. The study population was selected by patients aged ≥ 20 years who underwent long-term HD (≥ 3 months) during the observed period. Patients diagnosed with cancer or who performed renal transplantation before dialysis were excluded to avoid specific patient characteristics distorting the dialysis center's characteristic estimation. The Institutional Review Board of Kaohsiung Medical University Hospital reviewed and approved the study protocol (KMUHIRB-EXEMPT(I)-20150026). The Institutional Review Board of Kaohsiung Medical University Hospital approved our request to waive informed consent due to the use of anonymous personal identification numbers in the database. All study procedures adhered to the principles of the Declaration of Helsinki.

Percutaneous transluminal angioplasty and vascular access recreation

The PTA for dialysis VA was a fee-for-service reimbursement under the Taiwan Health Insurance System. To accurately reflect PTA for dialysis VA, only specific procedure codes (Supplementary Table S1) that appear in outpatient or inpatient admission claim records accompanying peripheral vascular disease diagnosis codes [the International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM codes): 250.7, 440.21, and 443)] were identified. Because patients' primary VA may fail after dialysis, we defined reimbursed codes for surgeries involving AVF, AVG, and permanent catheter creation, as well as for medical materials used for AVG and temporary double-lumen catheters 4 months after dialysis initiation, which were identified from outpatient or inpatient admission claim data as subsequent VA recreation (VAR) (Supplementary Table S1).

Covariates

The study HD centers were categorized based on their affiliation hospital level (tertiary hospital, regional hospital, local hospital, and clinic). This categorization reflects the care center's capacity to handle more severe cases, offer teaching opportunities for training healthcare workers, and frequently results in higher reimbursement from national health insurance. To accurately represent the characteristics of the study dialysis centers, patients were selected based on the center where they received HD most frequently during each observed year. We collected several characteristics of the HD centers to use as covariates in our analysis. These included average annual HD service volume, average patient age and HD vintage, proportion of male patients, proportion of patients with comorbidities, and proportion of cumulative mortality. Patients with selected comorbidities were identified by specific ICD-9-CM codes occurring at least twice in outpatient records or once in admission hospital records within one year before dialysis initiation, as documented during the study period (see Supplementary Table S1). The selected comorbidities included diabetes mellitus (DM), hypertension, myocardial infarction, congestive heart failure, stroke, gout, and peripheral vascular disease. We determined death based on the last date of dialysis reimbursed codes, with no further reimbursement records observed within 3 months thereafter. Patient-level information was aggregated annually to form HD center-level data, focusing on the center where the patient received HD most frequently. PTA and VAR occurrences were attributed to an HD center only if the patient received the most frequent HD sessions there during the month of the procedure, the preceding month, and the following month.

Statistical analysis

The study described overall patients' characteristics using the mean and standard deviation or median, the first and third quartiles for continuous variables, and counts and percentages for categorical variables. In addition, we described the HD center's characteristics using mean and standard deviation and tested different distributions of characteristics between affiliation hospital levels (tertiary hospital, regional hospital, local hospital, and clinic) by one-way ANOVA. The annual dialysis center's PTA and VAR rates were calculated by dividing annual PTA and VAR numbers by the total HD service sessions and displayed as per thousand. Box plots were used to visually describe the annual dialysis center's PTA and VAR rates. The proportion of each comorbidity was counted to represent the patient severity for each HD center.

Additionally, the cumulative mortality proportion was calculated by dividing the sum of the death number by the cared patient number in 2004–2012 to represent the care scale and quality of the HD center. We used linear regression models to examine annual PTA and VAR rates, both overall and specific to affiliation hospitals, to detect linear trends over time using the calendar year as the independent variable. Due to potential variability in the analyzed data from HD centers over time, we computed the average PTA and VAR rates, along with covariate values, from 2004 to 2012. We classified HD centers as frequent PTA users if their mean annual PTA rate fell within the fourth quartile and as infrequent PTA users if their mean annual PTA rate fell within the first quartile. Trend plots were used to compare annual PTA and VAR rates among overall and affiliation hospital-specific HD centers categorized by frequent and infrequent PTA use from 2004 to 2012. Statistical differences between groups were assessed using mixed models, with the HD center treated as a random effect. Additionally, mixed models were employed to investigate HD center factors associated with annual PTA and VAR rates. We assumed the annual PTA and VAR rates for dialysis centers followed a normal distribution. Results from univariable and multivariable models were presented as regression coefficients with their respective 95% confidence intervals (CI). Data management and analysis were conducted using SAS (version 9.4; SAS Institute Inc., Cary, NC, USA), with statistical significance set at a two-tailed p value of < 0.05.

Results

Patient and hemodialysis center characteristics

We initially identified 112,741 patients who underwent hemodialysis (HD) for over 3 months. After excluding patients under 20 years old and those with cancer diagnoses, 82,005 patients were included in the final analysis (Fig. 1). The mean age was 62.7 ± 13.9 years, with a balanced gender distribution, and a mean HD vintage of 4.3 ± 2.9 years (Table 1). Variations in patient characteristics were observed across different hospital levels, with regional and local hospitals having the highest average age and tertiary and regional hospitals exhibiting the longest median HD vintage. Regional hospitals treated a higher proportion of diabetic patients, while local hospitals had more patients with congestive heart failure, stroke, and peripheral vascular disease.

Fig. 1.

Study flow diagram.

Table 1.

Patient characteristics in hemodialysis centers stratified by hospital level.

Variables	Patient level	HD center by hospital level*
	Overall population	Tertiary number = 17	Regional number = 86	Local number = 251	Clinic number = 466	P value
Patient characteristics
Patient number	82,005	20,091	24,120	16,645	21,149
Gender, male, N (%) for patient level, Mean ± SD for HD center
	40,809 (49.8)	50.5 ± 2.7	49.1 ± 7.1	51.5 ± 14.6	52.7 ± 15.0	0.59
Age, year; Mean ± SD	62.7 ± 13.9	62.9 ± 1.8	64.8 ± 3.3	64.8 ± 5.4	61.9 ± 6.3	< 0.001
HD vintage, year ; Mean ± SD	4.3 ± 2.9	3.1 ± 0.2	2.8 ± 0.8	2.6 ± 0.9	2.9 ± 1.0	0.11
Median (Q1, Q3)	3.7 (1.7, 6.9)	3.0 (2.9, 3.2)	3.0 (2.3, 3.3)	2.7 (2.0, 3.2)	2.9 (2.4, 3.5)	0.01
Comorbidity, yes, N (%) for patient level, Mean ± SD for HD center
Diabetes	40,009 (48.8)	48.5 ± 4.4	53.0 ± 8.5	49.7 ± 16.2	46.5 ± 17.5	< 0.001
Hypertension	56,463 (68.9)	70.2 ± 5.5	71.7 ± 11.9	68.0 ± 20.0	67.9 ± 19.5	0.12
Myocardial infarction	1,958 (2.4)	2.8 ± 1.0	2.7 ± 1.6	4.1 ± 4.5	4.5 ± 9.0	0.68
Congestive heart failure	14,663 (17.9)	16.6 ± 4.8	20.5 ± 6.7	21.7 ± 13.4	18.8 ± 13.1	0.002
Stroke	5,919 (7.2)	6.5 ± 1.9	8.8 ± 5.5	9.7 ± 6.2	9.0 ± 8.3	0.007
Gout	9,345 (11.4)	12.3 ± 2.1	10.8 ± 3.7	14.0 ± 10.9	14.1 ± 10.3	0.97
Peripheral vascular disease	4,265 (5.2)	5.5 ± 1.6	6.0 ± 3.1	9.0 ± 9.9	6.8 ± 7.6	0.005
HD center's characteristics
HD service volume by quartile (total HD sessions per year during observation )						< 0.001
Q1 (< 13,862)	–	0 (0)	9 (10.5)	59 (23.5)	137 (29.4)
Q2 (≥ 13,862 ~  < 40,518)	–	0 (0)	9 (10.5)	62 (24.7)	134 (28.8)
Q3 (≥ 40,518 ~  < 84,940)	–	1 (5.9)	13 (15.1)	75 (29.9)	116 (24.9)
Q4 (≥ 84,940)	–	16 (94.1)	55 (64)	55 (21.9)	79 (17)
Cumulative mortality proportion (per 100 persons)	–	39.0 ± 5.7	50.3 ± 21.1	52.8 ± 24.1	40.2 ± 23.9	< 0.001
PTA rate (per 1000 HD sessions)	–	2.1 ± 1.1	2.3 ± 1.7	2.2 ± 1.7	2.2 ± 1.5	0.50
VA recreation rate (per 1000 HD sessions)	–	0.7 ± 0.3	0.9 ± 0.5	0.8 ± 1.0	0.7 ± 0.5	0.03$

*All data is expressed using each HD center's data as an analytic unit. The distributions of characteristics between affiliation hospital levels (tertiary hospital, regional hospital, local hospital, and clinic) were tested by one-way ANOVA.

^$Although the P value was less than 0.05 in the one-way ANOVA test, insignificant differences between groups were approved by the post hoc test.

The cumulative mortality proportion was calculated by the sum of the death number divided by the cared patient number in 2004–2012.

HD hemodialysis, SD standard deviation, Q1 first quartile, Q2 second quartile, Q3 third quartile, Q4 forth quartile, PTA percutaneous transluminal angioplasty, VA vascular access.

Temporal trends in annual percutaneous transluminal angioplasty rate and vascular access recreation rate

The box plots illustrate a notable increase in the median and variation of the PTA rate over the observed years (Fig. 2), while the change in median VAR rate was comparatively smaller. Trend analysis revealed significant increases in annual PTA and VAR rates across overall HD centers and by hospital level (Table 2). By 2012, the average HD center's PTA rate had risen to 3.23 per thousand sessions, a 1.6-fold increase from 2004 (1.24 ‰). Conversely, the average VAR rate slightly increased from 0.53‰ in 2004 to 0.81‰ in 2012. Notably, the smallest increase in PTA rate from 2004 to 2012 was observed at tertiary hospitals, with only a 0.8-fold increment from 1.46‰ in 2004. While all hospital levels exhibited significant increasing trends in PTA rate (all P values < 0.001), slight increases in VAR rate were only significant at local hospitals and clinics, with no significant linear trend change observed at tertiary and regional hospitals. Significant differences in annual PTA rates were observed between HD centers with frequent and infrequent PTA use, with maximum annual differences ranging from 2.1 to 4.0 per 1000 HD sessions (Fig. 3A). In 2012, PTA rates in centers with frequent use were fourfold higher (4.9 per 1000 HD sessions) compared to those with infrequent use (1.2 per 1000 HD sessions). However, no notable differences were observed in annual VAR rates between HD centers with frequent and infrequent PTA use across the observed years, with annual differences ranging from − 0.04 to 0.07 per 1000 sessions. A substantial increase in annual PTA rate differences was noted, accompanied by only slightly lower annual VAR rates when comparing tertiary and regional hospitals with frequent PTA use to infrequent use (see Fig. 3B,C). Similar trends were observed in HD centers affiliated with local hospitals and clinics, suggesting that while there were increasing trends in annual PTA rates, they did not result in significant changes in annual VAR rate differences between HD centers with frequent and infrequent PTA use (see Fig. 3D,E).

Fig. 2.

Distributions of (A) percutaneous transluminal angioplasty and (B) vascular access recreation rates across hemodialysis centers from 2004 to 2012. Detailed figure legend: Abbreviation: HD, hemodialysis; PTA, percutaneous transluminal angioplasty; VAR, vascular access recreation.

Table 2.

Annual rates of percutaneous transluminal angioplasty (PTA) and vascular access recreation (VAR) in hemodialysis centers by hospital level, 2004–2012.

(Per 1000 HD sessions)		2004	2005	2006	2007	2008	2009	2010	2011	2012	P value
Overall	PTA	1.24 (1.25)	1.47 (1.39)	1.75 (1.60)	1.88 (1.60)	2.16 (1.69)	2.59 (1.83)	2.81 (2.15)	3.14 (2.09)	3.23 (1.87)	< 0.001
	VA	0.53 (0.45)	0.72 (0.54)	0.66 (0.55)	0.74 (0.57)	0.82 (0.64)	0.83 (0.62)	0.83 (0.74)	0.87 (0.97)	0.81 (0.73)	< 0.001
By hospital level
Tertiary	PTA	1.46 (0.96)	1.51 (0.96)	1.78 (1.10)	1.94 (1.14)	1.96 (1.32)	2.1 (1.18)	2.39 (1.31)	2.52 (1.30)	2.64 (1.30)	< 0.001
	VA	0.54 (0.29)	0.74 (0.4)	0.79 (0.54)	0.83 (0.41)	0.81 (0.51)	0.8 (0.47)	0.82 (0.51)	0.72 (0.47)	0.75 (0.61)	0.27
Regional	PTA	1.48 (1.48)	1.91 (1.66)	2.22 (1.88)	2.22 (2.00)	2.5 (1.84)	2.81 (1.67)	3.11 (1.74)	3.49 (1.71)	3.56 (1.97)	< 0.001
	VA	0.54 (0.34)	0.91 (0.5)	0.92 (0.54)	0.95 (0.58)	1.01 (0.65)	1.00 (0.65)	0.94 (0.64)	0.98 (0.56)	0.88 (0.55)	0.14
Local	PTA	1.22 (1.34)	1.50 (1.41)	1.75 (1.7)	1.89 (1.72)	2.12 (1.85)	2.76 (2.16)	3.06 (2.99)	3.22 (2.36)	3.26 (2.03)	< 0.001
	VA	0.52 (0.41)	0.73 (0.56)	0.66 (0.6)	0.70 (0.60)	0.83 (0.70)	0.88 (0.71)	0.96 (0.98)	1.02 (1.55)	0.93 (1.01)	< 0.001
Clinic	PTA	1.15 (1.09)	1.32 (1.28)	1.62 (1.46)	1.78 (1.42)	2.12 (1.57)	2.47 (1.69)	2.64 (1.72)	3.05 (2.05)	3.17 (1.79)	< 0.001
	VA	0.54 (0.51)	0.65 (0.54)	0.59 (0.5)	0.71 (0.55)	0.76 (0.6)	0.76 (0.57)	0.73 (0.6)	0.78 (0.62)	0.74 (0.58)	0.006

Data represented as mean (SD). We inspected the changing linear trend of overall and affiliation hospital-level-specific annual PTA and VAR rates through a linear regression model by putting the calendar year as an independent variable.

PTA percutaneous transluminal angioplasty, VA vascular access.

Fig. 3.

Percutaneous transluminal angioplasty and vascular access recreation rates: A Comparative Analysis Based on PTA Frequency at Hemodialysis Centers (A) Overall, (B) Tertiary Hospitals, (C) Regional Hospitals, (D) Local Hospitals, and (E) Clinics, from 2004 to 2012. Detailed figure legend: We categorized hemodialysis centers into frequent and infrequent percutaneous transluminal angioplasty (PTA) use based on their mean annual PTA rates. Frequent PTA use centers were defined as those in the fourth quartile of mean annual PTA rates, while infrequent PTA use centers were those in the first quartile. Statistical comparisons between these groups were conducted using mixed models, considering the HD center as a random effect. Abbreviations: HD, hemodialysis; PTA, percutaneous transluminal angioplasty; VAR, vascular access recreation.

Factors associated with hemodialysis center's percutaneous transluminal angioplasty rate

In univariable analysis, the proportion of male patients (8.1; 95% CI 0.7; 15.5; p = 0.033), proportion of patients with DM (13.8; 95% CI 7.4; 20.3; p < 0.001), HD vintage (264.1; 95% CI 167.8; 360.5; p < 0.001), and HD center mortality (466.2; 95% CI 128.5; 804.0; p = 0.007) were positively associated with HD center PTA rate. However, the proportion of patients with stroke was negatively associated with HD center PTA rate (− 21.0; 95% CI − 38.0; − 3.9; p = 0.016) (Table 3). The multivariable analysis revealed positive associations between HD center PTA rate and the average age of patients (34.7; 95% CI 11.7; 57.6; p = 0.003), proportion of patients with diabetes mellitus (10.2; 95% CI 3.4; 16.9; p = 0.003), and HD vintage (114.1; 95% CI 3.3; 224.9; p = 0.044). Significant disparities in PTA rates persisted between cohorts with frequent and infrequent utilization of HD centers, even after stratifying based on key variables: average age (< and ≥ 63 years old), average proportion of individuals with DM (< and ≥ 45%), and average HD vintage (< 2.5 and ≥ 2.5 years). (Supplemental Table S2).

Table 3.

Factors associated with percutaneous transluminal angioplasty rate in hemodialysis centers.

(N = 820)	Univariable model			Multivariable model
	Coefficient	95% CI	P value	Coefficient	95% CI	P value
Cared patients' characteristics
Average age	5.7	(− 15.7; 27.1)	0.6	34.7	(11.7; 57.6)	0.003
Male proportion	8.1	(0.7; 15.5)	0.033	4.4	(− 2.8; 11.6)	0.23
Comorbidity proportion
DM	13.8	(7.4; 20.3)	< 0.001	10.2	(3.4; 16.9)	0.003
MI	0.2	(− 23.6; 24)	0.99	4.4	(− 19.5; 28.3)	0.72
CHF	1.3	(− 8.0; 10.5)	0.79	− 2.3	(− 12.2; 7.6)	0.64
Stroke	− 21.0	(− 38.0; − 3.9)	0.016	− 14.4	(− 31.1; 2.4)	0.09
HD vintage	264.1	(167.8; 360.5)	< 0.001	114.1	(3.3; 224.9)	0.044
HD center's characteristics
Hospital level
Tertiary	0.0 [Ref.]	−		0.0 [Ref.]	−
Regional	359.8	(− 1231.3; 1951)	0.66	485.5	(− 735.8; 1706.9)	0.44
Local	341.4	(− 1221.2; 1904)	0.67	512.2	(− 689.5; 1713.9)	0.40
Clinic	450.0	(− 1105.7; 2005.6)	0.57	524.8	(− 672.2; 1721.9)	0.39
HD service volume
1st quarter	0.0 [Ref.]	–		0.0 [Ref.]	–
2nd quarter	466.2	(128.5; 804.0)	0.007	234.3	(− 100.3; 568.9)	0.17
3rd quarter	288.5	(− 49.2; 626.2)	0.09	151.5	(− 191.5; 494.6)	0.39
4th quarter	216.5	(− 121.3; 554.2)	0.21	121.2	(− 241.5; 483.9)	0.51
Mortality	466.2	(128.5; 804.0)	0.007	234.3	(− 100.3; 568.9)	0.17

Uni- and multivariable analyses were conducted using mixed-effect models, assuming a normal distribution for the dependent variable. All covariates listed in the table were included in the multivariable analysis using a forced entry approach.

CI confidence interval, DM diabetes mellitus, MI myocardial infarction, CHF congestive heart failure.

Factors associated with hemodialysis center's vascular access recreation rate

We further explored the associations of study factors with HD center's VAR rate by univariable and multivariable analyses (Table 4). The results of univariable analyses revealed that the proportion of males (4.0; 95% CI 1.2; 6.8; p = 0.006) and HD vintage (55.8; 95% CI 18.2; 93.5; p = 0.004) were positively associated with the rate of VAR in HD centers, whereas the proportion of mortality (− 2.3; 95% CI − 4.2; − 0.5; p = 0.013) was negatively associated. These associations were consistent in the multivariable analysis for the proportion of males (3.8; 95% CI 0.9; 6.7; p = 0.01) and mortality proportion (− 2.3; 95% CI − 4.5; − 0.1; p = 0.039). Furthermore, average age (11.0; 95% CI 1.8; 20.3; p = 0.01), rather than HD vintage, was positively associated with the VAR rate in HD centers. Surprisingly, there was no significant linear association observed between the rate of PTA and VAR events in HD centers in both univariable (16.0; 95% CI 10.4; 42.4; p = 0.23) and multivariable analyses (− 2.6; 95% CI − 30.3; 25.0; p = 0.85).

Table 4.

Factors associated with vascular access recreation rate in hemodialysis centers.

(N = 820)	Univariable model			Multivariable model
	Coefficient	95% CI	P value	Coefficient	95% CI	P value
Cared patients' characteristics
Average age	8.3	(− 0.1;16.7)	0.055	11.0	(1.8; 20.3)	0.019
Male proportion	4.0	(1.2;6.8)	0.006	3.8	(0.9; 6.7)	0.01
Comorbidity proportion
DM	2.4	(− 0.1;5.0)	0.06	0.4	(− 2.3; 3.1)	0.76
MI	6.2	(− 3;15.3)	0.19	4.9	(− 4.6; 14.5)	0.31
CHF	2.9	(− 0.7;6.5)	0.11	1.9	(− 2.1; 5.8)	0.36
Stroke	2.0	(− 4.7;8.6)	0.56	2.1	(− 4.6; 8.9)	0.53
HD vintage	55.8	(18.2;93.5)	0.004	41.6	(− 2.9; 86.0)	0.07
HD center's characteristics
PTA rate	16.0	(− 10.4;42.4)	0.23	− 2.6	(− 30.3; 25.0)	0.85
Hospital level
Tertiary	0.0 [Ref.]	–		0.0 [Ref.]	–
Regional	233.8	(− 378.8; 846.5)	0.45	239.7	(− 110.2; 589.6)	0.18
Local	191.7	(− 410.0; 793.3)	0.53	202.0	(− 136.8; 540.9)	0.24
Clinic	67.1	(− 531.9; 666.1)	0.83	84.5	(− 251.8; 420.7)	0.62
HD service volume
1st quarter	0.0 [Ref.]	–		0.0 [Ref.]	–
2nd quarter	94.9	(− 35.6; 225.5)	0.15	43.6	(− 90.5; 177.7)	0.52
3rd quarter	70.8	(− 60.0; 201.7)	0.29	26.8	(− 110.6; 164.2)	0.7
4th quarter	38.2	(− 96.5; 173.0)	0.58	0.1	(− 145.1; 145.4)	> 0.99
Mortality	− 2.3	(− 4.2; − 0.5)	0.013	− 2.3	(− 4.5; − 0.1)	0.039

Uni- and multivariable analyses were conducted using mixed-effect models, assuming a normal distribution for the dependent variable. All covariates in the table were included in the multivariable analysis using a forced entry approach.

CI confidence interval, DM diabetes mellitus, MI myocardial infarction, CHF congestive heart failure.

Discussion

In this study, the efficacy of PTA was evaluated regarding the clinical outcome of VA patency, represented as VAR at the HD center level. We observed a significantly increased trend of PTA rate at all four hospital levels over time, while none showed an apparent reduction in VAR rate over the 9-year observation period. PTA rates varied among HD centers, with the frequent (the fourth quartile) and infrequent (the first quartile) groups showing nearly a fourfold difference, and this gap widened over time at all hospital levels. Surprisingly, HD centers with frequent use of PTA were not associated with a significantly lower VAR rate (coefficient: − 2.6, 95% CI − 30.3; 25.0). Similar results were observed even after considering the influences of hospital level, mean age, gender, and DM status. These findings suggest a need to reconsider recent indications of PTA and VAR as appropriate and advocate for developing more precise care guidelines for PTA indications.

As demonstrated in our study, an increased frequency of PTA use during the observation period is expected. This trend can be attributed to several factors, including the procedure's accessibility, continuous improvement in PTA skill, the cumulated effect of restenosis following previous PTA procedures, and certain demographic characteristics that have not been thoroughly examined in previous studies¹⁰. Our findings suggest that factors such as longer HD vintage, aging, and DM contribute to the increased use of PTA, as these conditions are associated with a higher risk of stenosis due to poorer vascular condition, inappropriate cannulation, hypotension during HD, among others, as reported by other investigators^8,11–14. These independent risk factors may partially contribute to the increased trend of PTA over time in both frequent and infrequent use groups. However, they cannot fully explain the variations in PTA rates among HD center at all four levels, raising concerns about the accuracy and consistency of PTA indications in clinical practice (Supplement Material). In other words, PTA’s highly diverse practice patterns suggest each HD center might have its own indications. The recent KDOQI guidelines do not support pre-emptive PTA intervention to improve access patency without clinical indicators³. Access flow and venous pressure changes have traditionally been common clinical indicators for detecting VA stenosis. However, both are influenced by various factors such as VA types, location, cannulation methods, and individualized patient blood pressure, which complicates their clinical application¹⁵. Several studies have attempted to identify indicators for PTA. A recent study found that weight-based VA flow could predict AVF functional failure and might serve as an indicator for PTA, but further large-scale clinical validation is needed¹⁶. Accurate and reliable clinical indicators of VA stenosis for guiding PTA, based on both short- and long-term outcomes, should be developed urgently in future research. From a policy perspective, Taiwan's healthcare system reimburses PTAs to ensure accessibility, yet effectiveness in reducing VA recreation remains unclear. Care teams view PTAs as proactive in relieving stenosis, but clinical evidence on sustained patency varies. Patients prefer PTAs due to concerns over surgery risks, despite uncertain long-term benefits. Demographic trends, like aging and comorbidities, drive PTA use, but our study does not establish a clear causal relationship between PTA frequency and VA recreation reduction. Future studies should focus on determining the optimal frequency of PTA to maximize VA patency.

Our findings do not support the efficacy of frequent PTA in improving VA patency, prompting a reconsideration of the role of PTA. In our study, the rate of PTA procedures in frequent PTA use HD centers in 2012 was nearly four times higher compared to infrequent centers. However, a similar VAR was observed in both types of centers in 2012. The finding suggests that some HD centers may rely on repeated PTA as the primary procedure to maintain VA function. However, repeated PTA is expensive and inefficient, as indicated by our findings. Chan et al. arrived at a conclusion similar to our study. In their matched cohort, balanced for vascular type, access age, intra-access blood flow, and Kt/V, they found that VA survival did not significantly differ between the non-intervention and PTA groups. The vascular survival benefit of PTA was only significant in participants with inadequate access flow or new VA¹⁷. However, they did not control the severity of VA stenosis. Another recent study observed patients who required a second treatment within 90 days after the initial PTA and found that those who underwent surgical reconstruction had significantly better one-year patency proportion than those who underwent a second PTA⁷. Neointimal hyperplasia was an important mechanism influencing AVF maturation and VA failure. Many complex pathophysiologic pathways involving neointimal hyperplasia, such as inflammation, tissue hypoxia, oxidative stress, endothelial mesenchymal transition, and fibrosis, were associated with VA stenosis and failure^18–20. PTA directly manages the stenosis site but cannot correct the origin causes of stenosis. On the other hand, although PTA manages VA stenosis, it probably induces vascular injury and restenosis²¹.

This study has several strengths. Firstly, we obtained data from a large national dataset comprising eighty-two thousand HD patients from eight hundred twenty HD units, ensuring the robustness and reliability of our findings. Secondly, we assessed the effect of PTA on VAR at the HD center level rather than the individual level. This approach was based on the assumption that the severity of a patient's VA stenosis is randomized across HD centers by patients conveniently choosing an HD center near their home. By adopting this study design, we could circumvent the influence of individual VA stenosis and quantify how PTA affected VAR. It is well known that the severity of VA stenosis is a major determinant of accumulated patency and is challenging to measure and compare among HD patients. This study also has several limitations. Firstly, we lack data on the severity of VA stenosis in patients receiving PTA or not. Then, the indications for each PTA are unclear, which may weaken the persuasiveness of further inferences. However, both limitations can be mitigated by analyzing the results using the HD center as the unit of analysis. Another consideration is the dataset's lack of contemporaneity. While there have been advancements in techniques and devices for VA stenosis intervention, such as stent grafts and drug-coated balloons, the guidelines for VA surveillance and PTA indications are inconclusive on using these new tools. Undoubtedly, one must interpret the results with caution considering the age of the dataset. Finally, many factors are associated with VAR, some of which are unknown or not fully understood. While we have made the best possible adjustments to the model, using the mortality proportion to represent quality care may not be sufficient. Further randomized clinical trials are needed to address these issues.

Conclusion

In summary, our study suggests that frequent use of PTA does not associate with longer vascular access patency at the center level, as evidenced by the absence of a significant association with lower vascular access recreation rates. These findings underscore the importance of reevaluating the indication for PTA in daily practice, particularly its efficacy in reducing vascular access recreation.

Abbreviations

HD

Hemodialysis

ESKD

End-stage kidney disease

VA

Vascular access

AVF

Arteriovenous fistula

AVG

Arteriovenous graft

PTA

Percutaneous transluminal angioplasty

VAR

VA recreation

DM

Diabetes mellitus

Author contributions

Research idea and study design: M.Y.L., P.H.W., and Y.W.C.; data acquisition: M.Y.L., F.X.C., and Y.W.C.; statistical analysis: M.Y.L. and F.X.C.; data analysis/interpretation: M.Y.L., F.X.C., and Y.W.C.; manuscript drafting: P.Y.W., M.Y.L., and Y.W.C.; supervision or mentorship: M.Y.L. and Y.W.C. M.Y.L. and Y.W.C. take responsibility for reporting this study honestly, accurately, and transparently, ensuring that no important aspects of the study have been omitted and that any discrepancies from the study, as planned, have been explained. All authors reviewed the manuscript.

Funding

This study was supported by the Taiwan Ministry of Science and Technology (grant number MOST: 104-2314-B-037-054), and Kaohsiung Medical University Hospital (grant numbers: KMUH104-4R11) to YWC. The funders had no role in the design and implementation of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.

Data availability

All data are included in the manuscript and/or supporting information for reference. The raw data could only be accessed and analyzed under the regulations of the Health and Welfare Data Science Centre, Ministry of Health and Welfare.

Competing interests

Ming-Yen Lin and Yi-Wen Chiu serve as editorial board members of Scientific Reports. This does not alter the authors’ adherence to Scientific Reports editorial policies and criteria. The other authors declared there are no competing interests.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-71158-z.