The Impact of Periprocedural Prosthetic Valve Leak After Transcatheter Aortic Valve Implantation

Abstract

A periprocedural prosthetic valve leak (PVL) after transcatheter aortic valve implantation (TAVI), a minimally invasive treatment modality for patients with severe, symptomatic aortic stenosis, may entail serious morbidity. Cohorts hospitalized for TAVI from a national database (2016–2020) were stratified on the presence of PVL post-TAVI. TAVI patients with and without PVL were selected for propensity score matching. Pearson’s x² test was used to compare outcomes. Among 319,448 TAVI patients over five years, 2043 had periprocedural PVL identified at index hospitalization, acute heart failure (49.61% vs. 41.15%, p < 0.001), acute kidney injury (20.40% vs. 11.77%, p < 0.001), cardiac tamponade (1.31% vs. 0.52%, p < 0.05), higher inpatient mortality (3.05% vs. 1.05%, p < 0.001), postprocedural bleeding (3.5% vs. 1.48%, p < 0.001), sudden cardiac arrest (15.34% vs. 8.54%, p < 0.001), and vascular complications (4.10% vs. 1.57%, p < 0.001). TAVI with PVL was associated with a significantly longer length of stay (p < 0.05) and total cost of hospitalization (p < 0.05). The 30-day (15.2% vs. 12%, p = 0.02), 90-day (24.4% vs. 19.9%, p < 0.01), and 180-day (34.7% vs. 24.8%, p < 0.01) readmission rates were significantly higher in the TAVI cohort with PVL. PVL in patients post-TAVI is associated with greater mortality and morbidity during index hospitalization, higher readmission rates, and increased burden on healthcare costs and infrastructure.

1. Introduction

Transcatheter aortic valve implantation (TAVI) has been consistently shown to be beneficial for the treatment of aortic stenosis (AS), the most common valvular pathology. Several landmark trials have shown the benefits of TAVI across the spectrum of surgical risk [1]. The catheter-based approach avoids the surgical risk but often leads to reintervention [2]. Although prosthetic valvular leak (PVL) is a risk factor for reintervention, its clinical and economic impact is uncertain [3]. A meta-analysis [4] confirmed high-quality trial data that PVL increases mortality risk in high-risk patients with aortic stenosis [5,6] and post hoc analyses in intermediate-risk patients [7,8] and extended these findings to cases of mild PVL [4]. Nevertheless, a lack of access to source data may have confounded meta-analyses [9]. We seek to provide clinicians with recommendations to minimize the mortality and morbidity of PVL for TAVI by means of a propensity-matched (PSM) analysis utilizing a national database [6,8].

2. Materials and Methods

2.1. Study Design and Population

The Nationwide Readmissions Database (NRD) of the Healthcare Cost and Utilization Project (HCUP) [10], a resource maintained by the Agency for Healthcare Research and Quality (AHRQ) of the Department of Health and Human Services of the United States, provides data on roughly 35 million weighted hospitalizations [10]. It is a nationally representative administrative database of the United States of America comprising discharge and readmission records of 58.2% of all hospitalizations. Patients who underwent TAVI were selected utilizing codes from the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) [11,12] (

Table 1. International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes for cohort identification and comorbidities [11].

Study Variables	ICD-10-CM Code
Acute kidney injury	N170, N171, N172, N178, N179, N1
Alcohol use	“F101, F1010, F1011, F1012, F10120, F10121, F10129, F1014, F1015, F10150, F10151, F10159, F1018, F10180, F10181, F10182, F10188, F1019, F102, F1020, F1021, F1022, F10220, F10221, F10229, F1023, F10230, F10231, F10232, F10239, F1024, F1025, F10250, F10251, F10259, F1026, F1027, F1028, F10280, F10281, F10282, F10288, F1029, F10920, F10921, F10929, F1094, F1095, F10950, F10951, F10959, F1096, F1097, F1098, F10980, F10981, F10982, F10988, F1099” [14] *
Anemia	D50, D51, D52, D53, D55, D56, D57, D58, D59, D60, D61, D62, D63, D64, D46.0, D46.1, D46.2, D46.4, O99.0
Aortic valve replacement/repair	02RF07, 02RF08, 02RF0K, 02RF0J, 02RF47, 02RF48, 02RF4J, 02RF4K, 02QF0Z, 02QF4Z, 02UF07, 02UF08, 02UF0J, 02UF0K, 02UF47, 02UF48, 02UF4J, 02UF4K
Atrial fibrillation/flutter	I48 family
Breakdown of prosthetic heart valve	T8201XA, T8201XD, T8201XS
Cardiogenic shock	R570
Chronic kidney disease stage ≥3	N183, N184, N185, E082, E132, I12, I13
Chronic obstructive pulmonary disease	J449
Coronary artery disease	“I2510, I25111, I25118, I25119, I252, I253, I254, I2541, I2542, I255, I256, I257, I2570, I25700, I25701, I25708, I25709, I2571, I25710, I25711, I25718, I25719, I2572, I25720, I25721, I25728, I25729, I2573, I25730, I25731, I25738, I25739, I2575, I25750, I25751, I25758, I25759, I2576, I25760, I25761, I25768, I25769, I2579, I25790, I25791, I25798, I25799, I258, I2581, I25810, I25811, I25812, I2582, I2583, I2584, I2589, I259” [14] *
Diabetes mellitus	E08-E13 family
Displacement of prosthetic heart valve	T8202XA, T8202XD, T8202XS
End stage renal disease	N186, Z992, Z4931, Z4901
Heart failure	I50, I501, I502, I5020, I5021, I5022, I5023, I503, I5030, I5031, I5032, I5033, I504, I5040, I5041, I5042, I5043, I508, I5081, I50810, I50811, I50812, I50813, I50814, I5082, I5083, I5084, I5089, I509
History of stroke	I69.3, Z86.73
Hypertension	I10, I1150, I1151, I1152, I1158, I1159
Mechanical circulatory support	5A02110, 5A02210, 5A0211D, 02HA3RZ, 5A02116, 5A0221D, 5A1522F, 5A1522G, 5A1522H, 5A15A2F, 5A15A2G, 5A15A2H
Mitral stenosis	Non-rheumatic (I342); Rheumatic (I050, I052)
Mitral valve replacement/repair	02RG07, 02RG08, 02RG0J, 02RG0K, 02RG47, 02RG48, 02RG4J, 02RG4K, 02QG0Z, 02QG4Z,02UG07, 02UG08, 02UG0J, 02UG0K, 02UG47, 02UG48, 02UG4J, 02UG4K
Obesity	E66, Z683, Z684, R939, Z6854, 09921
Obstructive sleep apnea	G47.33
Other non-specific valve complications	T8209XA, T8209XD, T8209XS
Peripheral vascular disease	E08.5, E09.5, E10.5, E11.5, E13.5, I73, T82.856, Z98.62, Z95.820, I25.2, I25.83
Prior coronary artery bypass graft	Z951
Prior myocardial infarction	1252
Prior percutaneous coronary intervention	Z955
Postprocedural bleeding	I97418, I9742, I97618, I97620, I97638, I97631, I97621, I9742, I97418, I97411, I97611, I97410, I97610, I97630, I97410
Prosthetic valve leak	T8203XA, T8203XD, T8203XS
Pulmonary valve replacement/repair	02RH07, 02RH08, 02RH0J, 02RH0K, 02RH47, 02RH48, 02RH4J, 02RH4K, 02QH0Z, 02QH4Z, 02UH07, 02UH08, 02UH0J, 02UH0K, 02UH47, 02UH48, 02UH4J, 02UH4K
Smoker	F17, Z87.891
Transcatheter aortic valve implantation	02RF37, 02RF37H, 02RF37Z, 02RF38, 02RF38H, 02RF38Z, 02RF3J, 02RF3JH, 02RF3JZ, 02RF3K, 02RF3KH, 02RF3KZ, 02RF4, 02RF47, 02RF47Z, 02RF48, 02RF48Z, 02RF4J, 02RF4JZ, 02RF4K, 02RF4KZ
Transcatheter pulmonary valve replacement	02RH37H, 02RH37Z, 02RH38H, 02RH38Z, 02RH3JH, 02RH3JZ, 02RH3KH, 02RH3KZ
Tricuspid valve replacement/repair	02RJ07, 02RJ08, 02RJ0J, 02RJ0K, 02RJ47, 02RJ480, 02RJ4J, 02RJ4K, 02QJ0Z, 02QJ4Z, 02UJ07, 02UJ08, 02UJ0J, 02UJ0K, 02UJ47, 02UJ48, 02UJ4J, 02UJ4K
Vascular complications	S15, S25, S35, S55, S65, S75, S85, T817, T8183XA, T8183XS, I770, I97621, I97630, I97631, I97638

* Used with permission of Springer Nature BV, from Proportion of Patients in the United States Who Fill Their Nirmatrelvir/Ritonavir Prescriptions; Rudolph, Abby E., Khan, Farid L., Singh, Tanya G., Valluri, Srinivas Rao, Puzniak, Laura A., McLaughlin, John M., Infectious Diseases and Therapy, volume 13, page(s) 2035–2052, 2024 [14]; permission conveyed through Copyright Clearance Center, Inc., Danvers, MA, USA.

) and the International Classification of Diseases, Tenth Revision, Procedure Coding System (ICD-10-PCS) [13]. Patients who underwent surgical aortic valve replacement (n = 415), mitral valve repair/replacement (n = 270), tricuspid valve repair/replacement (n = 14), and pulmonary valve repair/replacement (n = 0) were removed from the analysis to limit bias from other prosthetic valve leakages. All duplicates and minors (<18 years, n = 18) were excluded. Subsequently, we classified the patients into two categorical groups: TAVI patients who developed periprocedural prosthetic valve leakage (PVL) during index hospitalization and those without PVL during index hospitalization. For our subgroup analysis, we identified TAVI patients with other periprocedural mechanical complications, including prosthetic valve breakdown and displacement. Our study cannot fully generalize to the population of all patients who undergo TAVI because our data identify only the presence or absence of PVL and not the spectrum of PVL severity ranging from mild to severe. We employed variables associated with the number of days to intervention and length of stay (LOS) to determine readmission rates within the same patient cohort. The entire dataset from the index admission was included in the analysis. Since patients admitted within the same calendar year can only be identified by the National Readmission Database (NRD), which is arranged annually, we systematically included data from the first 11 months, 9 months, and 6 months of each year to ensure follow-up periods of 30, 90, and 180 days, respectively. In compliance with the reporting guidelines of the Healthcare Cost and Utilization Project (HCUP), observations containing fewer than 11 cases were omitted [10].

2.1.1. Baseline Characteristics

We identified adult (≥18 years old) TAVI patients. Baseline patient variables, including age, sex, comorbidities, and the Charlson Comorbidity Index [15], were analyzed. We also looked at other hospital characteristics, such as bed size, teaching status, and urban–rural classification.

2.1.2. Study Outcomes

The primary outcome was the difference in in-hospital mortality between TAVI patients with and without PVL. Secondary outcomes included other complications during the index hospitalization: acute kidney injury (AKI), acute heart failure (AHF), acute myocardial infarction (AMI), cardiogenic shock (CS), mechanical circulatory support (MCS), sudden cardiac arrest (SCA), postprocedural bleeding and vascular complications; length of stay (LOS), adjusted total charges, hospitalization costs on the index hospitalization, and 30-, 90-, and 180-day readmissions; propensity-matched 30-, 90-, and 180-day readmission rates, and causes of readmissions (

Table 2. Definitions of major outcomes after transcatheter aortic valve implantation (TAVI).

Outcomes	Definition
Acute kidney injury	Any acute kidney injury in TAVI patients during the hospital stay
Cardiogenic shock	Shock resulting from primary failure of the heart in its pumping function, as in myocardial infarction, severe cardiomyopathy, mechanical obstruction, or compression of the heart
Cost of hospitalization	The total adjusted amount that hospitals billed for their services for the duration of hospitalization
Length of stay (LOS)	The entire length of stay the patient spent in the hospital during the admission
Mechanical circulatory support	Any type of circulatory support with a balloon pump, impeller pump, extracorporeal membrane oxygenation (ECMO), or external heart assist system utilization during the index hospitalization
Mortality	All causes of death, including cardiovascular causes such as sudden cardiac death, death due to acute myocardial infarction, heart failure or cardiogenic shock, and non-cardiovascular causes
Postprocedural bleeding	Any major bleeding or hematoma during or after the procedure
Prosthetic valve leak	Leakage of prosthetic valve post-TAVI identified during the index hospitalization
Resource utilization	Calculated utilizing length of stay, inflation-adjusted total cost of the index hospitalization, including any reintervention or intensive care stays during index hospitalization. Subsequently, similar methodology was utilized to determine cost of hospitalization for readmission at 30, 90, and 180 days.
Valve complications	Valve breakdown, embolization, or other non-specified complications post-TAVI identified during the index hospitalization
Vascular complications	Any vascular injury as a complication of the procedure during the hospitalization post-TAVI

).

2.2. Statistical Analysis

To summarize both continuous and categorical variables, descriptive statistics were used. Categorical variables were reported as percentages and frequencies and compared using Pearson’s x² test. The distribution of data was evaluated with histogram analysis (Figure 1), and the independent sample t-test analysis was utilized to compare continuous variables which were normally distributed. The Mann–Whitney U test (Wilcoxon rank sum test) was used to compare continuous data which had non-parametric distribution. Differences in patient demographics, comorbidities, and study outcomes were examined between TAVI patients with and without PVL/other mechanical complications. The frequency of missing values was described, and Little’s missing completely at random (MCAR) was applied to detect missing data patterns [16]. A non-significant p-value (p > 0.05) indicated the data were missing at random, whereas a significant p-value (p < 0.05) suggested missing not at random (MNAR) [16]. The data were largely complete, with only randomly missing data patterns in the following variables: “Primary Expected Payer” missing (n = 334, 0.1%) and “Admission Status” missing (n = 629, 0.19%). As randomly missing data were less than 0.2% of our overall study population, we marked them missing and excluded them from analysis.

Following the management of missing data, univariate and multivariate logistic regression models were used to evaluate unadjusted and adjusted odds ratios for in-hospital outcomes within the study cohorts. The adjusted odds ratios of in-hospital outcomes were deemed significant with a p-value ≤ 0.05 and a confidence interval which does not include 1. Comorbidities and demographics were matched to control the confounders. The regression model was constructed using univariate screening, and a p-value threshold of ≤0.2 was determined as a cutoff for covariate inclusion in the final multivariate regression model (

Table 3. Multivariate regression variables obtained from univariate regression.

Variables Included in Multiple Regression
Age
Chronic kidney disease stage > 3
Chronic heart failure
Chronic pulmonary disease
Coronary artery disease
COVID-19
Deficiency anemia
Diabetes mellitus
End stage renal disease
Elective/non-elective admission
Family history of coronary artery disease
Gender
Hemodialysis
Hospital bed size
Hospital location and teaching status
Hospital region
Hyperlipidemia
Hypertension
Hypothyroidism
Obesity
Obstructive sleep apnea
Payer
Prior cerebral vascular accident
Prior coronary artery bypass graft
Prior myocardial infarction
Prior percutaneous coronary intervention
Pulmonary circulation disorder
Pulmonary hypertension
Rehab transfer
Resident
Weekend admission
Weight loss

). We computed the variance inflation factor (VIF) and tolerance (1/VIF) to evaluate multicollinearity. VIF > 5 and tolerance value < 0.2 were used as a significant correlation marker among independent variables. Therefore, deficiency anemia and weight loss were removed from the original variables (Table 3) for subsequent propensity score matching (PSM). A PSM model utilized the multivariate regression and implemented complete Mahalanobis Distance Matching within the Propensity Score Caliper (0.2) afterwards to create matched cohorts. The PSM-matched cohorts’ outcomes were compared using Pearson’s x² test. To evaluate the balance of variables in our propensity-matched cohorts, graphical box plots were generated (Figure 2). A similar PSM model was applied on 30-, 90-, and 180-day readmission analysis to compute readmission rates on matched cohorts. To avoid mortality readmission bias, index hospitalizations alive at discharge were retained for readmission analysis. Given the study population’s non-parametric distribution, trend analysis was performed using the Cochran–Armitage test for binary outcomes and the Jonckheere–Terpstra test or Cuzick’s test for continuous or ordered categorical variables. Total cost was adjusted for national inflation and integrated with cost–charge ratio (CCR) NRD files. In compliance with HCUP regulations, appropriate stratification, clustering, and weighting samples were conducted throughout all analyses. All statistical analyses were performed using Stata v. 18 software (Stata Corp, College Station, TX, USA) [17]. BioRender was used for the central illustrations [18].

3. Results

3.1. Demographic and Baseline Characteristics

A retrospective analysis was conducted on a large cohort of 319,448 TAVI patients, out of which 317,405 (99.36%) are in the TAVI without PVL group and 2043 (0.63%) in the TAVI with periprocedural PVL identified during the index hospitalization group. The demographics and baseline comorbidities comparison is presented in

Table 4. Baseline characteristics and comorbidity comparisons of patients who underwent transcatheter aortic valve implantation (TAVI) without and with periprocedural prosthetic valvular leak (PVL).

	TAVI Without PVL n = 317,405	TAVI with PVL n = 2043	p-Value
Age (median + interquartile range)
	80 (12)	79 (13)	<0.01
Indicator of sex
Male	176,444 (55.6%)	1353.10 (66.2%)	<0.001
Female	140,960 (44.4%)	690.60 (33.8%)
Insurance type
Medicare	284,526 (89.7%)	1793.80 (87.9%)	0.502
Medicaid	3820 (1.2%)	39 (1.9%)
Private insurance	21,250 (6.7%)	159.30 (7.8%)
Self-pay	1060 (0.3%)	10.3 (<1%)
Other	6326 (2.%)	39.30 (1.9%)
Type of admission
Non-elective	54,989.7 (17.4%)	544.40 (26.6%)	<0.001
Elective	261,786.1 (82.6%)	1499.30 (73.4%)
Bed size of the hospital
Small	15,728.5 (5.%)	47 (2.3%)	0.058
Medium	69,442 (21.9%)	490.90 (24.%)
Large	232,234 (73.2%)	1505.80 (73.7%)
Teaching status of urban hospitals
Metropolitan non-teaching	33,621.10 (10.6%)	190.80 (9.3%)	0.539
Metropolitan teaching	280,698.80 (88.4%)	1828.90 (89.5%)
Non-metropolitan hospital	3084.70 (1.00%%)	24 (1.2%)
Hospital urban–rural designation
Large metropolitan areas with at least 1 million residents	187,116.60 (59.%)	1392.70 (68.1%)	<0.001
Small metropolitan areas with less than 1 million residents	127,203.20 (40.1%)	627 (30.7%)
Micropolitan areas	3017.20 (1%)	21.2 (1%)
Admission day of the week
Mon–Fri	306,417.20 (96.5%)	1944.10 (95.1%)	0.101
Sat–Sun	10,985.40 (3.5%)	99.6 (4.9%)
Transfer flag indicating combination of discharges involving same-day events
Not a transfer or other same-day stay	304,517.80 (95.90%)	1888.10 (92.40%)	<0.001
Transfer involving two discharges from different hospitals	6723.10 (2.10%)	94.1 (4.60%)
Same-day stay involving two discharges from different hospitals	2743 (0.90%)	28.8 (1.40%)
Same-day stay involving two discharges at the same hospitals	2314.30 (0.70%)	17.4 (0.90%)
Same-day stay involving three or more discharges at the same or different hospitals	1106.50 (0.30%)	15.3 (0.70%)
Median household income national quartile for patient ZIP code
0–25th percentile	65,064.90 (20.70%)	372.6 (18.50%)	0.039
26th to 50th percentile	87,288.20 (27.80%)	549.3 (27.20%)
51st to 75th percentile	85,584.10 (27.30%)	512.9 (25.40%)
76th to 100th percentile	75,669.10 (24.10%)	582.1 (28.90%)
Control/ownership of hospital
Public	25,225.50 (7.90%)	198.8 (9.70%)	0.124
Private non-profit	262,429.40 (82.70%)	1695.80 (83.0%)
Private for profit	29,749.80 (9.40%)	149.1 (7.30%)
A combined record involving rehab transfer
Not a combined record or a combined record not involving rehabilitation, evaluation, or other aftercare	314,871 (99.20%)	2011.70 (98.40%)	0.017
Combined record involving transfer to rehabilitation, evaluation, or other aftercare	2533.6 (<1%)	32 (1.60%)
All patient refined diagnosis-related group: risk of mortality subclass
Minor likelihood of dying	36,806.10 (11.60%)	105.2 (5.10%)	<0.001
Moderate likelihood of dying	144,320.60 (45.50%)	588.2 (28.80%)
Major likelihood of dying	108,728.60 (34.30%)	967.8 (47.40%)
Extreme likelihood of dying	27,544.90 (8.70%)	382.4 (18.70%)
All patient refined diagnosis-related group: severity of illness subclass
Minor loss of function (includes cases with no comorbidity or complications)	34,115 (10.70%)	62.6 (3.10%)	<0.001
Moderate loss of function	77,171.20 (24.30%)	255.9 (12.50%)
Major loss of function	50,937.5 (16.0%)	577.9 (28.30%)
Extreme loss of function	155,176.40 (48.90%)	1147.4 (56.10%)
Comorbidities
Anemia	16,369.70 (5.2%)	132.10 (6.5%)	0.128
Diabetes mellitus	120,701.90 (38.%)	657.4 (32.2%)	0.001
Heart failure	229,229.10 (72.2%)	1620.10 (79.3%)	<0.001
Hyperlipidemia	232,502.90 (73.3%)	1423.10 (69.6%)	0.046
Hypertension	214,716.90 (67.6%)	1473.10 (72.1%)	0.018
Hypothyroid	59,188.40 (18.6%)	375.90 (18.4%)	0.891
Liver disease	4787.50 (1.5%)	29.8 (1.5%)	0.912
Obesity	65,920 (20.8%)	364.80 (17.9%)	0.064
Obstructive sleep apnea	46,637.70 (14.7%)	270.1 (13.2%)	0.252
Pneumonia	4430.40 (1.4%)	64.60 (3.2%)	<0.001
Prior coronary artery bypass graft	50,275.30 (15.8%)	444 (21.7%)	<0.001
Prior myocardial infarction	37,917.70 (11.9%)	258.60 (12.7%)	0.566
Prior percutaneous coronary intervention	61,625.30 (19.4%)	372.7 (18.2%)	0.541
Pulmonary disease	65,411.20 (20.6%)	410.60 (20.1%)	0.719
Pulmonary hypertension	36,927 (11.6%)	257(12.6%)	0.428
Right ventricular failure	1335 (0.4%)	33 (1.6%)	<0.001
Smoker	113,184.40 (35.7%)	706.60 (34.6%)	0.569

and Figure 3.

3.2. Statistical Analyses

3.2.1. Outcomes After Univariate and Multivariate Regression Analysis of TAVI with PVL

Patients with PVL had higher odds of in-hospital mortality (unadjusted odds ratio (uOR) = 2.52, 95% CI:1.6–3.97, p < 0.001), developing AKI (uOR = 2.54, 95% CI:2.0–3.2, p < 0.001), experiencing AHF (uOR = 1.50, 95% CI:1.28–1.7, p < 0.001), and requiring MCS (uOR = 3.81, 95% CI:2.16–6.73, p < 0.001). Additionally, PVL was associated with increased odds of SCA (uOR = 1.89, 95% CI:1.53–2.32, p < 0.001), post procedural bleeding (uOR = 1.79, 95% CI:1.20–2.66, p < 0.001), vascular complications (uOR = 1.81, 95% CI:1.24–2.64, p < 0.001), atrial fibrillation and flutter (uOR = 1.35, 95% CI:1.17–1.56, p < 0.001), and cardiogenic shock (CS) (uOR = 4.2, 95% CI:3.0–5.8, p < 0.001).

On multivariate analysis, several outcomes remained significantly associated with PVL. Notably, the odds of dying during hospitalization remained significantly higher in patients with PVL (adjusted odds ratio (aOR): 2.16, 95% CI:1.35–3.45, p < 0.001). Similarly, the risks of AKI (aOR:2.42, 95% CI:1.82–3.18, p < 0.001), CS (aOR 3.4, 95% CI:2.38–4.89, p < 0.001), and MCS (aOR: 2.94, 95% CI:1.58–5.47, p < 0.001) were significantly higher in patients with PVL. Furthermore, PVL was associated with higher odds of post procedural bleeding (aOR: 1.70, 95% CI:1.14–2.55, p < 0.001), cardiac tamponade (aOR: 2.95, 95% CI:1.22–7.09, p = 0.016), vascular complications (aOR: 1.76, 95% CI:1.20–2.57, p < 0.001), and atrial fibrillation/flutter (aOR: 1.36, 95% CI:1.18–1.57, p < 0.001). However, the association with AMI became non-significant in the multivariate analysis (aOR: 1.45, 95% CI:0.94–2.23, p = 0.089) (

Table 5. Univariate and multivariate regression analysis of TAVI with and without periprocedural prosthetic valve leak.

In-Hospital Outcomes	Univariate Regression Analysis			Multivariate Regression Analysis
	uOR	95% CI	p-Value	aOR	95% CI	p-Value
Acute congestive heart failure	1.50	1.28–1.7	<0.001	1.40	1.19–1.66	<0.001
Acute kidney injury	2.54	2.0–3.2	<0.001	2.42	1.82–3.18	<0.001
Acute myocardial infarction	1.74	1.17–2.59	<0.001	1.45	0.94–2.23	0.089
Atrial fibrillation and flutter	1.35	1.17–1.56	<0.001	1.36	1.18–1.57	<0.001
Cardiac tamponade	2.91	1.24–6.83	0.014	2.95	1.22–7.09	0.016
Cardiogenic shock	4.2	3.0–5.8	<0.001	3.4	2.38–4.89	<0.001
In-hospital mortality	2.52	1.6–3.97	<0.001	2.16	1.35–3.45	<0.001
Mechanical circulatory support	3.81	2.16–6.73	<0.001	2.94	1.58–5.47	<0.001
Post procedural bleeding	1.79	1.20–2.66	<0.001	1.70	1.14–2.55	<0.001
Vascular complications	1.81	1.24–2.64	<0.001	1.76	1.20–2.57	<0.001

Abbreviations: aOR: Adjusted odds ratio; CI: Confidence interval; uOR: Unadjusted odds ratio.

and Figure 4).

3.2.2. Outcomes of Unmatched and Propensity-Matched Cohorts of TAVI with PVL

The TAVI with PVL group demonstrated higher in-hospital mortality rates both in crude outcomes (unmatched group) (3.3% vs. 1.3%, p < 0.001) and after propensity matching (3.05% vs. 1.05%, p < 0.001). PVL presence was associated with more adverse events. Rates of AKI were substantially higher in the TAVI with PVL group, both in crude outcomes (21.6% vs. 9.8%, p < 0.01) and propensity-matched outcomes (20.4% vs. 11.7%, p < 0.01). Similarly, after propensity matching, rates of acute HF (49.6% vs. 41%, p < 0.01), SCA (15.3% vs. 8.5%, p < 0.01), CS (7.5% vs. 2.5%, p < 0.01), postprocedural bleeding (3.57% vs. 1.48%, p < 0.01), vascular complications (4.1% vs. 1.57%, p < 0.01), and atrial fibrillation/flutter (44.8% vs. 38.6%, p < 0.01) are markedly elevated in the TAVI with PVL cohort. Meanwhile, other outcomes, such as AMI and MCS, were insignificant in matched cohorts (

Table 6. In-hospital crude and propensity-matched outcomes comparison of transcatheter aortic valve implantation (TAVI) with and without periprocedural prosthetic valve leak (PVL).

Outcomes	Crude Outcomes			Propensity-Matched Outcomes
	TAVI Without PVL n = 317,405	TAVI with PVL n = 2043	p-Value	TAVI Without PVL n = 1147	TAVI with PVL n = 1147	p-Value
Acute congestive heart failure	129,153 (40.7%)	1036 (50.7%)	<0.001	472 (41.1%)	569 (49.6%)	<0.001
Acute kidney injury	31,027 (9.8%)	442 (21.6%)	<0.001	135 (11.7%)	234 (20.4%)	<0.001
Acute myocardial infarction	6732 (2.1%)	74 (3.6%)	0.005	35 (3.05%)	39 (3.4%)	0.636
Atrial fibrillation and flutter	122,011 (38.4%)	937 (45.9%)	<0.001	443 (38.6%)	514 (44.8%)	0.003
Cardiogenic shock	6561 (2.1%)	167 (8.2%)	<0.001	29 (2.5%)	86 (7.5%)	<0.001
In-hospital mortality	2823(1.3%)	66 (3.3%)	<0.001	12 (1.05%)	35 (3.05%)	<0.001
Mechanical circulatory support	2823 (0.89%)	67 (3.3%)	<0.001	20 (1.7%)	30 (2.6%)	0.153
Postprocedure bleeding	6246(2.0%)	71 (3.5%)	0.004	17 (1.48%)	41 (3.57%)	<0.001
Sudden cardiac arrest	27,651 (8.7%)	312 (15.3%)	<0.001	98 (8.5%)	176 (15.34%)	<0.001
Vascular complications	7800 (2.5%)	89 (4.4%)	0.002	18 (1.57%)	47 (4.1%)	<0.001

Abbreviations: TAVI: Transcatheter aortic valve implantation; PVL: Prosthetic valve leak.

and Figure 5).

3.2.3. Outcomes After Propensity Matching and Multivariate Regression of TAVI with Mechanical Complications

The cohort with mechanical complications (displacement or breakdown of the valve during the procedure) experienced a significantly higher in-hospital mortality rate of 4.60%, which was markedly elevated compared to the relatively lower rate of 1.34% observed in the complication-free group. The adjusted odds ratio (aOR) for mortality was 3.64, underscoring the critical impact these complications wield on patient survival. Similarly, AKI (aOR 2.58, 95% CI:2.15–3.08, 23% vs. 11.87%, p < 0.001), acute HF (aOR 2.25, 95% CI:1.94–2.63, 62% vs. 44%, p < 0.001), AMI (aOR 1.98, 95% CI:1.40–2.80, 4.7% vs. 2.5%, p < 0.001), MCS (aOR 3.35, 95% CI:2.39–4.71, 4% vs. 1.48%, p < 0.001), SCA (aOR 1.6, 95% CI:1.34–1.91, 13% vs. 8%, p < 0.001), postprocedural bleeding (aOR 1.67, 95% CI:1.12–2.5, 3% vs. 1.55, p < 0.008), vascular complications (aOR 2.03, 95% CI:1.5–2.74, 4.38% vs. 2.1%, p < 0.001), atrial fibrillation and flutter (aOR 1.34, 95% CI:1.17–1.54, 44% vs. 37.8%, p < 0.001), and CS (aOR 3.3, 95% CI:2.53–4.31, 8.8% vs. 3.6%, p < 0.001) showed higher adjusted odds on multivariate regression which persisted after propensity matching (

Table 7. Outcomes after propensity matching and multivariate regression analysis of transcatheter aortic valve implantation (TAVI) with and without mechanical complications.

Outcomes	Propensity-Matched Cohort			Adjusted Odds Ratios
	Without Mechanical Complications n = 1414	With Mechanical Complications n = 1414	p-Value	aOR	95% CI	p-Value
	n (%)	n (%)
Acute congestive heart failure	623 (44.03)	881 (62.26)	<0.001	2.25	1.94–2.63	<0.001
Acute kidney injury	168 (11.87)	326 (23.04)	<0.001	2.58	2.15–3.08	<0.001
Acute myocardial infarction	36 (2.54)	67 (4.73)	0.002	1.98	1.40–2.80	<0.001
Atrial fibrillation and flutter	536 (37.88)	628 (44.38)	<0.001	1.34	1.17–1.54	<0.001
Cardiogenic shock	51 (3.60)	125 (8.83)	<0.001	3.3	2.53–4.31	<0.001
In-hospital mortality	19 (1.34)	65 (4.60)	<0.001	3.64	2.57–5.16	<0.001
Mechanical circulatory support	21 (1.48)	57 (4.03)	<0.001	3.35	2.39–4.71	<0.001
Postprocedure bleeding	22 (1.55)	43 (3.04)	0.008	1.67	1.12–2.50	0.011
Sudden cardiac arrest	114 (8.06)	184 (13)	<0.001	1.60	1.34–1.91	<0.001
Vascular complications	30 (2.12)	62 (4.38)	0.001	2.03	1.50–2.74	<0.001

Abbreviations: aOR: Adjusted odds ratio; CI: Confidence interval; TAVI: Transcatheter aortic valve replacement; uOR: Unadjusted odds ratio.

).

3.2.4. Resource Utilization of TAVI Cohort with and Without PVL and Mechanical Complications

Patients with PVL and mechanical complications exhibited extended LOS, with a median LOS of 3 days (IQR: 6) compared to 2 days (IQR: 3) for those without PVL and a median LOS of 3 days (IQR: 7) compared to 2 days (IQR: 3) for those without mechanical complications. Furthermore, these groups experience higher adjusted total charges, with median charges of USD 219,953 (IQR: 169,018) for PVL patients and USD 204,511 (IQR: 166,063) for those with mechanical complications, in contrast to USD 176,354 (IQR: 129,426) and USD 176,392 (IQR: 129,328) for their respective counterparts. Similar patterns emerge in total costs, indicating the financial burden associated with complications (

Table 8. Resource utilization of TAVI cohorts with and without periprocedural prosthetic valve leak and mechanical complications.

Resource Utilization	Without PVL	With PVL	p-Value	Without Mechanical Complications	With Mechanical Complications	p-Value
	Median (IQR)	Median (IQR)		Median (IQR)	Median (IQR)
LOS in Days	2 (3)	3 (6)	<0.001	2 (3)	3 (7)	<0.001
Adjusted Total Charge	176,354 (129,426)	219,953 (169,018)	<0.001	176,392 (129,328)	204,511 (166,063)	<0.001
Total Cost	45,339 (23,547)	54,751 (35,037)	<0.001	45,341 (23,559)	50,986 (29,661)	<0.001

Abbreviations: IQR: Interquartile range.

). The yearly trend of adjusted cost has decreased for the TAVI cohort without PVL (p-trend < 0.05), while it had not significantly changed for TAVI patients with PVL over the study period (p-trend > 0.05). In contrast, the length of hospitalization had decreased for both cohorts (p-trend ≤ 0.05) from 2016 to 2020 (

Table 9. Yearly trend of adjusted total cost and length of hospital stay in TAVI cohorts with and without prosthetic valve leak.

Year	Total Cost Yearly Trend		Length of Stay Yearly Trend
	TAVI Without PVL	TAVI with PVL	TAVI Without PVL	TAVI with PVL
	Median	Median	Median	Median
2016	50,801 (65,675–39,179)	60,297 (77,712–48,221)	3 (6–2)	5 (10–3)
2017	47,131 (60,930–36,913)	55,562 (73,557–42,004)	2 (5–2)	3 (8–2)
2018	45,224 (58,590–35,538)	51,891 (74,054–37,348)	2 (4–1)	3 (7–2)
2019	43,533 (57,087–34,081)	51,525 (72,424–39,066)	2 (3–1)	3 (9–2)
2020	46,199 (61,312–35,942)	57,763 (80,647–42,591)	1 (3–1)	2 (7–1)
p-trend	<0.001	0.2802	<0.001	<0.001

).

3.2.5. Readmission Analysis on the Propensity-Matched TAVI Cohorts with and Without PVL

The TAVI cohort with PVL showed significantly higher readmission rates up to six months follow-up post discharge. The 30-day (15.2% vs. 12%, p = 0.02), 90-day (24.4% vs. 19.9%, p < 0.01), and 180-day (34.7% vs. 24.8%, p < 0.01) readmission rates were significantly higher in the presence of concomitant PVL (

Table 10. Readmission rates in propensity-matched cohorts.

Readmission Rates in Propensity-Matched Cohorts
30-day Readmissions	Without PVL	With PVL	p-value
	n = 1016	n = 1016
	n (%)	n (%)
Readmits	122 (12)	155 (15.2)	0.023
90-day Readmissions	Without PVL	With PVL	p-value
	n = 812	n = 812
	n (%)	n (%)
Readmits	162 (19.9)	198 (24.4)	<0.01
180-day Readmissions	Without PVL	With PVL	p-value
	n = 521	n = 521
	n (%)	n (%)
Readmits	129 (24.8)	181 (34.7)	<0.01

Abbreviation: PVL: Prosthetic valve leak.

).

Aortic stenosis contributed to the majority of these readmissions (>70%), followed by combined valvular pathology, stenosis of prosthetic devices, hypertensive heart, and CKD diseases at 30-, 90-, and 180-day follow-ups. The most common ICD diagnoses associated with readmission are presented in

Table 11. Most common ICD-10 causes of 30-day readmissions.

30-Day Readmission Causes
ICD-10 Code	ICD-10 Diagnosis	Count	Percentages (%)
I350	Nonrheumatic aortic (valve) stenosis	22,637.1	75.9
I352	Nonrheumatic aortic (valve) stenosis with insufficiency	1928.768	6.5
I080	Rheumatic disorders of both mitral and aortic valves	1479.659	5.0
I083	Combined rheumatic disorders of mitral, aortic and tricuspid valves	1041.29	3.5
T82857A	Stenosis of other cardiac prosthetic devices, implants and grafts, initial encounter	684.0428	2.3
Q231	Congenital insufficiency of aortic valve	249.5053	0.8
I130	Hypertensive heart and chronic kidney disease with heart failure and stage 1 through stage 4 chronic kidney disease, or unspecified chronic kidney disease	226.9943	0.8
I082	Rheumatic disorders of both aortic and tricuspid valves	162.404	0.5
I214	Non-ST elevation (NSTEMI) myocardial infarction	122.5465	0.4
I351	Nonrheumatic aortic (valve) insufficiency	120.8068	0.4
T82897A	Other specified complication of cardiac prosthetic devices, implants and grafts, initial encounter	100.2057	0.3
I110	Hypertensive heart disease with heart failure	91.08601	0.3
I060	Rheumatic aortic stenosis	62.9384	0.2
I132	Hypertensive heart and chronic kidney disease with heart failure and with stage 5 chronic kidney disease, or end stage renal disease	50.7093	0.2
T82228A	Other mechanical complication of biological heart valve graft, initial encounter	35.81099	0.1
I5043	Acute on chronic combined systolic (congestive) and diastolic (congestive) heart failure	34.5458	0.1
I5033	Acute on chronic diastolic (congestive) heart failure	32.08654	0.1
T8209XA	Other mechanical complication of heart valve prosthesis, initial encounter	30.77705	0.1
T8203XA	Leakage of heart valve prosthesis, initial encounter	29.33904	0.1
A419	Sepsis, unspecified organism	23.98208	0.1

,

Table 12. Most common ICD-10 causes of 90-day readmissions.

90-Day Readmission Causes
ICD-10 Code	ICD-10 Diagnosis	Count	Percentage (%)
I350	Nonrheumatic aortic (valve) stenosis	34,014.22	73.9
I352	Nonrheumatic aortic (valve) stenosis with insufficiency	3044.265	6.6
I080	Rheumatic disorders of both mitral and aortic valves	2297.723	5.0
I083	Combined rheumatic disorders of mitral, aortic and tricuspid valves	1570.149	3.4
T82857A	Stenosis of other cardiac prosthetic devices, implants and grafts, initial encounter	1249.885	2.7
I130	Hypertensive heart and chronic kidney disease with heart failure and stage 1 through stage 4 chronic kidney disease, or unspecified chronic kidney disease	466.6844	1.0
Q231	Congenital insufficiency of aortic valve	359.0492	0.8
I082	Rheumatic disorders of both aortic and tricuspid valves	284.2048	0.6
I214	Non-ST elevation (NSTEMI) myocardial infarction	228.4911	0.5
T82897A	Other specified complication of cardiac prosthetic devices, implants and grafts, initial encounter	199.5951	0.4
I351	Nonrheumatic aortic (valve) insufficiency	197.9469	0.4
I110	Hypertensive heart disease with heart failure	159.5987	0.3
I060	Rheumatic aortic stenosis	111.0938	0.2
I132	Hypertensive heart and chronic kidney disease with heart failure and with stage 5 chronic kidney disease, or end stage renal disease	95.46474	0.2
I5043	Acute on chronic combined systolic (congestive) and diastolic (congestive) heart failure	73.09896	0.2
T8209XA	Other mechanical complication of heart valve prosthesis, initial encounter	72.47526	0.2
T82228A	Other mechanical complication of biological heart valve graft, initial encounter	61.68306	0.1
A419	Sepsis, unspecified organism	60.80909	0.1
I5023	Acute on chronic systolic (congestive) heart failure	58.33688	0.1
I5033	Acute on chronic diastolic (congestive) heart failure	55.18985	0.1

and

Table 13. Most common ICD-10 causes of 180-day readmissions.

180-Day Readmission Causes
ICD-10 Code	ICD-10 Diagnosis	Count	Percentage (%)
I350	Nonrheumatic aortic (valve) stenosis	30,957.7	74.0
I352	Nonrheumatic aortic (valve) stenosis with insufficiency	2810.865	6.7
I080	Rheumatic disorders of both mitral and aortic valves	2003.486	4.8
I083	Combined rheumatic disorders of mitral, aortic and tricuspid valves	1408.717	3.4
T82857A	Stenosis of other cardiac prosthetic devices, implants and grafts, initial encounter	1169.29	2.8
I130	Hypertensive heart and chronic kidney disease with heart failure and stage 1 through stage 4 chronic kidney disease, or unspecified chronic kidney disease	404.2754	1.0
Q231	Congenital insufficiency of aortic valve	309.6377	0.7
I082	Rheumatic disorders of both aortic and tricuspid valves	265.4324	0.6
I214	Non-ST elevation (NSTEMI) myocardial infarction	225.5222	0.5
I351	Nonrheumatic aortic (valve) insufficiency	198.491	0.5
T82897A	Other specified complication of cardiac prosthetic devices, implants and grafts, initial encounter	183.0957	0.4
I110	Hypertensive heart disease with heart failure	150.0879	0.4
I132	Hypertensive heart and chronic kidney disease with heart failure and with stage 5 chronic kidney disease, or end stage renal disease	95.45225	0.2
I060	Rheumatic aortic stenosis	86.65321	0.2
T8209XA	Other mechanical complication of heart valve prosthesis, initial encounter	68.50824	0.2
I5043	Acute on chronic combined systolic (congestive) and diastolic (congestive) heart failure	65.36945	0.2
T82228A	Other mechanical complication of biological heart valve graft, initial encounter	60.17539	0.1
I5033	Acute on chronic diastolic (congestive) heart failure	59.84858	0.1
A419	Sepsis, unspecified organism	52.57824	0.1
I5023	Acute on chronic systolic (congestive) heart failure	50.53654	0.1

and Figure 6. The median total adjusted cost of hospitalization and LOS showed a decreasing trend over the years for the TAVI cohort without PVL for 30- and 90-day readmissions (p-trend < 0.05) (Figure 7). A significant decreasing trend for total adjusted cost was observed on 180-day readmits, while the trend for LOS was not significant in these patients (p-trend > 0.05). In contrast, the yearly trend of total adjusted cost and LOS for 30-, 90-, and 180-day readmissions in the TAVI cohort with PVL was insignificant (p-trend > 0.05) except for a decreasing trend for LOS on 180-day readmissions (p-trend < 0.05) (

Table 14. Yearly trend of total cost and length of stay of 30-day readmissions.

Year	Inflation-Adjusted Total Cost Yearly Trend		Length of Stay Yearly Trend
	TAVI Without PVL	TAVI with PVL	TAVI Without PVL	TAVI with PVL
	Median (IQR)	Median (IQR)	Median (IQR)	Median (IQR)
2016	53,401 (67,520–41,147)	49,921 (60,191–41,112)	4 (7–2)	4 (5–3)
2017	47,748 (61,654–37,354)	55,562 (65,246–44,192)	3 (6–2)	4 (10–2)
2018	46,527 (61,220–36,564)	50,024 (74,784–37,255)	3 (5–2)	5 (9–2)
2019	45,238 (58,813–35,143)	47,385 (78,422–36,637)	2 (5–1)	3 (6–2)
2020	48,440 (63,441–36,669)	65,430 (83,920–41,659)	2 (4–1)	3.5 (6.5–1.5)
p-trend	<0.001	0.7116	<0.001	0.3014

,

Table 15. Yearly trend of total cost and length of stay of 90-day readmissions.

Year	Inflation-Adjusted Total Cost Yearly Trend		Length of Stay Yearly Trend
	TAVI Without PVL	TAVI with PVL	TAVI Without PVL	TAVI with PVL
	Median (IQR)	Median (IQR)	Median (IQR)	Median (IQR)
2016	54,167 (70,107–41,343)	60,191 (103,561–39,737)	4 (8–2)	4 (18–3)
2017	49,084 (65,183–37,838)	55,562 (67,671–44,192)	3 (7–2)	4 (10–2)
2018	47,800 (63,195–36,930)	60,550 (80,142–38,036)	3 (7–2)	6 (15–2)
2019	45,856 (60,878–35,179)	51,897 (93,976–39,799)	2 (6–1)	4 (12–2)
2020	49,239 (65,833–37,084)	63,054 (87,055–45,033	2 (5–1)	3 (9–2)
p-trend	<0.001	0.8631	<0.001	0.0997

and

Table 16. Yearly trend of total cost and length of stay of 180-day readmissions.

Year	Inflation-Adjusted Total Cost Yearly Trend		Length of Stay Yearly Trend
	TAVI Without PVL	TAVI with PVL	TAVI Without PVL	TAVI with PVL
	Median (IQR)	Median (IQR)	Median (IQR)	Median (IQR)
2016	53,806 (69,992–41,117)	67,034 (102,446–47,702)	4 (8–2)	4.5 (18–3)
2017	48,770 (64,507–37,141)	59,523 (75,619–44,420)	3 (7–2)	4 (11–2)
2018	47,665 (62,790–36,927)	62,097 (94,344–41,768)	3 (6–2)	6 (15–3)
2019	45,384 (60,386–35,006)	50,899 (86,077–38,923)	2 (6–1)	4 (12–2)
2020	48,717 (64,272–37,101)	55,352 (83,162–37,752)	2 (5–1)	3 (9–1)
p-trend	<0.001	0.1581	0.0554	<0.001

and Figure 7 and Figure 8).

4. Discussion

In this retrospective cohort analysis of NRD (2016–2020) studying the effect of PVL in TAVI patients, the key findings are (1) more adverse in-hospital outcomes, including mortality, acute congestive heart failure, acute kidney injury, acute myocardial infarction, cardiogenic shock, postprocedural bleeding, sudden cardiac arrest, and vascular complications, among TAVI patients with periprocedural PVL than without PVL; (2) significantly higher readmission rates among patients with periprocedural PVL at 30, 90, and 180-day follow-ups; (3) longer lengths of stay and greater financial burden associated with PVL than those without PVL; (4) a decreasing trend in lengths of stay with no significant change in adjusted-cost trends for TAVI patients with PVL over the study period.

A previous observational study based on the Finnish database demonstrated the association of PVL post-TAVI with impaired mid-term (4-year) survival. A 15–21% decrease in 4-year survival rates with different grades of PVL was noted [19]. Similar results have been seen in prior studies. In the prespecified analysis of PVR in the placement of aortic transcatheter valves (PARTNER) trial, an association of increased 1-year mortality was noted in patients with PVL. Still, no difference in 30-day mortality was seen [20]. Similarly, a meta-analysis of 25 studies showed increased all-cause and cardiovascular mortality in patients with PVL after TAVI [21]. Another study using a prospective, non-randomized investigation model demonstrated an association between severe PVL and mortality [22]. Although these studies homogenously indicate the poor prognostic impact of PVL, the results are conflicting regarding PVL grades. Some studies observed poor outcomes with even mild PVL [20,21], while others resulted in such an association with only moderate–severe PVL [3,22,23,24]. Our findings reinforce the previously reported association of PVL with higher mortality, especially immediate leaks identified during the index hospitalization. Although discussion on the outcomes of different grades of PVL is beyond the scope of this study, future investigations will utilize a similar retrospective cohort design.

The higher acute kidney injury events are likely multifactorial but could be secondary to contrast use in aortic root angiographic assessment of PVL. Even though echocardiography is frequently used in PVL evaluation, there needs to be evidence for PVL quantification after TAVI as the parameters are recommended for surgical prosthetic valves and not validated for transcatheter heart valves [25]. On the other hand, aortic root angiography is convenient and easily available during the TAVI procedure. It can be performed to offer essential information and guide interventional measures to reduce PVL [26]. Bleeding is a common post-TAVI complication and is usually access-related [27]. The subsequent burden of another catheter-based PVL closure might add to these patients’ access-related vascular and bleeding complications [27].

Symptomatic PVL presents as heart failure in ~90% of cases and hemolytic anemia from shear stress on the red blood cells in 1/3rd–1/4th of patients [28,29]. The higher risk of cardiac complications (atrial fibrillation, atrial flutter, acute myocardial infarction, cardiogenic shock, and congestive heart failure) seen in this study could be secondary to volume overload due to PVL exacerbating the underlying cardiac disease [4]. However, most previous studies focus on mortality outcomes, and the effects on other individual outcomes still need to be determined [4,30,31,32]. The results of this study emphasize the harmful effects of periprocedural PVL, which warrant validation with more studies.

Some of the independent risk factors associated with periprocedural PVL in TAVI patients include higher aortic valve maximum gradient, aortic stenosis, aortic regurgitation, systolic pulmonary artery pressure > 55mmHg, predilation of the native valve, and self-expanding valve prosthesis. Our current study emphasizes the importance of identifying patients at high risk of PVL, such as those with severe valvular calcification or bicuspid aortic valve with calcified raphe [7,9,33]. Patients with decreased risk of PVL may benefit from surgical aortic valve replacement rather than TAVI [19]. In our study, the TAVI cohort with periprocedural PVL had higher right ventricular failure at baseline, but there was no significant difference in pulmonary hypertension or other pulmonary diseases. However, drawing a comparison regarding the rest of the baseline risk factors is beyond the scope of this study. The newer generation transcatheter prostheses [34] (Sapien 3, Evolut R and Pro, Lotus, and Acurate Neo) have a significantly decreased PVL risk compared to older prostheses [35] (Sapien, Sapien XT, Corevalve) [19].

The effect of periprocedural PVL on hospital readmissions remains uncertain. An analysis of randomized cohorts and continued access registries of the multicenter PARTNER trial showed a significant increase in 1-year rehospitalizations [20]. Similarly, a 2.7-fold increase in 1-year rehospitalization risk was seen in the PARTNER II trial [24]. In a French TAVI registry-based analysis of patients between 2010 and 2019, PVL was associated with higher rehospitalization rates for heart failure [29]. A recent meta-analysis to analyze the outcomes of TAVI patients with PVL observed a higher risk of rehospitalization [hazard ratio (HR) 1.81, 95% CI 1.54–2.12, p < 0.001] with PVL and a peak of rehospitalizations early in the postoperative period. This correlated with the inability of left ventricle compensation to volume overload initially. HR stabilization was noted in later periods (after three years) with eventual convergence to a neutral risk (HR = 1), likely due to the development of left ventricle compensation [4].

We noticed higher rates of 30-, 90-, and 180-day readmissions in patients with PVL. The higher rates could be explained by more adverse events necessitating additional healthcare or readmissions in patients with significant PVL at discharge. This uncertainty is beyond the scope of this article due to the nature of the NRD database, which lacks information on the course of hospitalization and patient condition at discharge.

The longer lengths of stay observed in this study align with more adverse events requiring additional treatments or clinical monitoring. Further, the treatment of PVL could add to the complexities of hospital stays. These factors could easily explain the longer lengths of stay and economic burden seen in the PVL cohort compared to those without PVL in this study. Over the study period, the decreasing trend in lengths of stay with a steady adjusted cost could also reflect the advancements in treatment and diagnostic approaches, yet a higher economic burden. Hypothetically, there could be a contribution of the COVID-19 pandemic, warranting earlier discharges due to higher infection risk during hospitalization and overall increased hospital burden requiring a faster turnover.

The subgroup analysis results of this study using mechanical complications, including valvular dehiscence and valvular displacement as stratification tools, strengthen the study’s primary results. A similar increase in adverse outcomes, lengths of stay, and financial burden was seen in TAVI patients with mechanical complications post-TAVI during the index hospitalization compared to those without complications.

Limitations

NRD’s information is based on physician-entered billing diagnoses and is subject to errors of under- or over-coding. The cross-sectional data preclude following the hospital course of the cohorts and studying the interventions and mechanical complications because they lack laboratory and echocardiographic parameters. Additionally, the current ICD-10 coding lacks specific codes for the severity of PVL. Therefore, our findings, which are limited to the index hospitalization, do not fully reflect the spectrum of PVL ranging from mild to severe seen in the general population of individuals undergoing TAVI. The type of prosthesis used during the TAVI procedures is also unavailable. Like any other retrospective study, there is an inherent tendency of selection bias. Residual PVL or residual impact of mechanical complications data at discharge were unavailable, precluding our ability to comment on the long-term impact of these periprocedural complications.

However, despite these limitations, this study provides valuable insight into the nationwide population. The analysis of a large database is an invaluable tool for the primary goal of our study to understand real-world practice and generate hypotheses for evolving minimally invasive valve therapy. Additionally, the NRD has been used widely and validated despite possibly including coding errors and documentation disparities.

5. Conclusions

This study illustrates the significant deleterious effects of periprocedural prosthetic valvular leaks on patients undergoing TAVI. Along with a higher risk of mortality, bleeding, renal failure, cardiac complications, and readmission, there is also an increased burden on healthcare costs and infrastructure. Even though the overall length of stay decreased for TAVI patients with PVL in recent years, the economic burden remains unchanged. However, the current retrospective study is limited by lack of specific data to facilitate stratification of PVL severity. Future studies will provide specific data about the severity of PVL, ranging from mild to severe, occurring in patients undergoing TAVI during and after index hospitalization. This calls for future prospective studies to investigate TAVI devices designed to minimize the occurrence of PVL and utilize emerging interventional technologies to lower PVL rates after TAVI.