Relationship Between Hospital Surgical Aortic Valve Replacement Volume and Transcatheter Aortic Valve Replacement Outcomes

Structured Abstract:

Objectives:

The authors sought to examine whether hospital surgical aortic valve replacement (SAVR) volume was associated with corresponding TAVR outcomes.

Background:

Recent studies have demonstrated a volume-outcome relationship for transcatheter aortic valve replacement (TAVR).

Methods:

We analyzed 208400 fee-for-service Medicare beneficiaries for all aortic valve replacement (AVR) procedures from 2012 – 2015. Claims for patients <65yo, concomitant CABG, other heart valve, or other major open-heart procedures were excluded, as were secondary admissions for AVR. Hospital SAVR volumes were stratified based on mean annual SAVR procedures during the study period. Our primary outcomes were 30-day and 1-year postoperative TAVR survival. Adjusted survival following TAVR was assessed by multivariable Cox regression.

Results:

A total of 65757 SAVR and 42967 TAVR admissions were evaluated. 21.7%(9324) of TAVR procedures were performed at hospitals with <100 (Group1), 35.6%(15298) at centers with 100–199 (Group2), 22.9% (9828) at centers with 200–299 (Group3), and 8517(19.8%) at hospitals performing ≥300 SAVR cases/year (Group4). Compared to Group4, 30-day TAVR mortality risk-adjusted odds ratio for Group1 was 1.32 (95%CI:1.18–1.47), 1.25 (95%CI:1.12–1.39) for Group2, and 1.08 (95%CI:0.82–1.25) for Group3. These adjusted survival differences in TAVR outcomes persisted at 1-year post-procedure.

Conclusions :

Total hospital SAVR volume appears to be correlated with TAVR outcomes, with higher 30-day and 1-year mortality observed at low volume centers. This data supports the importance of viable surgical program within the heart team, and the use of minimum SAVR hospital thresholds may be considered as an additional metric for TAVR performance.

Condensed Abstract:

The association between hospital surgical aortic valve replacement (SAVR) volume and corresponding transcatheter aortic valve replacement (TAVR) outcomes remains unknown. Medicare beneficiaries undergoing TAVR procedures from 2012 – 2015 were analyzed. 30-day and 1-year TAVR survival were assessed according to hospital mean annual SAVR volumes. Total hospital SAVR volume appears to be correlated with TAVR outcomes, with higher 30-day and 1-year mortality observed at low volume centers. This data supports the importance of viable surgical program within the heart team, and the use of minimum SAVR hospital thresholds may be considered as an additional metric for TAVR performance.

Introduction:

Transcatheter aortic valve replacement (TAVR) has emerged as an established treatment strategy for patients with symptomatic aortic stenosis who are either inoperable, high-risk or intermediate-risk.(1–5) This has brought practice change which is now reflected in contemporary studies showing an overall reduction in isolated surgical aortic valve replacement (SAVR) volumes and a decrease in comorbidities among SAVR patients.(6–8) The promising results of two recent randomized clinical trials in low-risk patients are monumental(9,10), and could inevitably shift the pendulum even further towards TAVR as the preferred choice over SAVR in all patients, despite an overall increase in total aortic valve replacement (AVR) volumes at TAVR sites.(11)

Concomitantly, we have witnessed an increasing confidence in TAVR utilization as accumulating operator experience and volume, innovations in valve design and technology, and improvements in patients selection have led to significant reductions in morbidity and mortality after TAVR.(1,12–16) These findings have further fueled the ongoing debate on minimum procedure volume requirements for TAVR operators and programs, and spurred the Centers for Medicare and Medicaid services (CMS) to re-examine its national coverage determination (NCD) for TAVR, which were previously developed in 2012 in an effort to support rational dispersion of TAVR technology in the United States.(17) This was a major discussion point in the most recent Medicare Evidence Development and Coverage Advisory Committee meeting, while evidence on the impact of percutaneous coronary intervention, SAVR, and TAVR volumes on TAVR outcomes was further investigated. As part of their efforts, the professional societies recently updated their consensus document to emphasize the importance of maintaining some volume requirements for TAVR while also shifting the focus to more direct measures of quality of care.(18) However, the relationship between hospital SAVR volume on TAVR outcomes, which is very germane in the current era of TAVR, remains unclear. In this study, we sought to determine whether hospital volume of SAVR was associated with corresponding TAVR outcomes. We hypothesized that increasing SAVR volume would predict better TAVR outcomes.

Patient and Methods:

Study population

We examined inpatient records using the Medicare Provider Analysis Review and Master Beneficiary Summary File data of all Medicare fee-for-service beneficiaries who underwent AVR procedures between January 1, 2012 through December 31, 2015. We utilized relevant International Classification of Disease Clinical Modification codes (ICD-9-CM and ICD-10-CM) codes for SAVR and TAVR to query the CMS inpatient claims file (Table S1). Records where these procedures were associated with diagnosis-related groups other than 216 – 221 (ICD-9-CM) or 306, 307 (ICD-10-CM) were sequestered from this study. The master file included a total of 250877 unique claims. To facilitate longitudinal follow-up, subjects who had more than 1 month of health maintenance organization (HMO) coverage were excluded resulting in 208400 unique claims of beneficiaries not enrolled in an HMO at any point during the study years. Concomitant CABG, other heart valve or other major open-heart procedures were excluded, as were secondary/subsequent admissions for AVR (Figure S1). This study was approved by our institutional review board and informed consent was waived.

Data collection and definitions

Preoperative comorbidities and chronic conditions were derived from the chronic conditions file, using coded diagnoses and procedures in the year prior to surgery. These covariates were chosen based on their comparability to the Society of Thoracic Surgeon’s risk factors(19), inclusion in the Charlson score(20), potential for confounding with our outcomes of interest, and clinical judgement (Table S1). Major Bleeding was defined as post-procedural bleeding, or the presence of any of the following not listed as present on admission: hemorrhage, hematemesis, gastrointestinal bleeding, acute post-hemorrhage anemia, adverse anti-coagulation reaction (excluding heparin-induced thrombocytopenia) or hemorrhage due to anticoagulation, epistaxis, cardiovascular bleeding or pulmonary bleeding/hemoptysis.

Outcomes of interest

Our primary outcomes of interest included 30-day and 1-year survival following TAVR. Survival was calculated in days from the procedure date until death date, or through December 31^st, 2017, if alive. Death of death was provided from national death index. Secondary outcomes were obtained from the CMS inpatient file, and included acute kidney injury, permanent stroke, permanent pacemaker implantation, hospital and intensive care unit (ICU) length of stay (LOS), discharge to skilled nursing facility, and early reoperation rates, defined as TAVR redo within 30 days of the index procedure date.

Statistical analyses

For comparative analysis, hospital SAVR experience was stratified into four groups (Groups 1, 2, 3, 4) based on the overall distribution and according to the annual mean SAVR procedures over all 4 years, after ensuring that hospital decile ranks for SAVR cases had remained stable over the study period. (Figure S2). To ensure reliable estimates, data from centers that did not perform at least 10 annual SAVR procedures and at least 1 annual TAVR procedure were sequestered. Continuous variables were tested for distribution and compared using ANOVA for normally distributed variables, or Kruskal-Wallis one-way ANOVA if non-normally distributed, and presented as mean +/− standard deviation or median with inter-quartile range, as appropriate. Binary variables are presented as number and percentage, and were compared using Chi-square tests with Bonferroni correction for multiple comparisons. Temporal trends in total AVR, and annual case volumes of TAVR within each SAVR group were assessed using Cochrane-Armitage trend test.

Logistic regression analysis were conducted to evaluate risk-adjusted 30-day and 1-year mortality rates. 30-day mortality was also evaluated by generalized linear mixed models, using institution as the source of random effects, to account for differences in baseline characteristics and clustering of patients within hospitals. Longitudinal survival was assessed by forward-entry Cox proportional hazard model. Because covariates were identified in part by examining the previous year’s claims, the Cox model excluded records from 2012. Variables evaluated for all logistic regression or Cox models include all those presented in Tables 1 and S1. An interaction term for surgery year and volume quintile was entered in all models to control for confounding due to institutional experience changes over time.

Table 1.

Characteristics and comorbidities of isolated TAVR patients according to SAVR volume groups.

Number of TAVR cases	Group 1 (10–99 SAVR/yr) n=9324	Group 2 (100–199 SAVR/yr) n=15298	Group 3 (200–299 SAVR/yr) n=9828	Group 4 (>/=300 SAVR/yr) n=8517	P-value
Age, years (mean±SD)	82.8 (7.4)*	82.8 (7.4)	82.7 (7.5)	83.3 (7.5)	0.001
>=85 years, (%)	4409 (47.3)*	7319 (47.8)*	4607 (46.9)*	4342 (51.0)	0.001
White race, (%)	8730 (93.6)	14363 (93.9)	9297 (94.6)	8039 (94.4)	0.001
Female, (%)	4565 (49.0)*	7214 (47.2)	4670 (47.5)	4042 (47.5)	0.025
Dyslipidemia, (%)	8045 (86.3)*	13390 (87.5)*	8550 (87.0)*	7590 (89.1)	0.001
Hypertension, (%)	8986 (96.4)	14790 (96.7)	9475 (96.4)	8243 (96.8)	0.327
Diabetes mellitus, (%)	4164 (44.7)*	6992 (45.7)	4455 (45.3)	3951 (46.4)	0.051
Peripheral vascular disease, (%)	88 (0.9)	124 (0.8)	77 (0.8)	69 (0.8)	0.327
Anemia, (%)	6545 (70.2)*	10856 (71)*	6992 (71.1)*	6339 (74.4)	0.001
COPD, (%)	3472 (37.2)*	5724 (37.4)*	3556 (36.2)*	3047 (35.8)	0.009
Chronic kidney disease without dialysis, (%)	4748 (50.9)	7908 (51.7)	5067 (51.6)	4367 (51.3)	0.734
Coronary artery disease, (%)	20 (0.2)	30 (0.2)	22 (0.2)	15 (0.2)	0.716
Atrial fibrillation, (%)	3397 (36.4)*	5796 (37.9)*	3841 (39.1)	3310 (38.9)	0.001
Ischemic heart disease, (%)	8832 (94.7)	14667 (95.9)	9307 (94.7)	8211 (96.4)	0.001
Previous myocardial infarction, (%)	544 (5.8)	963 (6.3)	611 (6.2)*	477 (5.6)	0.115
Congestive heart failure, (%)	7700 (82.6)*	13123 (85.8)*	8571 (87.2)*	7553 (88.7)	0.001
Liver disease, (%)	153 (1.6)	265 (1.7)	172 (1.8)	169 (2.0)	0.001
Home O2, (%)	44 (0.5)	75 (0.5)	45 (0.5)	27 (0.3)	0.001
Charlson score (Median, IQR)	6 (5–7)	6 (6–7)	6 (6–7)	6 (6–7)	0.001
History of Alzheimer’s disease, (%)	296 (3.2)	496 (3.2)	335 (3.4)	288 (3.4)	0.764
History of depression, (%)	1980 (21.2)*	3048 (19.9)*	2052 (20.9)	1658 (19.5)	0.039
Prior PCI, (%)	1717 (18.4)	3140 (20.5)	1972 (20.1)	1754 (20.6)	0.001
Prior CABG, (%)	1665 (17.9)	2974 (19.4)	1938 (19.7)	1732 (20.3)	0.001

CABG – coronary artery bypass graft; COPD – chronic obstructive pulmonary disease; PCI – percutaneous coronary intervention; SD – standard deviation.

^*Indicates statistical significance based on Wald chi-square test when compared pair-wise to the reference group of very-high SAVR volume center (>/=300 cases/yr

To assess the association between SAVR volumes and TAVR volumes, we first examined the correlation and partial correlation, controlling for institution. We found that both estimates were nearly identical (Pearson R = 0.726; by partial correlation = 0.725) suggesting that the correlation between the two variables (SAVR volume and TAVR volume) was nearly completely independent of the individual institution itself. Furthermore, given such a powerful predictive correlation between SAVR volumes and TAVR volumes, and the fact that up to 65% of the variability in TAVR volume was predicted by the SAVR volume, we elected not to adjust for TAVR volumes in our regression models because in doing so, we would violate the independence of observations assumption, among other flaws.

We also performed sensitivity analysis by including HMO patients, transfemoral approach TAVR patients only, and TAVR era before and after 2014 (to reflect Food and Drug Administration approval of TAVR in high-risk and intermediate risk patients). P-valves are presented, with a two-sided P-value <0.05 as the criterion of significance. All analyses were conducted using SPSS version 23.0 (IBM corporation, Armonk, NY) or R version 3.4.1 (R Foundation, Vienna Austria).

Results:

Patient characteristics

A total of 65,757 SAVR and 42,967 TAVR admissions were evaluated from 1,208 hospitals (Figure S2 and S3). Baseline characteristics for the four SAVR groups are listed in Table S1. Group1 (10–99 SAVR cases/year) comprised of 9234 TAVR patients (21.7%), Group2 (100–199 SAVR cases/year) consisted of 15298 TAVR patients (35.6%), Group3 (200–299 cases/year) consisted of 9828 TAVR patients (22.9%), and Group4 (>/=300 cases/year) consisted of 8517 TAVR patients (19.8%). The majority of the differences between the various groups were nominally statistically significant. In general, Group3 and Group4 patients exhibited worse baseline characteristics than Group1 and Group2 patients. The greatest baseline differences between Group4 and Group1 were in the proportion of patients who were 85 years or older (51% vs 47.3%), with atrial fibrillation (38.9% vs 36.4%), ischemic heart disease (96.4% vs 94.7%) and congestive heart failure (88.7% vs 82.6%, respectively; all P<0.05). Baseline characteristics by surgery year for each of the SAVR groups are summarized in Table S2–S5.

Postoperative outcomes

Transfemoral approach was utilized more frequently in Group4 (89.3%) versus Group1 patients (86.9%; Table S6). In Group4, the incidence of permanent pacemaker implantation (8.1% vs 5.7%), major bleeding (17.7% vs 16.4%) and early reoperation rates (0.3% vs 0.1%, respectively) were significantly higher compared to Group1 (all P<0.05). While there was no clinical difference in median ICU and hospital LOS between the groups, a higher proportion of patients in Group4 were discharged to a skill nursing facility compared to Group1 (27.4% vs 26.9%; P=0.001). In general, unadjusted 30-day mortality after TAVR was significantly higher in Group1 versus Group3 and Group4. The unadjusted 30-day TAVR mortality rates in Groups 1, 2, 3, 4 were 5.2%, 4.7%, 4.0%, and 3.7%, respectively (P=0.001; Figure 1).

Figure 1: Comparison of Postoperative Mortality and Readmissions after TAVR according to different SAVR volume groups.

Group1 includes <100 SAVR cases/year, Group2 = 100–199 SAVR cases/year, Group3 = 200–299 SAVR cases/year, Group4 = 300 or more SAVR cases/year. Unadjusted 30-day mortality and 1-year mortality was significantly higher in Group1 versus Group3 and Group4. There was no statistical difference in unadjusted 30-day readmissions between the groups.

Association between SAVR volume and TAVR outcomes

We observed an association between SAVR volume and TAVR outcomes at 30-days and 1-year. Risk-adjusted 30-day, 1-year mortality and overall all-cause mortality are shown in Table 2. Compared to Group4, 30-day risk-adjusted odds ratio for Group1 was 1.32 (95%CI:1.18–1.47), 1.25 (95%CI:1.12–1.39) for Group2, and 1.08 (95%CI:0.82–1.25) for Group3 When controlling for age, race, gender, the adjusted survival differences in TAVR outcomes persisted at 1-year post-procedure. (Table S7–S9). All relevant statistical interaction terms were tested within the model, and none proved statistically significant. In our multivariable Cox proportional hazards model, Group1 and Group2 had a 0.07 and 0.06-fold higher risk of all-cause mortality compared to Group4. (Figure 2)

Table 2:

SAVR volume Inter-group comparisons of risk-adjusted estimates of mortality in isolated TAVR patients

	ADJUSTED OUTCOME
Outcome variable	Effect Estimate	[95% CI]	P-value
All-cause mortality
SAVR Volume (Ref:>/=300 cases/yr or Group 4]	HR 1.000	-	0.001
Group 1 vs Group 4	HR 1.071	1.031–1.112	0.001
Group 2 vs Group 4	HR 1.057	1.007–1.109	0.026
Group 3 vs Group 4	HR 1.051	1.015–1.089	0.006
30-day mortality
SAVR Volume (Ref:>/=300 cases/yr or Group 4]	OR 1.000	-	0.001
Group 1 vs Group 4	OR 1.317	1.182–1.467	0.001
Group 2 vs Group 4	OR 1.246	1.119–1.387	0.001
Group 3 vs Group 4	OR 1.075	0.923–1.252	0.353
1-year mortality
SAVR Volume (Ref:>/=300 cases/yr or Group 4]	OR 1.000	-	0.001
Group 1 vs Group 4	OR 1.145	1.078–1.216	0.001
Group 2 vs Group 4	OR 1.125	1.062–1.191	0.001
Group 3 vs Group 4	OR 1.078	0.997–1.166	0.353

-OR – odds ratio; HR – hazard ratio

-Adjusted all-cause mortality based on multivariable cox regression model.

-Adjusted 30-day and 1-year mortality based on multivariable logistic regression models. Details of each model are summarized in Data supplement Tables S8–S10.

-Group1 includes <100 SAVR cases/year, Group2 = 100–199 SAVR cases/year, Group3 = 200–299 SAVR cases/year, Group4 = 300 or more SAVR cases/year.

Figure 2: Adjusted Cox survival curves for isolated TAVR patients, stratified by hospital SAVR volume.

Group1 includes <100 SAVR cases/year, Group2 = 100–199 SAVR cases/year, Group3 = 200–299 SAVR cases/year, Group4 = 300 or more SAVR cases/year. Group1 and Group2 are associated with significantly higher risk of all-cause mortality compared to Group4 (the reference group).

Temporal trends in volume and outcomes

Annual TAVR volume increased from 6427 cases in 2012 to 23209 cases in 2015, with a 261% growth (Figure S4). The ratio of annual TAVR/SAVR volume increased during the study period. There was also temporal variability in annual TAVR volumes in the different SAVR groups. (Figure S5) Likewise, we observed temporal decreases in annual unadjusted 30-day TAVR mortality in all SAVR groups (Figure S6)

Subgroup and Sensitivity Analysis

The subgroup analysis in excluded HMO patients showed results that were consistent with our main findings, although the effect estimates were attenuated (Table S10). We further stratified Group1 into <50 SAVR cases/year (1146 TAVR patients; 3%) and 50–99 SAVR cases/year (7569 TAVR patients; 20%). Both these groups had similar baseline characteristics, unadjusted 30-day mortality (7.7% vs 7.8%), and our overall outcomes did not change (Table S11, Figure S6). In our sensitivity analysis of transfemoral TAVR patients only, our results remained robust. (Figure S7, Table S12 and S13). Similar findings were observed when looking at surgical era in our time-dependent analysis (Figure S8). Finally, while the random effects from the institutions was small in our generalized liner mixed model, the fixed effects of SAVR volume on TAVR outcomes persisted (Table S14).

Discussion:

Current criteria for accreditation of TAVR centers is still an area of ongoing debate together with paucity of robust data examining the impact of hospital SAVR volume on TAVR outcomes. Understanding this intricate relationship is crucial as CMS re-examines minimum volume requirements in its NCD of TAVR. This nationally representative study demonstrated that SAVR volume alone was an independent predictor of mid-term TAVR outcomes, which persisted in the risk-adjusted analyses. This study also highlighted an inverse association between hospital SAVR experience and adverse TAVR outcomes, and which was pronounced beyond 100 SAVR cases. In our sensitivity analysis, our findings remained robust even after accounting for HMO patients, patients undergoing transfemoral TAVR only, and era of surgery (i.e. 2014). These findings not only shed light on a lingering question of the influence of SAVR volumes on TAVR outcomes, but also emphasizes the importance of sustaining a viable SAVR program within the heart team to achieve improved TAVR outcomes.

Several recent studies have demonstrated a strong inverse relationship between TAVR volume and outcomes.(14,16) For instance, a recent analysis of the Society of Thoracic Surgeon’s/American College of Cardiology Transcatheter Valve Therapy registry found that increasing TAVR site volume was associated with lower in-hospital risk-adjusted mortality, vascular complications and bleeding, and which appeared to persist up to 400 cases.(14) Similar relationships between procedure volume and outcomes have been reported for mitral and aortic valve surgery.(21–23) However, the relationship between SAVR volume and TAVR outcomes has not been well elucidated. A recent study, which also utilized the CMS claims data, found that hospital SAVR volume alone was not associated with better TAVR outcomes but rather the combination of high TAVR and SAVR volumes.(24) Our findings contradict their findings in that SAVR volume was directly associated with TAVR outcomes. The discrepancy in our findings could be attributed to a few reasons. First, their study included HMO patients which could have introduce bias and uncertainty in their estimation of longitudinal outcomes by design. Even though we excluded HMO patients in our study, our findings remained robust after including HMO patients in our sensitivity analysis. Secondly, to avoid confounding and reduce bias in interpretation, we only included isolated cases of TAVR procedures that were tied to specific DRG codes during the index hospitalization. Finally, unlike their study which utilized a binary cut-off of 97 SAVR cases, our study examined the impact of SAVR volumes beyond 100 and found that the association was more pronounced, and which persisted in our adjusted analysis.

Our analysis is unique and further adds to the literature by examining the relationship between the two surgeries (i.e. SAVR and TAVR) with different teams involved. How well both these teams interact in the context of a multidisciplinary heart team is essential, and in part, one of the factors we believe that has contributed towards the overall success of TAVR programs nationally. The ideal heart team also includes anesthesia, intensivists, nursing staff, specialist surgical and medical teams, who work closely with cardiac surgeons and cardiologists. Together, these teams provide an avenue for promoting interdisciplinary dialogue, creation of robust support networks, and dissemination of surgical and non-surgical knowledge as it pertains to patient care. We suspect that this intricate but important interaction between SAVR expertise and TAVR experience amongst the heart team stakeholders most likely explains the observed inverse association between SAVR volume and TAVR outcomes in our study. The importance of this concept was shown by the recent NCD to continue with the heart team requirement. Thus, relationship with SAVR volume may indicate that surgeon involvement in the heart team has an effect. However, this study was not designed to compare outcomes between non-heart team systems versus those employing heart teams.

This study has important practice and policy implications. In 2012, CMS proposed minimum volume requirements as part of its NCD, and for reimbursement purposes. For example, one of the requirements was to have centers perform a minimum of 50 SAVR procedures annually. More recently, some clinicians have proposed for revision of existing NCD guidelines due to couple of important concerns: First, there is concern that SAVR volumes could decrease further, and which may make it difficult to maintain TAVR accreditation, especially in the low-volume centers. Second, minimum requirements could also pose a barrier to entry for new centers interested in performing TAVR but who cannot meet SAVR requirements. Despite the ongoing debate surrounding the importance of heart teams in patient management, as some centers are thinking about transitioning into TAVR centers of excellence, the utility of these teams cannot be underestimated even though CMS revised the minimum volume requirements in the recent NCD. Nonetheless, our findings underscore the integral role of cardiac surgeons within the heart team and essentially imply that SAVR experience (through accumulating volume) closely determines the success of the TAVR program. Our findings are also pertinent to both existing hospitals already performing TAVR and new hospitals trying to establish a new TAVR practice. Hence, we strongly believe that continuing to invest resources and personnel to both the SAVR and TAVR programs will be important for overall TAVR outcomes.

While our findings couldn’t determine an exact cutoff for minimal SAVR volume requirement for institutions in the context of TAVR, our study provides useful data to help inform physicians, patients and CMS policy makers as we continue to seek to further improve patient mortality and morbidity. As previously pointed out, “TAVR is more than just a procedure. It is a party of a comprehensive treatment program that embraces team-based care by experienced clinicians with shared decision marking”.(25) As we anticipate additional growth of TAVR in the years ahead in light of recent FDA approval in low-risk patients(12,26), it becomes essential to monitor device and procedural performance, establish quality assurance and improvement initiatives, and importantly ensure successful integration of new TAVR programs especially in the low-volume centers. In this regard, quality control for TAVR procedures, especially given inter-site variability(27) in outcomes will be key through ongoing collaboration between TAVR device manufactures, professional societies, the FDA and CMS. Likewise, minimal SAVR volume thresholds will provide a framework to reliably benchmark outcomes over time in the setting of existing hospital practice variations to ensure the long-term success of emerging TAVR technologies.

Limitations

Our study had several limitations. First, the CMS is a hospital claims database without access to individual medical records and is subject to the shortcomings of other administrative datasets. Coding-related inconsistencies could overestimate or underestimate our findings, although we adjusted for several confounders. CMS precludes detailed assessment of patient presentation, procedural and echocardiographic details, and STS risk scores. Due to the nature of the database, frailty was difficult to assess formally. Given that the study period ended in 2015, and the fact that TAVR is a very dynamically changing field, it is possible that our findings may not adequately reflect current clinical practice and outcomes. Thus, our findings must be interpreted with context. Furthermore, although our analysis was based on a large number of patients, the results cannot be extrapolated to those who did not meet our inclusion criteria e.g. HMO patients, those undergoing concomitant coronary artery bypass grafting, or other valvular surgery.

Conclusions:

In summary, total hospital SAVR volume appears to be correlated with TAVR outcomes, with higher 30-day and 1-year mortality observed at low volume centers. This data supports the importance of viable surgical program within the heart team, and the use of minimum SAVR hospital thresholds may be considered as an additional metric for TAVR performance.

Central Illustration:

Association between surgical aortic valve replacement volumes and transcatheter aortic valve replacement outcomes.

Clinical Perspectives:

What is known?

There exists a volume-outcome relationship for transcatheter aortic valve replacement (TAVR).

What is new?

There appears to be an inverse association between hospital SAVR experience and adverse TAVR outcomes, with higher 30-day and 1-year mortality observed in low volume centers.

What is next?

While the importance of a viable surgical program within the heart team cannot be underestimated, further research on the use of minimum SAVR hospital thresholds should be explored as an additional metric to evaluate TAVR performance.

Abbreviations list:

AVR

Aortic valve replacement

TAVR

Transcatheter aortic valve replacement

SAVR

Surgical aortic valve replacement

CMS

Center for Medicare and Medicaid

ICD-X-CM

International classification of disease clinical modification

FDA

Food and drug administration

HMO

Health maintenance organization

LOS

Length of stay

ICU

Intensive care unit

PCI

Percutaneous coronary intervention

NCD

National coverage determination