Abstract
Objective:
Recent approval of transcatheter aortic valve replacement (TAVR) in patients at low surgical risk has resulted in a rapid real-world expansion of TAVR in patients not otherwise examined in recent low-risk trials. We sought to evaluate the outcomes of surgical aortic valve replacement (SAVR) vs TAVR in low-risk Medicare beneficiaries.
Methods:
Using the United States Centers for Medicare and Medicaid Services claims database, we evaluated all beneficiaries undergoing isolated SAVR (n=33,210) or TAVR (n=77,885) (2018-2020). International Classification of Diseases 10th revision codes were used to define variables and frailty was defined by the validated Kim Index. Doubly robust risk-adjustment was performed with inverse probability weighting and multilevel regression models, as well as competing-risk time to event analysis. A low risk cohort was identified to simulate recent low risk trials.
Results:
A total of 15,749 low risk patients (8,144 SAVR, 7,605 TAVR) were identified. Comparison was performed with doubly robust risk-adjustment accounting for all factors. TAVR was associated with lower perioperative stroke (OR 0.62, p<0.001) and hospital mortality (OR 0.16, p<0.001) compared to SAVR. However, risk-adjusted longitudinal analysis demonstrated TAVR was associated with higher late risk of stroke (HR 1.65, p < 0.001), readmission for valve reintervention (HR 1.88, p < 0.001) and all-cause mortality (HR 1.54, p < 0.001) compared to SAVR.
Conclusions:
Among low-risk Medicare beneficiaries under age 75 years undergoing isolated aortic valve replacement, SAVR was associated with higher index morbidity and mortality but improved 3-year risk-adjusted stroke, valve reintervention, and survival compared to TAVR.
Keywords: SAVR, TAVR, Low-Risk, Aortic Valve Replacement, Medicare
Introduction
Approval of transcatheter aortic valve replacement (TAVR) in patients at low surgical risk has resulted in a rapid expansion of TAVR.1-4 Surgical risk is defined by The Society of Thoracic Surgeons (STS) predicted risk of mortality (PROM).5-8 However, in real world practice, TAVR is being utilized in some populations not otherwise examined in recent low risk trials comparing TAVR to surgical aortic valve replacement (SAVR), such as in younger patients, aortic insufficiency and bicuspid valves.9-11 Availability of longitudinal data will be important to inform heart team decision making for this population with a life expectancy greater than 10 years.
Recent data from the STS Adult Cardiac Surgery Database (ACSD) has highlighted excellent long-term outcomes after SAVR with 95% survival at 8 years in low-risk patients under the age of 75.12 Furthermore, the longitudinal significance of structural valve degeneration after TAVR is less well understood then after SAVR.13, 14 This data, in combination with the highly selected and relatively small cohorts evaluated in prospective trials, highlights the need for larger real-world comparisons of SAVR vs TAVR in the low risk population.
The purpose of the current analysis was to evaluate the longitudinal contemporary outcomes of SAVR vs TAVR in Medicare beneficiaries. Furthermore, we sought to compare outcomes between SAVR vs TAVR in a real world low risk cohort. We hypothesized that SAVR would be associated with higher early morbidity and mortality but superior longitudinal freedom from death, stroke, or valve reintervention over the study period.
Methods
Patient Data
Using the United States Centers for Medicare and Medicaid Services (CMS) inpatient claims database, we evaluated all beneficiaries between age 65 and 85 years undergoing isolated SAVR (n=33,210) or TAVR (n=77,885) between January 1, 2018, and December 31, 2020. Patients with concomitant valvular procedures, coronary procedures, aortic procedures, prior heart surgery or valve procedure, endocarditis, mechanical support, cardiogenic shock, or emergency procedures were excluded (Supplemental Table 1). Patients over the age of 85 years were excluded a priori based on lack of clinical relevance of SAVR vs TAVR and valve durability. The West Virginia University Institutional Review Board approved this study with waiver of consent (Protocol #2210660362, 3/23/23).
Variables and Definitions
Procedures were identified through diagnosis-related group and International Classification of Diseases 10th Revision (ICD-10) Procedure Coding System codes. Variable adjudication followed standard methodology of ICD-10 coding (Supplemental Table 2).15 The Elixhauser comorbidity index, including 30 comorbidity variables, was utilized as a composite measure of baseline health using ICD-10 Clinical Modification codes.16 The Kim index, a validated claims based measurement of frailty was used for risk adjustment using 52 ICD diagnosis code-based variables, 25 CPT code-based variables, and 16 HCPCS code-based variables in the past 12 months to predict the value of a deficit-accumulation frailty index.17, 18 Present on Admission modifiers were used to delineate preoperative risk factors vs postoperative complications. Longitudinal data were obtained from CMS claims data for the primary indication for subsequent hospital admissions. Total direct hospital cost was assessed as reported in the database. There was no missing data. The low-risk cohort was defined as the lowest quartile of Elixhauser comorbidity index, lowest quartile of Kim frailty index, and age ≤75 years.
Statistical Analysis
Outcomes included all-cause mortality, and readmission for stroke, valve reintervention and heart failure. The primary outcome was the 3-year longitudinal risk of mortality. The secondary outcome was the composite of 3-year risk of death, stroke, or valve reintervention. Categorical variables were presented as n (%) while continuous variables were assessed for normality and presented as mean (standard deviation) or median (Q1, Q3) as appropriate. Univariable comparisons were performed using Chi Square for categorical comparisons while non-parametric tests assessed continuous variables given non-normal distribution. To account for confounding in this retrospective analysis we used inverse probability of treatment weighting (IPTW) of propensity scores. The IPTW models included all preoperative demographic, comorbidity, and frailty variables listed in Table 1. An acceptable balance threshold was set a priori at a standard mean difference <0.10 with good balance between groups (Figure 1). Doubly robust risk-adjustment was performed using multilevel regression analysis, Cox Proportional Hazards survival modeling, and Fine-Gray time to event analysis accounting for competing risk of death. All baseline covariates were included, and models were assessed for collinearity using variance inflation factor (Supplemental Methods). Longitudinal outcomes were evaluated to three years with censoring. Conditional longitudinal readmission was assessed before and after 90 days to account for the differences in perioperative morbidity and recovery. The proportional hazards assumption was assessed using Schoenfeld residuals. After identifying the low-risk cohort as defined above, the doubly robust methodology was repeated on this cohort for risk-adjustment.
Table 1.
Baseline Characteristics and Patient Demographics Entire Population
| Variable | Total (111,095) |
TAVR (77,885) |
SAVR (33,210) |
p |
|---|---|---|---|---|
| Age | 74(68, 79) | 76(70, 82) | 70(66, 76) | p<0.001 |
| Gender (Male) | 63,109(56.8%) | 42,699(54.8%) | 20,410(61.5%) | p<0.001 |
| Race (White) | 100,365(90.3%) | 71,390(91.7%) | 28,975(87.2%) | p<0.001 |
| Frailty Index | 0.173(+−0.046) | 0.177(+−0.047) | 0.161(+−0.042) | p<0.001 |
| Elixhauser score | 5.6(+−1.97) | 5.53(+−1.95) | 5.77(+−2.01) | p<0.001 |
| Diabetes mellitus | 39,929(35.9%) | 30,884(39.7%) | 9,045(27.2%) | p<0.001 |
| Peripheral artery disease | 14,536(13.1%) | 11,679(15%) | 2,857(8.6%) | p<0.001 |
| Chronic lung disease | 28,261(25.4%) | 21,319(27.4%) | 6,942(20.9%) | p<0.001 |
| GI bleeding, ulcer, reflux | 28,906(26%) | 20,287(26%) | 8,619(26%) | p=0.748 |
| Liver disease | 4,636(4.17%) | 3,088(3.96%) | 1,548(4.66%) | p<0.001 |
| Atrial Fibrillation | 38,736(34.9%) | 23,113(29.7%) | 15,623(47%) | p<0.001 |
| Coronary Artery Disease | 63,357(57%) | 49,292(63.3%) | 14,065(42.4%) | p<0.001 |
| Aortic Insufficiency | 5,879(5.29%) | 995(1.28%) | 4,884(14.7%) | p<0.001 |
| Bicuspid Valve | 6,851(6.17%) | 1,673(2.15%) | 5,178(15.6%) | p<0.001 |
| Anemia | 19,441(17.5%) | 14,842(19.1%) | 4,599(13.8%) | p<0.001 |
| Cancer | 4,721(4.25%) | 3,820(4.9%) | 901(2.71%) | p<0.001 |
| Cerebrovascular accident | 2,032(2.0%) | 1,579(2.0%) | 268(2.1%) | p=0.237 |
| Permanent Pacemaker | 7,332(7.1%) | 5,970(7.5%) | 774(6.0%) | p<0.001 |
| Heart Failure | 63,799(57.4%) | 51,864(66.6%) | 11,935(35.9%) | p<0.001 |
| Obesity | 28,208(25.4%) | 19,665(25.2%) | 8,543(25.7%) | p<0.001 |
| Malnutrition | 3,891(3.5%) | 1,918(2.46%) | 1,973(5.94%) | p<0.001 |
| Chronic Renal Disease | 17,749(16%) | 13,643(17.5%) | 4,106(12.4%) | p<0.001 |
| LOS (days) | 5.11(+−6.96) | 3.18(+−4.7) | 9.63(+−9) | p<0.001 |
| Total cost (Hospital) | 57,253(+−42,391) | 55,257(+−35,743) | 61,936(+−54,625) | p<0.001 |
| ARF(Hospital) | 45(0.1%) | 24(0.1%) | 21(0.1%) | p=0.092 |
| AKI (Hospital) | 13,463(12.1%) | 6,186(7.9%) | 7,277(21.9%) | p<0.001 |
| Stroke (Hospital) | 2,242(2.0%) | 1,180(1.5%) | 1,062(3.2%) | p=0.098 |
| Bleeding (Hospital) | 246(0.2%) | 119(0.2%) | 127(0.4%) | p<0.001 |
| New Pacemaker (Hospital) | 10,349(9.3%) | 9,655(12.4%) | 694(2.1%) | p<0.001 |
| Vascular Complication (Hospital) | 607(0.5%) | 574(0.7%) | 33(0.1%) | p<0.001 |
| Death (Hospital) | 1,894(1.7%) | 773(1.0%) | 1,121(3.4%) | p<0.001 |
| Readmit (30 days) | 12,733(11.5%) | 8,156(10.5%) | 4,577(13.8%) | p<0.001 |
| Readmit Bleed (30 day) | 269(0.2%) | 141(0.2%) | 128(0.4%) | p=0.122 |
| Readmit Stroke (30 day) | 551(0.5%) | 392(0.5%) | 159(0.5%) | p=0.168 |
| Readmit HF (30 day) | 7,678(6.9%) | 5,173(6.64%) | 2,505(7.54%) | p<0.001 |
| Any death (30 days) | 3,190 (2.87%) | 1,586 (2.04%) | 1,604 (4.83%) | p < 0.001 |
| Death (3 years) | 12,718(11.4%) | 10,205(13.1%) | 2,513(7.57%) | p<0.001 |
| Readmit (3 years) | 41,660(37.5%) | 29,865(38.3%) | 11,795(35.5%) | p<0.001 |
| Readmit Bleed (3 year) | 1,165(1.1%) | 774(1.0%) | 391(1.2%) | p=0.007 |
| Readmit Stroke (3 year) | 2,691(2.4%) | 2,033(2.6%) | 658(2.0%) | p<0.001 |
| Readmit HF (3 year) | 25,141(22.6%) | 19,064(24.5%) | 6,077(18.3%) | p<0.001 |
| Valve Reintervention (3 year) | 243(0.2%) | 191(0.3%) | 52(0.2%) | p=0.191 |
| Composite (3 year) | 14,637(13.2%) | 11,576(14.9%) | 3,061(9.2%) | p<0.001 |
ARF – Acute Renal Failure, AKI – Acute Kidney Injury, HF – Heart Failure
Figure 1: Balance Plots.
Highlights the pre and post adjustment weighted standard means of the Inverse probability of treatment weighting.
Results
Overall Cohort Characteristics
A total of 111,095 Medicare beneficiaries (33,210 SAVR, 77,885 TAVR) were included in the analysis. Prior to risk adjustment, the median age was 70 for SAVR and 76 for TAVR. TAVR patients had lower rates of Aortic Insufficiency (1.3% vs 14.7%, p<0.001), Bicuspid Valve (2.2% vs 15.6%, p<0.001), and Elixhauser comorbidity index but higher frailty index (Table 1). Compared to SAVR, TAVR was associated with lower operative mortality (1.0% vs 3.4%, p<0.001), but higher rates of new permanent pacemaker placement (12.4% vs 2.1%, p<0.001). At 3 years TAVR patients had lower longitudinal survival (p<0.001, Figure 2A) compared to SAVR with higher rates of the composite endpoint of death, stroke or valve reintervention (p<0.001, Figure 2C).
Figure 2: Overall Cohort.
A. Unadjusted survival curves for SAVR (blue) vs TAVR (red) in the overall population with 95% CI (p<0.001). B. Risk-Adjusted survival curves for SAVR (blue) vs TAVR (red) in the overall population with 95% CI (HR 1.28, p<0.001). C. Unadjusted freedom from death, stroke or valve reintervention for SAVR (blue) vs TAVR (red) in the overall population with 95% CI (p<0.001). D. Risk-Adjusted freedom from death, stroke or valve reintervention for SAVR (blue) vs TAVR (red) in the overall population with 95% CI (HR 1.79, p<0.001).
Overall Cohort Risk-Adjusted Outcomes
After comprehensive risk-adjustment, TAVR was associated with lower in hospital mortality (OR 0.25, p<0.001), stroke (OR 0.80, p<0.001), and bleeding (OR 0.73, p<0.001) but higher vascular complications (OR 1.54, p<0.001) and new permanent pacemaker implantation (OR 2.75, p<0.001, Table 2). Furthermore, TAVR was associated with higher risk-adjusted cost ($71,361 vs $68,928, p<0.001) but lower 30-day all-cause readmission (OR 0.66, p<0.001, Table 2).
Table 2.
Risk-Adjusted Outcomes Entire Cohort
| Variable | Hazards Ratio for TAVR (95% CI) [p-value] |
|---|---|
| AKI(Hospital)* | 0.71(0.45-1.10)[p=0.123] |
| Stroke (Hospital)* | 0.80(0.76-0.86)[p<0.001] |
| Bleeding (Hospital)* | 0.73(0.62-0.85)[p<0.001] |
| Vascular Complication (Hospital)* | 1.54(1.36-1.74)[p<0.001] |
| New Permanent Pacemaker (Hospital)* | 2.75(1.91-3.60)[p<0.001] |
| Death (Hospital)* | 0.25(0.23-0.26)[p<0.001] |
| Readmit (30 days)* | 0.66(0.64-0.67)[p<0.001] |
| Readmit Bleed (30 day)* | 0.46(0.38-0.55)[p<0.001] |
| Readmit Stroke (30 day)* | 0.88(0.78-0.99)[p=0.030] |
| Readmit HF (30 day)* | 0.67(0.65-0.70)[p<0.001] |
| Death (3 years)^ | 1.28(1.15-1.43)[p<0.001] |
| Readmit (3 years)# | 1.02(0.97-1.07)[p=0.504] |
| Readmit Bleed (3 year)# | 0.74(0.68-0.80)[p<0.001] |
| Readmit Stroke (3 year)# | 1.57(1.31-1.87)[p<0.001] |
| Readmit HF (3 year)# | 0.95(0.90-1.01)[p=0.093] |
| Valve Reintervention (3 Year)# | 1.62(1.33-1.97)[p<0.001] |
| Composite (3 year)# | 1.79(1.64-1.94)[p<0.001] |
Multivariable Logistic Regression, ^ Cox Proportional Hazards Model, # Fine Gray Multivariable Time to Event Model accounting for competing risk of death, ARF – Acute Renal Failure, AKI – Acute Kidney Injury, HF – Heart Failure, Composite – Death, Valve Reintervention or Stroke.
For the primary endpoint, TAVR was associated with increased risk-adjusted longitudinal all-cause mortality (HR 1.28, p < 0.001, Figure 2B). Similarly, TAVR was associated with higher readmission for stroke (HR 1.57, p<0.001) and valve reintervention (HR 1.62, p<0.001). There was no difference in risk-adjusted all-cause readmission or readmission for heart failure (Table 2), but the total risk-adjusted cost of readmissions was higher in TAVR vs SAVR ($37,006 vs $33,253, p<0.001). For the secondary endpoint, TAVR was associated with higher incidence of the composite endpoint of death, stroke, or valve reintervention (HR 1.79, p<0.001, Figure 2D).
Low-Risk Subgroup Characteristics
A total of 15,749 low-risk patients (8,144 SAVR, 7,605 TAVR) were identified. Prior to risk adjustment, the median age was 69 years in each group. The low-risk subgroup had significantly lower incidence of comorbid disease (Table 3). Compared to SAVR, TAVR was associated with lower hospital mortality (0.2% vs 0.9%) but no difference in all cause 30-day mortality (1.9% vs 2.1%). At 3 years TAVR patients had lower longitudinal survival (p<0.001, Figure 3A) and higher composite endpoint of death, stroke, or valve reintervention (p<0.001, Figure 3C).
Table 3.
Baseline Characteristics and Patient Demographics Low Risk Subgroup
| Variable | Total (15,749) |
TAVR (7,605) |
SAVR (8,144) |
p |
|---|---|---|---|---|
| Age | 69(67, 72) | 69(67, 73) | 69(67, 72) | p<0.001 |
| Gender (Male) | 10,057(63.9%) | 4,676(61.5%) | 5,381(66.1%) | p<0.001 |
| Race (White) | 14,042(89.2%) | 6,837(89.9%) | 7,205(88.5%) | p<0.001 |
| Frailty Index | 0.131(+−0.021) | 0.132(+−0.021) | 0.131(+−0.021) | p=0.005 |
| Elixhauser score | 3.85(+−1.05) | 3.79(+−1.07) | 3.91(+−1.03) | p<0.001 |
| Diabetes mellitus | 2,785(17.7%) | 1,742(22.9%) | 1,043(12.8%) | p<0.001 |
| Peripheral artery disease | 1,047(6.65%) | 639(8.4%) | 408(5.01%) | p<0.001 |
| Chronic lung disease | 1,721(10.9%) | 997(13.1%) | 724(8.89%) | p<0.001 |
| GI bleeding, ulcer, reflux | 3,428(21.8%) | 1,604(21.1%) | 1,824(22.4%) | p=0.049 |
| Liver disease | 275(1.75%) | 174(2.29%) | 101(1.24%) | p<0.001 |
| Atrial Fibrillation | 2,586(16.4%) | 1,128(14.8%) | 1,458(17.9%) | p<0.001 |
| Coronary Artery Disease | 6,471(41.1%) | 3,809(50.1%) | 2,662(32.7%) | p<0.001 |
| Aortic Insufficiency | 1,163(7.38%) | 115(1.51%) | 1,048(12.9%) | p<0.001 |
| Bicuspid Aortic Valve | 2,332(14.8%) | 464(6.1%) | 1,868(22.9%) | p<0.001 |
| Anemia | 1,179(7.49%) | 559(7.35%) | 620(7.61%) | p=0.552 |
| Cancer | 423(2.69%) | 281(3.69%) | 142(1.74%) | p<0.001 |
| Cerebrovascular accident | 1,130(7.18%) | 659(8.67%) | 471(5.78%) | p=0.237 |
| Permanent Pacemaker | 312(1.98%) | 176(2.31%) | 136(1.67%) | p=0.004 |
| Heart Failure | 5,228(33.2%) | 3,822(50.3%) | 1,406(17.3%) | p<0.001 |
| Obesity | 2,833(18.0%) | 1,569(20.6%) | 1,264(15.5%) | p<0.001 |
| Malnutrition | 162(1.0%) | 46(0.6%) | 116(1.4%) | p<0.001 |
| Chronic Renal Disease | 685(4.4%) | 433(5.7%) | 252(3.1%) | p<0.001 |
| LOS (days) | 4.36(+−4.41) | 1.93(+−2.37) | 6.63(+−4.66) | p<0.001 |
| Total cost (Hospital) | 48,702(+−29,419) | 51,234(+−30,000) | 46,338(+−28,666) | p<0.001 |
| ARF(Hospital) | 2(0.013%) | 0(0%) | 2(0.1%) | p=0.501 |
| AKI (Hospital) | 848(5.38%) | 135(1.78%) | 713(8.75%) | p<0.001 |
| Stroke (Hospital) | 173(1.1%) | 50(0.657%) | 123(1.51%) | p<0.001 |
| Bleeding (Hospital) | 16(0.1%) | 0(0%) | 16(0.2%) | p<0.001 |
| New Pacemaker (Hospital) | 649(4.1%) | 466(6.1%) | 183(2.3%) | p<0.001 |
| Vascular Complication (Hospital) | 59(0.4%) | 59(0.8%) | 0(0%) | p<0.001 |
| Death (Hospital) | 96(0.61%) | 18(0.237%) | 78(0.958%) | p<0.001 |
| Readmit (30 days) | 1,038(6.59%) | 323(4.25%) | 715(8.78%) | p<0.001 |
| Readmit Bleed (30 day) | 20(0.1%) | 3(0.1%) | 17(0.2%) | p=0.006 |
| Readmit Stroke (30 day) | 57(0.362%) | 20(0.263%) | 37(0.454%) | p=0.068 |
| Readmit HF (30 day) | 341(2.17%) | 115(1.51%) | 226(2.78%) | p<0.001 |
| Any death (30 days) | 150 (0.952%) | 40 (0.526%) | 110 (1.35%) | p < 0.001 |
| Death (3 years) | 658(4.18%) | 355(4.67%) | 303(3.72%) | p=0.003 |
| Readmit (3 years) | 3,385(21.5%) | 1,421(18.7%) | 1,964(24.1%) | p<0.001 |
| Readmit Bleed (3 year) | 69(0.4%) | 30(0.3%) | 39(0.4%) | p=0.496 |
| Readmit Stroke (3 year) | 243(1.33%) | 122(1.6%) | 121(1.5%) | p=0.075 |
| Readmit HF (3 year) | 1,110(7.05%) | 550(7.23%) | 560(6.88%) | p=0.401 |
| Valve Reintervention (3 year) | 44(0.279%) | 30(0.4%) | 14(0.2%) | p<0.001 |
| Composite (3 year) | 754(4.8%) | 409(5.4%) | 345(4.2%) | p<0.001 |
NSTEMI – Non-ST Elevation Myocardial Infarction, STEMI – ST Elevation Myocardial Infarction
Figure 3. Low-Risk Cohort.
A. Unadjusted survival curves for SAVR (blue) vs TAVR (red) in the low-risk cohort with 95% CI (p<0.001). B. Risk-Adjusted survival curves for SAVR (blue) vs TAVR (red) in the low-risk cohort with 95% CI (HR 1.54, p<0.001). C. Unadjusted freedom from death, stroke or valve reintervention for SAVR (blue) vs TAVR (red) in the low-risk cohort with 95% CI (p<0.001). D. Risk-Adjusted freedom from death, stroke or valve reintervention for SAVR (blue) vs TAVR (red) in the low-risk cohort with 95% CI (HR 1.57, p<0.001).
Low-Risk Subgroup Risk-Adjusted Outcomes
After doubly robust risk adjustment in the low-risk cohort, TAVR was associated with lower incidence of hospital mortality (OR 0.16, p<0.001), acute kidney injury (AKI) (OR 0.18, p<0.001), and stroke (OR 0.63, p<0.001) but higher rates of vascular complications (OR 1.84, p<0.001) and new permanent pacemaker (OR 2.21, p<0.001). Additionally, TAVR was associated with a higher risk-adjusted hospital cost ($65,062 vs $55,644, p<0.001) but lower all-cause 30-day readmission (0.45, p<0.001) (Table 4).
Table 4.
Risk-Adjusted Outcomes Low Risk Subgroup
| Variable | Hazards Ratio for TAVR (95% CI) [p-value] |
|---|---|
| AKI (Hospital)* | 0.18(0.15-0.20)[p<0.001] |
| Stroke (Hospital)* | 0.63(0.50-0.80)[p<0.001] |
| Vascular Complication (Hospital)* | 1.84(1.26-1.97)[p<0.001] |
| New Permanent Pacemaker (Hospital)* | 2.21(1.66-2.76)[p<0.001] |
| Death (Hospital)* | 0.16(0.11-0.24)[p<0.001] |
| Readmit (30 days)* | 0.45(0.41-0.50)[p<0.001] |
| Readmit Stroke (30 day)* | 0.47(0.29-0.71)[p<0.001] |
| Readmit HF (30 day)* | 0.41(0.34-0.47)[p<0.001] |
| Death (3 years)^ | 1.54(1.22-1.94)[p<0.001] |
| Readmit (3 years)# | 0.71(0.64-0.79)[p<0.001] |
| Readmit Bleed (3 year)# | 0.72(0.38-1.34)[p=0.296] |
| Readmit Stroke (3 year)# | 1.65(1.32-2.02)[p<0.001] |
| Readmit HF (3 year)# | 0.76(0.64-0.92)[p=0.005] |
| Valve Reintervention (3 Year)# | 1.88(1.46-2.38)[p<0.001] |
| Composite (3 year)# | 1.57(1.28-1.93)[p<0.001] |
Multivariable Logistic Regression
Cox Proportional Hazards Model
Fine Gray Multivariable Time to Event Model accounting for competing risk of death, ARF – Acute Renal Failure, AKI – Acute Kidney Injury, HF – Heart Failure, Composite – Death, Valve Reintervention or Stroke.
In risk-adjusted longitudinal analysis of the low-risk cohort, TAVR was associated with higher all-cause mortality (HR 1.54, p<0.001, Figure 3B) over the 3-year study period including the index admission. Similarly, TAVR was associated with higher readmission for stroke (HR 1.65, p<0.001) and valve reintervention (HR 1.88, p<0.001, Table 4). Finally, TAVR was associated with higher risk-adjusted total cost of readmission ($45,094 vs $41,066, p<0.001) and higher incidence of the composite endpoint of death, stroke, or valve reintervention (HR 1.57, p<0.001, Figure 3D).
Discussion
This contemporary real-world study of aortic valve replacement in Medicare beneficiaries demonstrates five important findings. First, these data highlight a majority of Medicare beneficiaries between the ages of 65 and 85 undergoing isolated aortic valve replacement receive TAVR compared to SAVR. Second, TAVR was associated with lower risk-adjusted in-hospital mortality, stroke, AKI and bleeding but higher vascular complications, new permanent pacemaker and cost. Third, in over 100,000 Medicare beneficiaries, after doubly robust risk adjustment inclusive of frailty, TAVR was associated with 28% higher longitudinal mortality and 79% higher occurrence of death, stroke or valve reintervention compared to SAVR at only 3 years. Fourth, in the lowest risk patient cohort of age 65-75 years, TAVR was associated with superior in-hospital outcomes despite higher rates of vascular complications, new permanent pacemaker, and cost. Finally, in this low-risk cohort, TAVR was associated with 54% higher risk-adjusted longitudinal mortality and 57% higher occurrence of death, stroke, or valve reintervention compared to SAVR at 3 years. Collectively, these findings may help to provide clarity for clinical decision-making by demonstrating the longitudinal advantages of SAVR over TAVR in low-risk patients.
In an important counter distinction to preliminary data presented by current trials comparing TAVR and SAVR in highly selected low risk patients, the present analysis confirms prior evidence demonstrating differences in early morbidity and mortality after TAVR vs SAVR in a contemporary low risk real-world Medicare population, but demonstrates a 3-year longitudinal outcome benefit of SAVR.1, 2, 9, 19 While SAVR is a more invasive therapy, the rapid expansion of minimally invasive techniques has given patients additional options.20, 21 Mid-term results of the Evolut low-risk and PARTNER3 trials in combination with intermediate risk studies has resulted in an exponential growth of TAVR beyond populations studied in these trials.1, 2, 11 In the current analysis, 70% of Medicare beneficiaries between 65-85 years old requiring aortic valve replacement undergo TAVR. There is growing concern that these highly selected clinical trial cohorts may not be generalizable to the general population given the increasing reports of valve thrombosis and elevated stroke risk in patients with TAVR valves.13, 14, 22 These concerns have been compounded by the mounting evidence of significant increased risk with re-operative aortic valve surgery after TAVR, further complicating valve durability issues.23, 24
When a patient requires aortic valve replacement there are many factors that influence the decision of SAVR vs TAVR and it is critical for the multidisciplinary heart team to provide patients a balanced interpretation of the evidence to allow patients to make the right choice for their individual needs. This includes consideration of concomitant treatment of comorbid disease like atrial fibrillation and coronary artery disease which our group has previously assessed.25 Despite varied interpretations of the recently published trials, there has been mounting registry evidence of longitudinal differences in survival driven by increased stroke and valve reintervention in the TAVR population.26 The present study further confirms these findings in over 100,000 Medicare beneficiaries accounting for age, comorbidities, and a validated measure of frailty using the most comprehensive longitudinal database in the US population. While registry data have limitations of selection bias that cannot be eliminated by statistical methodology, randomized trials have limitations of generalizability due to specific inclusion criteria. This is highlighted by major differences in midterm survival demonstrated in the low risk cohort of this study (3 year: 4.7% vs 3.7%) compared to the PARTNER3 (5 year: 10% vs 8.2%) and Evolute Low Risk (4 year: 10.7% vs 14.1%) trials. Both randomized and registry data must be carefully considered and critically evaluated when making treatment recommendations, particularly in lower risk patients.
The recent publication of midterm outcomes of SAVR vs TAVR in low-risk patients at 4 and 5 years has led some individuals to declare TAVR as a superior therapy, despite significant evidentiary questions. However, real world evidence from registries with multi-thousands of low-risk patients continue to conflict with these highly selected trials with less than one thousand patients in each group. Furthermore, a comprehensive analysis of longitudinal survival after SAVR in over 43,000 low risk patients from the STS ACSD highlights an overall 92.9% survival at 5 years.12 The present analysis of over 15,000 low-risk patients demonstrates TAVR was associated with higher longitudinal mortality (HR 1.54) and occurrence of death, stroke, or valve reintervention (HR 1.57). While the absolute differences in longitudinal outcomes were smaller in the low-risk cohort, the relative risk was equivalent to that seen in the overall population and these findings were statistically and clinically significant.
This study has several important limitations including the generalizability of these findings for patients under the age of 65. Additionally, the administrative nature of the CMS database precludes a comparative evaluation of valve gradients or anatomy. Despite doubly robust risk adjustment the possibility of confounding remains, the effect of which, however, may be lessened by the very large sample size and the low-risk subgroup comparison. However, we are unable to account for patient and surgeon factors not captured in the database preventing demonstration of causality. While the nature of the database precludes determination of timing of events during hospitalization, the CMS “present on admission” modifier assists in accurate categorization of predictor and outcome variables. Finally, we are unable to calculate an STS risk score based on claims data and instead use a novel definition of low risk based on Elixhauser comorbidity index, Kim Frailty index, and age which included the lowest risk 13.5% of the population. However, as highlighted in Table 3, the burden of comorbid disease and the 30-day all cause mortality (0.5% vs 1.3%) was very similar to that seen in the PARTNER3 and Evolute Low Risk Trials.1, 2
In conclusion, TAVR was associated with better in-hospital outcomes among Medicare beneficiaries, but worse risk-adjusted longitudinal freedom from death, stroke, or valve reintervention. Similarly, in the lowest risk cohort under age 75 years, TAVR was associated with worse 3-year risk-adjusted freedom from death, stroke, or valve reintervention compared to SAVR. These contemporary real-world data further clarify the advantages of SAVR over TAVR in low-risk patients, particularly in those less than age 75 years, and provide important information for heart teams to consider when making treatment decisions.
Supplementary Material
Central Figure:
Event Free Survival Low-Risk Surgical vs Transcatheter Aortic Valve Replacement
Central Message:
Among low-risk Medicare beneficiaries SAVR was associated with higher index morbidity and mortality but better 3-year risk-adjusted stroke, valve reintervention, and survival compared to TAVR.
Perspective Statement:
Approval of transcatheter aortic valve replacement (TAVR) in patients at low surgical risk has resulted in a rapid expansion of TAVR. However, in real world practice TAVR may not be generalizable to populations not evaluated in trials comparing TAVR to surgical aortic valve replacement (SAVR). Real-world longitudinal data will be critical to assess the lifetime benefits of SAVR vs TAVR.
Acknowledgment
We would like to thank Ricardo Pietrobon MD PhD, Aline Machiavelli MS, and Livia Eslabao MS for their statistical support. Additionally, we would like to thank Kalee Vincent and Mandy Moore for data and coding support.
Funding:
Supported by NIH NHLBI # 2UM1 HL088925 12 (JHM, JWH, VB)
Abbreviations:
- ACSD
Adult Cardiac Surgery Database
- AF
Atrial fibrillation
- AKI
Acute Kidney Injury
- ARF
Acute Renal Failure
- CABG
Coronary Artery Bypass Grafting
- CMS
Centers for Medicare and Medicaid Services
- HF
Heart Failure
- ICD-10
International Classification of Diseases 10th Revision
- IPTW
Inverse Probability of Treatment Weighting
- PROM
Predicted Risk of Mortality
- SAVR
Surgical Aortic Valve Replacement
- STS
Society of Thoracic Surgeons
- TAVR
Transcatheter Aortic Valve Replacement
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Disclosures: Dr. Rankin serves as a consultant to Atricure. No other co-authors have relevant disclosures.
Read at the 104th Annual Meeting of the American Association for Thoracic Surgery, Toronto Canada, April 27-30th 2024
References:
- 1.Mack MJ, Leon MB, Thourani VH, et al. Transcatheter Aortic-Valve Replacement in Low-Risk Patients at Five Years. N Engl J Med. 2023;389(21):1949–1960. [DOI] [PubMed] [Google Scholar]
- 2.Forrest JK, Deeb GM, Yakubov SJ, et al. 4-Year Outcomes of Patients With Aortic Stenosis in the Evolut Low Risk Trial. J Am Coll Cardiol. 2023;82(22):2163–2165. [DOI] [PubMed] [Google Scholar]
- 3.Makkar RR, Thourani VH, Mack MJ, et al. Five-Year Outcomes of Transcatheter or Surgical Aortic-Valve Replacement. N Engl J Med. 2020;382(9):799–809. [DOI] [PubMed] [Google Scholar]
- 4.Popma JJ, Deeb GM, Yakubov SJ, et al. Transcatheter Aortic-Valve Replacement with a Self-Expanding Valve in Low-Risk Patients. N Engl J Med. 2019;380(18):1706–1715. [DOI] [PubMed] [Google Scholar]
- 5.Johnston DR, Mahboubi R, Soltesz EG, et al. Redefining "low risk": Outcomes of surgical aortic valve replacement in low-risk patients in the transcatheter aortic valve replacement era. J Thorac Cardiovasc Surg. 2023;165(2):591–604 e593. [DOI] [PubMed] [Google Scholar]
- 6.Shahian DM, Badhwar V, Kurlansky PA, et al. The STS Participant-Level, Multiprocedural Composite Measure for Adult Cardiac Surgery. Ann Thorac Surg. 2022;114(2):467–475. [DOI] [PubMed] [Google Scholar]
- 7.Jacobs JP, Shahian DM, Grau-Sepulveda M, et al. Current Penetration, Completeness, and Representativeness of The Society of Thoracic Surgeons Adult Cardiac Surgery Database. Ann Thorac Surg. 2022;113(3):1461–1468. [DOI] [PubMed] [Google Scholar]
- 8.Jacobs JP, Shahian DM, Badhwar V, et al. The Society of Thoracic Surgeons 2021 Adult Cardiac Surgery Risk Models for Multiple Valve Operations. Ann Thorac Surg. 2022;113(2):511–518. [DOI] [PubMed] [Google Scholar]
- 9.Carroll JD, Mack MJ, Vemulapalli S, et al. STS-ACC TVT Registry of Transcatheter Aortic Valve Replacement. Ann Thorac Surg. 2021;111(2):701–722. [DOI] [PubMed] [Google Scholar]
- 10.Desai ND, O'Brien SM, Cohen DJ, et al. Composite Metric for Benchmarking Site Performance in Transcatheter Aortic Valve Replacement: Results From the STS/ACC TVT Registry. Circulation. 2021;144(3):186–194. [DOI] [PubMed] [Google Scholar]
- 11.Sharma T, Krishnan AM, Lahoud R, Polomsky M, Dauerman HL. National Trends in TAVR and SAVR for Patients With Severe Isolated Aortic Stenosis. J Am Coll Cardiol. 2022;80(21):2054–2056. [DOI] [PubMed] [Google Scholar]
- 12.Thourani VH, Habib R, Szeto WY, Sabik JF, Romano JC, MacGillivray TE, Badhwar V. Survival Following Surgical Aortic Valve Replacement in Low-Risk Patients: A Contemporary Trial Benchmark. Ann Thorac Surg. 2023;117(1):106–112. [DOI] [PubMed] [Google Scholar]
- 13.Imaeda S, Inohara T, Yoshijima N, et al. Natural History of Leaflet Thrombosis After Transcatheter Aortic Valve Replacement: A 5-Year Follow-Up Study. J Am Heart Assoc. 2022;11(23):e026334. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Hein M, Schoechlin S, Schulz U, et al. Long-Term Follow-Up of Hypoattenuated Leaflet Thickening After Transcatheter Aortic Valve Replacement. JACC Cardiovasc Interv. 2022;15(11):1113–1122. [DOI] [PubMed] [Google Scholar]
- 15.Likosky DS, Van Parys J, Zhou W, Borden WB, Weinstein MC, Skinner JS. Association Between Medicare Expenditure Growth and Mortality Rates in Patients With Acute Myocardial Infarction: A Comparison From 1999 Through 2014. JAMA Cardiol. 2018;3(2):114–122. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8–27. [DOI] [PubMed] [Google Scholar]
- 17.Kim DH, Patorno E, Pawar A, Lee H, Schneeweiss S, Glynn RJ. Measuring Frailty in Administrative Claims Data: Comparative Performance of Four Claims-Based Frailty Measures in the U.S. Medicare Data. J Gerontol A Biol Sci Med Sci. 2020;75(6):1120–1125. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Kim DH, Schneeweiss S, Glynn RJ, Lipsitz LA, Rockwood K, Avorn J. Measuring Frailty in Medicare Data: Development and Validation of a Claims-Based Frailty Index. J Gerontol A Biol Sci Med Sci. 2018;73(7):980–987. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Thourani VH, Brennan JM, Edelman JJ, et al. Association of Volume and Outcomes in 234 556 Patients Undergoing Surgical Aortic Valve Replacement. Ann Thorac Surg. 2022;114(4):1299–1306. [DOI] [PubMed] [Google Scholar]
- 20.Ghoreishi M, Thourani VH, Badhwar V, et al. Less-Invasive Aortic Valve Replacement: Trends and Outcomes From The Society of Thoracic Surgeons Database. Ann Thorac Surg. 2021;111(4):1216–1223. [DOI] [PubMed] [Google Scholar]
- 21.Wei LM, Cook CC, Hayanga JWA, Rankin JS, Mascio CE, Badhwar V. Robotic Aortic Valve Replacement: First 50 Cases. Ann Thorac Surg. 2022;114(3):720–726. [DOI] [PubMed] [Google Scholar]
- 22.Okuno T, Alaour B, Heg D, et al. Long-Term Risk of Stroke After Transcatheter Aortic Valve Replacement: Insights From the SwissTAVI Registry. JACC Cardiovasc Interv. 2023;16(24):2986–2996. [DOI] [PubMed] [Google Scholar]
- 23.Zaid S, Hirji SA, Bapat VN, et al. Surgical Explantation of Failed Transcatheter Aortic Valve Replacement. Ann Thorac Surg. 2023;116(5):933–942. [DOI] [PubMed] [Google Scholar]
- 24.Tang GHL, Zaid S, Kleiman NS, et al. Explant vs Redo-TAVR After Transcatheter Valve Failure: Mid-Term Outcomes From the EXPLANTORREDO-TAVR International Registry. JACC Cardiovasc Interv. 2023;16(8):927–941. [DOI] [PubMed] [Google Scholar]
- 25.Mehaffey JH, Kawsara M, Jagadeesan V, et al. Atrial Fibrillation Management During Surgical vs Transcatheter Aortic Valve Replacement. Annals of Thoracic Surgery. 2024;In Press. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Beyersdorf F, Bauer T, Freemantle N, et al. Five-year outcome in 18 010 patients from the German Aortic Valve Registry. Eur J Cardiothorac Surg. 2021;60(5):1139–1146. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.