- Email: [email protected]
Comparison of Inhospital Outcomes of Surgical Aortic Valve Replacement in Hospitals With and Without Availability of a Transcatheter Aortic Valve Implantation Program (from a Nationally Representative Database)
Comparison of Inhospital Outcomes of Surgical Aortic Valve Replacement in Hospitals With and Without Availability of a Transcatheter Aortic Valve Implantation Program (from a Nationally Representative Database) Vikas Singh, MDa, Apurva O. Badheka, MDb,*, Samir V. Patel, MD, MPHc, Nileshkumar J. Patel, MDa, Badal Thakkar, MDd, Nilay Patel, MDe, Shilpkumar Arora, MD, MPHf, Nish Patel, MDa, Achint Patel, MDg, Chirag Savani, MDh, Abhijit Ghatak, MDa, Sidakpal S. Panaich, MDi, Sunny Jhamnani, MDb, Abhishek Deshmukh, MDj, Ankit Chothani, MDk, Rajesh Sonani, MDl, Aashay Patel, MDm, Parth Bhatt, MDe, Abhishek Dave, MDn, Ronak Bhimani, MDo, Tamam Mohamad, MDi, Cindy Grines, MDi, Michael Cleman, MDp, John K. Forrest, MDp, and Abeel Mangi, MDp We hypothesized that the availability of a transcatheter aortic valve implantation (TAVI) program in hospitals impacts the overall management of patients with aortic valve disease and hence may also improve postprocedural outcomes of conventional surgical aortic valve replacement (SAVR). The aim of the present study was to compare the inhospital outcomes of SAVR in centers with versus without availability of a TAVI program in an unrestricted large nationwide patient population >50 years of age. SAVRs performed on patients aged >50 years were identified from the Nationwide Inpatient Sample (NIS) for the years 2011 and 2012 using the International Classification of Diseases, Ninth Revision, Clinical Modification procedure codes. SAVR cases were divided into 2 categories: those performed at hospitals with a TAVI program (SAVR-TAVI) and those without (SAVR-non-TAVI). A total of 9,674 SAVR procedures were identified: 4,526 (46.79%) in the SAVR-TAVI group and 5,148 (53.21%) in SAVR-non-TAVI group. The mean age of the study population was 70.2 – 0.1 years with majority (53%) of the patients aged >70 years. The mean Charlson’s comorbidity score for patients in SAVR-TAVI group was greater (greater percentage of patients were aged >80 years, had hypertension, congestive heart failure, renal failure, and peripheral arterial disease) than that of patients in SAVR-non-TAVI group (1.6 vs 1.4, p <0.001). The propensity score matching analysis showed a statistically significant lower inhospital mortality (1.25% vs 1.72%, p [ 0.001) and complications rate (35.6% vs 37.3%, p [ 0.004) in SAVR-TAVI group compared to SAVR-non-TAVI group. The mean length of hospital stay was similar in the 2 groups the cost of hospitalization was higher in the SAVRTAVI group ($43,894 – 483 vs $41,032 – 473, p <0.0001). Having a TAVI program was a significant predictor of reduced mortality and complications rate after SAVR in multivariate analysis. In conclusion, this largest direct comparative analysis demonstrates that SAVRs performed in centers with a TAVI program are associated with significantly lower mortality and complications rates compared to those performed in centers without a TAVI program. Ó 2015 Elsevier Inc. All rights reserved. (Am J Cardiol 2015;116:1229e1236)
a Cardiology Department, University of Miami Miller School of Medicine, Miami, Florida; bInterventional Cardiology Department, The Everett Clinic, Everett, Washington; cInternal Medicine Department, Western Reserve Health System, Youngstown, Ohio; dEpidemiology Department, Tulane School of Public Health and Tropical Medicine, New Orleans, Louisiana; eInternal Medicine Department, Saint Peter’s University Hospital, New Brunswick, New Jersey; fInternal Medicine Department, Mount Sinai St. Luke’s Roosevelt Hospital, New York, New York; gPublic Health Department, Icahn School of Medicine at Mount Sinai, New York, New York; hEpidemiology Department, New York Medical College, Valhalla, New York; iCardiology Department, Detroit Medical Center, Detroit, Michigan; jCardiology Department, Mayo Clinic, Rochester, Minnesota; k Internal Medicine Department, MedStar Washington Hospital Center,
0002-9149/15/$ - see front matter Ó 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.amjcard.2015.07.039
Washington, DC; lInternal Medicine Department, Emory University School of Medicine, Atlanta, Georgia; mInternal Medicine Department, Lankenau Institute for Medical Research, Wynnewood, Pennsylvania; nPublic Health Department, Texas A&M Medical Centre, College Station, Texas; oInternal Medicine Department, St. Vincent Charity Medical Centre, Cleveland, Ohio; and pCardiology Department, Yale School of Medicine, New Haven, Connecticut. Manuscript received June 1, 2015; revised manuscript received and accepted July 12, 2015. Drs. Singh, Badheka, and Patel share equal contribution to this manuscript. See page 1235 for disclosure information. *Corresponding author: Tel: (408) 324-4516; fax: (203) 737-2437. E-mail address: [email protected] (A.O. Badheka). www.ajconline.org
1230
The American Journal of Cardiology (www.ajconline.org)
Table 1 Baseline characteristics of the study population: A) unmatched, B) propensity matched SAVR in TAVI vs Non-TAVI hospital of 2011 - 2012
Demographic Variables
A) Unmatched Group
Non-TAVI capable TAVI capable hospital hospital
Total No. of Observations 5148 (53.21%) 4526 (46.79%) (unweighted) Total No. of Observations 25234.7 (53.23%) 22176.2 (weighted) (46.77%) Total No. of Hospitals 823 (74.14%) 287 (25.86%) No. of SAVR procedures per hospital (Hospital Volume) for 2011 Mean þ/- SE 30.7 (0.4) 113.3 (1.8) Median (Q1, Q3) 25 (14, 47) 89 (49, 142) Age Mean þ/- SE(Years) 69.80 (0.1) 70.67 (0.2) 50-59 18.1% 17.9% 60-69 30.6% 28.0% 70-79 31.8% 29.8% >80 19.6% 24.3% Sex Male 57.1% 59.5% Female 42.9% 40.5% Race Whites 78.6% 76.5% Blacks 4.4% 4.7% Hispanics 5.6% 4.9% Others 4.0% 5.7% Missing 7.5% 8.2% Comorbidities Charlson Comorbidities Score (%) Mean þ/- SE 1.4 (0.02) 1.6 (0.02) 0 30.4% 25.6% 1 32.8% 33.8% More than or equal to 2 36.8% 40.6% Obesity 20.6% 17.1% Hypertension 73.4% 75.2% Diabetes Mellitus 29.2% 26.7% Congestive Heart Failure 25.4% 33.5% History of Chronic Pulmonary 20.7% 20.9% Disease Peripheral Vascular Disease 17.4% 22.6% Fluid-electrolyte abnormalities & 33.9% 41.8% Renal Failure Neurological disorder or paralysis 6.1% 6.2% Anemia or coagulopathy 40.4% 42.2% Depression, Psychosis, or 9.6% 9.5% Substance Abuse Median Household Income Category for patient’s Zip code 1. 0-25th percentile 21.7% 20.1% 2. 26-50th percentile 25.6% 20.9% 3. 51-75th percentile 27.7% 25.3% 4. 76-100th percentile 23.0% 32.1% missing 1.9% 1.6% Primary Payer Medicare 65.2% 65.1% Medicaid 2.9% 2.8% Private including HMOs & PPOs 28.2% 28.4% Other/Self-pay/No charge 3.6% 3.6% missing 0.2% 0.1% Hospital Characteristics
B) Matched Group (1-1 Matching) for age, gender, Charlson’s score, bedsize, location and teaching status, type of admission, weekend
Overall
P-value
9674 47,410.9
TAVI capable hospital
2678 (50%)
2678 (50%)
Overall
P-value
5356
13149.2 (50.25%) 13017.06 (49.75%) 26,166.3
1110 63.4 (0.1) 38 (19, 76) 70.20 (0.1) 18.0% 29.4% 30.8% 21.8%
Non-TAVI capable hospital
<0.001 <0.001 <0.001
541 (73.81%)
192 (26.19%)
35.0 (0.5) 32 (19, 54) 70.2 (0.2) 17% 30.5% 31.4% 21.1%
111.0 (1.0) 89 (49, 142) 70.4(0.2) 16.9% 30.8% 31.2% 21.2%
733 75.3 (1.3) <0.001 54 (26, 89) 70.3(0.1) 0.786 17.0% 0.973 30.7% 31.3% 21.1%
58.2% 41.8%
<0.001
58.9% 41.1%
59.0% 41.0%
59.0% 41.0%
77.6% 4.5% 5.3% 4.8% 7.9%
<0.001
76.6% 5.2% 5.3% 3.9% 8.9%
77.1% 4.9% 4.6% 5.2% 8.2%
76.9% <0.001 5.0% 5.0% 4.6% 8.6%
1.5 (0.01) 28.1% 33.3% 38.6% 19.0% 74.2% 28.0% 29.2% 20.8%
<0.001
<0.001 <0.001 <0.001 <0.001 0.599
19.8% 37.6%
<0.001 <0.001
18.3% 36.7%
21.6% 38.3%
20.0% <0.001 37.5% 0.011
6.1% 41.2% 9.6%
0.810 0.000 0.639
6.5% 40.4% 9.5%
5.9% 42.2% 9.6%
6.2% 41.3% 9.6%
21.0% 23.4% 26.6% 27.3% 1.8%
<0.001
20.0% 24.0% 30.2% 24.0% 1.9%
19.0% 20.5% 25.2% 34.1% 1.2%
19.5% <0.001 22.3% 27.7% 29.0% 1.6%
65.2% 2.9% 28.3% 3.6% 0.1%
0.867
66.5% 3.0% 26.7% 3.7% 0.1%
64.5% 2.8% 29.6% 3.0% 0.1%
65.5% <0.001 2.9% 28.1% 3.4% 0.1%
1.40 (0.03) 29.8% 33.2% 37.0% 20.5% 73.0% 29.1% 24.2% 20.3%
1.46 (0.03) 29.8% 32.7% 37.5% 16.6% 74.3% 25.5% 31.9% 20.6%
0.897
1.43 (0.02) 0.524 29.8% 0.602 33.0% 37.2% 18.5% <0.001 73.7% 0.019 27.3% <0.001 28.0% <0.001 20.4% 0.501
0.036 0.003 0.767
Valvular Heart Disease/Inhospital Outcomes of SAVR
1231
Table 1 (continued) SAVR in TAVI vs Non-TAVI hospital of 2011 - 2012
Demographic Variables
A) Unmatched Group
Non-TAVI capable TAVI capable hospital hospital
B) Matched Group (1-1 Matching) for age, gender, Charlson’s score, bedsize, location and teaching status, type of admission, weekend
Overall
Bed size of Hospital depending on Location & Teaching Status Small 10.4% 2.0% 6.5% Medium 22.2% 15.1% 18.9% Large 67.1% 82.9% 74.5% Hospital Location & Teaching Status Rural 4.0% 0.7% 2.5% Urban Non-teaching 41.0% 9.7% 26.3% Urban Teaching 54.7% 89.6% 71.0% Hospital Region Northeast 18.0% 27.4% 22.4% Midwest 18.3% 20.6% 19.4% South 25.2% 30.8% 27.8% West 22.5% 14.8% 18.9% Type of Admission Non-elective 16.3% 17.2% 16.7% Elective 83.5% 82.6% 83.1% Admission Day Weekdays 97.1% 96.0% 96.6% Weekends 2.9% 4.0% 3.4% Disposition Home 39.2% 37.9% 38.6% Home Health Care 34.9% 38.3% 36.5% Transfer to Short-term 23.9% 22.3% 23.1% Hospital/other facilities Length of hospital-stay - Median 6 (5, 8) 7 (5, 9) 6 (5, 9) (Quartile 1 , 3), days Cost ($)* (Mean, SE) 41831.72 (339) 44840.75 (433) 43198.76 (270)
P-value
Non-TAVI capable hospital
TAVI capable hospital
Overall
P-value
<0.001
3.4% 24.8% 71.8%
3.5% 25.7% 70.9%
3.4% 25.2% 71.3%
0.275
<0.001
1.2% 16.0% 82.9%
1.2% 16.3% 82.5%
1.2% 16.1% 82.7%
0.694
<0.001
23.4% 18.1% 25.5% 20.2%
30.5% 18.1% 28.1% 17.6%
26.9% <0.001 18.1% 26.8% 18.9%
0.001
14.8% 85.2%
14.5% 85.5%
14.6% 85.4%
0.501
<0.001
97.1% 2.9%
97.0% 3.0%
97.0% 3.0%
0.821
<0.001
39.4% 35.3% 23.3%
36.9% 41.2% 20.7%
38.2% <0.001 38.2% 22.0%
6 (5, 8)
6 (5, 9)
41032.3 (473)
43894.14 (483)
<0.001
6 (5, 9) 42388.19 <0.001 (339)
SAVR-non-TAVI ¼ surgical aortic valve replacement performed in non-TAVI centers; SAVR-TAVI ¼ surgical aortic valve replacement performed in TAVI centers. * Cost calculated for 2012 after adjusting for inflation.
Transcatheter aortic valve implantation (TAVI) has emerged as a novel therapeutic option for aortic valvular stenosis in high-risk surgical patients with severe co-morbidities.1 The technology has rapidly gained worldwide acceptance, and since the Food and Drug Administration approval in November 2011, it has disseminated to a large number of hospitals throughout the United States. TAVI requires a complex decision-making and active involvement of an integrated “Heart Team” comprising physicians with diverse expertise. We hypothesized that the availability of a TAVI program in hospitals impacts the overall management of patients with aortic valve disease and hence may also improve postprocedural outcomes of conventional surgical aortic valve replacement (SAVR). The aim of the present study was to compare the inhospital outcomes of SAVR in centers with versus without availability of a TAVI program in an unrestricted large nationwide patient population aged >50 years.
Methods Data were obtained from the Nationwide Inpatient Sample (NIS) for the years 2011 and 2012. NIS is a part of a family of databases developed for the Healthcare Cost and Utilization Project (HCUP) and is sponsored by the Agency for Healthcare Research and Quality (AHRQ). Data from the NIS have previously been used to identify, track, and analyze national trends in health care usage, patterns of major procedures, access, disparity of care, trends in hospitalizations, charges, quality, and outcomes.2e5 We analyzed data from NIS 2011 and 2012 using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) procedure codes for implantation of a bioprosthetic (code 35.21) or mechanical (code 35.22) aortic valve. Only patients aged >50 years with isolated SAVR were included and those who underwent concomitant coronary artery bypass graft surgery, other valvular, or
1232
The American Journal of Cardiology (www.ajconline.org)
Table 2 In-hospital mortality and complications of SAVR in the study population: A) unmatched, B) propensity matched SAVR in TAVI vs Non-TAVI center
Complications (%) Death Any Complications Any Complications þ Death Stroke Transmural MI Deep sternal wound infection Hemorrhage requiring transfusion Cardiac Complications Iatrogenic Cardiovascular Complications Pericardial complications Respiratory failure Sepsis Acute Kidney Injury requiring Dialysis PE and DVT Anesthetics Complications Cardiac Device/Prosthetics/ Graft Complications
A) Unmatched dataset
B) Matched Group (1-1 Matching) for age, gender, Charlson’s score, bedsize, location and teaching status, type of admission, weekend
Non-TAVI capable
TAVI capable
Overall
P-value
Non-TAVI capable
TAVI capable
Overall
P-value
1.8% 36.0% 36.5% 0.9% 0.9% 0.6% 19.8% 12.8% 11.9% 0.9% 5.6% 1.9% 1.2% 0.7% 0.0% 0.5%
1.5% 36.5% 36.9% 1.2% 1.0% 0.6% 18.4% 13.2% 12.5% 0.6% 6.2% 2.9% 0.9% 1.0% 0.0% 0.5%
1.7% 36.2% 36.7% 1.0% 0.9% 0.6% 19.1% 13.0% 12.2% 0.8% 5.9% 2.4% 1.0% 0.8% 0.0% 0.5%
0.036 0.204 0.375 0.007 0.151 0.610 0.000 0.203 0.085 0.000 0.011 <0.001 0.003 <0.001 NA 0.752
1.7% 37.3% 38.0% 1.0% 0.9% 0.6% 22.2% 12.5% 11.8% 0.8% 5.2% 2.1% 1.1% 0.5% 0.0% 0.3%
1.3% 35.6% 35.9% 0.9% 1.1% 0.5% 18.0% 13.3% 12.7% 0.5% 5.6% 2.6% 0.8% 1.2% 0.0% 0.6%
1.5% 36.5% 36.9% 1.0% 1.0% 0.6% 20.1% 12.9% 12.3% 0.6% 5.4% 2.4% 1.0% 0.9% 0.0% 0.5%
0.002 0.004 0.001 0.803 0.139 0.186 <0.001 0.048 0.021 0.007 0.099 0.004 0.018 <0.001 NA 0.000
SAVR-non-TAVI ¼ surgical aortic valve replacement performed in non-TAVI centers; SAVR-TAVI ¼ surgical aortic valve replacement performed in TAVI centers; MI ¼ myocardial Infarction; PE ¼ pulmonary embolism; DVT ¼ deep venous thrombosis.
pericardial procedures were excluded from the analysis. The remaining SAVR cases were divided into 2 categories: those performed at hospitals with a TAVI program (SAVR-TAVI) and those without (SAVR-non-TAVI). The TAVI-capable hospitals were identified with ICD-9-CM procedure codes of 35.05 and 35.06. We defined severity of co-morbid conditions using Deyo modification of Charlson Comorbidity Index (CCI).6 This index contains 17 co-morbid conditions with differential weights. The score ranges from 0 to 33, with greater scores corresponding to greater burden of co-morbid diseases. The primary outcome was all-cause inhospital mortality. Procedural complications were identified by Patient Safety Indicators (PSIs), version 4.4, March 2012, which have been established by the AHRQ to monitor preventable adverse events during hospitalization. These indicators are based on ICD-9-CM codes and Medicare severity Diagnosis Related Groups, and each PSI has specific inclusion and exclusion criteria.6,7 This method of identifying patients undergoing procedures, co-morbid conditions, and associated complications has previously been used in several studies.2,3 Other outcomes studied were the length of hospital stay and cost of hospitalization. To estimate cost of hospitalization, the NIS data were merged with cost-to-charge ratios available from the HCUP. We estimated the cost of each inpatient stay by multiplying the total hospital charge with cost-to-charge ratios. Cost for each year was calculated in terms of the 2012 cost, after adjusting for inflation according to the latest consumer price index (CPI) data released by US government on January 16, 2013. Stata IC 11.0 (Stata-Corp, College Station, Texas) and SAS 9.3 (SAS Institute Inc, Cary, North Carolina) was used for analyses. Differences between categorical variables were
tested using the chi-square test and differences between continuous variables were tested using the Student t test, continuous variables with Gaussian distributions, and KruskaleWallis rank sum tests for continuous variables with nonGaussian distributions. Multivariate logistic regression was generated to identify the independent multivariate predictors of inhospital mortality. All interactions were thoroughly tested. Collinearity was assessed using variance inflation factor. We used propensity-scoring method to establish matched cohorts to control for imbalances of patients’ and hospitals’ characteristics between the 2 groups that may have influenced the primary outcome. A propensity score was assigned to each hospitalization. This was based on multivariate logistic regression model that examined the impact of 12 variables (patient demographics, co-morbidities, and hospital characteristics) on the likelihood of treatment assignment. Patients with similar propensity score in the 2 treatment groups were matched using a 1 to 1 scheme without replacement using greedy algorithm. Results A total of 9,674 SAVR procedures (which translates to an estimated 47,410 procedures performed in 1,110 hospitals) were identified of which 4,526 (46.79%) were performed in hospitals with availability of TAVI (SAVRTAVI group) and 5,148 (53.21%) in non-TAVI hospitals (SAVR-non-TAVI group). Table 1 demonstrates the baseline characteristics of the study population. The mean age of the study population was 70.2 0.1 years with majority (53%) of the patients aged >70 years; 58% were men and 78% were whites. There were significant
Valvular Heart Disease/Inhospital Outcomes of SAVR
1233
Figure 1. Complication rates during SAVR in TAVI versus non-TAVI centers in the propensity scoreematched cohort.
Table 3 Multivariate predictors of mortality and any complications Variable
Multivariate Simple logistic regression for In Hospital Mortality (Weighted) Unmatched Dataset
Multivariate Simple logistic regression for Any Complications (Weighted)
Propensity matched Dataset
Unmatched Dataset
Propensity matched Dataset
Odds Ratio LL UL P-Value Odds Ratio LL UL P-Value Odds Ratio LL UL P-Value Odds Ratio LL UL P-Value (95% CI) (95% CI) (95% CI) (95% CI) Status of TAVI Capability Non-TAVI capable Hospital TAVI Capable Hospital Age (in 10 year increment) Sex Male Female Charlson Score 0 1 More than or equal to 2 Bed size of Hospital depending on Location & Teaching Status Small Medium Large Type of Admission Non-elective Elective Admission Day Weekdays Weekends
1 0.79 1.26
Ref Ref Ref 0.68 0.92 0.00 1.18 1.36 <0.001
1 0.72 1.06
Ref Ref 0.58 0.88 0.96 1.17
Ref 0.00 0.27
1 0.98 1.21
Ref Ref Ref 0.94 1.02 0.25 1.19 1.23 <0.001
1 0.93 1.21
Ref Ref Ref 0.88 0.97 0.00 1.18 1.24 <0.001
1.00 0.95
Ref Ref 0.82 1.10
1.00 0.74
Ref Ref 0.60 0.92
Ref 0.01
1.00 0.97
Ref Ref 0.93 1.01
1.00 0.92
Ref Ref 0.88 0.97
1.00 1.23 3.54
Ref Ref Ref 0.94 1.61 0.13 2.82 4.45 <0.001
1.00 1.46 5.21
Ref Ref Ref 0.96 2.22 0.08 3.63 7.49 <0.001
1.00 1.11 1.60
Ref Ref Ref 1.05 1.17 <0.001 1.52 1.68 <0.001
1.00 1.12 1.72
Ref Ref Ref 1.05 1.20 0.00 1.61 1.83 <0.001
1.00 0.86 0.97
Ref Ref 0.62 1.18 0.73 1.29
1.00 0.51 0.62
Ref Ref 0.30 0.85 0.39 1.01
1.00 1.16 1.12
Ref Ref 1.07 1.27 1.04 1.22
1.00 1.01 1.03
Ref Ref 0.87 1.17 0.90 1.19
1.00 0.33
Ref Ref Ref 0.28 0.38 <0.001
1.00 0.26
Ref Ref Ref 0.21 0.32 <0.001
1 0.67
Ref Ref Ref 0.64 0.70 <0.001
1 0.66
Ref Ref Ref 0.61 0.71 <0.001
1.00 1.24
Ref Ref 0.94 1.64
1.00 0.96
Ref Ref 0.64 1.44
1.00 1.12
Ref Ref 1.01 1.24
1.00 1.34
Ref Ref 1.16 1.56
Ref 0.45
Ref 0.34 0.83
Ref 0.12
differences between the baseline characteristics of patients in the 2 groups (Table 1). A greater percentage of patients in SAVR-TAVI group were aged >80 years, had hypertension, congestive heart failure, renal failure, and peripheral arterial disease, whereas more patients in SAVR-nonTAVI group had diabetes and obesity. The mean
Ref 0.01 0.05
Ref 0.84
Ref 0.09
Ref 0.00 0.00
Ref 0.04
Ref 0.00
Ref 0.95 0.67
Ref 0.00
Charlson co-morbidity score for patients in SAVR-TAVI group was greater than that of patients in SAVR-nonTAVI group (1.6 vs 1.4, p <0.001). The outcomes of both groups are reported in Table 2. Propensity scoreematched analysis is reported in Table 1. In this analysis, multiple patient- and hospital-level
1234
The American Journal of Cardiology (www.ajconline.org)
variables that could have affected the results were adjusted for. These variables included age, gender, Charlson comorbidity score, bed size of the hospital, location of the hospital and teaching status, type, and timing of admissions (emergent vs nonemergent and weekend vs nonweekend). The propensity scoreematching analysis showed a statistically significant lower inhospital mortality (1.25% vs 1.72%, p ¼ 0.001) and complications rates (35.6% vs 37.3%, p ¼ 0.004) in SAVR-TAVI group compared to SAVR-non-TAVI group (Figure 1). The mean length of hospital stay was similar in the 2 groups; however, the cost of hospitalization was greater in the SAVR-TAVI group. In the multivariate analysis of propensity scoreematched population, having a TAVI program in the hospital was a significant predictor of reduced mortality (odds ratio [OR] 0.71, 95% confidence interval [CI] 0.6 to 0.8, p ¼ 0.0015), whereas a greater burden of co-morbidities (CCI >2) was an independent predictor of higher inhospital mortality (OR 5.2, 95% CI 3.6 to 7.5, p <0.0001) (Table 3). Similarly, having a TAVI program was an independent predictor of reduced complication rates in patients undergoing SAVR (Table 3). In another multivariate model that included hospital SAVR volume, presence of a TAVI program was a significant predictor of reduced mortality and complication rates (Supplementary Table 1). Discussion In this analysis of patients aged >50 years who underwent SAVR in the United States, we demonstrate that inhospital mortality and complication rates of SAVRs performed in centers with a TAVI program is significantly less than those performed in centers without a TAVI program. Despite having greater burden of risk factors and comorbidities, SAVRs performed in TAVI centers were associated with improved outcomes. This effect of TAVI on inhospital mortality and complication rates was also reproducible in the propensity scoreematched analysis and multivariate models. In 2012, the Society for Cardiovascular Angiography and Interventions, American Association for Thoracic Surgery, American College of Cardiology Foundation, and the Society of Thoracic Surgeons (STS) provided specific recommendations for institutional requirements for a TAVI program.8 It is recommended that institutions perform 50 SAVRs/year (at least 10 high risk with STS score 6).8 The multidisciplinary TAVI team must include a minimum of 2 institutional-based cardiac surgeons each of whom should have performed 100 SAVRs in their career, at least 10 of which are “high risk” (STS score 6) or 25 SAVRs/year or 50 SAVRs in 2 years.8 On the basis of these recommendations, TAVI centers are generally greater surgical volume centers and have highly experienced cardiothoracic surgeons which may explain the associated improved outcomes of SAVR performed in these centers. We noted a significantly high procedural volume (median Q1 to Q3) of 89 (49 to 142) versus 25 (14 to 47) in the SAVR-TAVI group compared to SAVR-non-TAVI group. The positive effect of larger hospital volume on reduction of inhospital mortality rate associated with surgical procedures including SAVR has been well reported in the previous medical reports.9,10
In an analysis of 6,270 procedures (3,487 SAVR and 2,783 SAVR þ coronary artery bypass grafting) using the statewide cardiothoracic surgical database (2008 to 2011) from Michigan, Patel et al11 reported a strong volumeeoutcome relation for SAVR particularly in high-risk patients. However, we noted a positive effect of TAVI program in the outcomes of SAVRs even in the multivariate models (with or without inclusion of hospital volume) and in the propensity scoreematched analysis in which the 2 groups were matched for multiple patient- and hospital-level variables including hospital size and teaching status. The introduction of TAVI program has been associated with a “halo effect” on SAVR in terms of referral and procedural volume especially for both low- and high-risk patients.12,13 Grant et al13 reported their single-center (UK) experience of initiating a TAVI program in a retrospective analysis of 815 consecutive patients undergoing isolated SAVR or coronary artery bypass grafting plus SAVR. Within the 2 years of introduction of TAVI program, they noted a 37% increase in surgical procedures.13 Interestingly, despite an increase in the mean logistic EuroSCORE (7.4 vs 7.9), they noted a decrease in the hospital mortality rate (2.9% vs 2.1%; p ¼ 0.48).13 An increased referral of low surgical risk patients to a TAVI center may also contribute to the improved SAVR outcomes. Inclusion of a large sample size in our study (>9,600 SAVR procedures) may also have enabled us to detect these differences compared to previous studies that may have been underpowered.14 More recently, Brennan et al15 used data from the STS/ American College of Cardiology transcatheter valve therapies registry to examine the association of TAVI availability in a hospital and overall AVR volumes (SAVR þ TAVI) and mortality from 2008 to 2013. During the study period, they noted that AVR (SAVR þ TAVI) inhospital mortality rate decreased from 3.4% to 2.9% (p <0.0001); however, the inhospital mortality rate for TAVI increased from 3.5% to 5.2% (O-to-E ratio 0.41 to 0.60, p ¼ 0.02).15 In that study, the inhospital mortality rate for SAVR decreased from 3.4% to 2.5% during the study period of 5 years, with the greatest absolute reduction in high-risk SAVR cases (11.7% to 9.3%). It is important to note that large-scale studies using STS database have previously shown that the outcomes of SAVR have been steadily improving in the United States despite worsening patient profiles since early 1990s, long before TAVI became available.16e18 Our present study was designed to eliminate the effect of temporal changes and specifically assess the impact of TAVI on SAVR (regardless of the risk profile) soon after TAVI was introduced (2011 to 2012) into the management of aortic valve disease. The outcomes of SAVR from NIS database have previously been shown to correlate with those from STS database.19 The propensity scoreematched analysis in our study also showed a benefit in terms of both mortality and complications when an SAVR is performed in a TAVI-capable center (absolute reduction of 0.47% for mortality and 1.7% for complications). The reduced complications rate in the SAVR-TAVI group (35.6% vs 37.3%) was small but statistically significant (p ¼ 0.004) and driven by a reduction in bleeding complications. No large-scale study has previously
Valvular Heart Disease/Inhospital Outcomes of SAVR
assessed differences in complications after SAVR in TAVI and non-TAVI center; however, in a study performed at a single center comparing outcomes before and after initiation of a TAVI program, Malaisrie et al12 noted a reduction in complications (43.8% vs 44.9%) after TAVI program. The salutary effect of TAVI may also be explained because of involvement of the multidisciplinary “heart team” in the management of complex valvular disorders. The rationale for team-based care is to improve clinical outcomes by facilitating shared decision-making in a timely and coordinated manner.20,21 Heart team associated with a TAVI center may improve inhospital outcomes of SAVR by optimizing patient selection, procedural performance, and complications management.22 This high-quality care however comes at a greater cost as witnessed in our study. The cost of SAVRs performed in TAVI centers was significantly greater compared to those in non-TAVI centers ($43,894 483 vs $41,032 473, p <0.0001) despite a similar length of hospital stay. The greatest cost of care may however be offset by a reduction in mortality and complications seen in the SAVR-TAVI group. The study limitations are inherent to the post hoc analysis of an administrative database. We have provided possible hypothesis for improved outcomes in the SAVT-TAVI group; however, the design of our study restricts any inference of causality or temporal associations. Compared to some of the previous studies, we lack information on the indication for SAVR, disease severity, cardiac functional class, frailty, and STS score. In 2012, NIS was redesigned to include a sample of discharges instead of a sample of hospitals, and therefore, hospital procedural volume could only be estimated for 2011. In addition, NIS does not provide any follow-up data. Although acknowledging these limitations, the present study has important strengths including a large unrestricted sample size from all-comers real-world data for patients undergoing SAVR and the use of standardized definitions of preventable adverse events/complications that are established by the AHRQ. Our study represents the largest direct comparative analysis of SAVR in TAVI and non-TAVI centers and adds to the growing reports on the “halo effect” of establishing a TAVR program in an institution. Disclosures
3.
4.
5.
6.
7. 8.
9. 10.
11.
12.
None of the authors have anything to disclose. 13.
Supplementary Data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.amjcard.2015.07.039. 1. Smith CR, Leon MB, Mack MJ, Miller DC, Moses JW, Svensson LG, Tuzcu EM, Webb JG, Fontana GP, Makkar RR, Williams M, Dewey T, Kapadia S, Babaliaros V, Thourani VH, Corso P, Pichard AD, Bavaria JE, Herrmann HC, Akin JJ, Anderson WN, Wang D, Pocock SJ. Transcatheter versus surgical aortic-valve replacement in high-risk patients. N Engl J Med 2011;364:2187e2198. 2. Badheka AO, Patel NJ, Grover P, Singh V, Patel N, Arora S, Chothani A, Mehta K, Deshmukh A, Savani GT, Patel A, Panaich SS, Shah N, Rathod A, Brown M, Mohamad T, Tamburrino FV, Kar S, Makkar R, O’Neill WW, De Marchena E, Schreiber T, Grines CL,
14.
15.
16.
1235
Rihal CS, Cohen MG. Impact of annual operator and institutional volume on percutaneous coronary intervention outcomes: a 5-year United States experience (2005-2009). Circulation 2014;130: 1392e1406. Singh V, Patel NJ, Badheka AO, Arora S, Patel N, Macon C, Savani GT, Manvar S, Patel J, Thakkar B, Panchal V, Solanki S, Chothani A, Panaich SS, Ram V, Kliger CA, Schreiber T, O’ Neill W, Cohen MG, Alfonso CE, Grines CL, Mangi A, Pfau S, Forrest JK, Cleman M, Makkar R. Comparison of outcomes of balloon aortic valvuloplasty plus percutaneous coronary intervention versus percutaneous aortic balloon valvuloplasty alone during the same hospitalization in the United States. Am J Cardiol 2015;115:480e486. Patel NJ, Deshmukh A, Pant S, Singh V, Patel N, Arora S, Shah N, Chothani A, Savani GT, Mehta K, Parikh V, Rathod A, Badheka AO, Lafferty J, Kowalski M, Mehta JL, Mitrani RD, Viles-Gonzalez JF, Paydak H. Contemporary trends of hospitalization for atrial fibrillation in the United States, 2000 through 2010: implications for healthcare planning. Circulation 2014;129:2371e2379. Deshmukh A, Patel NJ, Pant S, Shah N, Chothani A, Mehta K, Grover P, Singh V, Vallurupalli S, Savani GT, Badheka A, Tuliani T, Dabhadkar K, Dibu G, Reddy YM, Sewani A, Kowalski M, Mitrani R, Paydak H, Viles-Gonzalez JF. In-hospital complications associated with catheter ablation of atrial fibrillation in the United States between 2000 and 2010: analysis of 93 801 procedures. Circulation 2013;128: 2104e2112. McDonald KM, Romano PS, Geppert J, Davies SM, Duncan BW, Shojania KG, Hansen A. Measures of patient safety based on hospital administrative data—the patient safety indicators, Structured Abstract; Vol 02-0038. Rockville, MD. Agency Healthc Reseach Qual 2002. Romano PS, Geppert JJ, Davies S, Miller MR, Elixhauser A, McDonald KM. A national profile of patient safety in U.S. hospitals. Health Aff (Millwood) 2003;22:154e166. Tommaso CL, Bolman RM 3rd, Feldman T, Bavaria J, Acker MA, Aldea G, Cameron DE, Dean LS, Fullerton D, Hijazi ZM, Horlick E, Miller DC, Moon MR, Ringel R, Ruiz CE, Trento A, Weiner BH, Zahn EM. Multisociety (AATS, ACCF, SCAI, and STS) expert consensus statement: operator and institutional requirements for transcatheter valve repair and replacement, part 1: transcatheter aortic valve replacement. J Thorac Cardiovasc Surg 2012;143: 1254e1263. Birkmeyer JD, Siewers AE, Finlayson EV, Stukel TA, Lucas FL, Batista I, Welch HG, Wennberg DE. Hospital volume and surgical mortality in the United States. N Engl J Med 2002;346:1128e1137. Hannan EL, O’Donnell JF, Kilburn H Jr, Bernard HR, Yazici A. Investigation of the relationship between volume and mortality for surgical procedures performed in New York State hospitals. JAMA 1989;262:503e510. Patel HJ, Herbert MA, Drake DH, Hanson EC, Theurer PF, Bell GF, Prager RL. Aortic valve replacement: using a statewide cardiac surgical database identifies a procedural volume hinge point. Ann Thorac Surg 2013;96:1560e1565; discussion 1565e1566. Malaisrie SC, Tuday E, Lapin B, Wang E, Lee R, McGee EC, Davidson C, McCarthy PM. Transcatheter aortic valve implantation decreases the rate of unoperated aortic stenosis. Eur J Cardiothorac Surg 2011;40:43e48. Grant SW, Devbhandari MP, Grayson AD, Dimarakis I, Kadir I, Saravanan DM, Levy RD, Ray SG, Bridgewater B. What is the impact of providing a transcatheter aortic valve implantation service on conventional aortic valve surgical activity: patient risk factors and outcomes in the first 2 years. Heart 2010;96:1633e1637. Malaisrie SC, McCarthy PM, McGee EC, Lee R, Rigolin VH, Davidson CJ, Beohar N, Lapin B, Subacius H, Bonow RO. Contemporary perioperative results of isolated aortic valve replacement for aortic stenosis. Ann Thorac Surg 2010;89:751e756. Brennan JM, Holmes DR, Sherwood MW, Edwards FH, Carroll JD, Grover FL, Tuzcu EM, Thourani V, Brindis RG, Shahian DM, Svensson LG, O’Brien SM, Shewan CM, Hewitt K, Gammie JS, Rumsfeld JS, Peterson ED, Mack MJ. The association of transcatheter aortic valve replacement availability and hospital aortic valve replacement volume and mortality in the United States. Ann Thorac Surg 2014;98:2016e2022; discussion 2022. Barreto-Filho JA, Wang Y, Dodson JA, Desai MM, Sugeng L, Geirsson A, Krumholz HM. Trends in aortic valve replacement for
1236
The American Journal of Cardiology (www.ajconline.org)
elderly patients in the United States, 1999-2011. JAMA 2013;310: 2078e2085. 17. Brown JM, O’Brien SM, Wu C, Sikora JA, Griffith BP, Gammie JS. Isolated aortic valve replacement in North America comprising 108, 687 patients in 10 years: changes in risks, valve types, and outcomes in the Society of Thoracic Surgeons National Database. J Thorac Cardiovasc Surg 2009;137:82e90. 18. Lee R, Li S, Rankin JS, O’Brien SM, Gammie JS, Peterson ED, McCarthy PM, Edwards FH. Fifteen-year outcome trends for valve surgery in North America. Ann Thorac Surg 2011;91:677e684; discussion p 684.
19. Astor BC, Kaczmarek RG, Hefflin B, Daley WR. Mortality after aortic valve replacement: results from a nationally representative database. Ann Thorac Surg 2000;70:1939e1945. 20. Holmes DR Jr, Rich JB, Zoghbi WA, Mack MJ. The heart team of cardiovascular care. J Am Coll Cardiol 2013;61:903e907. 21. Nallamothu BK, Cohen DJ. No “i” in Heart Team: incentivizing multidisciplinary care in cardiovascular medicine. Circ Cardiovasc Qual Outcomes 2012;5:410e413. 22. Gonzalez AA, Dimick JB, Birkmeyer JD, Ghaferi AA. Understanding the volume-outcome effect in cardiovascular surgery: the role of failure to rescue. JAMA Surg 2014;149:119e123.
a Cardiology Department, University of Miami Miller School of Medicine, Miami, Florida; bInterventional Cardiology Department, The Everett Clinic, Everett, Washington; cInternal Medicine Department, Western Reserve Health System, Youngstown, Ohio; dEpidemiology Department, Tulane School of Public Health and Tropical Medicine, New Orleans, Louisiana; eInternal Medicine Department, Saint Peter’s University Hospital, New Brunswick, New Jersey; fInternal Medicine Department, Mount Sinai St. Luke’s Roosevelt Hospital, New York, New York; gPublic Health Department, Icahn School of Medicine at Mount Sinai, New York, New York; hEpidemiology Department, New York Medical College, Valhalla, New York; iCardiology Department, Detroit Medical Center, Detroit, Michigan; jCardiology Department, Mayo Clinic, Rochester, Minnesota; k Internal Medicine Department, MedStar Washington Hospital Center,
0002-9149/15/$ - see front matter Ó 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.amjcard.2015.07.039
Washington, DC; lInternal Medicine Department, Emory University School of Medicine, Atlanta, Georgia; mInternal Medicine Department, Lankenau Institute for Medical Research, Wynnewood, Pennsylvania; nPublic Health Department, Texas A&M Medical Centre, College Station, Texas; oInternal Medicine Department, St. Vincent Charity Medical Centre, Cleveland, Ohio; and pCardiology Department, Yale School of Medicine, New Haven, Connecticut. Manuscript received June 1, 2015; revised manuscript received and accepted July 12, 2015. Drs. Singh, Badheka, and Patel share equal contribution to this manuscript. See page 1235 for disclosure information. *Corresponding author: Tel: (408) 324-4516; fax: (203) 737-2437. E-mail address: [email protected] (A.O. Badheka). www.ajconline.org
1230
The American Journal of Cardiology (www.ajconline.org)
Table 1 Baseline characteristics of the study population: A) unmatched, B) propensity matched SAVR in TAVI vs Non-TAVI hospital of 2011 - 2012
Demographic Variables
A) Unmatched Group
Non-TAVI capable TAVI capable hospital hospital
Total No. of Observations 5148 (53.21%) 4526 (46.79%) (unweighted) Total No. of Observations 25234.7 (53.23%) 22176.2 (weighted) (46.77%) Total No. of Hospitals 823 (74.14%) 287 (25.86%) No. of SAVR procedures per hospital (Hospital Volume) for 2011 Mean þ/- SE 30.7 (0.4) 113.3 (1.8) Median (Q1, Q3) 25 (14, 47) 89 (49, 142) Age Mean þ/- SE(Years) 69.80 (0.1) 70.67 (0.2) 50-59 18.1% 17.9% 60-69 30.6% 28.0% 70-79 31.8% 29.8% >80 19.6% 24.3% Sex Male 57.1% 59.5% Female 42.9% 40.5% Race Whites 78.6% 76.5% Blacks 4.4% 4.7% Hispanics 5.6% 4.9% Others 4.0% 5.7% Missing 7.5% 8.2% Comorbidities Charlson Comorbidities Score (%) Mean þ/- SE 1.4 (0.02) 1.6 (0.02) 0 30.4% 25.6% 1 32.8% 33.8% More than or equal to 2 36.8% 40.6% Obesity 20.6% 17.1% Hypertension 73.4% 75.2% Diabetes Mellitus 29.2% 26.7% Congestive Heart Failure 25.4% 33.5% History of Chronic Pulmonary 20.7% 20.9% Disease Peripheral Vascular Disease 17.4% 22.6% Fluid-electrolyte abnormalities & 33.9% 41.8% Renal Failure Neurological disorder or paralysis 6.1% 6.2% Anemia or coagulopathy 40.4% 42.2% Depression, Psychosis, or 9.6% 9.5% Substance Abuse Median Household Income Category for patient’s Zip code 1. 0-25th percentile 21.7% 20.1% 2. 26-50th percentile 25.6% 20.9% 3. 51-75th percentile 27.7% 25.3% 4. 76-100th percentile 23.0% 32.1% missing 1.9% 1.6% Primary Payer Medicare 65.2% 65.1% Medicaid 2.9% 2.8% Private including HMOs & PPOs 28.2% 28.4% Other/Self-pay/No charge 3.6% 3.6% missing 0.2% 0.1% Hospital Characteristics
B) Matched Group (1-1 Matching) for age, gender, Charlson’s score, bedsize, location and teaching status, type of admission, weekend
Overall
P-value
9674 47,410.9
TAVI capable hospital
2678 (50%)
2678 (50%)
Overall
P-value
5356
13149.2 (50.25%) 13017.06 (49.75%) 26,166.3
1110 63.4 (0.1) 38 (19, 76) 70.20 (0.1) 18.0% 29.4% 30.8% 21.8%
Non-TAVI capable hospital
<0.001 <0.001 <0.001
541 (73.81%)
192 (26.19%)
35.0 (0.5) 32 (19, 54) 70.2 (0.2) 17% 30.5% 31.4% 21.1%
111.0 (1.0) 89 (49, 142) 70.4(0.2) 16.9% 30.8% 31.2% 21.2%
733 75.3 (1.3) <0.001 54 (26, 89) 70.3(0.1) 0.786 17.0% 0.973 30.7% 31.3% 21.1%
58.2% 41.8%
<0.001
58.9% 41.1%
59.0% 41.0%
59.0% 41.0%
77.6% 4.5% 5.3% 4.8% 7.9%
<0.001
76.6% 5.2% 5.3% 3.9% 8.9%
77.1% 4.9% 4.6% 5.2% 8.2%
76.9% <0.001 5.0% 5.0% 4.6% 8.6%
1.5 (0.01) 28.1% 33.3% 38.6% 19.0% 74.2% 28.0% 29.2% 20.8%
<0.001
<0.001 <0.001 <0.001 <0.001 0.599
19.8% 37.6%
<0.001 <0.001
18.3% 36.7%
21.6% 38.3%
20.0% <0.001 37.5% 0.011
6.1% 41.2% 9.6%
0.810 0.000 0.639
6.5% 40.4% 9.5%
5.9% 42.2% 9.6%
6.2% 41.3% 9.6%
21.0% 23.4% 26.6% 27.3% 1.8%
<0.001
20.0% 24.0% 30.2% 24.0% 1.9%
19.0% 20.5% 25.2% 34.1% 1.2%
19.5% <0.001 22.3% 27.7% 29.0% 1.6%
65.2% 2.9% 28.3% 3.6% 0.1%
0.867
66.5% 3.0% 26.7% 3.7% 0.1%
64.5% 2.8% 29.6% 3.0% 0.1%
65.5% <0.001 2.9% 28.1% 3.4% 0.1%
1.40 (0.03) 29.8% 33.2% 37.0% 20.5% 73.0% 29.1% 24.2% 20.3%
1.46 (0.03) 29.8% 32.7% 37.5% 16.6% 74.3% 25.5% 31.9% 20.6%
0.897
1.43 (0.02) 0.524 29.8% 0.602 33.0% 37.2% 18.5% <0.001 73.7% 0.019 27.3% <0.001 28.0% <0.001 20.4% 0.501
0.036 0.003 0.767
Valvular Heart Disease/Inhospital Outcomes of SAVR
1231
Table 1 (continued) SAVR in TAVI vs Non-TAVI hospital of 2011 - 2012
Demographic Variables
A) Unmatched Group
Non-TAVI capable TAVI capable hospital hospital
B) Matched Group (1-1 Matching) for age, gender, Charlson’s score, bedsize, location and teaching status, type of admission, weekend
Overall
Bed size of Hospital depending on Location & Teaching Status Small 10.4% 2.0% 6.5% Medium 22.2% 15.1% 18.9% Large 67.1% 82.9% 74.5% Hospital Location & Teaching Status Rural 4.0% 0.7% 2.5% Urban Non-teaching 41.0% 9.7% 26.3% Urban Teaching 54.7% 89.6% 71.0% Hospital Region Northeast 18.0% 27.4% 22.4% Midwest 18.3% 20.6% 19.4% South 25.2% 30.8% 27.8% West 22.5% 14.8% 18.9% Type of Admission Non-elective 16.3% 17.2% 16.7% Elective 83.5% 82.6% 83.1% Admission Day Weekdays 97.1% 96.0% 96.6% Weekends 2.9% 4.0% 3.4% Disposition Home 39.2% 37.9% 38.6% Home Health Care 34.9% 38.3% 36.5% Transfer to Short-term 23.9% 22.3% 23.1% Hospital/other facilities Length of hospital-stay - Median 6 (5, 8) 7 (5, 9) 6 (5, 9) (Quartile 1 , 3), days Cost ($)* (Mean, SE) 41831.72 (339) 44840.75 (433) 43198.76 (270)
P-value
Non-TAVI capable hospital
TAVI capable hospital
Overall
P-value
<0.001
3.4% 24.8% 71.8%
3.5% 25.7% 70.9%
3.4% 25.2% 71.3%
0.275
<0.001
1.2% 16.0% 82.9%
1.2% 16.3% 82.5%
1.2% 16.1% 82.7%
0.694
<0.001
23.4% 18.1% 25.5% 20.2%
30.5% 18.1% 28.1% 17.6%
26.9% <0.001 18.1% 26.8% 18.9%
0.001
14.8% 85.2%
14.5% 85.5%
14.6% 85.4%
0.501
<0.001
97.1% 2.9%
97.0% 3.0%
97.0% 3.0%
0.821
<0.001
39.4% 35.3% 23.3%
36.9% 41.2% 20.7%
38.2% <0.001 38.2% 22.0%
6 (5, 8)
6 (5, 9)
41032.3 (473)
43894.14 (483)
<0.001
6 (5, 9) 42388.19 <0.001 (339)
SAVR-non-TAVI ¼ surgical aortic valve replacement performed in non-TAVI centers; SAVR-TAVI ¼ surgical aortic valve replacement performed in TAVI centers. * Cost calculated for 2012 after adjusting for inflation.
Transcatheter aortic valve implantation (TAVI) has emerged as a novel therapeutic option for aortic valvular stenosis in high-risk surgical patients with severe co-morbidities.1 The technology has rapidly gained worldwide acceptance, and since the Food and Drug Administration approval in November 2011, it has disseminated to a large number of hospitals throughout the United States. TAVI requires a complex decision-making and active involvement of an integrated “Heart Team” comprising physicians with diverse expertise. We hypothesized that the availability of a TAVI program in hospitals impacts the overall management of patients with aortic valve disease and hence may also improve postprocedural outcomes of conventional surgical aortic valve replacement (SAVR). The aim of the present study was to compare the inhospital outcomes of SAVR in centers with versus without availability of a TAVI program in an unrestricted large nationwide patient population aged >50 years.
Methods Data were obtained from the Nationwide Inpatient Sample (NIS) for the years 2011 and 2012. NIS is a part of a family of databases developed for the Healthcare Cost and Utilization Project (HCUP) and is sponsored by the Agency for Healthcare Research and Quality (AHRQ). Data from the NIS have previously been used to identify, track, and analyze national trends in health care usage, patterns of major procedures, access, disparity of care, trends in hospitalizations, charges, quality, and outcomes.2e5 We analyzed data from NIS 2011 and 2012 using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) procedure codes for implantation of a bioprosthetic (code 35.21) or mechanical (code 35.22) aortic valve. Only patients aged >50 years with isolated SAVR were included and those who underwent concomitant coronary artery bypass graft surgery, other valvular, or
1232
The American Journal of Cardiology (www.ajconline.org)
Table 2 In-hospital mortality and complications of SAVR in the study population: A) unmatched, B) propensity matched SAVR in TAVI vs Non-TAVI center
Complications (%) Death Any Complications Any Complications þ Death Stroke Transmural MI Deep sternal wound infection Hemorrhage requiring transfusion Cardiac Complications Iatrogenic Cardiovascular Complications Pericardial complications Respiratory failure Sepsis Acute Kidney Injury requiring Dialysis PE and DVT Anesthetics Complications Cardiac Device/Prosthetics/ Graft Complications
A) Unmatched dataset
B) Matched Group (1-1 Matching) for age, gender, Charlson’s score, bedsize, location and teaching status, type of admission, weekend
Non-TAVI capable
TAVI capable
Overall
P-value
Non-TAVI capable
TAVI capable
Overall
P-value
1.8% 36.0% 36.5% 0.9% 0.9% 0.6% 19.8% 12.8% 11.9% 0.9% 5.6% 1.9% 1.2% 0.7% 0.0% 0.5%
1.5% 36.5% 36.9% 1.2% 1.0% 0.6% 18.4% 13.2% 12.5% 0.6% 6.2% 2.9% 0.9% 1.0% 0.0% 0.5%
1.7% 36.2% 36.7% 1.0% 0.9% 0.6% 19.1% 13.0% 12.2% 0.8% 5.9% 2.4% 1.0% 0.8% 0.0% 0.5%
0.036 0.204 0.375 0.007 0.151 0.610 0.000 0.203 0.085 0.000 0.011 <0.001 0.003 <0.001 NA 0.752
1.7% 37.3% 38.0% 1.0% 0.9% 0.6% 22.2% 12.5% 11.8% 0.8% 5.2% 2.1% 1.1% 0.5% 0.0% 0.3%
1.3% 35.6% 35.9% 0.9% 1.1% 0.5% 18.0% 13.3% 12.7% 0.5% 5.6% 2.6% 0.8% 1.2% 0.0% 0.6%
1.5% 36.5% 36.9% 1.0% 1.0% 0.6% 20.1% 12.9% 12.3% 0.6% 5.4% 2.4% 1.0% 0.9% 0.0% 0.5%
0.002 0.004 0.001 0.803 0.139 0.186 <0.001 0.048 0.021 0.007 0.099 0.004 0.018 <0.001 NA 0.000
SAVR-non-TAVI ¼ surgical aortic valve replacement performed in non-TAVI centers; SAVR-TAVI ¼ surgical aortic valve replacement performed in TAVI centers; MI ¼ myocardial Infarction; PE ¼ pulmonary embolism; DVT ¼ deep venous thrombosis.
pericardial procedures were excluded from the analysis. The remaining SAVR cases were divided into 2 categories: those performed at hospitals with a TAVI program (SAVR-TAVI) and those without (SAVR-non-TAVI). The TAVI-capable hospitals were identified with ICD-9-CM procedure codes of 35.05 and 35.06. We defined severity of co-morbid conditions using Deyo modification of Charlson Comorbidity Index (CCI).6 This index contains 17 co-morbid conditions with differential weights. The score ranges from 0 to 33, with greater scores corresponding to greater burden of co-morbid diseases. The primary outcome was all-cause inhospital mortality. Procedural complications were identified by Patient Safety Indicators (PSIs), version 4.4, March 2012, which have been established by the AHRQ to monitor preventable adverse events during hospitalization. These indicators are based on ICD-9-CM codes and Medicare severity Diagnosis Related Groups, and each PSI has specific inclusion and exclusion criteria.6,7 This method of identifying patients undergoing procedures, co-morbid conditions, and associated complications has previously been used in several studies.2,3 Other outcomes studied were the length of hospital stay and cost of hospitalization. To estimate cost of hospitalization, the NIS data were merged with cost-to-charge ratios available from the HCUP. We estimated the cost of each inpatient stay by multiplying the total hospital charge with cost-to-charge ratios. Cost for each year was calculated in terms of the 2012 cost, after adjusting for inflation according to the latest consumer price index (CPI) data released by US government on January 16, 2013. Stata IC 11.0 (Stata-Corp, College Station, Texas) and SAS 9.3 (SAS Institute Inc, Cary, North Carolina) was used for analyses. Differences between categorical variables were
tested using the chi-square test and differences between continuous variables were tested using the Student t test, continuous variables with Gaussian distributions, and KruskaleWallis rank sum tests for continuous variables with nonGaussian distributions. Multivariate logistic regression was generated to identify the independent multivariate predictors of inhospital mortality. All interactions were thoroughly tested. Collinearity was assessed using variance inflation factor. We used propensity-scoring method to establish matched cohorts to control for imbalances of patients’ and hospitals’ characteristics between the 2 groups that may have influenced the primary outcome. A propensity score was assigned to each hospitalization. This was based on multivariate logistic regression model that examined the impact of 12 variables (patient demographics, co-morbidities, and hospital characteristics) on the likelihood of treatment assignment. Patients with similar propensity score in the 2 treatment groups were matched using a 1 to 1 scheme without replacement using greedy algorithm. Results A total of 9,674 SAVR procedures (which translates to an estimated 47,410 procedures performed in 1,110 hospitals) were identified of which 4,526 (46.79%) were performed in hospitals with availability of TAVI (SAVRTAVI group) and 5,148 (53.21%) in non-TAVI hospitals (SAVR-non-TAVI group). Table 1 demonstrates the baseline characteristics of the study population. The mean age of the study population was 70.2 0.1 years with majority (53%) of the patients aged >70 years; 58% were men and 78% were whites. There were significant
Valvular Heart Disease/Inhospital Outcomes of SAVR
1233
Figure 1. Complication rates during SAVR in TAVI versus non-TAVI centers in the propensity scoreematched cohort.
Table 3 Multivariate predictors of mortality and any complications Variable
Multivariate Simple logistic regression for In Hospital Mortality (Weighted) Unmatched Dataset
Multivariate Simple logistic regression for Any Complications (Weighted)
Propensity matched Dataset
Unmatched Dataset
Propensity matched Dataset
Odds Ratio LL UL P-Value Odds Ratio LL UL P-Value Odds Ratio LL UL P-Value Odds Ratio LL UL P-Value (95% CI) (95% CI) (95% CI) (95% CI) Status of TAVI Capability Non-TAVI capable Hospital TAVI Capable Hospital Age (in 10 year increment) Sex Male Female Charlson Score 0 1 More than or equal to 2 Bed size of Hospital depending on Location & Teaching Status Small Medium Large Type of Admission Non-elective Elective Admission Day Weekdays Weekends
1 0.79 1.26
Ref Ref Ref 0.68 0.92 0.00 1.18 1.36 <0.001
1 0.72 1.06
Ref Ref 0.58 0.88 0.96 1.17
Ref 0.00 0.27
1 0.98 1.21
Ref Ref Ref 0.94 1.02 0.25 1.19 1.23 <0.001
1 0.93 1.21
Ref Ref Ref 0.88 0.97 0.00 1.18 1.24 <0.001
1.00 0.95
Ref Ref 0.82 1.10
1.00 0.74
Ref Ref 0.60 0.92
Ref 0.01
1.00 0.97
Ref Ref 0.93 1.01
1.00 0.92
Ref Ref 0.88 0.97
1.00 1.23 3.54
Ref Ref Ref 0.94 1.61 0.13 2.82 4.45 <0.001
1.00 1.46 5.21
Ref Ref Ref 0.96 2.22 0.08 3.63 7.49 <0.001
1.00 1.11 1.60
Ref Ref Ref 1.05 1.17 <0.001 1.52 1.68 <0.001
1.00 1.12 1.72
Ref Ref Ref 1.05 1.20 0.00 1.61 1.83 <0.001
1.00 0.86 0.97
Ref Ref 0.62 1.18 0.73 1.29
1.00 0.51 0.62
Ref Ref 0.30 0.85 0.39 1.01
1.00 1.16 1.12
Ref Ref 1.07 1.27 1.04 1.22
1.00 1.01 1.03
Ref Ref 0.87 1.17 0.90 1.19
1.00 0.33
Ref Ref Ref 0.28 0.38 <0.001
1.00 0.26
Ref Ref Ref 0.21 0.32 <0.001
1 0.67
Ref Ref Ref 0.64 0.70 <0.001
1 0.66
Ref Ref Ref 0.61 0.71 <0.001
1.00 1.24
Ref Ref 0.94 1.64
1.00 0.96
Ref Ref 0.64 1.44
1.00 1.12
Ref Ref 1.01 1.24
1.00 1.34
Ref Ref 1.16 1.56
Ref 0.45
Ref 0.34 0.83
Ref 0.12
differences between the baseline characteristics of patients in the 2 groups (Table 1). A greater percentage of patients in SAVR-TAVI group were aged >80 years, had hypertension, congestive heart failure, renal failure, and peripheral arterial disease, whereas more patients in SAVR-nonTAVI group had diabetes and obesity. The mean
Ref 0.01 0.05
Ref 0.84
Ref 0.09
Ref 0.00 0.00
Ref 0.04
Ref 0.00
Ref 0.95 0.67
Ref 0.00
Charlson co-morbidity score for patients in SAVR-TAVI group was greater than that of patients in SAVR-nonTAVI group (1.6 vs 1.4, p <0.001). The outcomes of both groups are reported in Table 2. Propensity scoreematched analysis is reported in Table 1. In this analysis, multiple patient- and hospital-level
1234
The American Journal of Cardiology (www.ajconline.org)
variables that could have affected the results were adjusted for. These variables included age, gender, Charlson comorbidity score, bed size of the hospital, location of the hospital and teaching status, type, and timing of admissions (emergent vs nonemergent and weekend vs nonweekend). The propensity scoreematching analysis showed a statistically significant lower inhospital mortality (1.25% vs 1.72%, p ¼ 0.001) and complications rates (35.6% vs 37.3%, p ¼ 0.004) in SAVR-TAVI group compared to SAVR-non-TAVI group (Figure 1). The mean length of hospital stay was similar in the 2 groups; however, the cost of hospitalization was greater in the SAVR-TAVI group. In the multivariate analysis of propensity scoreematched population, having a TAVI program in the hospital was a significant predictor of reduced mortality (odds ratio [OR] 0.71, 95% confidence interval [CI] 0.6 to 0.8, p ¼ 0.0015), whereas a greater burden of co-morbidities (CCI >2) was an independent predictor of higher inhospital mortality (OR 5.2, 95% CI 3.6 to 7.5, p <0.0001) (Table 3). Similarly, having a TAVI program was an independent predictor of reduced complication rates in patients undergoing SAVR (Table 3). In another multivariate model that included hospital SAVR volume, presence of a TAVI program was a significant predictor of reduced mortality and complication rates (Supplementary Table 1). Discussion In this analysis of patients aged >50 years who underwent SAVR in the United States, we demonstrate that inhospital mortality and complication rates of SAVRs performed in centers with a TAVI program is significantly less than those performed in centers without a TAVI program. Despite having greater burden of risk factors and comorbidities, SAVRs performed in TAVI centers were associated with improved outcomes. This effect of TAVI on inhospital mortality and complication rates was also reproducible in the propensity scoreematched analysis and multivariate models. In 2012, the Society for Cardiovascular Angiography and Interventions, American Association for Thoracic Surgery, American College of Cardiology Foundation, and the Society of Thoracic Surgeons (STS) provided specific recommendations for institutional requirements for a TAVI program.8 It is recommended that institutions perform 50 SAVRs/year (at least 10 high risk with STS score 6).8 The multidisciplinary TAVI team must include a minimum of 2 institutional-based cardiac surgeons each of whom should have performed 100 SAVRs in their career, at least 10 of which are “high risk” (STS score 6) or 25 SAVRs/year or 50 SAVRs in 2 years.8 On the basis of these recommendations, TAVI centers are generally greater surgical volume centers and have highly experienced cardiothoracic surgeons which may explain the associated improved outcomes of SAVR performed in these centers. We noted a significantly high procedural volume (median Q1 to Q3) of 89 (49 to 142) versus 25 (14 to 47) in the SAVR-TAVI group compared to SAVR-non-TAVI group. The positive effect of larger hospital volume on reduction of inhospital mortality rate associated with surgical procedures including SAVR has been well reported in the previous medical reports.9,10
In an analysis of 6,270 procedures (3,487 SAVR and 2,783 SAVR þ coronary artery bypass grafting) using the statewide cardiothoracic surgical database (2008 to 2011) from Michigan, Patel et al11 reported a strong volumeeoutcome relation for SAVR particularly in high-risk patients. However, we noted a positive effect of TAVI program in the outcomes of SAVRs even in the multivariate models (with or without inclusion of hospital volume) and in the propensity scoreematched analysis in which the 2 groups were matched for multiple patient- and hospital-level variables including hospital size and teaching status. The introduction of TAVI program has been associated with a “halo effect” on SAVR in terms of referral and procedural volume especially for both low- and high-risk patients.12,13 Grant et al13 reported their single-center (UK) experience of initiating a TAVI program in a retrospective analysis of 815 consecutive patients undergoing isolated SAVR or coronary artery bypass grafting plus SAVR. Within the 2 years of introduction of TAVI program, they noted a 37% increase in surgical procedures.13 Interestingly, despite an increase in the mean logistic EuroSCORE (7.4 vs 7.9), they noted a decrease in the hospital mortality rate (2.9% vs 2.1%; p ¼ 0.48).13 An increased referral of low surgical risk patients to a TAVI center may also contribute to the improved SAVR outcomes. Inclusion of a large sample size in our study (>9,600 SAVR procedures) may also have enabled us to detect these differences compared to previous studies that may have been underpowered.14 More recently, Brennan et al15 used data from the STS/ American College of Cardiology transcatheter valve therapies registry to examine the association of TAVI availability in a hospital and overall AVR volumes (SAVR þ TAVI) and mortality from 2008 to 2013. During the study period, they noted that AVR (SAVR þ TAVI) inhospital mortality rate decreased from 3.4% to 2.9% (p <0.0001); however, the inhospital mortality rate for TAVI increased from 3.5% to 5.2% (O-to-E ratio 0.41 to 0.60, p ¼ 0.02).15 In that study, the inhospital mortality rate for SAVR decreased from 3.4% to 2.5% during the study period of 5 years, with the greatest absolute reduction in high-risk SAVR cases (11.7% to 9.3%). It is important to note that large-scale studies using STS database have previously shown that the outcomes of SAVR have been steadily improving in the United States despite worsening patient profiles since early 1990s, long before TAVI became available.16e18 Our present study was designed to eliminate the effect of temporal changes and specifically assess the impact of TAVI on SAVR (regardless of the risk profile) soon after TAVI was introduced (2011 to 2012) into the management of aortic valve disease. The outcomes of SAVR from NIS database have previously been shown to correlate with those from STS database.19 The propensity scoreematched analysis in our study also showed a benefit in terms of both mortality and complications when an SAVR is performed in a TAVI-capable center (absolute reduction of 0.47% for mortality and 1.7% for complications). The reduced complications rate in the SAVR-TAVI group (35.6% vs 37.3%) was small but statistically significant (p ¼ 0.004) and driven by a reduction in bleeding complications. No large-scale study has previously
Valvular Heart Disease/Inhospital Outcomes of SAVR
assessed differences in complications after SAVR in TAVI and non-TAVI center; however, in a study performed at a single center comparing outcomes before and after initiation of a TAVI program, Malaisrie et al12 noted a reduction in complications (43.8% vs 44.9%) after TAVI program. The salutary effect of TAVI may also be explained because of involvement of the multidisciplinary “heart team” in the management of complex valvular disorders. The rationale for team-based care is to improve clinical outcomes by facilitating shared decision-making in a timely and coordinated manner.20,21 Heart team associated with a TAVI center may improve inhospital outcomes of SAVR by optimizing patient selection, procedural performance, and complications management.22 This high-quality care however comes at a greater cost as witnessed in our study. The cost of SAVRs performed in TAVI centers was significantly greater compared to those in non-TAVI centers ($43,894 483 vs $41,032 473, p <0.0001) despite a similar length of hospital stay. The greatest cost of care may however be offset by a reduction in mortality and complications seen in the SAVR-TAVI group. The study limitations are inherent to the post hoc analysis of an administrative database. We have provided possible hypothesis for improved outcomes in the SAVT-TAVI group; however, the design of our study restricts any inference of causality or temporal associations. Compared to some of the previous studies, we lack information on the indication for SAVR, disease severity, cardiac functional class, frailty, and STS score. In 2012, NIS was redesigned to include a sample of discharges instead of a sample of hospitals, and therefore, hospital procedural volume could only be estimated for 2011. In addition, NIS does not provide any follow-up data. Although acknowledging these limitations, the present study has important strengths including a large unrestricted sample size from all-comers real-world data for patients undergoing SAVR and the use of standardized definitions of preventable adverse events/complications that are established by the AHRQ. Our study represents the largest direct comparative analysis of SAVR in TAVI and non-TAVI centers and adds to the growing reports on the “halo effect” of establishing a TAVR program in an institution. Disclosures
3.
4.
5.
6.
7. 8.
9. 10.
11.
12.
None of the authors have anything to disclose. 13.
Supplementary Data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.amjcard.2015.07.039. 1. Smith CR, Leon MB, Mack MJ, Miller DC, Moses JW, Svensson LG, Tuzcu EM, Webb JG, Fontana GP, Makkar RR, Williams M, Dewey T, Kapadia S, Babaliaros V, Thourani VH, Corso P, Pichard AD, Bavaria JE, Herrmann HC, Akin JJ, Anderson WN, Wang D, Pocock SJ. Transcatheter versus surgical aortic-valve replacement in high-risk patients. N Engl J Med 2011;364:2187e2198. 2. Badheka AO, Patel NJ, Grover P, Singh V, Patel N, Arora S, Chothani A, Mehta K, Deshmukh A, Savani GT, Patel A, Panaich SS, Shah N, Rathod A, Brown M, Mohamad T, Tamburrino FV, Kar S, Makkar R, O’Neill WW, De Marchena E, Schreiber T, Grines CL,
14.
15.
16.
1235
Rihal CS, Cohen MG. Impact of annual operator and institutional volume on percutaneous coronary intervention outcomes: a 5-year United States experience (2005-2009). Circulation 2014;130: 1392e1406. Singh V, Patel NJ, Badheka AO, Arora S, Patel N, Macon C, Savani GT, Manvar S, Patel J, Thakkar B, Panchal V, Solanki S, Chothani A, Panaich SS, Ram V, Kliger CA, Schreiber T, O’ Neill W, Cohen MG, Alfonso CE, Grines CL, Mangi A, Pfau S, Forrest JK, Cleman M, Makkar R. Comparison of outcomes of balloon aortic valvuloplasty plus percutaneous coronary intervention versus percutaneous aortic balloon valvuloplasty alone during the same hospitalization in the United States. Am J Cardiol 2015;115:480e486. Patel NJ, Deshmukh A, Pant S, Singh V, Patel N, Arora S, Shah N, Chothani A, Savani GT, Mehta K, Parikh V, Rathod A, Badheka AO, Lafferty J, Kowalski M, Mehta JL, Mitrani RD, Viles-Gonzalez JF, Paydak H. Contemporary trends of hospitalization for atrial fibrillation in the United States, 2000 through 2010: implications for healthcare planning. Circulation 2014;129:2371e2379. Deshmukh A, Patel NJ, Pant S, Shah N, Chothani A, Mehta K, Grover P, Singh V, Vallurupalli S, Savani GT, Badheka A, Tuliani T, Dabhadkar K, Dibu G, Reddy YM, Sewani A, Kowalski M, Mitrani R, Paydak H, Viles-Gonzalez JF. In-hospital complications associated with catheter ablation of atrial fibrillation in the United States between 2000 and 2010: analysis of 93 801 procedures. Circulation 2013;128: 2104e2112. McDonald KM, Romano PS, Geppert J, Davies SM, Duncan BW, Shojania KG, Hansen A. Measures of patient safety based on hospital administrative data—the patient safety indicators, Structured Abstract; Vol 02-0038. Rockville, MD. Agency Healthc Reseach Qual 2002. Romano PS, Geppert JJ, Davies S, Miller MR, Elixhauser A, McDonald KM. A national profile of patient safety in U.S. hospitals. Health Aff (Millwood) 2003;22:154e166. Tommaso CL, Bolman RM 3rd, Feldman T, Bavaria J, Acker MA, Aldea G, Cameron DE, Dean LS, Fullerton D, Hijazi ZM, Horlick E, Miller DC, Moon MR, Ringel R, Ruiz CE, Trento A, Weiner BH, Zahn EM. Multisociety (AATS, ACCF, SCAI, and STS) expert consensus statement: operator and institutional requirements for transcatheter valve repair and replacement, part 1: transcatheter aortic valve replacement. J Thorac Cardiovasc Surg 2012;143: 1254e1263. Birkmeyer JD, Siewers AE, Finlayson EV, Stukel TA, Lucas FL, Batista I, Welch HG, Wennberg DE. Hospital volume and surgical mortality in the United States. N Engl J Med 2002;346:1128e1137. Hannan EL, O’Donnell JF, Kilburn H Jr, Bernard HR, Yazici A. Investigation of the relationship between volume and mortality for surgical procedures performed in New York State hospitals. JAMA 1989;262:503e510. Patel HJ, Herbert MA, Drake DH, Hanson EC, Theurer PF, Bell GF, Prager RL. Aortic valve replacement: using a statewide cardiac surgical database identifies a procedural volume hinge point. Ann Thorac Surg 2013;96:1560e1565; discussion 1565e1566. Malaisrie SC, Tuday E, Lapin B, Wang E, Lee R, McGee EC, Davidson C, McCarthy PM. Transcatheter aortic valve implantation decreases the rate of unoperated aortic stenosis. Eur J Cardiothorac Surg 2011;40:43e48. Grant SW, Devbhandari MP, Grayson AD, Dimarakis I, Kadir I, Saravanan DM, Levy RD, Ray SG, Bridgewater B. What is the impact of providing a transcatheter aortic valve implantation service on conventional aortic valve surgical activity: patient risk factors and outcomes in the first 2 years. Heart 2010;96:1633e1637. Malaisrie SC, McCarthy PM, McGee EC, Lee R, Rigolin VH, Davidson CJ, Beohar N, Lapin B, Subacius H, Bonow RO. Contemporary perioperative results of isolated aortic valve replacement for aortic stenosis. Ann Thorac Surg 2010;89:751e756. Brennan JM, Holmes DR, Sherwood MW, Edwards FH, Carroll JD, Grover FL, Tuzcu EM, Thourani V, Brindis RG, Shahian DM, Svensson LG, O’Brien SM, Shewan CM, Hewitt K, Gammie JS, Rumsfeld JS, Peterson ED, Mack MJ. The association of transcatheter aortic valve replacement availability and hospital aortic valve replacement volume and mortality in the United States. Ann Thorac Surg 2014;98:2016e2022; discussion 2022. Barreto-Filho JA, Wang Y, Dodson JA, Desai MM, Sugeng L, Geirsson A, Krumholz HM. Trends in aortic valve replacement for
1236
The American Journal of Cardiology (www.ajconline.org)
elderly patients in the United States, 1999-2011. JAMA 2013;310: 2078e2085. 17. Brown JM, O’Brien SM, Wu C, Sikora JA, Griffith BP, Gammie JS. Isolated aortic valve replacement in North America comprising 108, 687 patients in 10 years: changes in risks, valve types, and outcomes in the Society of Thoracic Surgeons National Database. J Thorac Cardiovasc Surg 2009;137:82e90. 18. Lee R, Li S, Rankin JS, O’Brien SM, Gammie JS, Peterson ED, McCarthy PM, Edwards FH. Fifteen-year outcome trends for valve surgery in North America. Ann Thorac Surg 2011;91:677e684; discussion p 684.
19. Astor BC, Kaczmarek RG, Hefflin B, Daley WR. Mortality after aortic valve replacement: results from a nationally representative database. Ann Thorac Surg 2000;70:1939e1945. 20. Holmes DR Jr, Rich JB, Zoghbi WA, Mack MJ. The heart team of cardiovascular care. J Am Coll Cardiol 2013;61:903e907. 21. Nallamothu BK, Cohen DJ. No “i” in Heart Team: incentivizing multidisciplinary care in cardiovascular medicine. Circ Cardiovasc Qual Outcomes 2012;5:410e413. 22. Gonzalez AA, Dimick JB, Birkmeyer JD, Ghaferi AA. Understanding the volume-outcome effect in cardiovascular surgery: the role of failure to rescue. JAMA Surg 2014;149:119e123.