Five-year Clinical and Economic Outcomes Among Patients With Medically Managed Severe Aortic Stenosis Results From a Medicare Claims Analysis

Transcription

1 Five-year Clinical and Economic Outcomes Among Patients With Medically Managed Severe Aortic Stenosis Results From a Medicare Claims Analysis Mary Ann Clark, MHA; Suzanne V. Arnold, MD, MHA; Francis G. Duhay, MD; Ann K. Thompson, BSN, MBA; Michelle J. Keyes, PhD; Lars G. Svensson, MD, PhD; Robert O. Bonow, MD; Benjamin T. Stockwell, BA; David J. Cohen, MD, MSc Background Patients with severe, symptomatic aortic stenosis, who do not undergo valve replacement surgery have a poor long-term prognosis. Limited data exist on the medical resource utilization and costs during the final stages of the disease. Methods and Results We used data from the 2003 Medicare 5% standard analytic files to identify patients with aortic stenosis and a recent hospitalization for heart failure, who did not undergo valve replacement surgery within the ensuing 2 calendar quarters. These patients (n=2150) were considered to have medically managed severe aortic stenosis and were tracked over 5 years to measure clinical outcomes, medical resource use, and costs (from the perspective of the Medicare Program). The mean age of the cohort was 82 years, 64% were female, and the estimated logistic European System for Cardiac Operative Risk Evaluation (EuroSCORE) (a measure of predicted mortality with cardiac surgery) was 17%. During 5 years of follow-up, overall mortality was 88.4% with a mean survival duration of 1.8 years. During this time period, patients experienced an average of 4.4 hospital admissions, 52% were admitted to skilled nursing care, and 28% were admitted to hospice care. The total 5-year costs were $ per patient, whereas mean annual follow-up costs (excluding the index quarter) per year alive were $ Conclusions Elderly patients with severe aortic stenosis undergoing medical management have limited long-term survival and incur substantial costs to the Medicare Program. These results have important implications for policy makers interested in better understanding the cost-effectiveness of emerging treatment options such as transcatheter aortic valve replacement. (Circ Cardiovasc Qual Outcomes. 2012;5: ) Key Words: epidemiology valves heart failure cost Medicare Aortic stenosis (AS) is the most common form of valvular heart disease, affecting 2% to 3% of the adult US population with greater prevalence among the elderly. 1 3 Patients may tolerate severe AS for many years, but once symptoms develop the prognosis is poor without aortic valve replacement (AVR). 4 7 Although most patients with severe symptomatic AS currently undergo AVR, a sizable proportion of patients do not undergo treatment. 1,7 9 In recent years, transcatheter AVR (TAVR) has been developed as a treatment option for patients with severe AS and high surgical risk and has been shown to improve survival in inoperable patients and reduce the need for rehospitalization among such patients. 10 Whether TAVR represents a rational use of limited societal resources requires a thorough understanding of the longterm outcomes and costs of untreated patients with severe symptomatic AS. Although several studies have documented long-term outcomes for this condition, these studies have been based on select patients presenting for cardiac evaluation at a limited number of centers. 7,9,11,12 In addition, no studies to date have documented the healthcare resources and costs associated with the long-term care of such patients. We, therefore, investigated outcomes and costs among unselected patients with medically managed severe, symptomatic AS from the perspective of the US Medicare Program. Methods Data Sources and Study Population The data sources for this study were the Medicare standard analytic files (SAFs) for 2001 through The SAFs contain all claims (physician, inpatient, outpatient, skilled nursing, home health, hospice, durable medical equipment suppliers, and other suppliers) Received January 2, 2012; accepted August 3, From the The Neocure Group, LLC, Washington, DC (M.A.C.); Saint Luke s Mid America Heart Institute University of Missouri-Kansas City, Kansas City, MO (S.V.A., D.J.C.); Edwards Lifesciences, LLC, Irvine, CA (A.K.T., F.G.D.); The Burgess Group, LLC, Alexandria, VA (M.J.K., B.T.S.); Department of Thoracic and Cardiovascular Surgery, The Cleveland Clinic, Cleveland, OH (L.G.S.); and the Center for Cardiovascular Innovation, Northwestern University Feinberg School of Medicine, Chicago IL (R.O.B.). The online-only Data Supplement is available at Correspondence to David J. Cohen, MD, MSc, Saint Luke s Mid America Heart Institute, 4401 Wornall Rd., Kansas City, MO dcohen@saint-lukes.org 2012 American Heart Association, Inc. Circ Cardiovasc Qual Outcomes is available at DOI: /CIRCOUTCOMES

2 698 Circ Cardiovasc Qual Outcomes September 2012 WHAT IS KNOWN Patients with medically managed symptomatic, severe aortic stenosis have a poor long-term prognosis, but the costs during the final stages of the disease have not been estimated. WHAT THE STUDY ADDS Elderly patients with medically managed severe, symptomatic aortic stenosis not only have a poor prognosis (1-year mortality was 49% and average life-span was 1.8 years in our study) but they also are intensive users of healthcare resources. Using Medicare data, we found that these patients had a high rate of rehospitalization (1.9 admissions per patient-year) with prolonged lengths of stay (11.5 hospital days per patient-year). As a result, they incurred substantial healthcare related costs, with mean annual costs of ~$30 000/year and cumulative costs of $63 844/patient over the 5 years of follow-up. These findings have important implications for policy makers interested in better understanding the costs of medically managing patients with severe, symptomatic aortic stenosis and a basis from which to consider the impact of emerging treatment options on overall health expenditures. from a 5% random sample of Medicare beneficiaries enrolled in the fee-for-service Medicare Program. Unique patient identifiers allow claims to be linked across time for longitudinal analysis of health resource use, costs (Medicare payments), and outcomes (procedure, diagnosis codes, death date). The patient population for our study was designed to approximate patients who would be candidates for TAVR based on the presence of severe, symptomatic AS who were considered to be at high risk for surgical AVR and were undergoing medical management. Because claims data do not contain results from diagnostic testing or other information typically available in medical records, it is not possible to identify the severity or symptomatic status of patients directly using claims data. However, we were able to use various diagnosis and procedure codes and other information to select our patient population. Patients with AS were identified using physician, inpatient, and outpatient claims from 2003 containing an International Classification of Diseases 9th Clinical Modification (ICD9-CM) diagnosis (DX) or procedure (PX) code for AS. These included any of the following AS codes: ICD9-CM DX 395.0, 395.2, 424.1, or 746.3; or any of the following aortic valve procedure codes: ICD9-CM PX 35.01, 35.96, or CPT To define a group of patients with severe symptomatic AS, we required that patients also have either an acute inpatient hospitalization in 2003 with a principal diagnosis of heart failure (ICD-9 DX , , , , , , , , , , 428.0, 428.1, , , , , , , , , , , , 428.9), or a balloon aortic valvuloplasty procedure (ICD9-CM PX 35.01, 35.96, CPT 92986). To ensure that patients were intended for medical management rather than surgery, patients were excluded if they underwent surgical AVR (ICD-9 PX 35.21, 35.22, CPT 33405), coronary artery bypass graft, surgery (ICD-9 PX 36.1x-36.2x; CPT , 33140, 33141), or percutaneous coronary intervention (ICD-9 PX 36.0x; CPT 92980, 92982, 92995) within 2 calendar quarters of their index heart failure hospitalization. The index date was defined as the quarter of the first hospitalization for heart failure or balloon valvuloplasty in Continuous enrollment in both Medicare Parts A and B for the entire follow-up period or until death was also required, and patients who were enrolled in a Medicare health management organization plan at any time during follow-up were excluded. To assess the accuracy of our inclusion criteria for delineating a group of patients with symptomatic AS, we performed a chart review of patients admitted to Saint Luke s Hospital of Kansas City from 2010 to 2011, who met the inclusion criteria and examined their echocardiographic and cardiac catheterization data to assess the severity of AS. Among the 136 patients who met the inclusion criteria, 88 had severe AS and an additional 13 had at least moderately severe AS; the positive predictive values for severe and moderately severe AS were thus 65% (95% confidence interval 57% 73%) and 74% (95% confidence interval 67% 81%), respectively. Outcome Measures Clinical outcomes of interest included death, myocardial infarction (MI), and stroke. Death was ascertained from the SAF denominator file. An MI was diagnosed if there was at least 1 hospital admission with a principal diagnosis of MI (ICD-9 DX , , , , , , , , , ). A stroke was identified if there was a hospitalization with a principal diagnosis of stroke (ICD-9 DX 430, 431, , , , , , , 435.0, 435.1, 435.3, 435.8, 435.9, 436, ). Other outcomes tracked included the occurrence of surgical AVR (ICD-9 PX 35.21, 35.22, CPT 33405), balloon aortic valvuloplasty (ICD-9 PX 35.01, 35.96, CPT 92986), admission to hospice, admission to a skilled nursing facility, number of acute inpatient hospitalizations, hospital days, and number of skilled nursing facility days. Costs were assessed from a Medicare perspective and measured as payments by the Medicare Program for services reported on Medicare Parts A and B claims. Medicare payments exclude beneficiary cost sharing such as coinsurance and deductibles. Costs were computed on the basis of all types of claims and accumulated over the 5-year follow-up period. Additionally, we examined costs on an annual basis calculated as the mean cost during the ensuing year among patients who were alive at the beginning of each time period. The robustness of the SAF data allowed us to examine and report on other various components of cost and resource use, which are presented in Table 4. These components were determined on the basis of Medicare claim type, bill type, and provider number. Cardiovascular specific costs were defined on the basis of prespecified diagnosis and procedure codes. 13 Risk Stratification and Adjustment The logistic European System for Cardiac Operative Risk Evaluation (EuroSCORE) and Medicare s Hierarchical Condition Categories (HCCs) were used to stratify patients according to their predicted risk of adverse clinical outcome and costs. The logistic EuroSCORE ( was used to estimate predicted operative mortality for each patient to stratify patients into high-risk (HR) and nonhigh-risk (NHR) groups. 14,15 Although the Society for Thoracic Surgeons risk model is generally viewed as more valid than the logistic EuroSCORE for patients undergoing AVR, it is proprietary and requires several clinical covariates that cannot be ascertained from claims data. 16 For the EuroSCORE calculations, age was derived from the Medicare denominator file for Each of the other risk factors was determined by the presence or absence of ICD-9 DX and PX codes on certain claims from the 2 years before each patient s index event or index hospitalization (see online-only Data Supplement Appendix A for details). The process of identifying and defining each EuroSCORE component was iterative and involved substantial clinical review. To our knowledge, this method has not been attempted previously. HR patients (HR group) were defined as a logistic EuroSCORE of 20% or greater. All other patients were considered NHR (NHR group). Additional covariates to account for comorbid conditions not already included within the EuroSCORE components were identified

3 Clark et al Outcomes of Medically Managed Aortic Stenosis 699 Table 1. Baseline Patient Characteristics Characteristic All Patients High Risk Nonhigh Risk n=2150 n=651 n=1499 Age, mean±sd 82.0± ± ±9.2 <0.001 Female sex, % <0.001 White race, % East region (CMS regions 1 4), % Central region (CMS regions 5 7), % West region (CMS regions 8 10), % EuroSCORE comorbidity components, % Chronic pulmonary disease <0.001 Extracardiac arteriopathy <0.001 Neurologic dysfunction <0.001 Previous cardiac surgery <0.001 End-stage renal disease Active endocarditis 0.7 * * NA Critical preoperative state <0.001 Unstable angina <0.001 Left ventricular dysfunction <0.001 Recent myocardial infarction <0.001 Pulmonary hypertension <0.001 EuroSCORE, mean±sd, % 17±14 34±14 10±5 <0.001 EuroSCORE 20%, % NA Comorbid condition (Medicare HCC group), % Metastatic cancer and acute leukemia Lung, upper digestive tract, and other severe cancers Lymphatic, head and neck, brain, and other major cancers Breast, prostate, colorectal and other cancers and tumors Diabetes with renal or peripheral circulatory manifestation Diabetes with neurologic or other specified manifestation Diabetes with ophthalmologic or unspecified manifestation Diabetes without complication Protein-calorie malnutrition <0.001 Rheumatoid arthritis and inflammatory connective tissue Major depressive, bipolar, and paranoid disorders Cardio-respiratory failure and shock <0.001 Congestive heart failure Acute myocardial infarction * * * NA Acute ischemic heart disease Specified heart arrhythmias <0.001 Ischemic or unspecified stroke <0.001 Hemiplegia/hemiparesis Vascular disease with complications Vascular disease <0.001 Chronic obstructive pulmonary disease Proliferative diabetic retinopathy and vitreous hemorrhage Renal dialysis Renal failure <0.001 HCC score, mean±sd 3.4± ± ±1.39 <0.001 CMS indicates Center for Medicare and Medicaid Services; EuroSCORE, European System for Cardiac Operative Risk Evaluation; HCC, Medicare Hierarchical Condition Category. *Fewer than 11 patients. P Value

4 700 Circ Cardiovasc Qual Outcomes September 2012 using HCCs. The HCC methodology is used by Medicare to calculate risk-adjusted costs and payments to Medicare Advantage plans. 17 The methodology uses diagnosis codes on inpatient, outpatient, and physician claims from the previous year to identify comorbid conditions. These comorbid conditions along with other demographic variables are used to calculate an HCC score to adjust Medicare Advantage monthly capitation payments. The risk scores are indexed to a value of 1.0, which represents the cost of the average Medicare beneficiary. Therefore, a calculated risk score of 2.0 means that a patient would be expected to be twice as costly as the average Medicare patient. In our study, we used the HCC score only to describe baseline patient characteristics; however, we used several clinically relevant HCC groups to adjust for comorbid conditions in the multivariable models, including cancer (HCC7, 8, and 9), diabetes mellitus with complications (HCC16), malnutrition (HCC21), cardio-respiratory failure and shock (HCC79), vascular disease with complication (HCC104), hemodialysis (HCC130), and renal failure (HCC131). Statistical Analysis Continuous variables are expressed as mean±sd values and were compared using t tests. Categorical variables are presented as frequencies and percentages and were compared using Fisher s exact test. In accordance with the standard Medicare Data Use Agreements, we are unable to report any data with values or calculations resulting in cell sizes <11. When this occurs, values are reported as *. To determine the independent predictors of mortality during the 5-year follow-up period, we used Cox regression modeling techniques where mortality was the dependent variable, and the independent variables included age, sex, race, geographic region, individual EuroSCORE components, and clinically relevant HCC comorbid conditions. All analyses were performed with SAS version 9.2 (SAS Institute, Inc., Cary, NC), and a 2-sided P value of 0.05 was considered statistically significant. Results Patient Population A total of 3624 patients with severe, symptomatic AS were identified in the 2003 Medicare SAF. From this group, 1474 underwent AVR surgery within 2 calendar quarters and were excluded from our analytic cohort; the remaining 2150 (59.3%) patients did not undergo AVR within 2 calendar quarters and constituted the analytic cohort for our study. Baseline characteristics of the study population are summarized in Table 1. The mean age was 82.0 years, 63.6% were female, and 90.0% were white. The mean estimated EuroSCORE for the overall population was 17%, and 30.3% of the patients had an estimated score 20% and were therefore considered to be at high risk for AVR (HR group). The study population also had a significant number of comorbid conditions. The mean HCC score for the group was 3.4 indicating that their expected health care costs were over 3 times those of the average Medicare beneficiary. Table 2. Five-Year Clinical Outcomes of Medically Managed Severe Aortic Stenosis Patients All Patients High Risk Nonhigh Risk n=2150 n=651 n=1499 P Value Survival, % <0.001 Follow-up years, mean±sd Myocardial infarction, % 1.8± ± ±1.7 < Stroke, % In multivariable analyses, independent predictors of mortality included demographic factors (higher age, male sex, white race), several EuroSCORE components (neurologic dysfunction, active endocarditis, critical preoperative state, recent MI), and comorbid conditions (Table 3). A patient with active endocarditis had nearly a 3-fold increase in risk of death (hazard ratio 2.95, P<0.001), whereas patients with metastatic cancer and leukemia, on renal dialysis, or those with lung, upper digestive tract, and other severe cancers had increased hazard ratios of 1.76 (P<0.001), 1.72 (P=0.008), and 1.61 (P=0.004), respectively. Interestingly, women were less likely to die (hazard ratio 0.88, P=0.009) and white patients were more likely to die (hazard ratio 1.27, P=0.003) during the 5 years of follow-up. Resource Use and Costs Measures of health care resource use are summarized in Table 4. Over the 5-year follow-up period, there was an average of 4.4 acute inpatient hospitalizations per patient (including the index admission) with a mean total length of stay of 26.7 days. After excluding the index admission, patients experienced an average of 1.9 hospital admissions per year. Overall, 2.1% of the patients who were initially managed medically eventually underwent surgical AVR, and 1.9% of patients underwent balloon aortic valvuloplasty during the follow-up period. Approximately half (52.0%) of all patients required skilled nursing home care, whereas 27.6% and 57.4% received hospice and home health care, respectively. Compared with non-hr patients, HR patients were less likely Clinical Outcomes Of the initial cohort of 2150 patients, only 249 (11.6%) were alive at the end of the 5-year follow-up period (Table 2 and Figure 1), with modestly lower survival among HR versus NHR patients (5.1% versus 14.4%, P<0.001). Mean survival duration for the overall cohort was 1.8±1.6 years: 1.4±1.4 years in the HR group and 2.0±1.7 in the NHR group (P<0.001). Rates of MI (7.5% versus 7.7%, P=0.866) and stroke (6.0% versus 5.7%, P=0.817) were similar for the HR and NHR groups. Figure 1. Survival rates over 5 years for the overall population (all patients), high-risk patients, and nonhigh-risk patients.

5 Clark et al Outcomes of Medically Managed Aortic Stenosis 701 Table 3. Predictors of 5-Year Mortality* Patient demographics to undergo AVR and more likely to be admitted to a skilled nursing facility during follow-up. Mean total medical costs per patient during the 5-year period were $63 844±$ (Figure 2A). Most of the medical costs were due to acute inpatient hospitalizations Adjusted HR (95% CI) P Value Age 1.04 ( ) <0.001 Female sex 0.88 ( ) White race 1.27 ( ) Center for Medicare and Medicaid services region West Reference East 0.84 ( ) Central 0.78 ( ) EuroSCORE components Neurologic dysfunction 1.18 ( ) Active endocarditis 2.95 ( ) <0.001 Critical preoperative state 1.39 ( ) <0.001 Recent myocardial infarction 1.29 ( ) <0.001 Comorbid conditions not already included in EuroSCORE Metastatic cancer and acute leukemia 1.76 ( ) <0.001 Lung, upper digestive tract, and other severe cancers 1.61 ( ) Lymphatic, head and neck, brain, and other major cancers 1.54 ( ) Diabetes with neurologic or other specified manifestation 1.24 ( ) Protein-calorie malnutrition 1.37 ( ) <0.001 Cardio-respiratory failure and shock 1.27 ( ) <0.001 Vascular disease with complications 1.24 ( ) Renal dialysis 1.72 ( ) Renal failure 1.39 ( ) <0.001 HR indicates hazard ratio; CI, confidence interval; EuroSCORE, European System for Cardiac Operative Risk Evaluation. *Independent predictors of all-cause mortality based on Cox Proportional Hazards regression. Table 4. Five-Year Health Care Resource Use (49.7%, $31 714±$35 129), physician services (16.5%, $10 531±$15 366), and skilled nursing care (10.4%, $6668±$11 525). The remaining costs (23.4%) were fairly evenly distributed among other types of health care services. Cardiovascular costs accounted for 41% of total medical costs All Patients High Risk Nonhigh Risk n=2150 n=651 n=1499 Acute inpatient hospitalization, % NA P Value Hospitalizations, mean±sd 4.4± ± ± Hospital days, mean±sd 26.7± ± ± Aortic valve replacement, % 2.1 * * Balloon aortic valvuloplasty, % Long-term care hospital, % Inpatient rehab facility, % Skilled nursing facility, % Skilled nursing days, mean±sd 25.5± ± ± Hospice care, % Home health care, % Outpatient hospital care, % Physician services, % NA Durable medical equipment use, % Dialysis services, %

6 702 Circ Cardiovasc Qual Outcomes September 2012 Figure 2. Mean 5-year cumulative total medical (A) and cardiovascular (B) costs per patient by major cost category. during the 5-year follow-up ($26 035; Figure 2B). Mean total medical costs per patient were $ in year 1 and ranged from $ to $ in years 2 to 5 (calculated on the basis of the number of patients alive at the start of each year; Figure 3A). Mean cardiovascular costs per patient were $ in year 1 and ranged from $6534 to $8216 in years 2 to 5 (Figure 3B). After excluding costs that had incurred during the index quarter (ie, costs associated with the initial hospital admission), mean annual healthcare costs over the remaining 4.75 follow-up years were $29 278/patient-year for the overall population. Average annual healthcare costs per patient were somewhat higher for the HR versus Non-HR group ($ versus $27 891). Discussion Although numerous studies have examined the outcomes of AVR in the elderly, less is known about the outcomes and costs of medical management in such patients. With the recent introduction of effective, less-invasive approaches for the management of AS such as TAVR, there is increasing interest in understanding the outcomes of medically managed AS. To address this gap in knowledge, we used Medicare data to identify a large cohort of patients with AS and evidence of significant heart failure who were managed conservatively and followed this cohort through serial claims data for 5 years. Our study of >2100 patients represents the largest cohort of patients with medically managed severe AS and provides several key insights into this increasingly prevalent condition. Consistent with previous studies, we found that a substantial proportion of elderly patients with severe symptomatic AS do not undergo surgical valve replacement. Indeed, in the Medicare population, only 41% of patients with severe AS and a hospital admission for heart failure underwent surgical AVR within the ensuing 6 months. These findings are similar to those of Iung and colleagues who found that in a European population of patients >75 years of age with severe symptomatic AS (n=216), approximately one third did not undergo subsequent AVR. 8 A retrospective analysis of 740 patients with echocardiographically documented severe AS (defined as an aortic valve area 0.8 cm 2 ) diagnosed between 1993 and 2003 demonstrated that 61% did not undergo AVR. 4 Finally, a more recent US-based analysis demonstrated that among 369 elderly patients with echocardiographically documented severe AS, only 48% underwent AVR over a 16-month median follow-up period a rate that decreased substantially with age but was relatively independent of the predicted operative mortality. 11 The most common reasons why AVR was not performed in this series included patient comorbidities and high perceived operative risk, patient refusal, and concern that the patient s symptoms may not have been related to AS. The second important finding of our study was that the prognosis of severe, symptomatic AS remains dismal without valve replacement. In the Medicare population, 1-year mortality was 49% among patients treated medically, which increased to 88% by 5 years. Survival was even worse among patients who were classified as high risk for surgical AVR with 1-year mortality of 58% and 5-year mortality of 95%. These findings are consistent with numerous studies of the natural history of untreated AS from the presurgical era, 18 as well as series of patients treated medically in the era of valve replacement surgery. 4,11 In the largest previous study of patients with medically managed AS, Varadarajan and colleagues found that survival was somewhat higher than in our population (62% at 1 year and 32% at 5 years). 4 However, their study population was considerably younger than ours (mean age 75 versus 82 years), and the symptomatic status of the patients was unknown. Finally, we found that despite their poor prognosis, elderly patients with severe, symptomatic AS undergoing medical management incurred substantial healthcare related costs. Over the full 5-year follow-up period, mean annual costs (excluding index quarter costs) averaged $30 000/y, with cumulative costs of $63 844/patient for the overall population and $61 146/patient for the HR subgroup. In addition to the high rate of rehospitalization (1.9 admissions per patient-year, in this population) and prolonged length of stay (11.5 hospital days per patient-year), these patients experienced a high rate of admissions to skilled nursing facilities (52%) and use of hospice care (28%). If one assumes that all of these patients represent incident cases of symptomatic AS, extrapolation of these results to the full Medicare population would suggest that the annual cost of patients with medically managed AS is $1.3 billion/y. Even if only half of these cases are truly incident (a realistic lower bound given the observed life

7 Clark et al Outcomes of Medically Managed Aortic Stenosis 703 Figure 3. Mean annual total medical (A) and cardiovascular (B) cost per patient alive at the beginning of each year. Numbers under each year indicate the number of patients who survived to that year. Note that costs in year 1 include costs associated with the index hospitalization. expectancy of 1.8 years), the annual cost of these patients would be $600 million/y. It is important to note, however, that these are the costs associated with the management of patients with AS, rather than the costs of managing the disease itself. Nonetheless, understanding these costs is important particularly when considering the downstream consequences of successful correction of AS by either AVR or TAVR among such patients. Whether these costs would be altered by correction of the underlying valve pathology by either surgical or transcatheter AVR is unknown at this time and requires an understanding of the complex interplay of multiple competing risks in this highly complex patient population. Given the extensive comorbidities present in the study population, it is likely that their medical care costs would remain substantial even if their AS were corrected. Indeed, economic analysis of the Placement of AoRTic TraNscathetER Valve Trial (Cohort B) (PARTNER B) comparing TAVR with conservative management in select patients with severe AS who were not candidates for surgery has demonstrated that TAVR actually increases overall costs to society for this population, both due to the high cost of the TAVR procedure and the costs associated with increased life expectancy. 19 Study Limitations This study has several important limitations. First, the Medicare claims data that formed the basis of our study were not originally collected for the purposes of conducting research but for paying claims. As a result, it was not possible to identify the results of diagnostic tests or severity of illness. We, therefore, used a combination of ICD-9 diagnosis and procedure codes and CPT-4 procedure codes as the main method for identifying our AS patient population as well as comorbid conditions and several clinical outcomes. Consequently, it is possible that some of our population had only mild AS with an underlying cardiomyopathy unrelated to AS as the explanation for their symptoms. However, we did a chart review of 136 patients who met inclusion criteria for our study and found that our approach had a positive predictive value for identifying patients with at least moderately severe AS of 74%. Similarly, it is possible that some patients with severe, symptomatic AS were missed if their symptoms were not severe enough to result in hospitalization for heart failure or because they had other AS-related symptoms such as angina or syncope. Although it would have been possible to include patients with only an outpatient diagnosis of heart failure, we felt that the use of a more restrictive definition requiring both a diagnosis of AS and at least 1 hospital admission for heart failure would define a highly symptomatic population that would have been considered for surgical valve replacement. We, therefore, believe that this approach represents a reasonable compromise between sensitivity and specificity given the limitation of the underlying claims database. Finally, although our goal was to estimate the costs of treating medically managed patients with severe AS who would be potentially eligible for TAVR, given the limitations of our administrative dataset, we were not able to restrict the analysis to patients who would truly be eligible for TAVR in current practice. Indeed, it is likely that some patients included in our analysis would either have been too sick or too healthy for TAVR in current practice. The similarity of our 1- and 5-year mortality results to those from previous studies derived from echocardiographic databases 4,5 and to the 1- and 2-year outcomes of the medical therapy arm of the PARTNER B trial 10 provides reassurance that our approach was reasonable. Despite our inability to provide a definitive diagnosis of severe, symptomatic AS, our use of claims data as the basis for our study provides other important advantages. Most importantly, use of claims data provides near complete capture of important medical resource utilization and is an ideal method for identifying the costs of care to the Medicare program a key goal of our study. Nonetheless, it is likely that we have underestimated the overall economic burden of untreated AS for 2 reasons. First, despite the fact that Medicare Part D pharmaceutical benefits were introduced in 2006, the standard analytic files do not currently include claims for prescription drug use. In addition, the Medicare Program pays only a portion of all skilled nursing home care, the majority of which is covered by state Medicaid programs. Thus, our estimates of annual and cumulative healthcare costs should be considered a lower bound of the true costs for the study population.

8 704 Circ Cardiovasc Qual Outcomes September 2012 Conclusions Elderly patients with severe symptomatic AS undergoing medical management have a limited long-term survival, but still incur substantial costs to the Medicare Program. These findings have important implications for policy makers interested in better understanding the cost-effectiveness of emerging treatment options such as TAVR for patients with severe, symptomatic AS who are currently managed medically. Acknowledgments No other persons have made substantial contributions to this manuscript. Sources of Funding Edwards Lifesciences, LLC. Disclosures A.K. Thompson and Dr Duhay are employees of Edwards Lifesciences. M.A. Clark, B.T. Stockwell, and Dr Keyes are consultants on projects funded by industry including Edwards Lifesciences. Dr Cohen receives research grant support from Edwards Lifesciences and Medtronic. Dr Bonow has a consulting relationship with Edwards Lifesciences. Dr Svensson serves as a clinical investigator in studies funded by industry but has no direct financial relationship with industry to disclose. Dr Arnold has no relevant financial disclosures. References 1. Bach DS, Radeva JI, Birnbaum HG, Fournier AA, Tuttle EG. Prevalence, referral patterns, testing, and surgery in aortic valve disease: leaving women and elderly patients behind? J Heart Valve Dis. 2007;16: Nkomo VT, Gardin JM, Skelton TN, Gottdiener JS, Scott CG, Enriquez- Sarano M. Burden of valvular heart diseases: a population-based study. Lancet. 2006;368: Lindroos M, Kupari M, Heikkilä J, Tilvis R. Prevalence of aortic valve abnormalities in the elderly: an echocardiographic study of a random population sample. J Am Coll Cardiol. 1993;21: Varadarajan P, Kapoor N, Bansal RC, Pai RG. Clinical profile and natural history of 453 nonsurgically managed patients with severe aortic stenosis. Ann Thorac Surg. 2006;82: Varadarajan P, Kapoor N, Bansal RC, Pai RG. Survival in elderly patients with severe aortic stenosis is dramatically improved by aortic valve replacement: Results from a cohort of 277 patients aged > or =80 years. Eur J Cardiothorac Surg. 2006;30: Kang DH, Park SJ, Rim JH, Yun SC, Kim DH, Song JM, Choo SJ, Park SW, Song JK, Lee JW, Park PW. Early surgery versus conventional treatment in asymptomatic very severe aortic stenosis. Circulation. 2010;121: Pellikka PA, Sarano ME, Nishimura RA, Malouf JF, Bailey KR, Scott CG, Barnes ME, Tajik AJ. Outcome of 622 adults with asymptomatic, hemodynamically significant aortic stenosis during prolonged follow-up. Circulation. 2005;111: Iung B, Cachier A, Baron G, Messika-Zeitoun D, Delahaye F, Tornos P, Gohlke-Barwolf C, Boersma E, Ravaud P, Vahanian A. Decision-making in elderly patients with severe aortic stenosis: why are so many denied surgery? Eur Heart J. 2005;26: Bouma BJ, van Den Brink RB, van Der Meulen JH, Verheul HA, Cheriex EC, Hamer HP, Dekker E, Lie KI, Tijssen JG. To operate or not on elderly patients with aortic stenosis: the decision and its consequences. Heart. 1999;82: Leon MB, Smith CR, Mack M, Miller DC, Moses JW, Svensson LG, Tuzcu EM, Webb JG, Fontana GP, Makkar RR, Brown DL, Block PC, Guyton RA, Pichard AD, Bavaria JE, Herrmann HC, Douglas PS, Petersen JL, Akin JJ, Anderson WN, Wang D, Pocock S; PARTNER Trial Investigators. Transcatheter aortic-valve implantation for aortic stenosis in patients who cannot undergo surgery. N Engl J Med. 2010;363: Bach DS, Siao D, Girard SE, Duvernoy C, McCallister BD, Jr., Gualano SK. Evaluation of patients with severe symptomatic aortic stenosis who do not undergo aortic valve replacement: the potential role of subjectively overestimated operative risk. Circ Cardiovasc Qual Outcomes. 2009;2: Bakaeen FG, Chu D, Ratcliffe M, Gopaldas RR, Blaustein AS, Venkat R, Huh J, LeMaire SA, Coselli JS, Carabello BA. Severe aortic stenosis in a veteran population: treatment considerations and survival. Ann Thorac Surg. 2010;89: Clark MA, Bakhai A, Lacey MJ, Pelletier EM, Cohen DJ. Clinical and economic outcomes of percutaneous coronary interventions in the elderly: an analysis of medicare claims data. Circulation. 2004;110: Toumpoulis IK, Anagnostopoulos CE, Toumpoulis SK, DeRose JJ, Jr., Swistel DG. EuroSCORE predicts long-term mortality after heart valve surgery. Ann Thorac Surg. 2005;79: Wendt D, Osswald B, Thielmann M, Kayser K, Tossios P, Massoudy P, Kamler M, Jakob H. The EuroSCORE - still helpful in patients undergoing isolated aortic valve replacement? Interact Cardiovasc Thorac Surg. 2010;10: Dewey TM, Brown D, Ryan WH, Herbert MA, Prince SL, Mack MJ. Reliability of risk algorithms in predicting early and late operative outcomes in high-risk patients undergoing aortic valve replacement. J Thorac Cardiovasc Surg. 2008;135: Pope GC, Kautter J, Ellis RP, Ash AS, Ayanian JZ, Lezzoni LI, Ingber MJ, Levy JM, Robst J. Risk adjustment of Medicare capitation payments using the CMS-HCC model. Health Care Financ Rev. 2004;25: Ross J, Jr., Braunwald E. Aortic stenosis. Circulation. 1968;38(1 Suppl): Reynolds MR, Magnuson EA, Wang K, Lei Y, Vilain K, Walczak J, Kodali SK, Lasala JM, O'Neill WW, Davidson CJ, Smith CR, Leon MB, Cohen DJ. PARTNER Investigators. Cost-effectiveness of transcatheter aortic valve replacement compared with standard care among inoperable patients with severe aortic stenosis: results from the placement of aortic transcatheter valves (PARTNER) trial (Cohort B). Circulation. 2012;125: