Specialist-directed care can improve outcomes across a

Thirty-Day Outcomes in Medicare Patients With Heart Failure at Heart Transplant Centers Scott L. Hummel, MD, MS; Natalie P. Pauli, MD; Harlan M. Krumholz, MD, SM; Yun Wang, PhD; Jersey Chen, MD, MPH; Sharon-Lise T. Normand, PhD; Brahmajee K. Nallamothu, MD, MPH Background Heart transplant centers are generally considered centers of excellence for heart failure care. However, their overall performance has not previously been evaluated in a broad population of elderly patients with heart failure, many of whom are not transplant candidates. Methods and Results We identified 1 million elderly Medicare beneficiaries who were hospitalized for heart failure between 2004 and 2006 at 4500 hospitals. We calculated 30-day risk-standardized mortality rates and standardized mortality ratios as well as 30-day risk-standardized readmission rates and standardized readmission ratios at heart transplant centers and non heart transplant hospitals using risk-standardization models that the Centers for Medicare & Medicaid Services uses for public reporting. The 30-day risk-standardized mortality rates were lower at heart transplant centers than non heart transplant hospitals nationally (10.6% versus 11.5%, P 0.001) but were similar at peer institutions offering coronary artery bypass grafting within the same geographical region (10.6% versus 10.6%, P 0.96). The mean standardized mortality ratio for heart transplant centers was 0.9 (SD, 0.1; range, 0.7 to 1.3). No differences were noted in 30-day risk-standardized readmission rates between heart transplant centers and non heart transplant hospitals nationally (23.6% versus 23.8%, P 0.55). The mean standardized readmission ratio for heart transplant centers was 1.0 (SD, 0.1; range, 0.8 to 1.2). Conclusions In elderly Medicare patients with heart failure, heart transplant centers have lower 30-day risk-standardized mortality rates than non heart transplant hospitals nationally; however, this difference is not present in comparison with peer institutions or for 30-day risk-standardized readmission rates. (Circ Heart Fail. 2010;3:244-252.) Key Words: heart failure morbidity mortality Medicare hospitals outcomes Specialist-directed care can improve outcomes across a wide range of medical conditions, including heart failure (HF). 1 4 Hospitals performing heart transplants require dedicated infrastructure to evaluate and treat the sickest patients with HF and, in many ways, represent the ultimate specialistdirected care for this condition. However, HF outcomes at heart transplant centers have not been previously reported across a broad population of elderly patients, most of whom are not transplant candidates. Clinical Perspective on p 252 Elderly patients who are hospitalized for HF have a high risk for death and suffer frequent readmissions after discharge, 5 and the risk for these outcomes varies markedly across hospitals. 1,6 8 This observation is not entirely explained by patient characteristics, 1,6 8 but little is known about the hospital-level factors that contribute to outcomes in the elderly HF population. It is plausible that the expertise at heart transplant centers improves outcomes for patients with HF hospitalized there, even those not eligible for transplantation. If this is the case, adopting some of the related processes of care at the much larger number of nontransplant hospitals might improve HF outcomes in elderly patients. Accordingly, we used recently developed and validated risk-adjustment models to assess short-term HF outcomes in Medicare patients with HF hospitalized at heart transplant centers. We hypothesized that heart transplant centers would have lower risk-standardized 30-day mortality and readmission rates than hospitals that did not perform transplantation. Methods Study Sample We used 100% of the Medicare Provider Analysis and Review Part A inpatient files between 2004 and 2006 obtained from the Centers for Received June 5, 2009; accepted December 17, 2009. From the Division of Cardiovascular Medicine (S.L.H., B.K.N.), Department of Internal Medicine, University of Michigan, Ann Arbor, Mich; Division of General Medicine (N.P.P.), Department of Internal Medicine, Emory University, Atlanta, Ga; Division of Cardiovascular Medicine (H.M.K., J.C.), Department of Internal Medicine, Yale University School of Medicine, New Haven, Conn; Center for Outcomes Research and Evaluation (H.M.K., Y.W.), Yale New Haven Hospital, New Haven, Conn; Department of Health Care Policy (S.-L.T.N.), Harvard Medical School and Department of Biostatistics, Harvard School of Public Health, Boston, Mass. Guest Editor for this article was Salvador Borges-Neto, MD. Correspondence to Scott L. Hummel, MD, MS, University of Michigan, 1500 E Medical Center Dr/SPC 5853, Ann Arbor, MI 48109-5853, E-mail scothumm@med.umich.edu 2010 American Heart Association, Inc. Circ Heart Fail is available at http://circheartfailure.ahajournals.org DOI: 10.1161/CIRCHEARTFAILURE.109.884098 244

Hummel et al Medicare HF Outcomes at Transplant Centers 245 Medicare & Medicaid Services (CMS). These files included patientlevel demographic information, principal and secondary International Classification of Diseases, Ninth Revision, Clinical Modification (ICD- 9-CM) diagnosis codes, and procedure codes on each hospitalization for the Medicare population. The study population included Medicare patients aged 65 years or older hospitalized with a principal discharge diagnosis of HF by ICD-9-CM diagnostic codes 402.01, 402.11, 402.91, 404.01, 404.03, 404.11, 404.13, 404.91, 404.93, or 428.xx. We included discharges for patients enrolled in Medicare fee-forservice for at least 1 year because we used 12 months of previous utilization data to obtain information on comorbidities (see below). We excluded all discharges in patients who were discharged alive, not against medical advice, and within the first day of admission because we deemed these patients unlikely to have had truly decompensated HF (6% of total discharges for mortality cohort). Because we sought to focus on outcomes in routine patients with HF, we also excluded discharges for patients with a history of heart transplantation (ICD-9-CM procedure code 37.51) or mechanical circulatory support (ICD-9-CM procedure codes 37.52 to 37.54 and 37.62 to 37.68) (n 799 for the mortality analysis; n 1110 for the readmission analysis). We also excluded discharges from hospitals with 10 or fewer HF cases during the study period (n 1050 for the mortality analysis; n 823 for the readmission analysis). By using publicly available data from the United Network of Organ Sharing, we identified 110 heart transplant centers that operated during the study period, including 14 small-volume ( 10 transplants per year) heart transplant programs that were not approved by Medicare for heart transplant-related care but provided inpatient services for elderly Medicare patients with HF. We included these facilities because we judged them likely to have HF-related infrastructure similar to that at other heart transplant centers. We then identified all other non heart transplant centers that cared for Medicare patients with HF during the study period. These were further stratified into nonpeer or peer institutions, with the latter defined as those with 300 beds and providing coronary artery bypass grafting (CABG). For hospitalizations that involved interhospital transfer during an admission, we linked both hospitalizations into a single episode of care. For analyses related to mortality, outcomes were attributed to the initial admitting hospital; for analyses related to readmission, outcomes were attributed to the discharging hospital. In all cases, comorbidities were identified from data collected at the initial admitting hospital to avoid classifying in-hospital complications as comorbidities. For analyses related to mortality, if a patient had multiple HF hospitalizations within a single calendar year, 1 admission was randomly selected for the analysis. For readmission analyses, 1 discharge was randomly selected if the multiple HF hospitalizations occurred within a 30-day period and at the same hospital. This selection process led to different sample sizes between mortality and readmission (Tables 1 and 2). Outcome Variables and Risk-Standardization Models Our primary outcomes were (1) all-cause mortality within 30 days from the date of the index HF admission and (2) all-cause hospital readmission within 30 days of the date of discharge from the index HF hospitalization for in-hospital survivors. For each outcome, separate study populations were constructed as described previously here. We used risk-standardization models that CMS currently uses for public reporting. Details have been previously published of the mortality 6 and readmission 7 risk standardization models. In summary, they are 2-level (patient and hospital) hierarchical random intercept models with patient characteristics as fixed effects at the patient level and random effects at the hospital level. No hospital characteristics were adjusted for random effects. 9 Importantly, these administrative data-based models perform well compared with models that use abstracted clinical chart data and have been endorsed by the National Quality Forum. In brief, the risk-standardization models used patient-level administrative data in 2 demographic, 12 cardiovascular, and 29 comorbidity categories (Tables 1 and 2) obtained from Medicare Part A (ie, inpatient and hospital outpatient data) and B (ie, physician office outpatient data) files. These files include facility claims for inpatient care, ambulatory surgery, surgical or diagnostic procedures, and outpatient provider encounters. We used data on comorbidities from the index admission and utilization over the 12 months before the index admission to adjust for patient-specific risk. We obtained mortality data through the Medicare enrollment database. Statistical Analysis We compared baseline characteristics of patients by hospital type heart transplant centers and non heart transplant hospitals (including both nonpeer and peer institutions) using 2 tests for categorical variables and ANOVA for continuous variables. By using data from all hospitals included in this study, we then calculated a 30-day standardized mortality ratio (SMR) for each hospital using hierarchical generalized linear models. 6 This technique accounts for patient characteristics and addresses the potential effects of patient clustering within hospitals. The SMR was calculated as the predicted mortality in each hospital, given its patient mix and estimated hospital-specific effect, divided by the expected mortality in that hospital, given the same patient mix and the average hospitalspecific effect. We used a bootstrapping algorithm within each institution to generate the 95% CI estimates of SMRs and their variances. To investigate hospital performance at heart transplant centers and nontransplant hospitals, each hospital was categorized on the basis of the 95% CI estimates of the SMRs into 1 of the following 3 groups: predicted mortality significantly better than expected mortality (ie, SMR significantly 1.0), predicted mortality no different than expected, or predicted mortality significantly worse than expected (ie, SMR significantly 1.0). We calculated hospital-specific risk-standardized mortality rates (s) by multiplying each hospital s SMR by the national unadjusted rate. This form of indirect standardization allowed for assessments of overall quality in patients with HF at a hospital for the type of patients it treats, although direct hospital-to-hospital comparisons may be limited. 10 To further examine hospital performance at heart transplant and non heart transplant hospitals nationwide, we categorized each hospital based on its into 1 of the following 3 groups: s below the 25th percentile, s between the 25th and 75th percentiles, and s above the 75th percentile. We then calculated the numbers of heart transplant and non heart transplant hospitals within these 3 categories. For non heart transplant hospitals, we separately calculated the numbers of nonpeer and peer institutions (as defined previously here) within each category. To further account for potential differences in patients across healthcare markets, we repeated these analyses after limiting the study population to patients at non heart transplant hospitals that were located in the same hospital referral regions (HRRs) as a heart transplant center according to the Dartmouth Atlas of Health Care. In the same manner, 30-day standard readmission ratios (SRRs) and risk-standardized readmission rates (s) were calculated for each hospital. We again categorized hospitals based on their SRR (significantly lower than expected, as expected, or significantly worse than expected) and ( 25th percentile, 25th to 75th percentile, and 75th percentile). We conducted all analyses using SAS statistical software version 9.1.3. Results Between 2004 and 2006, we identified 1 318 899 discharges in Medicare patients with HF at 4570 hospitals in our study sample for assessing 30-day mortality and 1 610 803 discharges in Medicare patients with HF at 4607 hospitals in our study sample for assessing 30-day readmission rates. Among the study sample for 30-day mortality, 73 778 (5.6%) patients were admitted at 110 heart transplant centers, 289 710 (22.0%) were admitted at peer institutions with CABG, and 955 411 (72.4%) were admitted at nonpeer institutions. This pattern of admission across hospital types was similar in the study sample for 30-day readmission rates.

246 Circ Heart Fail March 2010 Table 1. Patient Characteristics: 30-Day Mortality Cohort Overall (n 1 318 899) Heart Transplant Centers (n 73 778) Nontransplant Hospitals (n 1 245 121) Characteristic n % n % n % Demographic Mean (SD) age, y 80.5 (7.9) 79.2 (7.9) 80.5 (7.9) Female gender 765 507 58.0 39 868 54.0 725 639 58.3 Cardiovascular History of PTCA 88 213 6.7 5880 8.0 82 333 6.6 History of CABG 155 517 11.8 9363 12.7 146 154 11.7 History of HF 964 237 73.1 56 138 76.1 908 099 72.9 History of AMI 125 536 9.5 7670 10.4 117 866 9.5 Unstable angina 198 369 15.0 12 308 16.7 186 061 14.9 Chronic atherosclerosis 936 559 71.0 53 744 72.9 882 815 70.9 Cardiopulmonary failure and shock 231 393 17.5 14 165 19.2 217 228 17.5 Valvular heart disease 621 677 47.1 38 832 52.6 582 845 46.8 Comorbidity Hypertension 1 091 295 82.7 61 470 83.3 1 029 825 82.7 Stroke 139 271 10.6 7805 10.6 131 466 10.6 Renal failure 370 408 28.1 24 266 32.9 346 142 27.8 COPD 619 421 47.0 29 682 40.2 589 739 47.4 Pneumonia 507 818 38.5 27 773 37.6 480 045 38.6 Diabetes 640 219 48.5 35 926 48.7 604 293 48.5 Protein-calorie malnutrition 70 058 5.3 4001 5.4 66 057 5.3 Dementia 267 671 20.3 12 825 17.4 254 846 20.5 Paralysis/functional disability 90 006 6.8 5151 7.0 84 855 6.8 Peripheral vascular disease 423 348 32.1 25 104 34.0 398 244 32.0 Metastatic cancer 53 233 4.0 3336 4.5 49 897 4.0 Trauma in past year 450 620 34.2 23 943 32.5 426 677 34.3 Major psychiatric disorders 113 363 8.6 6124 8.3 107 239 8.6 Chronic liver disease 25 191 1.9 1844 2.5 23 347 1.9 Observed mortality rates 30-day rate 150 160 11.4 7334 9.9 142 826 11.5 PTCA indicates percutaneous transluminal coronary angioplasty; AMI, acute myocardial infarction; COPD, chronic obstructive pulmonary disease. For the study samples examining 30-day mortality and readmission, baseline patient characteristics found in each hospital type are summarized in Tables 1 and 2. In general, patients admitted at heart transplant centers were younger, more likely to be male gender, and had higher prevalence of previous heart disease than those at nontransplant hospitals. There was also a higher prevalence of cardiopulmonary respiratory failure or shock and renal failure in the heart transplant center cohort but a lower prevalence of chronic obstructive pulmonary disease, pneumonia, and dementia (Tables 1 and 2). Patients admitted to peer institutions with CABG were more similar to those at heart transplant centers but still had slightly lower prevalence of previous heart disease, shock, and renal failure (data available from authors by request). The observed 30-day mortality rates were 9.9% versus 11.5% (P 0.001), respectively, for Medicare patients with HF at heart transplant centers compared with those at non heart transplant hospitals nationally. After adjustment for patient differences and hospital-level effects, the 30-day s remained significantly lower at heart transplant centers (10.6% versus 11.5%; P 0.001). When heart transplant centers were compared with peer institutions with CABG nationally, s remained significantly lower at heart transplant centers (10.6% versus 11.2%; P 0.015). However, no differences in s were noted when heart transplant centers were compared with peer institutions with CABG within the same HRRs (10.6% versus 10.6%; P 0.96). These findings are summarized in Table 3. The percentage of each hospital type within categories ( 25th percentile, between the 25th and 75th percentiles, and 75th percentile) is shown in Figure 1. Inspection of the SMRs for heart transplant centers revealed 25 institutions (22.7%) with SMRs significantly lower

Hummel et al Medicare HF Outcomes at Transplant Centers 247 Table 2. Patient Characteristics: 30-Day Readmission Cohort Overall (n 1 610 803) Heart Transplant Centers (n 98 631) Nontransplant Hospitals (n 1 512 172) Characteristic n % n % n % Demographic Mean (SD) age, y 80.1 (7.8) 78.7 (7.7) 80.2 (7.8) Female gender 922 968 57.3 51 718 52.4 871 250 57.6 Cardiovascular History of CABG 207 878 12.9 13 978 14.2 193 900 12.8 History of HF 977 257 60.7 63 277 64.2 913 980 60.4 History of AMI 252 579 15.7 16 875 17.1 235 704 15.6 Arrhythmias 744 859 46.2 49 685 50.4 695 174 46.0 Cardiopulmonary failure and shock 204 241 12.7 13 259 13.4 190 982 12.6 Valvular heart disease 630 810 39.2 43 518 44.1 587 292 38.8 Vascular or circulatory disease 451 116 28.0 28 736 29.1 422 380 27.9 Angina pectoris, old MI 1 075 376 66.8 68 233 69.2 1 007 143 66.6 Other heart disease 255 083 15.8 16 153 16.4 238 930 15.8 Comorbidity Stroke 87 418 5.4 5202 5.3 82 216 5.4 Renal failure 394 160 24.5 28 228 28.6 365 932 24.2 COPD 679 354 42.2 35 693 36.2 643 661 42.6 Pneumonia 464 043 28.8 26 430 26.8 437 613 28.9 Diabetes 714 673 44.4 43 635 44.2 671 038 44.4 Other endocrine disease 506 163 31.4 32 014 32.5 474 149 31.4 Other urinary tract disorders 456 342 28.3 30 276 30.7 426 066 28.2 Decubitus ulcer or skin ulcer 127 908 7.9 7,697 7.8 120 211 8.0 Other gastrointestinal disorders 598 795 37.2 35 316 35.8 563 479 37.3 Peptic ulcer, hemorrhage, other 186 264 11.6 11 238 11.4 175 026 11.6 Severe hematologic disorders 41 729 2.6 2876 2.9 38 853 2.6 Nephritis 44 293 2.8 2997 3.0 41 296 2.7 Protein-calorie malnutrition 71 452 4.4 4302 4.4 67 150 4.4 Dementia 233 982 14.5 11 442 11.6 222 540 14.7 Paralysis/functional disability 71 215 4.4 4277 4.3 66 938 4.4 Metastatic cancer 26 892 1.7 1899 1.9 24 993 1.7 Cancer 190 606 11.8 12 422 12.6 178 184 11.8 Major psychiatric disorders 73 552 4.6 4329 4.4 69 223 4.6 End-stage liver disease 84 557 5.3 5795 5.9 78 762 5.2 End-stage renal disease 41 213 2.6 3218 3.3 37 995 2.5 Asthma 79 834 5.0 5644 5.7 74 190 4.9 Iron deficiency and anemia 580 285 36.0 34 190 34.7 546 095 36.1 Drug alcohol abuse 129 747 8.1 8164 8.3 121 583 8.0 Depression 167 017 10.4 9368 9.5 157 649 10.4 Other psychiatric disorders 101 346 6.3 5011 5.1 96 335 6.4 Lung fibrosis 122 439 7.6 6680 6.8 115 759 7.7 Observed readmission rates 30-day rate 382 522 23.8 23 481 23.8 359 041 23.7 PTCA indicates percutaneous transluminal coronary angioplasty; AMI, acute myocardial infarction; MI, myocardial infarction; COPD, chronic obstructive pulmonary disease. than the average expected mortality ratio for all hospitals (defined by 95% CIs with an upper limit excluding 1.0). Of the remaining heart transplant centers, 2 (1.8%) had SMRs significantly 1.0. The mean SMR for heart transplant centers was 0.9 (SD, 0.1; range, 0.7 to 1.3) [Figure 2A]. In comparison, 197 (4.4%) hospitals without heart transplantation had SMRs significantly 1.0, and 187 (4.2%) had SMRs significantly 1.0. The mean SMR for non heart

248 Circ Heart Fail March 2010 Table 3. Distributions of s by Hospital Type Mean (SD) Range (Min Max) Overall Low ( 25th) Middle (25th 75th) High ( 75th) Low High ( 25th) ( 75th) Hospital Group 30-Day Mortality Total (n) (SE) Total (SE) Total (SE) Total (SE) All HRRs Overall 4570 11.5 (1.5) 1142 9.7 (0.7) 2285 11.4 (0.5) 1143 13.4 (1.0) 5.9 10.5 10.5 12.4 12.4 19.7 Heart transplant center 110 10.6 (1.5) 47 9.2 (0.8) 49 11.4 (0.5) 14 13.0 (0.5) 7.4 10.4 10.6 12.3 12.5 14.2 Non heart transplant hospital 4460 11.5 (1.5) 1095 9.7 (0.7) 2236 11.3 (0.5) 1129 13.4 (1.0) 5.9 10.5 10.5 12.4 12.4 19.7 Hospital with CABG 431 11.2 (1.7) 153 9.4 (0.9) 168 11.3 (0.5) 110 13.4 (0.9) 6.3 10.5 10.5 12.3 12.4 17.4 Other hospital 4029 11.5 (1.5) 942 9.7 (0.6) 2068 11.4 (0.5) 1019 13.4 (1.0) 5.9 10.5 10.5 12.4 12.4 19.7 Heart transplant center HRRs Overall 2094 11.3 (1.5) 631 9.6 (0.8) 1025 11.4 (0.5) 438 13.4 (1.0) 5.9 10.5 10.5 12.4 12.4 18.0 Non heart transplant hospital 1984 11.3 (1.5) 584 9.6 (0.7) 976 11.4 (0.5) 424 13.5 (1.0) 5.9 10.5 10.5 12.4 12.4 18.0 Hospital with CABG 196 10.6 (1.5) 90 9.3 (1.0) 76 11.2 (0.6) 30 13.5 (1.0) 6.3 10.5 10.5 12.3 10.5 14.2 Other hospital 1788 11.4 (1.5) 494 9.6 (0.7) 900 11.4 (0.5) 394 13.5 (1.0) 5.8 10.5 10.5 12.4 12.4 18.0 Middle (25th 75th) transplant hospitals was 1.01 (SD, 0.1; range, 0.5 to 1.7). In the subset of peer institutions with CABG, 110 (11.2%) had SMRs significantly 1.0, and 85 (8.6%) had SMRs significantly 1.0. In contrast to the these findings, observed 30-day readmission rates in Medicare patients with HF were not different between heart transplant centers and other nontransplant hospitals nationally (23.8% versus 23.8%; P 0.65). After adjustment for patient differences and hospital-level effects, 30-day s remained similar between heart transplant centers and other hospitals (23.6% versus 23.8%; P 0.55). 100% 80% 60% 40% 12.7% 44.5% These findings are summarized in Table 4. The percentage of each hospital type within categories ( 25th percentile, between the 25th and 75th percentiles, and 75th percentile) is shown in Figure 3. Inspection of the SRRs for heart transplant centers revealed 18 (16.4%) hospitals with SRRs significantly lower than the average expected readmission ratio for all hospitals (defined by 95% CIs with an upper limit excluding 1.0). Of the remaining heart transplant centers, 16 (14.5%) had SRRs significantly 1.0. The mean SRR for heart transplant centers was 1.0 (SD, 0.1; range, 0.8 and 1.2) [Figure 2B]. In <25th percentile 25-75th percentile >75th percentile 25.5% 25.3% 39.0% 51.3% 15.3% 38.8% 22.0% 50.3% 20% 42.7% 35.5% 5% 23.4% 45.9% 27.6% 0% Heart Transplant All HRR: Large Center (N=110) teaching hospital with CABG (N=431) All HRR: Other Heart Transplant Heart Transplant (N=4,029) Center HRR: Large Center HRR: Other teaching hospital (N=1,788) with CABG (N=196) Hospital Type Figure 1. Percentage of facilities in percentile groups by hospital type. This figure reflects aggregate information by hospital type, and no significant differences between individual hospitals of each type should be assumed. All HRR indicates national comparisons; Heart transplant center HRR, comparisons between hospital types among facilities located in the same geographic area as a heart transplant center.

Hummel et al Medicare HF Outcomes at Transplant Centers 249 Figure 2. Distributions of SMRs (A) and SRRs (B) at heart transplant centers. Heart transplant center number (1 to 110, denoted on x-axis) is consistent across both panels to enable graphical comparison of mortality and readmission ratios at each individual center. comparison, 196 (4.4%) non heart transplant hospitals had SRRs significantly 1.0, whereas 237 (5.3%) had SRRs significantly 1.0. The mean SRR for non heart transplant hospitals was 1.0 (SD, 0.1; range, 0.7 to 1.4). Table 4. Hospital Group 30-Day Readmission All HRRs Distributions of s by Hospital Type Total (n) Mean (SD) Overall Low ( 25th) Middle (25th to 75th) High ( 75th) (SE) Total (n) (SE) Total (n) Discussion Across a broad population of elderly Medicare patients with HF, we found that heart transplant centers had lower 30-day mortality rates than non heart transplant hospitals nationally. (SE) Total (n) (SE) Low ( 25th) Range (Min Max) Middle (25th to 75th) High ( 75th) Overall 4607 23.8 (1.9) 1151 21.5 (0.9) 2305 23.7 (0.6) 1151 26.3 (1.3) 17.1 22.6 22.6 24.8 24.8 32.6 Heart transplant 110 23.6 (2.2) 43 21.5 (0.9) 38 23.8 (0.7) 29 26.6 (1.3) 18.9 22.6 22.7 24.8 24.8 29.3 center Non heart 4497 23.8 (1.9) 1108 21.5 (0.9) 2267 23.7 (0.6) 1122 26.3 (1.3) 17.1 22.6 22.5 24.8 24.8 32.6 transplant center Hospital with 432 23.3 (2.2) 167 21.2 (1.0) 164 23.6 (0.6) 101 26.3 (1.3) 17.3 22.6 22.6 24.8 24.9 30.4 CABG Other hospital 4065 23.8 (1.8) 941 21.6 (0.9) 2103 23.7 (0.6) 1021 26.3 (1.3) 17.1 22.6 22.6 24.8 24.8 32.6 Heart Transplant Center HRRs Overall 2113 24.0 (2.0) 470 21.6 (0.8) 1023 23.7 (0.6) 620 26.4 (1.3) 17.5 22.6 22.6 24.8 24.8 31.6 Non heart 2003 24.0 (1.9) 427 21.6 (0.8) 985 23.7 (0.6) 591 26.4 (1.3) 17.5 22.6 22.6 24.8 24.8 31.6 transplant hospital Hospital with 197 24.0 (2.2) 55 21.4 (0.9) 78 23.7 (0.7) 64 26.5 (1.4) 18.5 22.6 22.6 24.8 24.9 30.4 CABG Other hospital 1806 24.0 (1.9) 372 21.6 (0.8) 907 23.7 (0.6) 527 26.3 (1.3) 17.5 22.6 22.6 24.8 24.8 31.6

250 Circ Heart Fail March 2010 100% <25th percentile 25-75th percentile >75th percentile 80% 26.4% 23.4% 25.1% 32.5% 29.2% 60% 34.5% 38.0% 40% 51.7% 39.6% 50.2% 20% 39.1% 38.7% 23.1% 27.9% 20.6% 0% Heart Transplant All HRR: Large Center (N=110) teaching hospital with CABG (N=432) All HRR: Other Heart Transplant (N=4,065) Center HRR: Large teaching hospital with CABG (N=197) Hospital Type Heart Transplant Center HRR: Other (N=1,806) Figure 3. Percentage of facilities in percentile groups by hospital type. This figure reflects aggregate information by hospital type, and no significant differences between individual hospitals of each type should be assumed. All HRR indicates national comparisons; Heart transplant center HRR, comparisons between hospital types among facilities located in the same geographic area as a heart transplant center. This difference diminished after adjusting for patient characteristics and between-hospital variance across facilities. We noted no significant differences in 30-day risk-adjusted mortality between heart transplant centers and peer institutions providing CABG surgery in the same geographic area. Finally, 30-day risk-adjusted readmission rates were similar in Medicare patients with HF discharged from heart transplant centers and nontransplant hospitals. Within 30 days of hospital discharge for HF, 10% to 15% of elderly Medicare patients are readmitted to an acute care hospital, and 10% have died. 5 In our national sample of elderly Medicare patients with HF, we observed large between-hospital variations in 30-day s and s. Previous work by our group 6 and others 11,12 has suggested that much of this variation is due to hospital-level characteristics rather than to patient characteristics or guideline adherence. Despite the lack of clinical trial evidence guiding the management of acutely decompensated HF, 13 a significant number of short-term deaths and rehospitalizations seem to be preventable. 14,15 Understanding the hospital-level factors that contribute to early mortality and readmission is critical to improving outcomes in the elderly HF population. The main findings of our study are somewhat unexpected. Heart transplant centers are generally considered centers of excellence for HF care because of the dedicated infrastructure needed to evaluate and treat complex and often critically ill patients with HF. We anticipated that the expertise present within successful heart transplant programs also would translate into better short-term outcomes in the general elderly HF population. Indeed, we did find lower 30-day mortality rates at heart transplant centers compared with all nontransplant hospitals. However, large, comparably resourced peer institutions offering CABG surgery in the same geographic area had similar mortality outcomes to heart transplant centers. This finding suggests that the lower mortality rates in elderly Medicare patients with HF at heart transplant centers are unrelated to the presence of a transplant program and may be due to more traditional hospital-level factors. These characteristics could include better adherence to guideline-based performance measures, 16 higher hospital HF case volumes, 17 and increased likelihood of cardiologist-directed HF care, 4,18 none of which are specific to heart transplant centers. When we evaluated 30-day readmission rates, the results suggested even more equivalence between heart transplant centers and non heart transplant hospitals perhaps because short-term hospital readmission in elderly patients with HF is often multifactorial and may depend, in part, on issues that are only indirectly related to disease severity or inpatient HF management. 14,19,20 The administrative risk-standardization model that we used for 30-day readmission is less predictive than that for 30-day mortality, 6,7 which could limit the ability of heart transplant centers to demonstrate improved readmission outcomes. Nonetheless, the overlap of the point estimates for SRR and at heart transplant centers and nontransplant hospitals is striking. Transplant-eligible patients with HF are generally younger and have fewer comorbidities than the population we studied. Accordingly, these patients may have different predictors of outcomes or responses to interventions. However, it is likely that heart transplant centers use specific resources in this group that could potentially benefit elderly patients with HF, who often present with multiple medical comorbidities, polypharmacy, functional limitations, and suboptimal social support systems. As others have noted, 14,19 up to half of

Hummel et al Medicare HF Outcomes at Transplant Centers 251 rehospitalizations in this population are related to failures in HF self-care measures and postdischarge care coordination. All CMS-approved solid-organ transplant programs (including heart transplant programs) currently are required to provide multidisciplinary care and hospital discharge planning for pre- and posttransplant patients. The physician-led team must include a care coordinator and representatives from nursing, social work, nutritional services, and pharmacology. 21 Such a care model would seem ideally suited for elderly patients with HF because several studies have shown that multidisciplinary hospital discharge education, care transition planning, and outpatient management can improve outcomes in this population. 22 24 Promoting such spillover effects into nontransplant candidates is not currently required by CMS for heart transplant centers. Kidney transplant centers that provide outpatient dialysis provide an interesting contrast because multidisciplinary care plans are mandated for all patients with end-stage renal disease at these centers whether or not they are listed for transplantation. 21 We speculate that many heart transplant centers apply their extensive expertise and infrastructure for HF care primarily in patients who may be transplant candidates and less so in the general population of patients with HF. If the specialized resources available at heart transplant centers are indeed focused only on the small number of transplant evaluations, an important opportunity to understand and improve postdischarge outcomes in the larger elderly HF population may be missed. Limitations Our findings should be considered in the context of the following study design issues and limitations. The riskstandardization models we used in this study rely on administrative data and may not fully account for all factors that affect mortality and readmission. However, the models have been extensively validated using clinical data from chart review and are supported by CMS for comparison and public reporting of hospital performance. 6,7 Another related concern is the potential for referral to heart transplant centers for acute care hospitalization and with follow-up care provided locally, which could lead to challenges in attributing overall performance in HF outcomes to a single facility. Our focus on 30-day events was meant to minimize this possibility because these short-term outcomes are most likely to be influenced by immediate care provided during the inpatient setting. Although the CMS models and public outcomes reporting are based on this time frame, it is conceivable that a longer duration of follow-up could reveal further outcome differences between heart transplant centers and nontransplant hospitals. The administrative data sources we used did not allow us to directly assess the quality of HF care delivered at heart transplant centers and nontransplant hospitals. Future work will need to focus on these areas because it could provide details that better explain our findings. In addition, it will be important in subsequent analyses to understand what type of resources at heart transplant centers may be available to a broad population of patients with HF. Finally, it is important to note that the results of our study cannot be extrapolated to other populations, in particular younger and potentially transplant-eligible patients with HF. Conclusions Heart transplant centers offer extensive and highly specialized infrastructure designed to care for the sickest and most complex patients with HF. However, in elderly Medicare patients with HF, we found little evidence that short-term outcomes are consistently better at these centers than at other nontransplant hospitals, particularly peer institutions within the same geographic area. Precise reasons for these findings are unclear but may be related to heart transplant centers specifically targeting their specialized resources to the transplant-eligible HF population. Additional studies are needed to better understand which hospital characteristics are associated with improved inpatient and postdischarge HF outcomes. Acknowledgments We thank Geoffrey Schreiner for his assistance in data analysis and the preparation of tables and figures for this article. Source of Funding The analyses on which this article is based were performed under contract number HHSM-500-2005-CO001C, entitled Utilization and Quality Control Quality Improvement Organization for the State (Commonwealth) of Colorado, funded by CMS, an agency of the US Department of Health and Human Services. Disclosures Drs Pauli, Wang, Chen, and Nallamothu report no relevant conflicts of interest. Dr Krumholz reports receiving funding from CMS contracts, including the contract funding this work, and serves as the chair of the United Health Care Scientific Advisory Board. Dr Normand serves on the advisory boards for Blue Cross/Blue Shield, Kaiser Permanente, and Medicare. Dr Hummel is a transplant cardiologist employed at a heart transplant center. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government. The authors assume full responsibility for the accuracy and completeness of the ideas presented. CMS did not participate in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation of the manuscript. CMS reviewed and approved the use of its data for this work and approved submission of the manuscript. References 1. Nallamothu BK, Wang Y, Cram P, Birkmeyer JD, Ross JS, Normand S-LT, Krumholz HM. Acute myocardial infarction and congestive heart failure outcomes at specialty cardiac hospitals. Circulation. 2007;116: 2280 2287. 2. Reis SE, Holubkov R, Edmundowicz D, McNamara DM, Zell KA, Detre KM, Feldman AM. Treatment of patients admitted to the hospital with congestive heart failure: specialty-related disparities in practice patterns and outcomes. J Am Coll Cardiol. 1997;30:733 738. 3. Jong P, Gong Y, Liu PP, Austin PC, Lee DS, Tu JV. Care and outcomes of patients newly hospitalized for heart failure in the community treated by cardiologists compared with other specialists. Circulation. 2003;108: 184 191. 4. Foody JM, Rathore SS, Wang Y, Herrin J, Masoudi FA, Havranek EP, Krumholz HM. Physician specialty and mortality among elderly patients hospitalized with heart failure. Am J Med. 2005;118:1120 1125. 5. Kosiborod M, Lichtman JH, Heidenreich PA, Normand SL, Wang Y, Brass LM, Krumholz HM. National trends in outcomes among elderly patients with heart failure. Am J Med. 2006;19:616 e611 e617.

252 Circ Heart Fail March 2010 6. Krumholz HM, Wang Y, Mattera JA, Wang Y, Han LF, Ingber MJ, Roman S, Normand S-LT. An administrative claims model suitable for profiling hospital performance based on 30-day mortality rates among patients with heart failure. Circulation. 2006;113:1693 1701. 7. Keenan PS, Normand S-LT, Lin Z, Drye EE, Bhat KR, Ross JS, Schuur JD, Stauffer BD, Bernheim SM, Epstein AJ, Wang Y, Herrin J, Chen J, Federer JJ, Mattera JA, Wang Y, Krumholz HM. An administrative claims measure suitable for profiling hospital performance on the basis of 30-day all-cause readmission rates among patients with heart failure. Circ Cardiovasc Qual Outcomes. 2008;1:29 37. 8. Philbin EF, DiSalvo TG. Prediction of hospital readmission for heart failure: development of a simple risk score based on administrative data. J Am Coll Cardiol. 1999;33:1560 1566. 9. Normand S-LT, Wang Y, Krumholz HM. Assessing surrogacy of data sources for institutional comparisons. Health Serv Outcomes Res Methodol. 2007;7:79 96. 10. Shahian DM, Normand S-LT. Comparison of risk-adjusted hospital outcomes. Circulation. 2008;117:1955 1963. 11. Fonarow GC, Abraham WT, Albert NM, Stough WG, Gheorghiade M, Greenberg BH, O Connor CM, Pieper K, Sun JL, Yancy C, Young JB; for the OPTIMIZE-HF investigators and hospitals. Association between performance measures and clinical outcomes for patients hospitalized with heart failure. J Am Med Assoc. 2007;297:61 70. 12. Werner RM, Bradlow ET. Relationship between Medicare s Hospital Compare performance measures and mortality rates. J Am Med Assoc. 2006;296:2694 2702. 13. Jessup M, Abraham WT, Casey DE, Feldman AM, Francis GS, Ganiats TG, Konstam MA, Mancini DM, Rahko PS, Silver MA, Stevenson LW, Yancy CW. 2009 Focused Update: ACCF/AHA guidelines for the diagnosis and management of heart failure in adults: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines: developed in collaboration with the International Society for Heart and Lung Transplantation. Circulation. 2009;119:1977 2016. 14. Michalsen A, Konig G, Thimme W. Preventable causative factors leading to hospital admission with decompensated heart failure [see comments]. Heart. 1998;80:437 441. 15. Medicare Payment Advisory Committee. Report to the Congress: improving efficiency in Medicare. http://www.medpac.gov/documents/ Jun07_EntireReport.pdf. Published June 2007. Accessed August 6, 2009. 16. Fonarow GC, Yancy CW, Heywood JT; the ADHERE Scientific Advisory Committee Study Group and Investigators. Adherence to heart failure quality-of-care indicators in U.S. hospitals: analysis of the ADHERE registry [see comment]. Arch Intern Med. 2005;165: 1469 1477. 17. Williams SC, Koss RG, Morton DJ, Schmaltz SP, Loeb JM. Case volume and hospital compliance with evidence-based processes of care. Int J Qual Health Care. 2008;20:79 87. 18. Foody JM, Rathore SS, Wang Y, Herrin J, Masoudi FA, Havranek EP, Radford MJ, Krumholz HM. Predictors of cardiologist care for older patients hospitalized for heart failure. Am Heart J. 2004;147:66 73. 19. Vinson JM, Rich MW, Sperry JC, Shah AS, McNamara T. Early readmission of elderly patients with congestive heart failure [see comments]. J Am Geriatr Soc. 1990;38:1290 1295. 20. Krumholz HM, Chen YT, Wang Y, Vaccarino V, Radford MJ, Horwitz RI. Predictors of readmission among elderly survivors of admission with heart failure. Am Heart J. 2000;139:72 77. 21. Department of Health and Human Services and Centers for Medicare & Medicaid Services. Medicare program: Hospital conditions of participation: requirements for approval and re-approval of transplant centers to perform organ transplants; final rule (2007). http://www.cms.hhs.gov/ CertificationandComplianc/Downloads/Transplantfinal.pdf. Published March 30, 2007. Accessed August 6, 2009. 22. Koelling TM, Johnson ML, Cody RJ, Aaronson KD. Discharge education improves clinical outcomes in patients with chronic heart failure. Circulation. 2005;111:179 185. 23. Rich MW, Beckham V, Wittenberg C, Leven CL, Freedland KE, Carney RM. A multidisciplinary intervention to prevent the readmission of elderly patients with congestive heart failure [see comment]. N Engl J Med. 1995;333:1190 1195. 24. Naylor M, Brooten D, Campbell R, Maislin G, McCauley K, Schwartz JS. Transitional care of older adults hospitalized with heart failure: a randomized, controlled trial. J Am Geriatr Soc. 2004;52:675 684. CLINICAL PERSPECTIVE Elderly patients hospitalized for heart failure (HF) have a high risk for death and suffer frequent readmissions after discharge. The rates vary widely across hospitals, and a significant number of these adverse events seem preventable. Understanding the hospital-level factors that contribute to early mortality and readmission may be key to improving outcomes in the elderly HF population. Heart transplant centers dedicate extensive multidisciplinary resources to HF care. We hypothesized that this expertise might improve outcomes in elderly Medicare patients with HF, even though most are not transplant candidates. We used hierarchical general linear modeling and validated administrative data-based risk-standardization algorithms for 30-day death and readmission to compare outcomes at heart transplant centers and nontransplant hospitals for all fee-for-service Medicare beneficiaries hospitalized for HF from 2004 to 2006. We found that heart transplant centers had slightly lower risk-standardized 30-day mortality rates than all other hospitals nationally but similar mortality rates to other large hospitals offering coronary bypass surgery within the same geographic area. Heart transplant centers and nontransplant hospitals had nearly identical risk-standardized 30-day readmission rates. Precise reasons for these findings are unclear but may be related to heart transplant centers specifically targeting their specialized resources to the transplant-eligible HF population. Additional studies are needed to better understand which hospital characteristics are associated with improved postdischarge outcomes in elderly patients with HF.