Design and rationale of a retrospective clinical effectiveness study of aldosterone antagonist therapy in patients with heart failure

Design and rationale of a retrospective clinical effectiveness study of aldosterone antagonist therapy in patients with heart failure Lesley H. Curtis, PhD, a,b Xiaojuan Mi, PhD, a Laura G. Qualls, MS, a Bradley G. Hammill, MS, a Stephen C. Hammill, MD, c Paul A. Heidenreich, MD, d Frederick A. Masoudi, MD, MSPH, e Soko Setoguchi, MD, DrPH, a,b Adrian F. Hernandez, MD, MHS, a,b and Gregg C. Fonarow, MD f Durham, NC; Rochester, MN; Palo Alto, and Los Angeles, CA; and Aurora, CO

Background Despite demonstrated efficacy in randomized trials, aldosterone antagonist therapy is not used in many eligible patients with heart failure. Questions remain about its clinical effectiveness and safety for patients who are underrepresented in randomized trials and those at risk for hyperkalemia. Methods The proposed study will evaluate the effectiveness of aldosterone antagonist therapy in eligible Medicare beneficiaries ≥65 years old hospitalized for heart failure between 2005 and 2008. Data are from the GWTG-HF registry linked with Medicare inpatient and prescription drug event files. We will use inverse probability-weighted estimators to assess differences in mortality, cardiovascular readmission, and readmission for hyperkalemia between patients who receive or do not receive aldosterone antagonist therapy. Results

The initial data set included 33,652 patients; 5,463 (16.2%) met all inclusion criteria. Compared with patients who did not meet the inclusion criteria, patients in the final cohort were more likely to be younger (77.3 vs 80.3 years) and male (63.8% vs 41.3%) and to have ischemic heart failure (74.2% vs 59.5%) (all P b .001). Mortality rates were 24.7% at 1 year and 50.7% at 3 years; cardiovascular readmission rates were 50.1% at 1 year and 65.2% at 3 years.

Conclusions The proposed study will evaluate the clinical effectiveness of aldosterone antagonist therapy in Medicare beneficiaries hospitalized for heart failure with reduced ejection fraction, an underrepresented population in clinical trials. By addressing this evidence gap, the study has the potential to inform clinical decision making and improve patient outcomes. (Am Heart J 2012;163:946-953.e1.)

Heart failure is a major public health problem in the United States, affecting close to 6 million people 1 and accounting for N3% of all Medicare payments. 2 Outcomes of patients with heart failure are poor. Nearly 23% of Medicare beneficiaries with heart failure are readmitted within 30 days of an index hospitalization, and nearly 40% die within 1 year. 3 Despite significant

From the aDuke Clinical Research Institute, Duke University School of Medicine, Durham, NC, bDepartment of Medicine, Duke University School of Medicine, Durham, NC, c Division of Cardiovascular Disease, Mayo Clinic, Rochester, MN, dVA Palo Alto Healthcare System, Palo Alto, CA, eUniversity of Colorado, Aurora, CO, and fDepartment of Medicine, Ahmanson-UCLA Cardiomyopathy Center, University of California, Los Angeles, CA. Philip F. Binkley, MD, MPH served as guest editor for this article. Submitted October 3, 2011; accepted March 2, 2012. Reprint requests: Lesley H. Curtis, PhD, Duke Clinical Research Institute, PO Box 17969,

Durham, NC 27715. E-mail: [email protected] 0002-8703/$ - see front matter © 2012, Mosby, Inc. All rights reserved. doi:10.1016/j.ahj.2012.03.007

therapeutic advances in heart failure, questions remain about the translation of safety and efficacy findings from clinical trials to clinical practice. 4,5 Consequently, the Institute of Medicine recently identified cardiovascular disease as a national priority for comparative effectiveness research. 6 Aldosterone antagonist therapy is a class I recommendation in the American College of Cardiology/American Heart Association guidelines for the treatment for patients with moderately to severely symptomatic heart failure, reduced ejection fraction, and serum creatinine level ≤2.5 mg/dL in men or ≤2.0 mg/dL in women. 7 However, use of aldosterone antagonist therapy varies widely and falls far below other class I evidence-based therapies, 8,9 in part, because of questions about its effectiveness and safety outside clinical trial settings. 10 Observational comparative effectiveness research has the potential to inform clinical decision making, especially where gaps exist in evidence from clinical trials. However, this research is a subject of debate because of

American Heart Journal Volume 163, Number 6

the potential for selection bias, confounding bias, and concerns about the selective reporting of results. 11 Careful attention to study design and data quality can minimize these concerns. Best practice guidelines for comparative effectiveness research recommend a priori specification of study objectives and study design. 12 Therefore, we describe the design of a retrospective clinical effectiveness study of aldosterone antagonist therapy in a large national sample of eligible Medicare beneficiaries. In addition, we use existing observational data to describe the study cohort and to show how the study will address gaps in evidence.

Methods The proposed study is a retrospective cohort study comparing outcomes of eligible Medicare beneficiaries who receive aldosterone antagonist therapy with outcomes of eligible beneficiaries who do not receive this therapy, overall and by clinically relevant subgroups. The null hypothesis is that receipt of aldosterone antagonist therapy at discharge has no effect on mortality and readmission rates at 1 and 3 years among Medicare beneficiaries aged ≥65 years who were hospitalized for heart failure, had a history of heart failure before hospitalization, and were eligible for aldosterone antagonist therapy.

Data sources Data for the analysis are from the American Heart Association's GWTG-HF registry and the Centers for Medicare and Medicaid Services. The GWTG-HF registry is a hospitalbased heart failure registry that uses a Web-based data collection system. The program is implemented voluntarily by hospitals and is the successor to the OPTIMIZE-HF registry, which has been described previously. 13 Patients are eligible for the GWTG-HF registry if they were hospitalized for heart failure as a primary cause of admission or developed significant heart failure symptoms during hospitalization when heart failure was not the initial reason for admission. Outcome Sciences, Inc (Cambridge, MA), serves as the data collection and coordination center for GWTG-HF. The Duke Clinical Research Institute (Durham, NC) serves as the data analysis center. The Medicare data for the study include the 100% Medicare Part A inpatient claims files and the corresponding denominator files for 2005 through 2009. The inpatient files contain institutional claims for facility costs covered under Medicare Part A. The corresponding denominator files include dates of birth, sex, race/ethnicity, dates of death, and information about program eligibility and enrollment. We used combinations of indirect identifiers—admission date, discharge date, patient sex, and patient age or date of birth—to link registry records with inpatient claims. 14 We then used an encrypted Medicare beneficiary identifier to gather information about subsequent events.

Study cohort The initial data set includes patients who were aged ≥65 years with a GWTG-HF hospitalization linked to Medicare claims; were discharged alive between January 1, 2005, and December 31, 2009; and were enrolled in fee-for-service Medicare at

Curtis et al 947

discharge. To ensure that each registry record meets a minimum standard of quality, we require that vital signs at admission and discharge medications are recorded. We considered the first hospitalization during the study period to be the index hospitalization. To construct the study cohort, we restricted the initial data set to patients who were discharged home and had a documented history of heart failure before the index hospitalization. We defined aldosterone antagonist therapy eligibility as left ventricular ejection fraction (LVEF) ≤35% or a qualitative description of moderate or severe left ventricular systolic dysfunction, serum creatinine level at admission ≤2.5 mg/dL in men and ≤2.0 mg/dL in women, and no documented contraindication to aldosterone antagonist therapy. (Serum potassium level is not included because it was not collected in the GWTG-HF registry until 2008.) Patients were excluded if they had a documented contraindication to aldosterone antagonist therapy (ie, allergy to aldosterone antagonist, hyperkalemia, renal dysfunction defined as creatinine N2.5 mg/dL in men and N2.0 mg/dL in women, etc). In the clinical effectiveness analysis, we will further require that patients were naive to aldosterone antagonist therapy at admission using information about medications at admission and discharge from the registry record and prescription drug event data. 15

Exposure and comparison groups The treatment of interest is aldosterone antagonist therapy prescribed at discharge, as recorded in the GWTG-HF registry. The comparison is no aldosterone antagonist therapy prescribed at discharge, according to the GWTG-HF registry. This comparison reflects the decision faced by clinicians in daily practice.

Outcomes The outcomes of interest are death from any cause, cardiovascular readmission, and readmission for hyperkalemia. Death is determined based on the death date recorded in the Medicare denominator files, and readmissions are based on Medicare inpatient claims. We will identify cardiovascular readmission using the following diagnosis-related group codes: 104-112, 115-118, 121-145, 479, 514-518, 525-527, 535, 536, and 547-558 before October 1, 2007, and codes 215-238, 242-254, 258-262, and 280-316 on or after October 1, 2007. Hospitalizations for rehabilitation are not considered to be readmissions. Readmission for hyperkalemia is defined by any diagnosis of hyperkalemia (International Classification of Diseases, Ninth Revision, Clinical Modification code 276.7). In a secondary analysis, we will rely on a primary diagnosis of hyperkalemia to identify a hyperkalemia readmission. The primary period of follow-up for ascertainment of these outcomes is 1 year after discharge from the index hospitalization, as recorded in the GWTG-HF registry. We will also perform secondary analyses at 3 years after discharge. For patients who do not experience a particular outcome, we will define a censoring date as the earliest of (a) 1 year (or 3 years) after discharge, (b) the end of the study period, or (c) the date when the patient enrolled in a Medicare-managed care plan. For the readmission outcomes, we will treat death as a competing risk.

Covariates Data collected in the GWTG-HF registry include demographic characteristics, medical history, results of admission laboratory

American Heart Journal June 2012

948 Curtis et al

tests, vital signs at admission, and medications at admission and discharge. In addition, we will use Hierarchical Condition Category codes from Medicare claims to identify protein-energy malnutrition, dementia, disability, major psychiatric disorders, and chronic liver disease. The independent prognostic value of these variables has been established for modeling mortality and readmission after hospitalization for heart failure. 16,17

Statistical analysis We will use an inverse probability-weighted estimator—an extension of the propensity score—to assess differences in outcomes among the treatment groups in both analyses. We will fit the treatment selection model as a logistic regression model with treatment as the dependent variable and the baseline characteristics described above as the independent variables. We will then weight each patient in the sample by the inverse of the estimated probability of the treatment received. To estimate the unadjusted effect of treatment on each outcome, we will use unweighted proportional hazards regression models. We will estimate adjusted effects using weighted proportional hazards regression models and the weights described above. The primary analysis will rely on the registry record to establish the use of aldosterone antagonist therapy. In addition to estimating an overall treatment effect, we will estimate the effect of treatment in subgroups formed based on age (ie, 65-79 vs ≥80 years), sex, race (ie, black, white, and other/unknown), ischemic etiology of heart failure (as recorded in the GWTG-HF registry), and the presence of diabetes mellitus (as recorded in the GWTG-HF registry).

Sensitivity analysis Because the discharge prescription flag recorded in the GWTG-HF registry is an imperfect measure of exposure to aldosterone antagonist therapy, 18 we will perform a sensitivity analysis that relies on the Medicare prescription drug event data available for a subset of the study cohort.

Baseline characteristics of the study cohort In selecting the appropriate population for the clinical effectiveness study, we undertook an analysis of existing observational data. We created the study cohort as described above and summarized the baseline characteristics of the study population. Registry data through 2008 and claims data through 2009 were available at the time of the analysis. Medicare prescription drug event data were not yet available, so we could not apply the restriction that patients be naive to aldosterone antagonist therapy. We present frequencies with percentages for categorical variables and means with SDs for continuous variables. We used the standard normal z test to test for differences in categorical variables. We used the Satterthwaite t test to test for differences between population means. We used Kaplan-Meier methods to estimate the cumulative incidence of mortality at 1 and 3 years, overall and by prespecified subgroups. We tested for differences between subgroups using log-rank tests. We estimated cardiovascular readmission at 1 and 3 years using the cumulative incidence function, which accounts for the competing risk of mortality. We tested for differences between subgroups on these outcomes using Gray tests. 19

We used a significance level of α = .01 for all hypotheses. All P values were based on 2-sided tests. We used R software version 2.6 (R Foundation for Statistical Computing, Vienna, Austria) to perform the cumulative incidence analyses. For all other analyses, we used SAS software version 9.2 (SAS Institute Inc, Cary, NC). T h i s s tu d y w a s fu n d e d u n de r c o n t r a c t n u m b e r HHSA29020050032I (Duke University DEcIDE Center) from the Agency for Healthcare Research and Quality, US Department of Health and Human Services, as part of the Developing Evidence to Inform Decisions About Effectiveness (DEcIDE) program.

Results Figure 1 shows the derivation of the study cohort. The initial linked data set included 33,652 patients, from which we excluded 8,899 patients who were not discharged home and 4,192 patients who did not have a documented history of heart failure before the index hospitalization. The restricted data set included 20,656 patients, from which we further excluded 14,722 patients who were not eligible for aldosterone antagonist therapy. After all exclusions, the study cohort included 5,934 patients who were treated at 238 hospitals. Of these patients, 471 (7.9%) were missing values for smoking within the past year, systolic blood pressure, respiratory rate, heart rate, LVEF, and/or serum urea nitrogen. The subset of the cohort with complete data was comparable with the overall population (online Appendix), so we conducted all remaining analyses of eligible patients using the population with complete data (n = 5,463). Patients who met the inclusion criteria differed from those who did not in that they were younger and predominantly male and nearly three-quarters had heart failure of ischemic origin (Table I). More than half of eligible patients had LVEF ≤25%, and the proportion with impaired renal function was low. Most eligible patients were prescribed an angiotensin-converting enzyme (ACE) inhibitor or angiotensin receptor blocker (ARB), β-blocker, diuretic, and/or lipid-lowering agent at discharge. Aldosterone antagonist therapy was prescribed to 28% of eligible patients. Mortality rates among eligible patients were 24.7% at 1 year and 50.7% at 3 years (Table II and Figure 2). One- and 3year mortality rates were higher in eligible patients aged ≥80 years, and the 3-year mortality rate was higher in eligible patients with ischemic heart failure. Cardiovascular readmission rates among eligible patients were 50.1% at 1 year and 65.2% at 3 years (Table II and Figure 3) and differed by etiology of heart failure and diabetic status.

Discussion In establishing priorities for comparative effectiveness research, the Institute of Medicine described the breadth of study designs, including “systematic reviews, database research, observational studies, and randomized trials,”

American Heart Journal Volume 163, Number 6

Curtis et al 949

Figure 1

Derivation of the study cohort.

that could be used to assess real-world outcomes. 6 The need for openness and transparency in the registration and reporting of clinical trials is widely recognized, yet similar expectations have not been established for the range of nonrandomized study designs used under the banner of comparative effectiveness research. Concerns about selective reporting of post hoc analyses led journal editors to call for increased transparency in the design and reporting of comparative effectiveness studies, including

laying out intended analyses in advance. 11 We address this concern by describing the design of a clinical effectiveness study, including the study hypothesis, outcomes, and intended analyses. To our knowledge, this is the first article to prospectively describe a proposed clinical effectiveness analysis of a retrospectively identified cohort. We strongly advocate for mandatory registration of observational clinical effectiveness studies using a platform similar to ClinicalTrials.gov. Until such a system

American Heart Journal June 2012

950 Curtis et al

Table I. (continued )

Table I. Baseline characteristics

Met inclusion criteria

Met inclusion criteria

Characteristic

Yes (n = 5463)

Demographic characteristics Age, y, mean (SD) 77.3 (7.5) Age group, y, n (%) 65-79 3310 (60.6) ≥80 2153 (39.4) Female, n (%) 1975 (36.2) Race, n (%) Black 721 (13.2) White 4459 (81.6) Other/unknown 283 (5.2) Medical history, n (%) Anemia 672 (12.3) Atrial fibrillation 1854 (33.9) Chronic obstructive 1430 (26.2) pulmonary disease Depression 369 (6.8) Diabetes mellitus 2054 (37.6) Heart failure with 4051 (74.2) ischemic etiology Hyperlipidemia 2567 (47.0) Hypertension 3826 (70.0) Smoker in the 642 (11.8) past year Pacemaker 1022 (18.7) Peripheral vascular 698 (12.8) disease 791 (14.5) Prior cerebrovascular accident/transient ischemic attack Renal insufficiency 669 (12.2) Claims-based history at admission, n (%) Chronic liver 35 (0.6) disease Dementia 198 (3.6) Disability 76 (1.4) Malnutrition 73 (1.3) Psychiatric disorder 38 (0.7) Vital signs at admission Heart rate, 83.6 (19.2) mean (SD) Respiratory rate 304 (5.6) ≥30 breath/min, n (%) Systolic blood 134 (26.6) pressure, mm Hg, mean (SD) Tests at admission LVEF, median (IQR) 25.0 (20.0-30.0) LVEF ≤25%, n (%) 2963 (54.2) Serum creatinine, 1.3 (0.4) mg/dL, mean (SD) Serum urea nitrogen, mg/dL, n (%) b20 1729 (31.6) 20-50 3380 (61.9) ≥50 354 (6.5) Medications at discharge, n (%) ACE inhibitor 3356 (61.4) ARB 982 (18.0) ACE inhibitor 4258 (77.9)

No (n = 27 718)

80.3 (8.0)

P

b.001

12 316 (44.4) 15 402 (55.6) 16 264 (58.7)

b.001 b.001 b.001

2838 (10.2) 23 083 (83.3) 1797 (6.5)

b.001 .004 b.001

5302 (19.1) 9670 (34.9) 7805 (28.2)

b.001 .18 .002

2811 (10.1) 10 918 (39.4) 16 480 (59.5)

b.001 .01 b.001

10 723 (38.7) 20 622 (74.4) 2299 (8.3)

b.001 b.001 b.001

2678 (9.7) 3582 (12.9)

b.001 .77

4462 (16.1)

.002

5373 (19.4)

b.001

241 (0.9)

.06

2,285 604 717 332

(8.2) (2.2) (2.6) (1.2)

83.4 (20.4)

b.001 b.001 b.001 b.001 .61

1998 (7.2)

b.001

141.8 (29.7)

b.001

50.0 (35.0-60.0) b.001 3215 (11.6) b.001 1.8 (3.2) b.001

7873 (28.4) 15 294 (55.2) 3955 (14.3)

b.001 b.001 b.001

12 048 (43.5) 4406 (15.9) 15 949 (57.5)

b.001 b.001 b.001

Characteristic and/or ARB Aldosterone antagonist β-Blocker Digoxin Diuretic Lipid-lowering agent Warfarin

Yes (n = 5463)

1526 (27.9) 4706 1721 4257 3267 1905

(86.1) (31.5) (77.9) (59.8) (34.9)

No (n = 27 718)

3031 (10.9) 19 155 5211 20 204 12 384 7845

(69.1) (18.8) (72.9) (44.7) (28.3)

P

b.001 b.001 b.001 b.001 b.001 b.001

IQR, Interquartile range.

exists, we hope that others will follow a similar approach to the one we have taken. The real-world applicability of the study cohort we describe is clear, in that patients aged ≥65 years constitute the large majority of the population with heart failure. 20 Table III shows selected baseline characteristics of research participants randomized to the placebo group in the RALES 21 and the EMPHASIS-HF trials. 22 The mean age of patients eligible for inclusion in the planned comparative effectiveness study is nearly 10 years greater than in either trial population, and women comprise a larger proportion of the sample. Left ventricular ejection fraction is similar between the trials and the planned study, but ischemic heart failure is more common in the planned study (74.2% vs 55% in RALES and 69% in EMPHASIS-HF). The use of ACE inhibitors and/or ARBs is lower in the planned study (77.9% vs 94% in RALES and 92.9% in EMPHASIS-HF), but the use of βblockers, digoxin, and lipid-lowering agents is similar. The mortality rate among patients meeting the eligibility criteria in the planned study is similar to mortality in the placebo arm of RALES, likely reflecting the underlying severity of heart failure in both study populations. In contrast, mortality in the placebo arm of EMPHASIS-HF is markedly lower and consistent with mild symptomatic heart failure. Other findings are noteworthy. First, the demographic characteristics of the patients who met the guidelinebased inclusion criteria for our study differ markedly from those who did not. Women and patients ≥80 years old constituted most patients who did not meet the inclusion criteria. Currently, aldosterone antagonist therapy is not recommended for patients with heart failure with preserved LVEF, which increases with age and is more prevalent in women. 23 In addition, women and older patients are more likely to be discharged to skilled nursing facilities. 7 Our study has some limitations. First, the analysis is restricted to patients ≥65 years old enrolled in fee-forservice Medicare, so the results may not generalize to younger patients and those enrolled in Medicare-

American Heart Journal Volume 163, Number 6

Curtis et al 951

Table II. Cumulative incidence of mortality and cardiovascular readmission among patients eligible for aldosterone antagonist therapy (n = 5463) 1y

Mortality Overall 1215 (24.7) Age group 65-79 y 595 (20.0) ≥80 y 620 (31.8) Sex Male 788 (25.2) Female 427 (23.7) Race Black 137 (21.3) White 1018 (25.3) Other/unknown 60 (23.1) Heart failure etiology Ischemic 925 (25.3) Nonischemic 290 (22.8) Diabetes mellitus No 761 (24.7) Yes 454 (24.6) Cardiovascular readmission Overall 2554 (50.1) Age group 65-79 y 1558 (50.6) ≥80 y 996 (49.5) Sex Male 1622 (50.1) Female 932 (50.1) Race Black 371 (55.0) White 2062 (49.7) Other/ 121 (44.8) unknown Heart failure etiology Ischemic 1950 (51.6) Nonischemic 604 (45.8) Diabetes mellitus No 1533 (48.2) Yes 1021 (53.3)

Characteristic

3y

P

No. of events (cumulative incidence)⁎

P

1893 (50.7) b.001

b.001 975 (44.2) 918 (60.6)

.32

.09 1230 (52.4) 663 (47.9)

.08

.05 219 (47.7) 1587 (51.6) 87 (43.4)

.14

.007 1456 (52.5) 437 (45.4)

.87

.44 1166 (49.4) 727 (52.7) 2985 (65.2)

.44

.29 1825 (66.0) 1160 (63.7)

.92

.94 1893 (65.1) 1092 (65.1)

.02

.02 420 (69.1) 2421 (64.7) 144 (61.4)

Age, mean (SD) Female, % White race, % LVEF, mean (SD) Mean (SD) Median (IQR)

RALES EMPHASIS-HF GWTG-HF (N = 841) (N = 1373) (N = 5463)⁎ 65 (12) 27 86

68.6 (7.6) 21.9 83.1

77.3 (7.5) 36.2 81.6

25.2 (6.8) –

26.1 (4.7) –

Systolic blood pressure, 122 (20) mm Hg, mean (SD) Heart rate, mean (SD) 81 (15) Serum creatinine, mg/dL Mean (SD) – Median 1.2 Heart failure with 54 ischemic etiology, % Diabetes mellitus, % – Hypertension, % – Stroke, % – Medication at randomization visit, % ACE inhibitor 94 ARB – ACE inhibitor 94 and/or ARB Antiarrhythmic agent – Antiplatelet agent or – oral anticoagulant β-Blocker 10 Digoxin 72 Diuretic 100 Lipid-lowering agent – All-cause mortality rate † 1y 24.5 3y 55.5

124 (17)

– 25.0 (20.0-30.0) 134 (26.6)

72 (13)

83.6 (19.2)

1.16 (0.31) – 68.1

1.3 (0.4)

39.1 66.2 9.2

37.6 70.0 14.5

76.8 19.4 92.9

61.4 18.0 77.9

14.0 88.4

34.9⁎

86.9 27.5 85.7 62.3

86.1 31.5 77.9 59.8

8.0 25.0

24.7 50.7

74.2

⁎ Participants from the GWTG-HF registry who met the inclusion criteria. † Estimated from published Kaplan-Meier curves.

b.001

.002 2281 (67.5) 704 (58.3) b.001

b.001

Figure 2

1809 (63.8) 1176 (67.5)

⁎ Unadjusted cumulative incidence per 100 patients at risk.

managed care plans. Second, the analysis is limited to hospitals participating in the GWTG-HF registry, so the findings may not generalize to all hospitals. However, the GWTG-HF includes a broad range of hospitals, and previous studies have suggested that Medicare beneficiaries in national heart failure registries have similar demographic characteristics, clinical characteristics, and outcomes compared with other Medicare beneficiaries. 24 Third, contraindications to aldosterone antagonist therapy were abstracted from the medical record. Furthermore, although documented hyperkalemia as a contraindication to aldosterone antagonist therapy was

70%

Cumulative Incidence of Death

Group

No. of events (cumulative incidence)⁎

Table III. Characteristics of participants randomized to the placebo group in the RALES and EMPHASIS-HF trials

60% 50% 40% 30% 20% 10% 0% 0

1

2

Years After Index Hospitalization Cumulative incidence of death in the study cohort.

3

American Heart Journal June 2012

952 Curtis et al

References

Cumulative Incidence of Readmission

Figure 3 70% 60% 50% 40% 30% 20% 10% 0% 0

1

2

3

Years After Index Hospitalization Cumulative incidence of cardiovascular readmission in the study cohort.

collected, serum potassium level was not recorded in the registry until 2008 and could not be included in the eligibility criteria. Therefore, a proportion of patients reported to be eligible for aldosterone antagonist therapy may have had undocumented contraindications. The proposed study will evaluate the clinical effectiveness of aldosterone antagonist therapy in Medicare beneficiaries hospitalized for heart failure with reduced ejection fraction, a patient group that is underrepresented in clinical trials and for which the potential risks and benefits of treatment may differ from those observed among patients enrolled in clinical trials. By addressing a critical gap in existing evidence, the proposed study has the potential to inform clinical decision making and improve patient outcomes.

Disclosures This study was funded under contract number HHSA29020050032I (Duke University DEcIDE Center) from the Agency for Healthcare Research and Quality, US Department of Health and Human Services, as part of the Developing Evidence to Inform Decisions About Effectiveness (DEcIDE) program. The GWTG-HF program is provided by the American Heart Association and is currently supported by an unrestricted educational grant from Medtronic, Inc. GWTG-HF has been supported previously by GlaxoSmithKline, Ortho-McNeil, and the American Heart Association Pharmaceutical Roundtable. Disclaimer: The authors of this report are responsible for its content. Statements in the report should not be construed as endorsement by the Agency for Healthcare Research and Quality or the US Department of Health and Human Services.

1. Lloyd-Jones D, Adams R, Carnethon M, et al. Heart disease and stroke statistics—2009 update: a report from the American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Circulation 2009;119:e21-e181. 2. Centers for Medicare & Medicaid Services. Medicare & Medicaid Statistical Supplement, 2010 Edition. http://www.cms.gov/ MedicareMedicaidStatSupp/09_2010.asp. Last accessed May 12, 2011. 3. Curtis LH, Greiner MA, Hammill BG, et al. Early and long-term outcomes of heart failure in elderly persons, 2001-2005. Arch Intern Med 2008;168:2481-8. 4. Cherubini A, Oristrell J, Pla X, et al. The persistent exclusion of older patients from ongoing clinical trials regarding heart failure. Arch Intern Med 2011;171:550-6. 5. Shah KB, Rao K, Sawyer R, et al. The adequacy of laboratory monitoring in patients treated with spironolactone for congestive heart failure. J Am Coll Cardiol 2005;46:845-9. 6. Institute of Medicine. Initial national priorities for comparative effectiveness research. Washington, DC: National Academies Press; 2009. 7. Hunt SA; American College of Cardiology, American Heart Association Task Force on Practice Guidelines (Writing Committee to Update the 2001 Guidelines for the Evaluation and Management of Heart Failure). ACC/AHA 2005 guideline update for the diagnosis and management of chronic heart failure in the adult: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Update the 2001 Guidelines for the Evaluation and Management of Heart Failure). J Am Coll Cardiol 2005;46:e1-e82. 8. Albert NM, Yancy CW, Liang L, et al. Use of aldosterone antagonists in heart failure. JAMA 2009;302:1658-65. 9. Fonarow GC, Yancy CW, Albert NM, et al. Heart failure care in the outpatient cardiology practice setting: findings from IMPROVE HF. Circ Heart Fail 2008;1:98-106. 10. Juurlink DN, Mamdani MM, Lee DS, et al. Rates of hyperkalemia after publication of the Randomized Aldactone Evaluation Study. N Engl J Med 2004;351:543-51. 11. Ensuring integrity in comparative effectiveness research: accentuate the negative. PLoS Med 2009;6:e1000152. 12. Vandenbroucke JP, von Elm E, Altman DG, et al. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration. Epidemiology 2007;18:805-35. 13. Hong Y, LaBresh KA. Overview of the American Heart Association “Get with the Guidelines” programs: coronary heart disease, stroke, and heart failure. Crit Pathw Cardiol 2006;5:179-86. 14. Hammill BG, Hernandez AF, Peterson ED, et al. Linking inpatient clinical registry data to Medicare claims data using indirect identifiers. Am Heart J 2009;157:995-1000. 15. Ray WA. Evaluating medication effects outside of clinical trials: newuser designs. Am J Epidemiol 2003;158:915-20. 16. Keenan PS, Normand SL, Lin Z, et al. 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. 17. Krumholz HM, Wang Y, Mattera JA, et al. An administrative claims model suitable for profiling hospital performance based on 30-day mortality rates among patients with heart failure. Circulation 2006; 113:1693-701. 18. Butler J, Arbogast PG, Daugherty J, et al. Outpatient utilization of angiotensin-converting enzyme inhibitors among heart failure

American Heart Journal Volume 163, Number 6

patients after hospital discharge. J Am Coll Cardiol 2004;43: 2036-43. 19. Gray RJ. A class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat 1988;16:1141-54. 20. Roger VL, Go AS, Lloyd-Jones DM, et al. Heart disease and stroke statistics—2011 update: a report from the American Heart Association. Circulation 2011;123:e18-e209. 21. Pitt B, Zannad F, Remme WJ, et al. The effect of spironolactone on morbidity and mortality in patients with severe heart failure. Randomized Aldactone Evaluation Study Investigators. N Engl J Med 1999;341:709-17.

Curtis et al 953

22. Zannad F, McMurray JJ, Krum H, et al. Eplerenone in patients with systolic heart failure and mild symptoms. N Engl J Med 2011;364: 11-21. 23. Vasan RS, Larson MG, Benjamin EJ, et al. Congestive heart failure in subjects with normal versus reduced left ventricular ejection fraction: prevalence and mortality in a population-based cohort. J Am Coll Cardiol 1999;33:1948-55. 24. Curtis LH, Greiner MA, Hammill BG, et al. Representativeness of a national heart failure quality-of-care registry: comparison of OPTIMIZE-HF and non-OPTIMIZE-HF Medicare patients. Circ Cardiovasc Qual Outcomes 2009;2:377-84.

American Heart Journal Volume 163, Number 6

Curtis et al 953.e1

Appendix Appendix Table. Baseline characteristics of the population eligible for aldosterone antagonist therapy and the population eligible with complete data

Variable

Eligible population (n = 5934)

Eligible population with complete data (n = 5463)

Demographic characteristics Age (y), mean (SD) 77.3 (7.5) 77.3 Age group, n (%) 65-79 y 3612 (60.9) 3310 ≥80 y 2322 (39.1) 2153 Female, n (%) 2163 (36.5) 1975 Race, n (%) Black 802 (13.5) 721 White 4800 (80.9) 4459 Other / unknown 332 (5.6) 283 Medical history, n (%) Anemia 717 (12.1) 672 Atrial fibrillation 1996 (33.6) 1854 Chronic obstructive 1541 (26.0) 1430 pulmonary disease Depression 401 (6.8) 369 Diabetes mellitus 2231 (37.6) 2054 Heart failure with 4379 (73.8) 4051 ischemic etiology Hyperlipidemia 2765 (46.6) 2567 Hypertension 4159 (70.1) 3826 Smoker in past year 696 (11.7) 642 Pacemaker 1091 (18.4) 1022 Peripheral vascular 743 (12.5) 698 disease Prior cerebrovascular 843 (14.2) 791 accident/transient ischemic attack Renal insufficiency 732 (12.3) 669 Claims-based history (at index hospitalization), n (%) Chronic liver disease 38 (0.6) 35 Dementia 219 (3.7) 198

(7.5) (60.6) (39.4) (36.2) (13.2) (81.6) (5.2) (12.3) (33.9) (26.2) (6.8) (37.6) (74.2) (47.0) (70.0) (11.8) (18.7) (12.8) (14.5)

(12.2) (0.6) (3.6)

Appendix Table. (continued )

Variable

Eligible population (n = 5934)

Disability 80 (1.3) Malnutrition 75 (1.3) Psychiatric disorder 46 (0.8) Vital signs at admission Systolic blood pressure 135 (26.7) (mm Hg), mean (SD) Respiratory rate ≥30 331 (5.6) breath/min, n (%) Heart rate, mean (SD) 83.6 (19.4) Tests at admission Left ventricular ejection 25.0 (20.0, 30.0) fraction, median (IQR) Left ventricular ejection 3079 (51.9) fraction ≤25%, n (%) Serum urea nitrogen, n (%) b20 mg/dL 1821 (30.7) 20-50 mg/dL 3613 (60.9) ≥50 mg/dL 381 (6.4) Serum creatinine, n (%) b1.5 mg/dL 3780 (63.7) 1.5-2.0 mg/dL 1555 (26.2) ≥2.0 mg/dL 599 (10.1) Medication at discharge, n (%) ACE inhibitor 3650 (61.5) Aldosterone 1633 (27.5) antagonist Angiotensin receptor 1065 (17.9) blocker β-blocker 5105 (86.0) Digoxin 1844 (31.1) Diuretic 4619 (77.8) Lipid-lowering agent 3529 (59.5) Warfarin 2050 (34.5)

Eligible population with complete data (n = 5463) 76 (1.4) 73 (1.3) 38 (0.7) 134 (26.6) 304 (5.6) 83.6 (19.2) 25.0 (20.0, 30.0) 2963 (54.2)

1729 (31.6) 3380 (61.9) 354 (6.5) 3457 (63.3) 1454 (26.6) 552 (10.1) 3356 (61.4) 1526 (27.9) 982 (18.0) 4706 1721 4257 3267 1905

(86.1) (31.5) (77.9) (59.8) (34.9)