Abstract
Background:
A prior hospitalization due to heart failure is associated with poor outcomes in ambulatory patients with heart failure. Less is known about this association in hospitalized patients with heart failure and whether it varies by ejection fraction.
Methods:
Of the 25,345 hospitalized patients in Medicare-linked OPTIMIZE-HF registry, 22,491 had known heart failure, of whom 7648 and 9558 had heart failure with preserved (≥50%) and reduced (≤40%) ejection fraction (HFpEF and HFrEF), respectively. Overall, 927 and 1862 patients with HFpEF and HFrEF had heart failure hospitalizations during six months before the index hospitalization, respectively. Using propensity scores for prior heart failure hospitalization, we assembled two matched cohorts of 924 pairs and 1844 pairs of patients with HFpEF and HFrEF, respectively, each balanced for 58 baseline characteristics. Cox regression models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for outcomes during six years of follow-up.
Results:
Among 1848 matched patients with HFpEF, HRs (95% CIs) for all-cause mortality, all-cause readmission and heart failure readmission were 1.35 (1.21–1.50; p<0.001), 1.34 (1.21–1.47; p<0.001) and 1.90 (1.67–2.16; p<0.001), respectively. Respective HRs (95% CIs) in 3688 matched patients with HFrEF were 1.17 (1.09 –1.26; p<0.001), 1.32 (1.23 –1.41; p<0.001), and 1.48 (1.37 –1.61; p<0.001).
Conclusions:
Among hospitalized patients with heart failure, a prior heart failure hospitalization is associated with higher risks of mortality and readmission in both HFpEF and HFrEF. The relative risks of death and heart failure readmission appear to be higher in HFpEF than in HFrEF.
Keywords: Heart failure, Mortality, Prior hospitalization, Preserved ejection fraction, Readmission
Introduction
Heart failure hospitalization has been shown to be associated with poor outcomes in ambulatory younger patients with chronic heart failure, mostly among those with heart failure with reduced ejection fraction (HFrEF).1–5 However, less is known about this association among hospitalized older patients with acute heart failure, in general, and especially in the subset with preserved ejection fraction (HFpEF), which is the predominant phenotype in older adults.6 The objective of this study was to examine the association of a prior heart failure hospitalization with outcomes separately in propensity score-matched cohorts of hospitalized older patients with HFpEF and HFrEF.
Methods
Data Source and Study Patients
We used data from the Organized Program to Initiate Lifesaving Treatment in Hospitalized Patients with Heart Failure (OPTIMIZE-HF), a national hospital-based registry designed to improve the standard of heart failure care and increase adherence to guideline directed medical therapy.7,8 The OPTIMIZE-HF registry is comprised of 48,612 heart failure hospitalizations in 259 hospitals in 48 states between March 1, 2003, and December 31, 2004. It includes data on demographics, characteristics of patients, quality of care, and outcomes, which were accessed using an internet-based information system.
Assembly of Study Cohorts
Of 48,612 heart failure hospitalizations in the OPTIMIZE-HF registry, 26,376 were Medicare-linked and 25,345 of these patients were discharged alive from the hospital. We excluded 2854 patients with a new diagnosis of heart failure as these patients could not be categorized as having a prior heart failure hospitalization. Of the 22,491 patients with known heart failure, 7648 had HFpEF (ejection fraction ≥50%), of whom 927 had a prior heart failure hospitalization within six months of index hospitalization which was ascertained from Medicare data (Figure 1). Of the remaining 6721 patients who did not have a heart failure hospitalization in the six months prior to the index hospitalization, 1094 (16%) had a heart failure hospitalization longer than six months prior. Using a multivariable logistic regression model, we estimated propensity scores for prior heart failure hospitalization within six months of the index heart failure hospitalization for each of the 7648 patients with HFpEF using 58 baseline characteristics as covariates.9–11 Using a matching algorithm described elsewhere,12 we matched 924 of the 927 patients with HFpEF who had a prior heart failure hospitalization with another 924 who did not, thus assembling a matched cohort of 1848 patients with HFpEF (Figure 1), balanced on 58 baseline characteristics (Figure 2). We then repeated the above process in 9558 patients with HFrEF (ejection fraction ≤40%), of whom 1862 had a prior heart failure hospitalization, and assembled a matched cohort of 1844 pairs of patients (Figure 1). Patients with ejection fraction from 41% to 49% (n=5285) were not analyzed.
Figure 1.
Flow chart displaying assembly of propensity score-matched cohorts of patients by a heart failure hospitalization (HFH) in the six months prior to the index HFH (HFH = Heart failure hospitalization; HFpEF = Heart failure with preserved ejection fraction; HFrEF = Heart failure with reduced ejection fraction; LVEF = Left ventricular ejection fraction; OPTIMIZE-HF = Organized Program to Initiate Lifesaving Treatment in Hospitalized Patients with Heart Failure)
Figure 2.
Love plot displaying absolute standardized differences comparing 58 baseline characteristics of 7648 pre-match and 1848 matched patients with heart failure with preserved ejection fraction, by prior versus no prior heart failure hospitalization in past six months, before and after propensity score matching. An absolute standardized difference of 0% indicates no residual bias and values <10% indicate inconsequential bias (ACE = Angiotensin-converting enzyme; ARB = Angiotensin receptor blockers; CCBs = Calcium channel blockers; COPD = Chronic obstructive pulmonary disease; ICD =Implantable cardioverter-defibrillator; JVP = Jugular venous pressure; LVEF = Left ventricular ejection fraction)
Outcomes Data
The primary outcome studied was time to all-cause mortality during six (median, two) years of follow-up. Secondary outcomes included times to all-cause and heart failure readmissions. All outcome data, including events and time to events, were collected from the Medicare 100% MedPAR File and the 100% Beneficiary Summary File between January 1, 2002 and December 31, 2008.13
Statistical Analysis
We compared between-group baseline characteristics using the Pearson chi-square and Wilcoxon rank sum tests, as appropriate. All outcome analyses were conducted on propensity score-matched data. All-cause mortality associated with prior heart failure hospitalization in HFpEF and HFrEF is displayed as Kaplan-Meier plots (Figure 3). Hazard ratios (HRs) and 95% confidence intervals (CIs) for the risk of mortality and readmission associated with prior heart failure hospitalization were estimated using Cox regression models, separately in HFpEF and HFrEF. Time to mortality was defined as the number of days to death for patients who died and number of days to study end as censoring time for those patients who did not die. Time to readmission was defined as the number of days to readmission for patients who had a subsequent readmission. For patients without readmission, censoring time was defined as the number of days to death or study end, whichever occurred first.
Figure 3.
Kaplan-Meier plots for all-cause mortality during six years of post-discharge follow up by heart failure hospitalization during six months prior to the index hospitalization in 924 and 1844 pairs of propensity score-matched patients with heart failure with preserved ejection fraction (HFpEF; top panel) and heart failure with reduced ejection fraction (HFrEF; bottom panel), respectively (CI = Confidence interval; HR = Hazard ratio).
We conducted formal sensitivity analyses using Rosenbaum’s approach to determine if significant associations observed in our matched data could be explained by an unmeasured covariate.14 From the 924 pairs of matched patients, we identified pairs in which we could determine which member of the pair clearly had a longer survival or event-free survival time. For example, if one member of a pair died at six months and other member was censored at 12 months, then clearly the second member outlived the other member. In contrast, if one member of a pair is censored at six months and other member died at 12 months, then one could not definitively determine which member of the pair has survived longer. We then tested whether, in the absence of a hidden bias, if patients in the prior heart failure hospitalization group had shorter survival times than their matched counterparts. A significant sign-score test would provide strong evidence of a relationship between prior heart failure hospitalization and time to a particular event. The sign-score test was used to calculate “sensitivity bounds” for a hypothetical unmeasured confounder to determine how much it would need to increase the odds of prior heart failure hospitalization to explain away the significant associations of prior heart failure hospitalization with outcomes. Our sensitivity analysis assumed that the potential unmeasured confounder was a binary baseline characteristic that was a near perfect predictor of the outcomes, which was also not strongly correlated with any of 58 baseline characteristics used in our propensity score model. However, sensitivity analysis cannot determine if such an unmeasured confounder exists. All statistical tests were 2-tailed, and a p value <0.05 was considered significant. All statistical analyses were conducted using IBM SPSS Statistics for Windows software, version 24 (IBM, Armonk, New York).
Results
The 1848 matched patients with HFpEF had a mean age of 78 (±11) years; 90% were ≥65 years, 69% were women, and 15% were African American. Before matching, patients with a prior heart failure hospitalization generally had higher comorbidity and symptom burdens, and a greater proportion was receiving diuretics and digoxin, which were balanced after matching (Table 1a). Absolute standardized differences for all 58 baseline characteristics were <10% in the matched data, suggesting no residual between-group imbalances of confounding consequence (Figure 2).15 Pre-match and post-match baseline characteristics of patients with HFrEF are displayed in Table 1b.
Table 1a.
Baseline Characteristics of Hospitalized Older Patients with Heart Failure with Preserved Ejection Fraction (≥50%), by Hospitalization due to Heart Failure in the Past Six Months
| Mean (±standard deviation) or n (%) | Before propensity score matching (n=7648) | After propensity score matching (n=1848) | ||||
|---|---|---|---|---|---|---|
| Heart failure hospitalization in six months prior to the index hospitalization | Heart failure hospitalization in six months prior to the index hospitalization | |||||
| No (n=6721) | Yes (n=927) | P value | No (n=924) | Yes (n=924) | P value | |
| Age (years) | 78 (±11) | 78 (±11) | 0.183 | 78 (±11) | 78 (±11) | 0.848 |
| Women | 4444 (66%) | 643 (69%) | 0.050 | 643 (70%) | 640 (69%) | 0.880 |
| African American | 797 (12%) | 134 (14%) | 0.023 | 150 (16%) | 131 (14%) | 0.218 |
| Left ventricular ejection fraction (%) | 57 (±6) | 56 (±6) | 0.180 | 57 (±6) | 56 (±6) | 0.726 |
| Smoker in past 1 year | 645 (10%) | 90 (10%) | 0.914 | 91 (10%) | 90 (10%) | 0.938 |
| Admission from nursing home | 122 (2%) | 35 (4%) | <0.001 | 28 (3%) | 33 (4%) | 0.515 |
| Past medical history | ||||||
| Hypertension | 5214 (78%) | 702 (76%) | 0.207 | 691 (75%) | 700 (76%) | 0.627 |
| Coronary artery disease | 3037 (45%) | 476 (51%) | <0.001 | 475 (51%) | 473 (51%) | 0.926 |
| Diabetes mellitus | 2801 (42%) | 442 (48%) | 0.001 | 442 (48%) | 440 (48%) | 0.926 |
| Cerebrovascular disease | 1200 (18%) | 173 (19%) | 0.548 | 184 (20%) | 172 (19%) | 0.479 |
| Peripheral vascular disease | 971 (14%) | 150 (16%) | 0.162 | 151 (16%) | 147 (16%) | 0.800 |
| Atrial fibrillation | 2378 (35%) | 373 (40%) | 0.004 | 354 (38%) | 370 (40%) | 0.446 |
| Chronic obstructive pulmonary disease | 2054 (31%) | 328 (35%) | 0.003 | 330 (36%) | 326 (35%) | 0.846 |
| Anemia | 1458 (22%) | 235 (25%) | 0.012 | 220 (24%) | 234 (25%) | 0.449 |
| Depression | 853 (13%) | 113 (12%) | 0.666 | 110 (12%) | 112 (12%) | 0.886 |
| Admission findings | ||||||
| Dyspnea on exertion | 4239 (63%) | 542 (58%) | 0.007 | 548 (59%) | 542 (59%) | 0.777 |
| Dyspnea at rest | 2874 (43%) | 442 (48%) | 0.005 | 418 (45%) | 440 (48%) | 0.305 |
| Orthopnea | 1686 (25%) | 231 (25%) | 0.913 | 234 (25%) | 231 (25%) | 0.872 |
| Paroxysmal nocturnal dyspnea | 891 (13%) | 114 (12%) | 0.418 | 114 (12%) | 114 (12%) | 1.000 |
| Chest pain | 1506 (22%) | 206 (22%) | 0.899 | 215 (23%) | 204 (22%) | 0.541 |
| Jugular venous pressure elevation | 1661 (25%) | 260 (28%) | 0.028 | 253 (27%) | 257 (28%) | 0.835 |
| Third heart sound | 384 (6%) | 62 (7%) | 0.235 | 54 (6%) | 61 (7%) | 0.500 |
| Pulmonary rales | 4299 (64%) | 604 (65%) | 0.478 | 591 (64%) | 602 (65%) | 0.593 |
| Peripheral edema | 4452 (66%) | 622 (67%) | 0.604 | 593 (64%) | 621 (67%) | 0.170 |
| Serum sodium (mEq/L) | 137 (±10) | 137 (±11) | 0.540 | 136 (±13) | 137 (±11) | 0.408 |
| Hemoglobin (g/dL) | 11.8 (±2.8) | 11.4 (±3.5) | <0.001 | 11.5 (±3.4) | 11.4 (±3.5) | 0.594 |
| Discharge findings | ||||||
| Heart rate (bpm) | 74 (±14) | 74 (±14) | 0.736 | 73 (±14) | 74 (±14) | 0.254 |
| Systolic blood pressure (mmHg) | 130 (±22) | 129 (±22) | 0.089 | 129 (±22) | 129 (±22) | 0.874 |
| Diastolic blood pressure (mmHg) | 66 (±12) | 65 (±12) | 0.076 | 65 (±12) | 65 (±12) | 0.370 |
| Serum creatinine (mg/dL) | 1.7 (±1.4) | 1.8 (±1. 4) | 0.025 | 1.8 (±1.6) | 1.8 (±1. 4) | 0.789 |
| Discharge medications | ||||||
| ACE inhibitors or ARBs | 3811 (57%) | 527 (57%) | 0.932 | 519 (56%) | 526 (57%) | 0.743 |
| Beta blockers | 3864 (57%) | 561 (61%) | 0.080 | 527 (57%) | 558 (60%) | 0.143 |
| Aldosterone antagonists | 512 (8%) | 84 (9%) | 0.124 | 78 (8%) | 84 (9%) | 0.622 |
| Digoxin | 1237 (18%) | 199 (21%) | 0.025 | 180 (19%) | 198 (21%) | 0.299 |
| Loop diuretics | 5049 (75%) | 734 (79%) | 0.007 | 728 (79%) | 732 (79%) | 0.819 |
| Nitrates | 1627 (24%) | 290 (31%) | <0.001 | 314 (34%) | 288 (31%) | 0.197 |
| Amlodipine | 781 (12%) | 107 (12%) | 0.945 | 96 (10%) | 106 (11%) | 0.456 |
| Other calcium channel blockers | 1236 (18%) | 167 (18%) | 0.782 | 182 (20%) | 167 (18%) | 0.373 |
| Hospital length of stay (days) | 5.7 (±4.8) | 6.3 (±5.1) | 0.001 | 6.3 (±6.0) | 6.3 (±5.1) | 0.874 |
| Hospital characteristics | ||||||
| Beds (numbers) | 395 (±247) | 398 (±221) | 0.708 | 395 (±241) | 398 (±221) | 0.737 |
| Academic center | 2830 (42%) | 434 (47%) | 0.007 | 434 (47%) | 432 (47%) | 0.926 |
| Interventional center | 5152 (77%) | 736 (79%) | 0.063 | 740 (80%) | 733 (79%) | 0.686 |
| Heart transplant center | 971 (14%) | 148 (16%) | 0.220 | 149 (16%) | 147 (16%) | 0.899 |
ACE= Angiotensin-converting enzyme; ARBs = Angiotensin receptor blockers
Table 1b.
Baseline Characteristics of Hospitalized Older Patients with Heart Failure with Reduced Ejection Fraction (≤40%), by Hospitalization due to Heart Failure in Prior Six Months
| Mean (±standard deviation) or n (%) | Before propensity score matching (n=9558) | After propensity score matching (n=3688) | ||||
|---|---|---|---|---|---|---|
| Heart failure hospitalization in six months prior to the index hospitalization | Heart failure hospitalization in six months prior to the index hospitalization | |||||
| No (n=7696) | Yes (n=1862) | P value | No (n=1844) | Yes (n=1844) | P value | |
| Age (years) | 76 (±10) | 74 (±12) | <0.001 | 74 (±11) | 74 (±12) | 0.759 |
| Women | 3135 (41%) | 800 (43%) | 0.079 | 812 (44%) | 793 (43%) | 0.528 |
| African American | 993 (13%) | 400 (22%) | <0.001 | 405 (22%) | 389 (21%) | 0.522 |
| Left ventricular ejection fraction (%) | 27 (±8) | 25 (±9) | <0.001 | 25 (±8) | 25 (±8) | 0.705 |
| Smoker in past 1 year | 1011 (13%) | 265 (14%) | 0.212 | 263 (14%) | 261 (14%) | 0.925 |
| Admission from nursing home | 97 (1%) | 40 (2%) | 0.004 | 34 (2%) | 38 (2%) | 0.634 |
| Past medical history | ||||||
| Hypertension | 5186 (67%) | 1229 (66%) | 0.255 | 1223 (66%) | 1218 (66%) | 0.862 |
| Coronary artery disease | 4644 (60%) | 1179 (63%) | 0.018 | 1150 (62%) | 1169 (63%) | 0.517 |
| Diabetes mellitus | 3083 (40%) | 871 (47%) | <0.001 | 863 (47%) | 861 (47%) | 0.947 |
| Cerebrovascular disease | 1248 (16%) | 281 (15%) | 0.235 | 277 (15%) | 278 (15%) | 0.963 |
| Peripheral vascular disease | 1248 (16%) | 298 (16%) | 0.824 | 306 (17%) | 296 (16%) | 0.656 |
| Atrial fibrillation | 2603 (34%) | 664 (36%) | 0.134 | 656 (36%) | 657 (36%) | 0.973 |
| Chronic obstructive pulmonary disease | 2073 (27%) | 554 (30%) | 0.015 | 536 (29%) | 548 (30%) | 0.664 |
| Anemia | 1231 (16%) | 323 (17%) | 0.156 | 319 (17%) | 321 (17%) | 0.931 |
| Depression | 757 (10%) | 187 (10%) | 0.789 | 187 (10%) | 186 (10%) | 0.956 |
| Admission findings | ||||||
| Dyspnea on exertion | 4941 (64%) | 1123 (60%) | 0.002 | 1108 (60%) | 1120 (61%) | 0.686 |
| Dyspnea at rest | 3279 (43%) | 870 (47%) | 0.001 | 855 (46%) | 855 (46%) | 1.000 |
| Orthopnea | 2093 (27%) | 556 (30%) | 0.021 | 544 (30%) | 546 (30%) | 0.942 |
| Paroxysmal nocturnal dyspnea | 1245 (16%) | 312 (17%) | 0.544 | 302 (16%) | 308 (17%) | 0.790 |
| Chest pain | 1643 (21%) | 382 (21%) | 0.430 | 371 (20%) | 380 (21%) | 0.713 |
| Jugular venous pressure elevation | 2398 (31%) | 688 (37%) | <0.001 | 699 (38%) | 678 (37%) | 0.475 |
| Third heart sound | 872 (11%) | 259 (14%) | 0.002 | 279 (15%) | 253 (14%) | 0.223 |
| Pulmonary rales | 4850 (63%) | 1184 (64%) | 0.649 | 1183 (64%) | 1172 (64%) | 0.706 |
| Peripheral edema | 4674 (61%) | 1210 (65%) | 0.001 | 1190 (65%) | 1195 (65%) | 0.863 |
| Serum sodium (mEq/L) | 137 (±10) | 136 (±12) | 0.007 | 136 (±12) | 136 (±12) | 0.965 |
| Hemoglobin (g/dL) | 12.4 (±2.3) | 12.1 (±3.8) | <0.001 | 12.2 (±2.8) | 12.1 (±3.8) | 0.579 |
| Discharge findings | ||||||
| Heart rate (bpm) | 75 (±13) | 76 (±13) | 0.018 | 76 (±13) | 76 (±13) | 0.743 |
| Systolic blood pressure (mmHg) | 121 (±21) | 118 (±20) | <0.001 | 118 (±20) | 118 (±20) | 0.974 |
| Diastolic blood pressure (mmHg) | 66 (±12) | 65 (±12) | 0.005 | 65 (±12) | 65 (±12) | 0.541 |
| Serum creatinine (mg/dL) | 1.7 (±1.2) | 1.9 (±1.4) | <0.001 | 1.9 (±1.3) | 1.9 (±1.4) | 0.853 |
| Discharge medications | ||||||
| ACE inhibitors or ARBs | 5294 (69%) | 1240 (67%) | 0.068 | 1235 (67%) | 1227 (67%) | 0.780 |
| Beta blockers | 5505 (72%) | 1323 (71%) | 0.682 | 1310 (71%) | 1312 (71%) | 0.942 |
| Aldosterone antagonists | 1235 (16%) | 340 (18%) | 0.021 | 361 (20%) | 336 (18%) | 0.293 |
| Digoxin | 2886 (38%) | 800 (43%) | <0.001 | 756 (41%) | 788 (43%) | 0.285 |
| Loop diuretics | 6120 (80%) | 1502 (81%) | 0.270 | 1471 (80%) | 1486 (81%) | 0.536 |
| Nitrates | 2067 (27%) | 558 (30%) | 0.007 | 568 (31%) | 549 (30%) | 0.496 |
| Amlodipine | 417 (5%) | 90 (5%) | 0.312 | 93 (5%) | 90 (5%) | 0.820 |
| Other calcium channel blockers | 527 (7%) | 103 (6%) | 0.040 | 105 (6%) | 103 (6%) | 0.886 |
| Hospital length of stay (days)* | 5.9 (5.5) | 6.5 (9.7) | <0.001 | 6.1 (6.0) | 6.5 (9.7) | 0.234 |
| Hospital characteristics* | ||||||
| Number of beds | 426 (255) | 442 (251) | 0.015 | 442 (252) | 441 (250) | 0.918 |
| Academic center | 3634 (47%) | 1005 (54%) | <0.001 | 1015 (55%) | 993 (54%) | 0.467 |
| Interventional center | 6070 (79%) | 1527 (82%) | 0.003 | 1533 (83%) | 1509 (82%) | 0.298 |
| Heart transplant center | 1273 (17%) | 399 (21%) | <0.001 | 376 (20%) | 391 (21%) | 0.543 |
ACE = Angiotensin-converting enzyme; ARB = Angiotensin receptor blocker
Mortality in the Matched HFpEF Cohort
All-cause mortality occurred in 81% and 71% of patients with and without a prior heart failure hospitalization, respectively during six (median, two) years of follow-up (HR associated with time to a prior heart failure hospitalization, 1.35; 95% CI, 1.21–1.50; p<0.001; Table 2, Figure 3). In 93% (861/924) of the matched pairs, we were able to determine which member of the pair had a shorter time to death, and in 59% (510/861) of those pairs, patients with a shorter survival time belonged to the group with prior heart failure hospitalization (sign-score test p <0.0001). A hidden covariate that is a near-perfect predictor of mortality would need to increase the odds of prior heart failure hospitalization by 21% before it could explain away this association. There was no heterogeneity in the association between prior heart failure hospitalization and all-cause mortality among various clinically relevant subgroups (Figure 4). HRs (95% CIs) for 30-day and 1-year all-cause mortality were 1.65 (1.15–2.39; p=0.007) and 1.44 (1.23–1.68; p<0.001), respectively (Table 2).
Table 2.
Outcomes in 1848 Propensity Score-Matched Hospitalized Older Patients with Heart Failure with Preserved Ejection Fraction (≥50%), by Hospitalization due to Heart Failure in the Past Six Months
| Events (%) by heart failure hospitalization in six months prior to the index hospitalization | Hazard ratio associated with prior hospitalization due to heart failure (95% CIs); p value | ||
|---|---|---|---|
| No (n=924) | Yes (n=924) | ||
| 30-day outcomes | |||
| All-cause mortality | 46 (5%) | 75 (8%) | 1.65 (1.15–2.39); p=0.007 |
| All-cause readmission | 221 (24%) | 308 (33%) | 1.48 (1.24–1.75); p<0.001 |
| Heart failure readmission | 65 (7%) | 137 (15%) | 2.21 (1.65–2.97); p<0.001 |
| 1-year outcomes | |||
| All-cause mortality | 276 (30%) | 369 (40%) | 1.44 (1.23–1.68); p<0.001 |
| All-cause readmission | 656 (71%) | 723 (78%) | 1.37 (1.23–1.52); p<0.001 |
| Heart failure readmission | 266 (29%) | 423 (46%) | 1.99 (1.70–2.32); p<0.001 |
| Overall outcomes | |||
| All-cause mortality | 656 (71%) | 747 (81%) | 1.35 (1.21–1.50); p<0.001 |
| All-cause readmission | 827 (90%) | 830 (90%) | 1.34 (1.21–1.47); p<0.001 |
| Heart failure readmission | 407 (44%) | 552 (60%) | 1.90 (1.67–2.16); p<0.001 |
Figure 4.
Forest plots displaying associations of heart failure hospitalization in the six months prior to index hospitalization and all-cause mortality during six years of follow-up in clinically relevant subgroups of propensity score-matched patients with heart failure with preserved ejection fraction (≥50%) (ACE = Angiotensin-converting enzyme; ARB = Angiotensin receptor blocker; CI = Confidence interval)
Readmissions in the Matched HFpEF Cohort
All-cause readmission occurred in 90% of patients in both groups (HR associated with time to prior heart failure hospitalization, 1.34; 95% CI, 1.21–1.47; p<0.001; Table 2). In 829 of 924 matched pairs, we were able to determine which member of the pair had a shorter time to all-cause readmission, and in 60% (495/829) of those pairs, those with a shorter readmission-free survival belonged to the group with prior heart failure hospitalization (sign-score test p <0.001). A hidden covariate that is a near-perfect predictor of all-cause readmission would need to increase the odds of prior heart failure hospitalization by 22% before it could explain away this association.
Heart failure readmission occurred in 60% and 44% of patients with and without a prior heart failure hospitalization, respectively (HR, 1.90; 95% CI, 1.67–2.16; p<0.001; Table 2). Of the 861 matched pairs with time to heart failure readmission data for both members of the pair, this time was shorter in 68% (393/861) of those in the prior heart failure hospitalization group (sign-score test p <0.001). A hidden covariate that is a near-perfect predictor of heart failure readmission would need to increase the odds of prior heart failure hospitalization by 44% before it could explain away this association.
Mortality in the Matched HFrEF Cohort
All-cause mortality occurred in 79% and 73% of patients with and without prior heart failure hospitalization, respectively (HR 1.17; 95% CI 1.09–1.26; p<0.001; Table 3, Figure 3). In 1728 out of 1844 matched pairs, we were able to determine which patient within a pair had a shorter time to death, and 55% (951/1728) of those pairs belonged to the group with a prior heart failure hospitalization (sign-score test p <0.001). A hidden covariate that is a near-perfect predictor of mortality would need to increase the odds of prior heart failure hospitalization by 10% before it could explain away this association.
Table 3.
Outcomes in 3688 Propensity Score-Matched Hospitalized Older Patients with Heart Failure with Reduced Ejection Fraction (≤40%), by Hospitalization due to Heart Failure in the Past Six Months
| Events (%) by heart failure hospitalization in six months prior to the index hospitalization | Hazard ratio associated with prior hospitalization due to heart failure (95% CIs); p value | ||
|---|---|---|---|
| No (n=1844) | Yes (n=1844) | ||
| 30-day outcomes | |||
| All-cause mortality | 131 (7%) | 158 (9%) | 1.22 (0.96 –1.53); p=0.099 |
| All-cause readmission | 462 (25%) | 623 (34%) | 1.44 (1.28–1.63); p<0.001 |
| Heart failure readmission | 225 (12%) | 365 (20%) | 1.72 (1.46–2.03); p<0.001 |
| 1-year outcomes | |||
| All-cause mortality | 647 (35%) | 748 (41%) | 1.21 (1.09 –1.34); p=0.001 |
| All-cause readmission | 1273 (69%) | 1455 (79%) | 1.40 (1.30 –1.51); p<0.001 |
| Heart failure readmission | 763 (41%) | 1007 (55%) | 1.56 (1.42 –1.72); p<0.001 |
| Overall outcomes | |||
| All-cause mortality | 1354 (73%) | 1463 (79%) | 1.17 (1.09 –1.26); p<0.001 |
| All-cause readmission | 1595 (86%) | 1637 (89%) | 1.32 (1.23 –1.41); p<0.001 |
| Heart failure readmission | 1061 (58%) | 1249 (68%) | 1.48 (1.37 –1.61); p<0.001 |
Readmissions in the Matched HFrEF Cohort
HRs (95% CIs) for all-cause readmission and heart failure readmission associated with a prior heart failure hospitalization were 1.32 (1.23–1.41; p<0.001) and 1.48 (1.37–1.61; p<0.001; Table 3). Of the 1844 matched pairs, we were able to determine time to all-cause and heart failure readmission in 1615 and 1268 pairs of patients, respectively, and in 58% (935/1615) and 62% (785/1268) of those pairs, patients in the prior heart failure hospitalization group had shorter readmission-free survival (sign-score test p for both <0.001). A hidden covariate would need increase the odds of prior heart failure hospitalization by 20% and 31%, respectively, to explain away these associations.
Discussion
Findings from our study demonstrate that a prior heart failure hospitalization is associated with higher risks for short- and long-term all-cause mortality, all-cause readmission and heart failure readmission in patients with HFpEF. We also observed similar associations in patients with HFrEF. Interestingly, the relative risks for all-cause mortality and heart failure readmission appeared to be higher in HFpEF than in HFrEF. To the best of our knowledge, this is the first study to report adverse outcomes associated with a prior heart failure hospitalization in propensity score-matched cohorts of hospitalized older patients with HFpEF and HFrEF. These results suggest that a prior heart failure hospitalization is prognostically as, or perhaps even more significant in HFpEF than in HFrEF.
There are several potential explanations for the findings of our study. Patients with a prior heart failure hospitalization likely represent those with more advanced disease and a higher burden of comorbidities. Before matching, more patients with a prior heart failure hospitalization had dyspnea at rest, jugular venous pressure elevation, and received digoxin and loop diuretics, which are all markers of worse outcomes.16–18 These patients also had a higher burden of comorbidities such as coronary artery disease, diabetes mellitus, atrial fibrillation, chronic obstructive pulmonary disease and anemia. Although between-group distribution of these and other characteristics were balanced in our matched cohort, matching may not balance for their severity. For example, 48% of patients in our matched cohort had diabetes, but it is possible that those in the prior heart failure hospitalization group had more severe or advanced diabetes. Furthermore, if the reasons for the higher prevalence of diabetes or use of diuretics persisted during follow-up, then they would be expected to become imbalanced and have poor outcomes.
Hospitalization due to acute decompensation of heart failure is often associated with a relatively heightened and more sustained neurohormonal activation, which may further contribute to the poor outcomes in the group with a prior heart failure hospitalization.5,19,20 Furthermore, more aggressive therapeutic strategies with the use of intravenous loop diuretics and inotropes can lead to further neurohormonal activation, electrolyte imbalance and worsening kidney function, thus increasing the risk of poor outcomes among patients with prior heart failure hospitalization.18,21–23 Finally, non-adherence to prescribed diet and therapy is known to contribute to heart failure hospitalizations.24 If patients with a prior heart failure hospitalization continue to be less adherent to their diet and drugs, that would be expected to contribute to poor outcomes.
An interesting observation of our study is that prior heart failure hospitalization-associated risks of all-cause mortality and heart failure readmission were higher in patients with HFpEF than in those with HFrEF. A potential explanation is that adverse effects are often less pronounced in subsets of patients that have higher baseline risk or event rates. For example, the absolute risk of 30-day heart failure readmission was 8% higher in both HFrEF (20% – 12% = 8%) and HFpEF (15% – 7% = 8%). However, because of the higher baseline risk in HFrEF (12%) than in HFpEF (7%), the relative risk was lower in HFrEF. In contrast, baseline risks of 30-day all-cause readmission were similar (25% and 24%) for HFrEF and HFpEF, and relative risks were also similar: 44% and 48% for HFrEF and HFpEF, respectively. Furthermore, competing risk factors for poor outcomes in HFrEF might be stronger than in HFpEF. For example, fatal ventricular arrhythmias are more common in HFrEF than in HFpEF and a higher incidence of sudden cardiac death in patients with HFrEF would attenuate the risk of death associated with a prior heart failure hospitalization.25,26
Several previous studies have examined the association of prior heart failure hospitalization with outcomes in patients with chronic heart failure.1–5 A prior heart failure hospitalization is associated with higher risks of subsequent poor outcomes among ambulatory patients with HFrEF or HFpEF.1,2 The current study is distinguished by its large sample size, national representation, use of propensity scores to assemble balanced cohorts, sensitivity analyses, subgroup analyses, and comparison with HFrEF. Our study findings have important clinical implications. The HFpEF phenotype is characteristically, prognostically and therapeutically different form HFrEF. Findings from our study suggest that a heart failure hospitalization in the past six months can be used to risk stratify patients with HFpEF who might be at a higher risk for worse outcomes.
There are several limitations of this study. Despite our use of propensity score-matching, confounding resulting from residual bias or to unmeasured covariates is possible. However, findings from post-match absolute standardized differences suggest that residual biases of 58 variables used in the propensity score model were inconsequential and findings from the sensitivity analyses suggest that these results were relatively immune to bias by unmeasured confounders. Patients in the prior heart failure hospitalization group had to survive up to six months to have a heart failure readmission during index hospitalization. However, this survivor cohort phenomenon may have attenuated the true association between a prior heart failure hospitalization and subsequent outcomes. Our study is based on fee-for-service Medicare beneficiaries, which may limit generalizability.
Conclusions
Among hospitalized older patients with HFpEF, a heart failure hospitalization in the past six months is associated with higher risks of all-cause mortality, all-cause readmission and heart failure readmission, with relative risks which were similar to or perhaps greater than those observed in HFrEF. Future studies need to replicate these findings in more contemporary patient populations, and to develop and test interventions that may improve outcomes in patients with HFpEF and a prior heart failure hospitalization.
Clinical Significance.
Among hospitalized older patients with heart failure with preserved ejection fraction (HFpEF), a heart failure hospitalization within prior six months is associated with higher risks of mortality and readmissions.
These risks are similar to those observed in patients with heart failure with reduced ejection fraction (HFrEF).
The relative risks of mortality and heart failure readmission appear to be higher in HFpEF than in HFrEF.
Acknowledgments
Funding: Dr. Ali Ahmed was in part supported by the National Institutes of Health through grants (R01-HL085561, R01-HL085561-S and R01-HL097047) from the National Heart, Lung, and Blood Institute (NHLBI). OPTIMIZE-HF was sponsored by GlaxoSmithKline, but played no role in the design, conduct, analyses or interpretation of the current study.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Disclosures: Dr. Fonarow reports consulting with Abbott, Amgen, Bayer, Janssen, Medtronic, Novartis, and was the Principle Investigator of OPTIMIZE-HF. None of the other authors report any conflicts of interest related to this manuscript. This content is solely the responsibility of the authors and does not necessarily represent the official views of the Department of Veterans Affairs.
References:
- 1.Bello NA, Claggett B, Desai AS, et al. Influence of previous heart failure hospitalization on cardiovascular events in patients with reduced and preserved ejection fraction. Circ Heart Fail 2014;7(4):590–595. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Ahmed A, Allman RM, Fonarow GC, et al. Incident heart failure hospitalization and subsequent mortality in chronic heart failure: a propensity-matched study. J Card Fail 2008;14(3):211–218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Solomon SD, Dobson J, Pocock S, et al. Influence of nonfatal hospitalization for heart failure on subsequent mortality in patients with chronic heart failure. Circulation 2007;116(13):1482–1487. [DOI] [PubMed] [Google Scholar]
- 4.Kommuri NV, Koelling TM, Hummel SL. The impact of prior heart failure hospitalizations on long-term mortality differs by baseline risk of death. Am J Med 2012;125(2):209e209–209e215. [DOI] [PubMed] [Google Scholar]
- 5.Setoguchi S, Stevenson LW, Schneeweiss S. Repeated hospitalizations predict mortality in the community population with heart failure. Am Heart J 2007;154(2):260–266. [DOI] [PubMed] [Google Scholar]
- 6.Dunlay SM, Roger VL, Redfield MM. Epidemiology of heart failure with preserved ejection fraction. Nat Rev Cardiol 2017;14(10):591–602. [DOI] [PubMed] [Google Scholar]
- 7.Fonarow GC, Stough WG, Abraham WT, et al. Characteristics, treatments, and outcomes of patients with preserved systolic function hospitalized for heart failure: a report from the OPTIMIZE-HF Registry. J Am Coll Cardiol 2007;50(8):768–777. [DOI] [PubMed] [Google Scholar]
- 8.Fonarow GC, Abraham WT, Albert NM, et al. Organized Program to Initiate Lifesaving Treatment in Hospitalized Patients with Heart Failure (OPTIMIZE-HF): rationale and design. Am Heart J 2004;148(1):43–51. [DOI] [PubMed] [Google Scholar]
- 9.Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983;70(1):41–55. [Google Scholar]
- 10.Rubin DB. Using propensity scores to help design observational studies: application to the tobacco litigation. Health Services and Outcomes Research Methodology 2001;2(3–4):169–188. [Google Scholar]
- 11.Ahmed A, Husain A, Love TE, et al. Heart failure, chronic diuretic use, and increase in mortality and hospitalization: an observational study using propensity score methods. Eur Heart J 2006;27(12):1431–1439. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Ahmed MI, White M, Ekundayo OJ, et al. A history of atrial fibrillation and outcomes in chronic advanced systolic heart failure: a propensity-matched study. Eur Heart J 2009;30(16):2029–2037. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Zhang Y, Kilgore ML, Arora T, et al. Design and rationale of studies of neurohormonal blockade and outcomes in diastolic heart failure using OPTIMIZE-HF registry linked to Medicare data. Int J Cardiol 2013;166(1):230–235. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Rosenbaum PR. Sensitivity to hidden bias. . In: Rosenbaum PR, ed. Observational Studies Vol 1. 2nd ed. New York: Springer-Verlag; 2002:105–170. [Google Scholar]
- 15.Wahle C, Adamopoulos C, Ekundayo OJ, Mujib M, Aronow WS, Ahmed A. A propensity-matched study of outcomes of chronic heart failure (HF) in younger and older adults. Arch Gerontol Geriatr 2009;49(1):165–171. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Gheorghiade M, Zannad F, Sopko G, et al. Acute heart failure syndromes: current state and framework for future research. Circulation 2005;112(25):3958–3968. [DOI] [PubMed] [Google Scholar]
- 17.Ahmed A, Rich MW, Fleg JL, et al. Effects of digoxin on morbidity and mortality in diastolic heart failure: the ancillary digitalis investigation group trial. Circulation 2006;114(5):397–403. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Hasselblad V, Gattis Stough W, Shah MR, et al. Relation between dose of loop diuretics and outcomes in a heart failure population: results of the ESCAPE trial. Eur J Heart Fail 2007;9(10):1064–1069. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Schrier RW, Abraham WT. Hormones and hemodynamics in heart failure. N Engl J Med 1999;341(8):577–585. [DOI] [PubMed] [Google Scholar]
- 20.Lin AH, Chin JC, Sicignano NM, Evans AM. Repeat Hospitalizations Predict Mortality in Patients With Heart Failure. Mil Med 2017;182(9):e1932–e1937. [DOI] [PubMed] [Google Scholar]
- 21.Francis GS, Siegel RM, Goldsmith SR, Olivari MT, Levine TB, Cohn JN. Acute vasoconstrictor response to intravenous furosemide in patients with chronic congestive heart failure. Activation of the neurohumoral axis. Ann Intern Med 1985;103(1):1–6. [DOI] [PubMed] [Google Scholar]
- 22.Cooper HA, Dries DL, Davis CE, Shen YL, Domanski MJ. Diuretics and risk of arrhythmic death in patients with left ventricular dysfunction. Circulation 1999;100(12):1311–1315. [DOI] [PubMed] [Google Scholar]
- 23.Butler J, Forman DE, Abraham WT, et al. Relationship between heart failure treatment and development of worsening renal function among hospitalized patients. Am Heart J 2004;147(2):331–338. [DOI] [PubMed] [Google Scholar]
- 24.Michalsen A, Konig G, Thimme W. Preventable causative factors leading to hospital admission with decompensated heart failure. Heart 1998;80(5):437–441. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Zile MR, Gaasch WH, Anand IS, et al. Mode of death in patients with heart failure and a preserved ejection fraction: results from the Irbesartan in Heart Failure With Preserved Ejection Fraction Study (I-Preserve) trial. Circulation 2010;121(12):1393–1405. [DOI] [PubMed] [Google Scholar]
- 26.Carson P, Massie BM, McKelvie R, et al. The irbesartan in heart failure with preserved systolic function (I-PRESERVE) trial: rationale and design. J Card Fail 2005;11(8):576–585. [DOI] [PubMed] [Google Scholar]