Abstract
BACKGROUND
Heart failure (HF) is a major source of morbidity and mortality. Fluid retention and shortness of breath are its cardinal manifestations for which loop diuretics are used. Although their usefulness is well accepted, less is known about their role in improving clinical outcomes.
OBJECTIVES
To determine the relationship between loop diuretics and clinical outcomes in patients with HF.
METHODS
Of the 25,345 older patients hospitalized for HF in Medicare-linked OPTIMIZE-HF registry, 9,866 (39%) received no pre-admission diuretics. We excluded 1,083 patients receiving dialysis and 847 discharged on thiazide diuretics. Of the remaining 7,936 patients, 5,568 (70%) were prescribed loop diuretics at discharge. Using propensity scores for receipt of loop diuretics estimated for each of the 7,936 patients, we assembled a matched cohort of 2,191 pairs of patients, balanced on 74 baseline characteristics. Hazard ratios (HRs) and 95% confidence intervals (CIs) for outcomes were estimated in the matched cohort.
RESULTS
Matched patients (N=4,382) had a mean age of 78 years, 54% were women, and 11% African American. 30-day all-cause mortality occurred in 4.9% (107/2,191) and 6.6% (144/2,191) of patients in the loop diuretic and no loop diuretic groups, respectively (HR when the use of loop diuretics was compared to their nonuse, 0.73; 95% CI, 0.57–0.94; p=0.016). Patients in the loop diuretic group had a significantly lower risk of 30-day HF readmission (HR, 0.79; 95% CI, 0.63–0.99; p=0.037) but not of 30-day all-cause readmission (HR, 0.89; 95% CI, 0.79–1.01; p=0.081). None of the associations was statistically significant during 60 days of follow-up.
CONCLUSION
Hospitalized older patients not taking diuretics prior to hospitalization for HF decompensation who received a discharge prescription for loop diuretics had significantly better 30-day clinical outcomes than those not discharged on loop diuretics. These findings provide new information about short-term clinical benefits associated with loop diuretic use in HF.
Keywords: Heart failure, Loop diuretics, Outcomes
Tweet:
Discharge prescription of loop diuretics is linked to improved 30-day outcomes in hospitalized patients with heart failure.
Condensed Abstract:
Loop diuretics improve fluid retention and dyspnea and are essential for the management of heart failure (HF). However, less is known about the association between loop diuretic use and clinical outcomes. Findings from the current propensity score-matched study suggest that in hospitalized older patients not taking diuretics prior to hospitalization for HF decompensation, loop diuretic prescription at discharge is associated with a lower risk of 30-day mortality and readmission.
Heart failure (HF) is a major cause of morbidity and mortality and is a leading cause for hospitalization (1). Fluid retention is central to the pathophysiology of HF and underlies the cardinal manifestations of HF which are shortness of breath and edema (2). Loop diuretics frequently are the only drugs that can adequately control fluid retention in HF (1,3). According to the American College of Cardiology Foundation / American Heart Association (ACCF/AHA) HF guideline, “diuretics have been shown to improve symptoms and exercise tolerance in patients with heart failure”, however, “diuretic effects on morbidity and mortality are not known” (1). The objective of the current study was to examine the association between loop diuretics and clinical outcomes in patients with HF with reduced and preserved ejection fraction (HFrEF and HFpEF).
Methods
Data Source and Study Population
We used data from the Organized Program to Initiate Lifesaving Treatment in Hospitalized Patients with Heart Failure (OPTIMIZE-HF) registry for the current analysis. OPTIMIZE-HF is a web-based registry of 48,612 HF hospitalizations in 259 hospitals from 48 states between March 1, 2003 and December 31, 2004 (4–8). Extensive baseline data were collected, and data on long-term outcomes were later obtained for 26,376 unique patients by linking with the Medicare data (Figure 1) (9). Of the 26,376 patients, 25,345 were discharged alive. The OPTIMIZE-HF protocol was approved by each participating center’s institutional review board (RB) or by a central IRB (5). Data used for the current analysis was approved by the IRB and Research and Development Committee of the Veterans Affairs Medical Center, Washington, DC.
Figure 1. Assembly of the Study Cohort.
Flow chart displaying assembly of a propensity score-matched cohort of older patients hospitalized for heart failure decompensation who were not taking diuretics prior to hospitalization, by loop diuretic prescription at discharge. OPTIMIZE-HF = Organized Program to Initiate Lifesaving Treatment in Hospitalized Patients with Heart Failure.
Assembly of an Inception Cohort
To minimize prevalent-user bias, we assembled an inception cohort by excluding 15,479 patients who were taking diuretics prior to hospitalization for HF decompensation (10,11). We then excluded 728 patients with a history of dialysis and 355 who received dialysis during hospitalization (Figure 1). Because thiazide diuretics have less diuretic effect, have a different mechanism of action, and may augment the effect of loop diuretics (1,12), we excluded 847 patients discharged on thiazide diuretics (Figure 1). Thus, our study sample included 7,936 patients. Of the 7,936 patients, 5,568 received a discharge prescription for loop diuretics and 2,368 did not.
Assembly of a Balanced Cohort
In a randomized controlled trial (RCT) of diuretics, all patients will have a 50% probability of receiving the drug regardless of whether one received it or not. The probability of receiving a prescription for diuretics in the clinical practice setting, however, would be influenced by measured and unmeasured baseline characteristics and would vary between 0 and 100%. This conditional probability, also known as a propensity score (13,14), can be estimated to assemble a matched cohort in which patients receiving and not receiving a prescription for diuretics will be balanced on measured baseline characteristics. This balance is measured as an absolute standardized difference, and baseline characteristics with values <10% are considered balanced (0% implies no bias). Although within a matched pair patients receiving and not receiving a prescription for diuretics may not have the same baseline characteristics, they will have a similar probability of receiving the drug (15–17). Thus, like randomization, propensity score matching is a study design tool. As in an RCT, the process of assembling a propensity score-matched cohort is outcome blinded (18,19), but unlike in an RCT, it may not balance unmeasured baseline characteristics. However, sensitivity analysis (described later) can determine their impact on observed significant associations.
We used a non-parsimonious multivariable logistic regression model to estimate propensity scores for the receipt of loop diuretics for each of the 7,936 patients. We used 74 baseline patient and care characteristics listed in Supplemental Figure 1 as covariates in the model. Using a greedy matching algorithm described elsewhere in detail (6–8,20), we matched 2,191 patients receiving a prescription for loop diuretics with 2,191 patients not receiving one based on their propensity scores. Among the 4,382 matched patients, those receiving and not receiving a prescription for loop diuretics had the same 67% probability of receiving those drugs (mean propensity score, 0.67; ±standard deviation, 0.13 for both study groups; p=0.950). We then estimated absolute standardized differences for all 74 baseline characteristics to assess their post-match balance.
Outcomes Data
Our outcomes of interest were HF readmission, all-cause readmission, and all-cause mortality. We also examined two combined endpoints of either readmission or mortality. We examined these outcomes at 30 and 60 days after hospital discharge. Data on all events and time to events were collected from the Medicare data (9).
Statistical Analyses
Descriptive analyses were conducted using the Pearson chi-square and Wilcoxon rank sum tests. All outcome analyses were conducted in the matched cohort in which patients receiving and not receiving a prescription for loop diuretics were balanced on 74 baseline characteristics. Kaplan-Meier survival analysis were used to generate plots for all-cause mortality and HF readmission. Cox regression models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) associated with loop diuretic use. Because patients in the matched cohort were balanced on 74 baseline characteristics, the Cox regression model was not adjusted for these variables. Assumption of the proportional hazard was assessed by visual examinations of the log (minus log) curves. Formal sensitivity analyses were conducted using Rosenbaum’s approach (21) described elsewhere in detail (6,7). All outcomes were analyzed separately among patients with HFrEF and HFpEF, defined as EF <45% and ≥45%, respectively. Additional subgroup analyses were conducted to determine the homogeneity of the association in other clinically relevant subgroups. All statistical tests were 2-tailed, and a p value <0.05 was considered significant. SPSS for Windows version 26 (IBM Corp., Armonk, NY) and SAS for Window version 9.2 (Cary, NC) were used for data analyses.
Results
Baseline Characteristics
Patients in the propensity score-matched cohort (n=4,382) had a mean (±standard deviation) age of 78 (±10) years, 54% were women, and 11% were African American. Before matching, patients in the loop diuretic group were older and had a higher prevalence of signs and symptoms of HF (Table 1). After propensity score matching, all 74 baseline characteristics had an absolute standardized difference <5%, 56 had values <2%, and 41 had values <1% (0% indicates no residual bias; Supplemental Figure 1).
Table 1.
Baseline Characteristics by Discharge Prescription for Loop Diuretics in Older Patients With Heart Failure not Taking Diuretics Prior to Hospitalization for Heart Failure Decompensation
| Before propensity score matching (n=7,936) | After propensity score matching (n=4,382) | |||||
|---|---|---|---|---|---|---|
| Discharge prescription of loop diuretics | P value | Discharge prescription of loop diuretics | P value | |||
| No (n=2,368) | Yes (n=5,568) | No (n=2,191) | Yes (n=2,191) | |||
| Age, yrs | 77.7 (±10.7) | 78.3 (±10.2) | 0.010 | 77.9 (±10.4) | 77.8 (±10.5) | 0.841 |
| Age ≥70 years | 1504 (64%) | 3729 (67%) | 0.003 | 1114 (65%) | 1444 (66%) | 0.341 |
| Women | 1274 (54%) | 3061 (55%) | 0.336 | 1178 (54%) | 1202 (55%) | 0.467 |
| African American | 242 (10%) | 632 (11%) | 0.141 | 233 (11%) | 239 (11%) | 0.770 |
| Past medical history | ||||||
| HF diagnosis before admission | 1934 (82%) | 4467 (80%) | 0.136 | 1787 (82%) | 1783 (81%) | 0.876 |
| HF hospitalization (prior 6 mos.) | 288 (12%) | 632 (11%) | 0.301 | 263 (12%) | 277 (13%) | 0.520 |
| Hypertension | 1676 (71%) | 3969 (71%) | 0.649 | 1549 (71%) | 1541 (70%) | 0.791 |
| Myocardial infarction | 510 (22%) | 1199 (22%) | 0.997 | 470 (21%) | 479 (22%) | 0.741 |
| Diabetes mellitus | 769 (32%) | 1862 (33%) | 0.403 | 700 (32%) | 710 (32%) | 0.746 |
| Peripheral vascular disease | 307 (13%) | 684 (12%) | 0.402 | 280 (13%) | 283 (13%) | 0.892 |
| Atrial fibrillation | 641 (27%) | 1658 (30%) | 0.015 | 602 (27%) | 605 (28%) | 0.919 |
| COPD | 614 (26%) | 1346 (24%) | 0.097 | 567 (26%) | 572 (26%) | 0.863 |
| Acute kidney insufficiency | 66 (3%) | 94 (2%) | 0.001 | 57 (3%) | 52 (2%) | 0.628 |
| Admission symptoms and signs | ||||||
| Dyspnea on exertion | 1351 (57%) | 3530 (63%) | <0.001 | 1275 (58%) | 1265 (58%) | 0.760 |
| Orthopnea | 474 (20%) | 1420 (26%) | <0.001 | 459 (21%) | 449 (20%) | 0.709 |
| Paroxysmal nocturnal dyspnea | 259 (11%) | 797 (14%) | <0.001 | 250 (11%) | 249 (11%) | 0.962 |
| Dyspnea at rest | 951 (40%) | 2434 (44%) | 0.003 | 895 (41%) | 890 (41%) | 0.878 |
| JVP elevation | 499 (21%) | 1501 (27%) | <0.001 | 478 (22%) | 472 (22%) | 0.826 |
| Third heart sound | 150 (6%) | 445 (8%) | 0.010 | 141 (6%) | 138 (6%) | 0.853 |
| Pulmonary rales | 1362 (58%) | 3641 (65%) | <0.001 | 1291 (59%) | 1315 (60%) | 0.460 |
| Lower extremity edema | ||||||
| None to trace | 1536 (65%) | 2924 (53%) | 1387 (63%) | 1393 (64%) | ||
| Mild to moderate (1+ to 2+) | 659 (28%) | 2002 (36%) | <0.001 | 635 (29%) | 628 (29%) | 0.973 |
| Severe (3+ to 4+) | 173 (7%) | 642 (12%) | 169 (8%) | 170 (8%) | ||
| Other admission clinical findings | ||||||
| Weight (kilogram) | 75 (±18) | 76 (±18) | 0.100 | 75 (±18) | 75 (±18) | 0.348 |
| Heart rate, beat/min | 89 (±22) | 88 (±22) | 0.394 | 88 (±22) | 88 (±22) | 0.378 |
| Systolic blood pressure, mmHg | 147 (±32) | 147 (±31) | 0.466 | 147 (±32) | 148 (±32) | 0.910 |
| Diastolic blood pressure, mmHg | 76.5 (±16.4) | 77.4 (±16.0) | 0.022 | 76.8 (±16.3) | 76.9 (±15.9) | 0.847 |
| Admission laboratory findings | ||||||
| Serum creatinine, mg/dL | 1.4 (±0.6) | 1.3 (±0.5) | <0.001 | 1.3 (±0.6) | 1.3 (±0.6) | 0.358 |
| Serum sodium, mEq/L | 138 (±5) | 138 (±5) | 0.111 | 138 (±5) | 138 (±5) | 0.922 |
| Serum hemoglobin, g/dL | 12 (±2) | 12 (±2) | 0.905 | 12 (±2) | 12 (±2) | 0.747 |
| Serum pro-BNP, pg/mL | 1035 (507–1173) | 1053 (556–1273) | 0.011 | 1039 (509–1179) | 1034 (517–1220) | 0.763 |
| Serum troponin level elevation | 458 (19%) | 1027 (18%) | 0.349 | 411 (19%) | 430 (20%) | 0.466 |
| LVEF, % | 43 (±14) | 42 (±15) | 0.013 | 43 (±14) | 43 (±15) | 0.945 |
| LVEF ≤40% | 1027 (43%) | 2641 (47%) | 943 (43%) | 963 (44%) | ||
| LVEF 41–49% | 386 (16%) | 789 (14%) | 0.002 | 354 (16%) | 350 (16%) | 0.828 |
| LVEF ≥50% | 955 (40%) | 2138 (38%) | 894 (41%) | 878 (40%) | ||
| In-hospital medications and interventions | ||||||
| Dobutamine parenteral infusion | 59 (2%) | 127 (2%) | 0.570 | 52 (2%) | 58 (3%) | 0.562 |
| Dopamine parenteral infusion | 75 (3%) | 120 (2%) | 0.008 | 57 (3%) | 58 (3%) | 0.925 |
| Nesiritide parenteral infusion | 166 (7%) | 494 (9%) | 0.006 | 154 (7%) | 168 (8%) | 0.418 |
| Mechanical ventilation | 68 (3%) | 130 (2%) | 0.161 | 63 (3%) | 62 (3%) | 0.928 |
| Discharge medications | ||||||
| ACE inhibitors or ARBs | 1241 (52%) | 3781 (68%) | <0.001 | 1232 (56%) | 1234 (56%) | 0.951 |
| Beta blockers | 1369 (58%) | 3662 (66%) | <0.001 | 1329 (61%) | 1326 (61%) | 0.926 |
| Aldosterone antagonists | 147 (6%) | 604 (11%) | <0.001 | 145 (7%) | 147 (7%) | 0.904 |
| Digoxin | 483 (20%) | 1418 (25%) | <0.001 | 471 (21%) | 450 (21%) | 0.436 |
| Amlodipine | 175 (7%) | 398 (7%) | 0.703 | 167 (8%) | 163 (7%) | 0.819 |
| Anti-arrhythmic drugs | 252 (11%) | 659 (12%) | 0.127 | 242 (11%) | 237 (11%) | 0.809 |
| Warfarin | 456 (19%) | 1347 (24%) | <0.001 | 451 (21%) | 448 (20%) | 0.911 |
| Aspirin | 1069 (45%) | 2838 (51%) | <0.001 | 1031 (47%) | 1064 (49%) | 0.318 |
| Discharge instructions | ||||||
| Diet | 1909 (81%) | 4928 (89%) | <0.001 | 1850 (84%) | 1826 (83%) | 0.324 |
| Medications | 2005 (85%) | 5182 (93%) | <0.001 | 1952 (89%) | 1927 (88%) | 0.236 |
| Worsening symptoms | 1442 (61%) | 3729 (67%) | <0.001 | 1391 (63%) | 1377 (63%) | 0.661 |
| Weight monitoring | 1007 (43%) | 2994 (54%) | <0.001 | 986 (45%) | 994 (45%) | 0.808 |
| Follow-up | 1979 (84%) | 5044 (91%) | <0.001 | 1916 (87%) | 1913 (87%) | 0.891 |
| Hospital length of stay, days | 4 (2–7) | 4 (3–7) | 0.243 | 4 (2–7) | 4 (3–7) | 0.762 |
| Hospital academic center | 977 (41%) | 2353 (42%) | 0.408 | 905 (41%) | 915 (42%) | 0.759 |
Values are mean ±SD, n (%), or median (interquartile range). The p values comparing medians are based on nonparametric independent sample median test. ACE = Angiotensin Converting Enzyme; ARB = Angiotensin Receptor Blocker; BNP = B-type Natriuretic Peptide; COPD = Chronic Obstructive Pulmonary Disease; HF = Heart Failure; JVP = Jugular Venous Pressure; LVEF = Left Ventricular Ejection Fraction
30-Day All-Cause Mortality
Among the 2,191 pairs of propensity score-matched patients who were not taking diuretics prior to hospitalization for HF decompensation, 30-day all-cause mortality occurred in 4.9% (107/2,191) and 6.6% (144/2,191) of the patients receiving and not receiving a discharge prescription for loop diuretics, respectively (HR when receipt of a discharge prescription for loop diuretics was compared with its non-receipt, 0.73; 95% CI, 0.57–0.94; p=0.016; Table 2, Central Illustration). Findings of the formal sensitivity analysis are presented in Table 2 (footnote). The association between discharge prescription for loop diuretics and mortality attenuated during 60 days of follow-up and lost statistical significance (Table 2, Central Illustration).
Table 2.
Outcomes by Discharge Prescription for Loop Diuretics in 4,382 Propensity Score-Matched Older Patients with Heart Failure Not Taking Diuretics Prior to Hospitalization for Heart Failure Decompensation
| Outcomes by event type and follow-up duration | Events (%), by discharge prescription of loop diuretics | Hazard ratio associated with initiation of loop diuretics (95% confidence interval); p value | |
|---|---|---|---|
| No (n=2,191) | Yes (n=2,191) | ||
| 30 days | |||
| All-cause mortality* | 144 (6.6%) | 107 (4.9%) | 0.73 (0.57–0.94); p=0.016 |
| Heart failure readmission† | 168 (7.7%) | 135 (6.2%) | 0.79 (0.63–0.99); p=0.037 |
| All-cause readmission | 509 (23%) | 468 (21%) | 0.89 (0.79–1.01); p=0.081 |
| HF readmission or all-cause mortality‡ | 299 (14%) | 233 (11%) | 0.76 (0.64–0.91); p=0.002 |
| All-cause readmission or all-cause mortality§ | 604 (28%) | 530 (24%) | 0.85 (0.76–0.96); p=0.008 |
| 60 days | |||
| All-cause mortality | 232 (11%) | 201 (9%) | 0.86 (0.71–1.03); p=0.103 |
| Heart failure readmission | 251 (12%) | 236 (11%) | 0.92 (0.77–1.09); p=0.334 |
| All-cause readmission | 712 (33%) | 693 (32%) | 0.94 (0.85–1.05); p=0.267 |
| HF readmission or all-cause mortality | 455 (21%) | 410 (19%) | 0.88 (0.77–1.00); p=0.057 |
| All-cause readmission or all-cause mortality | 827 (38%) | 782 (36%) | 0.92 (0.83–1.01); p=0.080 |
Because formal sensitivity analyses can only be conducted when associations are significant in the matched cohort, only the results for significant 30-day associations are presented below:
For 30-day all-cause mortality, in 11% (242/2,191) of matched pairs we were able to determine which patients within a pair clearly had longer 30-day survival, and in 58% (140/242) of those pairs, these patients belonged to the loop diuretic group (p=0.015). This significant association could be explained away by a hidden covariate that is a near-perfect predictor of 30-day mortality, if it increased the odds of a discharge prescription for loop diuretics by 6.4%.
For 30-day HF readmission, in 13% (279/2,191) of matched pairs we were able to determine which patients within a pair clearly had a longer 30-day event-free survival, and in 56% (157/279) of the pairs, these patients belonged to the loop diuretics group (p=0.036). An unmeasured confounder that is a near-perfect predictor of 30-day HF readmission could explain away this association if it also increased the odds of a discharge prescription for loop diuretic by 1.6%.
For the combined endpoint of 30-day HF readmission or death, in 23% (496/2,191) of matched pairs we were able to determine which patients within a pair clearly had a longer 30-day event-free survival, and in 57% (283/ 496) of the pairs, these patients belonged to the loop diuretics group (p=0.002). An unmeasured confounder that is a near-perfect predictor of 30-day HF readmission could explain away this association if it also increased the odds of a discharge prescription for loop diuretic by 11.3%.
For the combined endpoint of 30-day total readmission or death, in 45% (976/2,191) of matched pairs we were able to determine which patients within a pair clearly had a longer 30-day event-free survival, and in 54% (522/ 976) of the pairs, these patients belonged to the loop diuretics group (p=0.030). An unmeasured confounder that is a near-perfect predictor of 30-day HF readmission could explain away this association if it also increased the odds of a discharge prescription for loop diuretic by 1.4%.
Central Illustration: Kaplan-Meier Plots by Loop Diuretic Prescription.
This study assessed the relationship of prescription of loop diuretics at the time of hospital discharge with all-cause mortality (top panel) and heart failure readmission (bottom panel) in 2191 pairs of propensity score-matched older patients hospitalized for heart failure decompensation who were not taking diuretics prior to hospitalization. During the first 30 days of follow-up after hospital discharge, a discharge prescription for loop diuretics was associated with a significantly lower risk of both outcomes. Both associations lost statistical significance during 60 days of follow-up. CI = confidence interval; HR = hazard ratio.
30-day all-cause mortality occurred in 5.7% (62/1,084) and 6.6% (78/1,075) of the patients with HFrEF receiving and not receiving loop diuretics, respectively (HR, 0.78; 95% CI, 0.56–1.09; P=0.147), and 4.1% (45/1,107) and 5.9% (66/1,116) of the patients with HFpEF receiving and not receiving loop diuretics, respectively (HR, 0.68; 95% CI, 0.46–0.99; p=0.043; p for interaction=0.582; data not presented). The associations were not different when EF was used as a continuous variable (p for interaction=0.889).
30-Day Readmissions
Among the 2,191 pairs of propensity score-matched patients, 30-day HF readmission occurred in 6.2% (135/2,191) and 7.7% (168/2,191) of the patients in the loop diuretic and no loop diuretic groups, respectively (HR associated with loop diuretic prescription, 0.79; 95% CI, 0.63–0.99; p=0.037; Table 2, Central Illustration). Findings of the formal sensitivity analysis are presented in Table 2 (footnote). Loop diuretic prescription had no significant association with 30-day non-HF readmission (HR, 0.84; 95% CI, 0.63–1.13; p=0.729; data not presented in Table 2) and 30-day all-cause readmission (HR, 0.89; 95% CI, 0.79–1.01; p=0.081; Table 2). The association of loop diuretic prescription at discharge had no significant association with HF or all-cause readmissions during 60 days of follow-up (Table 2).
30-day HF readmission occurred in 6.4% (69/1,084) and 8.2% (88/1,075) of HFrEF patients receiving and not receiving loop diuretics, respectively (HR, 0.76; 95% CI, 0.56–1.04; p=0.090), and 6.0% (66/1,107) and 7.2% (80/1,116) of HFpEF patients receiving and not receiving loop diuretics, respectively (HR, 0.81; 95% CI, 0.59–1.12; p=0.210; data not presented). The associations were not different between HFrEF versus HFpEF (p for interaction=0.937) or when EF was used as a continuous variable (p for interaction=0.909).
30-Day Combined Endpoints
The combined endpoint of 30-day HF readmission or all-cause mortality occurred in 11% (233/2,191) and 14% (299/2,191) of the patients receiving and not receiving loop diuretics, respectively (HR associated with loop diuretic prescription, 0.76; 95% CI, 0.64–0.91; p=0.002; Table 2, Figure 2). There was also an associated lower risk for the combined endpoint of 30-day all-cause readmission or all-cause mortality (Table 2). Findings of the formal sensitivity analysis are presented in Table 2 (footnote). Associations with both combined endpoints were attenuated during 60 days of follow-up and lost statistical significance (Table 2).
Figure 2: Forest Plots for Subgroup Analyses by Loop Diuretic Initiation.
In all subgroups analyzed, older patients hospitalized for heart failure decompensation who were not taking diuretics prior to hospitalization, those who received a prescription for loop diuretics at discharge had a lower risk of the combined endpoint of heart failure readmission or all-cause mortality during the first 30 days of follow-up after hospital discharge compared with patients who did not receive a loop diuretic prescription, except by history of prior heart failure, hypertension, diabetes mellitus admission pulmonary rales, and admission lower extremity edema. Note: Results of subgroup analyses need to be interpreted with caution as they may be false positive due to multiple comparisons and false negative due to inadequate power. ACE = angiotensin-converting enzyme, ARB = angiotensin receptor blocker, CI = confidence interval.
Among patients with HFrEF, the combined endpoint of 30-day HF readmission or all-cause mortality occurred in 11.6% (126/1,084) and 15.1% (162/1,075) of the patients in the loop diuretic and no loop diuretic groups, respectively (HR, 0.76; 95% CI, 0.60–0.95; p=0.019; Figure 2). Among patients with HFpEF, these events occurred in 9.7% (107 /1,107) and 12.3% (137/1,116) of the patients in the loop diuretic and no loop diuretic groups, respectively (HR, 0.77; 95% CI, 0.60–0.99; p=0.042; p for interaction=0.956; Figure 2).
Subgroup Analyses
Findings from other subgroup analyses of the combined endpoint of 30-day HF readmission or all-cause mortality are displayed in Figure 2. The association of discharge prescription for loop diuretics with the combined endpoint of 30-day HF readmission or all-cause mortality was generally homogeneous in other clinically relevant subgroups except that it was significantly stronger in subgroups with admission pulmonary rales and lower extremity edema (Figure 2). HRs (95% CIs) for the combined endpoint in subgroups with and without pulmonary rales were 0.62 (0.50–0.77; p<0.001) and 1.08 (0.81–1.43; p=0.604), respectively (p for interaction=0.003; Figure 2). This association was also significantly different between subgroups with and without lower extremity edema (p for interaction=0.001; Figure 2).
Discussion
Findings from the current study demonstrate that a loop diuretic prescription at discharge was associated with a significant albeit modest reduction in the risk of 30-day HF readmission in older patients hospitalized for HF decompensation who were not taking diuretics prior to hospitalization (Central Illustration). A loop diuretic prescription was also associated with a lower risk of 30-day all-cause mortality as well as of the combined endpoint of 30-day HF readmission or all-cause mortality. We also observed that these associations were homogeneous in subgroups with reduced and preserved EF. All associations attenuated during 60 days of follow-up and were no longer statistically significant (Central Illustration). To the best of our knowledge, this is the first report on the association between loop diuretic prescription and improved 30-day clinical outcomes in patients with HFrEF and HFpEF. These results suggest that the clinical benefits associated with loop diuretic use in patients with HF may extend beyond mere symptom alleviation to improved clinical outcomes.
Loop diuretics increase urinary sodium excretion in the loop of Henle by inhibiting the sodium-potassium-chloride cotransporter 2 (22). Findings from small RCTs with short follow-up suggest that loop diuretics may improve signs and symptoms of fluid retention in patients with HF (23–26). These findings may in part explain the lower risk of 30-day HF readmission observed in our study. This is supported by the findings from our subgroup analysis that suggest diuretic-associated clinical benefits were greater in subgroups with evidence of congestion such as pulmonary rales and lower extremity edema. This is also supported by our observation that the use of loops diuretics was not associated with a lower risk of non-HF related readmissions. The lower risk of 30-day all-cause mortality in the diuretic group is intriguing but may be mediated by improved HF symptoms and lower HF readmission risk. Continued congestion and hospitalization after discharge have been shown to be associated with a higher risk of death in patients with HF (27–30).
Several observations from our study suggest that the associations observed in our study may be an underestimation of the true associations of outcomes with loop diuretics. Because patients in our study were hospitalized for decompensated HF, presumably all were initiated on loop diuretics during hospitalization, which likely attenuated between-group differences in congestive symptoms and signs before discharge. Patients in the loop diuretics group also had a higher burden of congestion before admission (Table 1). Even though these and other measured baseline characteristics were balanced after matching, residual confounding and unmeasured confounding may have further attenuated the true associations. Finally, if some patients were restarted on loop diuretics due to congestion during the first 30 days, then the resultant misclassification would even further dilute the true associations. If more patients were restarted on diuretics during the second month that may explain the loss of significance of the 60-day associations. Taken together, the inpatient use of loop diuretics in all patients, the post-discharge resumption of loop diuretics in the no-diuretic group, and the potential residual/unmeasured confounding by a higher disease/symptom burden of the loop diuretic group suggest that the actual associations of loop diuretics with 30-day outcomes may be even greater than those detected in our study.
Information provided by the current study has practical implications for clinicians involved in HF care. Despite the general impression that most clinicians would use loop diuretics to relieve symptoms in nearly all patients with HF, findings from our study suggest that many patients hospitalized for HF decompensation were not receiving diuretics before hospitalization. Furthermore, a substantial portion of these patients was discharged without a loop diuretic prescription. A potential explanation for this is that HF symptoms of these patients appeared resolved, and loop diuretics were not considered necessary as these drugs are currently recommended only for improving symptoms (1). HF remains a leading cause for 30-day hospital readmission for older adults. The new message from our study is that prescription of loop diuretics at discharge may be associated with a lower risk of short-term rehospitalizations and mortality in these patients. These findings are expected to clarify the role of loop diuretics in HF and strengthen the evidence for the guideline recommendation on loop diuretics.
Limitations
Several limitations of our study need to be acknowledged. Even though patients receiving and not receiving a discharge prescription for loop diuretics were balanced on 74 measured characteristics at the time of the prescription (at study baseline) it is possible that as discussed above, observed significant associations are underestimated by residual and unmeasured confounding. The findings of our sensitivity analyses suggest that significant 30-day associations observed in our study may be sensitive to an unmeasured confounder. However, sensitivity analysis cannot determine whether an unmeasured confounder exists or not. Further, such a confounder would need to be a near-perfect predictor of the outcomes of the significant 30-day associations and also not have strong associations with any of the 74 measured baseline characteristics used in our study. We did not have access to data on loop diuretic doses or start/ re-start/ discontinuation after hospital discharge. Prior studies have suggested frequent adjustment of diuretics after hospital discharge (1). Like other guideline-recommended therapies, it is not only their use that may be important but also attention to adequate dosing at discharge, titration after discharge, and close monitoring. Although the management of HF has evolved in the past several decades, the role of loop diuretics has not. Our study is based on fee-for-service Medicare beneficiaries, which may limit generalizability. Findings from the current study may not be generalized to patients with renal failure requiring dialysis.
Conclusions
Among older patients hospitalized for decompensated heart failure who were not taking diuretics before hospitalization, a loop diuretic prescription at discharge is associated with a significantly lower risk of 30-day all-cause mortality and heart failure readmission. These findings provide new information that may strengthen guideline recommendations and improve short-term clinical outcomes in patients with heart failure.
Supplementary Material
Perspectives.
Competency in Patient Care: For older patients hospitalized with decompensated heart failure not taking diuretics prior to admission, prescription of a loop diuretic at discharge is associated with a lower 30-day risk of re-admission and mortality.
Translational Outlook: Future studies should examine the relationship of loop diuretic dosage at discharge and titration after discharge with longer-term clinical outcomes.
Funding:
Dr. Ahmed was in part supported by grants from the National Heart, Lung, and Blood Institute (NHLBI) (R01-HL085561, R01-HL085561-S and R01-HL097047) and the Department of Veterans Affairs Office of Research and Development (IIR 17-095). OPTIMIZE-HF was sponsored by GlaxoSmithKline, but played no role in the design, conduct, analyses or interpretation of the current study.
Abbreviations
- EF
ejection fraction
- HF
heart failure
- HFpEF
heart failure with preserved ejection fraction
- HFrEF
heart failure with reduced ejection fraction
- OPTIMIZE-HF
Organized Program to Initiate Lifesaving Treatment in Hospitalized Patients with Heart Failure
- RCT
randomized controlled trial
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 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: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the Department of Veterans Affairs. Dr. Fonarow reports consulting with Abbott, Amgen, Astra Zeneca, Bayer, CHF Solutions, Janssen, Medtronic, Merck, and Novartis, and was the Principal Investigator of OPTIMIZE-HF. Dr Filippatos participated in committees of trials and registries sponsored by Medtronic, Vifor, Boehringer Ingelheim, and Servier, and has received research grants from the European Union. None of the other authors report any conflicts of interest related to this manuscript.
References
- 1.Yancy CW, Jessup M, Bozkurt B et al. 2013 ACCF/AHA guideline for the management of heart failure: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2013;62:e147–239. [DOI] [PubMed] [Google Scholar]
- 2.Felker GM, Ellison DH, Mullens W, Cox ZL, Testani JM. Diuretic Therapy for Patients With Heart Failure: JACC State-of-the-Art Review. J Am Coll Cardiol 2020;75:1178–1195. [DOI] [PubMed] [Google Scholar]
- 3.Fonarow GC. Comparative effectiveness of diuretic regimens. N Engl J Med 2011;364:877–8. [DOI] [PubMed] [Google Scholar]
- 4.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;148:43–51. [DOI] [PubMed] [Google Scholar]
- 5.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;50:768–77. [DOI] [PubMed] [Google Scholar]
- 6.Malik A, Masson R, Singh S et al. Digoxin Discontinuation and Outcomes in Patients With Heart Failure With Reduced Ejection Fraction. J Am Coll Cardiol 2019;74:617–627. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Arundel C, Lam PH, Gill GS et al. Systolic Blood Pressure and Outcomes in Patients With Heart Failure With Reduced Ejection Fraction. J Am Coll Cardiol 2019;73:3054–3063. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Lam PH, Dooley DJ, Deedwania P et al. Heart Rate and Outcomes in Hospitalized Patients With Heart Failure With Preserved Ejection Fraction. J Am Coll Cardiol 2017;70:1861–1871. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.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;166:230–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Danaei G, Tavakkoli M, Hernan MA. Bias in observational studies of prevalent users: lessons for comparative effectiveness research from a meta-analysis of statins. Am J Epidemiol 2012;175:250–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Ray WA. Evaluating medication effects outside of clinical trials: new-user designs. Am J Epidemiol 2003;158:915–20. [DOI] [PubMed] [Google Scholar]
- 12.Jentzer JC, DeWald TA, Hernandez AF. Combination of loop diuretics with thiazide-type diuretics in heart failure. J Am Coll Cardiol 2010;56:1527–34. [DOI] [PubMed] [Google Scholar]
- 13.Rosenbaum PR, Rubin DB. The central role of propensity score in observational studies for causal effects. Biometrika 1983;70:41–55. [Google Scholar]
- 14.Rubin DB. Using propensity score to help design observational studies: Application to the tobacco litigation. Health Services and Outcomes Research Methodology 2001;2:169–188. [Google Scholar]
- 15.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;27:1431–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Ahmed A, Rich MW, Love TE et al. Digoxin and reduction in mortality and hospitalization in heart failure: a comprehensive post hoc analysis of the DIG trial. Eur Heart J 2006;27:178–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Austin PC. An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies. Multivariate Behav Res 2011;46:399–424. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Rubin DB. On principles for modeling propensity scores in medical research. Pharmacoepidemiol Drug Saf 2004;13:855–7. [DOI] [PubMed] [Google Scholar]
- 19.Austin PC, Stuart EA. Estimating the effect of treatment on binary outcomes using full matching on the propensity score. Stat Methods Med Res 2017;26:2505–2525. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.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;30:2029–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Rosenbaum PR. Sensitivity to hidden bias. In: Rosenbaum PR, editor Observational Studies. New York: Springer-Verlag, 2002:105–170. [Google Scholar]
- 22.Ellison DH, Felker GM. Diuretic Treatment in Heart Failure. N Engl J Med 2017;377:1964–1975. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Sherman LG, Liang CS, Baumgardner S, Charuzi Y, Chardo F, Kim CS. Piretanide, a potent diuretic with potassium-sparing properties, for the treatment of congestive heart failure. Clin Pharmacol Ther 1986;40:587–94. [DOI] [PubMed] [Google Scholar]
- 24.Patterson JH, Adams KF Jr., Applefeld MM, Corder CN, Masse BR. Oral torsemide in patients with chronic congestive heart failure: effects on body weight, edema, and electrolyte excretion. Torsemide Investigators Group. Pharmacotherapy 1994;14:514–21. [PubMed] [Google Scholar]
- 25.Wilson JR, Reichek N, Dunkman WB, Goldberg S. Effect of diuresis on the performance of the failing left ventricle in man. Am J Med 1981;70:234–9. [DOI] [PubMed] [Google Scholar]
- 26.Parker JO. The effects of oral ibopamine in patients with mild heart failure--a double blind placebo controlled comparison to furosemide. The Ibopamine Study Group. Int J Cardiol 1993;40:221–7. [DOI] [PubMed] [Google Scholar]
- 27.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;14:211–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Malik A, Gill GS, Lodhi FK et al. Prior Heart Failure Hospitalization and Outcomes in Patients with Heart Failure with Preserved and Reduced Ejection Fraction. Am J Med 2020;133:84–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Cooper LB, Lippmann SJ, DiBello JR et al. The Burden of Congestion in Patients Hospitalized With Acute Decompensated Heart Failure. Am J Cardiol 2019;124:545–553. [DOI] [PubMed] [Google Scholar]
- 30.Ambrosy AP, Pang PS, Khan S et al. Clinical course and predictive value of congestion during hospitalization in patients admitted for worsening signs and symptoms of heart failure with reduced ejection fraction: findings from the EVEREST trial. Eur Heart J 2013;34:835–43. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.