Diuretic dose and long-term outcomes in elderly patients with heart failure after hospitalization
Article Outline
- Abstract
- Background
- Methodology
- Results
- Discussion
- Disclosures
- Appendix A. Coding of diagnoses using ICD coding system
- Appendix B. Online Table A—Sensitivity analysis of adjusted risk of death relative to cumulative furosemide exposure
- Appendix C. Online Table B—Reclassification based on furosemide dose at end of year 1 for sensitivity analysis for death
- References
- Copyright
Background
The array of outcomes according to longitudinal furosemide doses in heart failure (HF) have not been evaluated. We examined the relationship of dynamic furosemide dose with mortality and hospitalizations for cardiovascular disease and renal dysfunction.
Methods
Among elderly patients with HF (≥65 years) newly discharged from hospital, dynamic furosemide exposure was determined by examining dose fluctuations up to 5 years of follow-up using the Ontario Drug Benefit pharmacare database. Dynamic furosemide exposures were classified as low dose (LD; 1-59 mg/d), medium dose (MD; 60-119 mg/d), or high dose (HD; ≥120 mg/d). Outcomes were assessed by modeling furosemide exposure as a time-dependent covariate.
Results
Among 4,406 patients (78.4 ± 7.0 years; 50.5% male), 46% changed furosemide dose categories within 1 year, and 63% changed dose categories over the follow-up period. High-dose furosemide patients were younger, were mostly male, and exhibited more ischemic or valvular disease, diabetes, atrial fibrillation, hypotension, hyponatremia, and higher baseline creatinine than LD. Compared with LD, MD exposure was associated with increased mortality with adjusted hazard ratio 1.96 (95% CI 1.79-2.15), whereas HD exposure conferred greater mortality risk with hazard ratio 3.00 (95% CI 2.72-3.31) after multiple covariate adjustment (both P < .001). Adjusted risks of hospitalization for HF (MD: 1.24 [95% CI 1.12-1.38] and HD: 1.43 [95% CI 1.26-1.63]), renal dysfunction (MD: 1.56 [95% CI 1.38-1.76] and HD: 2.16 [95% CI 1.88-2.49]), and arrhythmias (MD: 1.15 [95% CI 1.03-1.30] and HD: 1.45 [95% CI 1.27-1.66]) were also higher with increasing furosemide exposure.
Conclusion
Exposure to higher furosemide doses is associated with worsened outcomes and is broadly predictive of death and morbidity.
Background
Heart failure (HF) is a leading cause of hospitalization and mortality.1 The use of diuretics is central to HF management, particularly during episodes of acute decompensation, with more than 80% of patients receiving a loop diuretic.2 However, the use of loop diuretics in HF has come under recent scrutiny,3, 4, 5, 6 with studies of static drug exposures suggesting an increased mortality risk with their use.7, 8
There is a paucity of data on patterns of diuretic use in the nonacute setting, and the prognostic effects of diuretic exposure after accounting for temporal changes in drug dosages are unknown. After discharge from hospital, the dose of loop diuretics may change, but the magnitude and impact of these fluctuations have not been previously reported. Furthermore, although there may be morbidity associated with loop diuretic use, including renal compromise,9, 10, 11 fractures,12, 13 and sudden death,14 these risks have not been quantified in population-based studies of HF.
In this study, we evaluated the association of “dynamic furosemide dose” with the risk of mortality and recurrent hospitalization in a population-based cohort of patients who were discharged from HF hospitalization. Our secondary objective was to explore the association of dynamic furosemide dose with the risk of the above-mentioned adverse events. We hypothesized that increased dynamic doses would be independently associated with a higher risk of death and morbid adverse outcomes.
Methodology
Patients
We examined patients with HF in the Enhanced Feedback for Effective Cardiac Treatment (EFFECT) Study, which has been described previously.15 Briefly, a clinical chart abstraction study was performed in patients newly hospitalized for HF at 1 of the 86 hospital corporations in Ontario, Canada, between April 1, 1999, and March 30, 2001, if they met the Framingham HF criteria.16 Newly hospitalized patients were defined as those with no HF hospitalizations in the years before index hospitalization.17 All study patients had a diagnosis of HF coded in the Canadian Institute of Health Information discharge abstract database. Using the Ontario Drug Benefit (ODB) program database, we examined drug use in patients ≥65 years of age who were discharged home.18 Patients discharged without furosemide and those who did not fill furosemide prescriptions during follow-up were excluded. To ensure that patients were prescribed loop diuretics for HF, patients with dialysis dependence, nephrotic syndrome, and cirrhosis were excluded. Institutional research ethics board approval was obtained from all participating institutions.
Baseline data collection
As part of the EFFECT program, details of the patients' index admission were extracted from chart records by highly trained cardiology nurse abstractors. This included demographic characteristics, preadmission medical history, presenting symptoms, signs, investigations, and prescribed medications preadmission, in-hospital, and at discharge. Extremes of diuretic dosage (>160 mg) and aberrant dosages were verified by 2 investigators (H.M.A., D.S.L.) by comparing with abstracted free-text records.
Outcomes
There were 4 coprimary outcomes: all-cause mortality, hospitalization for all causes, HF-related hospitalizations, and cardiovascular (CV) hospitalization. Secondary outcomes were the occurrence of (a) renal dysfunction, (b) arrhythmias, and (c) fractures, as noted in subsequent hospitalizations. Outcome data were obtained by linking the clinical data described previously with administrative databases using the unique, encrypted health card number. Mortality data were obtained from the Registered Persons Database, which contains information on vital status and demographic characteristics on all residents of Ontario. Subsequent hospitalizations after the index HF discharge and the documented diagnoses were identified in the Canadian Institute of Health Information discharge abstract database using the International Classification of Diseases 9th (ICD-9-CM) or 10th (ICD-10-CA) revision systems (see Appendix A, online). Diagnoses for renal dysfunction included acute kidney injury, chronic kidney disease, dialysis initiation, electrolyte abnormalities, and complications related to renal dysfunction. Diagnoses for arrhythmias included the composite of sudden death, atrial, and ventricular arrhythmias.
Dynamic furosemide dose
To determine the dynamic furosemide dose, we searched the ODB database for all ambulatory prescription claims beginning from the time of discharge up to 5 years of follow-up or until death. We examined all furosemide prescriptions at intervals up to 90 days, which corresponded to the maximum prescribable period of drugs before refills are required. For each furosemide prescription, the duration of exposure was determined by the lesser of duration of the supplied prescription or the interval between 2 furosemide prescriptions. The total daily dose was determined by multiplying the dose per tablet with the amount supplied, divided by the prescription duration. After consultation with cardiologists specializing in HF, we decided a priori that furosemide doses would be categorized into the following 3 groups: low dose (LD, 1-59 mg/d), medium dose (MD, 60-119 mg/d), or high dose (HD, ≥120 mg/d). We updated the dose category (eg, low, medium, or high) variable for each interval as a time-dependent covariate to determine the dynamic furosemide dose. The proportion of time that patients were at LD, MD, or HD during the study period was also calculated for each patient.
Statistical analysis
Continuous variables are presented as mean (SD) and were compared using analysis of variance. Categorical variables were compared using the χ2 statistic. In separate longitudinal analyses, furosemide exposure was analyzed as both a static variable (based on discharge dose) and a time-varying covariate (based on ODB prescription claims), and effects were determined relative to LD. Time-to-event analyses were performed using the Kaplan-Meier method. Multiple Cox regression analyses were used to assess for the independent relationship between furosemide dose and outcomes. We adjusted for the following covariates in all models: age, sex, ischemic heart disease (IHD), prior myocardial infarction (MI), aortic/mitral valve disease, hypertension, diabetes, atrial fibrillation, cancer, cerebrovascular disease, chronic obstructive lung disease, dementia, smoking, New York Heart Association (NYHA) class, left ventricular ejection fraction (LVEF), hypo/normo/hypernatremia, hypo/normo/hyperkalemia, creatinine, urea, cardiomegaly, heart rate, respiratory rate, preadmission use of furosemide, presenting and discharge systolic blood pressure, and discharge medications including angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, β-adrenoreceptor antagonists, spironolactone, digoxin, potassium supplements, metolazone, and other diuretics.
All analyses were performed using SAS software, version 9.1 (SAS Institute Inc, Cary, NC).
Statistical significance was defined by a 2-sided P < .05. Data cells involving ≤5 patients were censored from the results in accordance with privacy regulations.
The Institute for Clinical Evaluative Sciences is supported in part by a grant from the Ontario Ministry of Health and Long Term Care. This study was funded by a Canadian Institutes of Health Research (CIHR) grant (MOP 86718) and a CIHR Team Grant in Cardiovascular Outcomes Research. The study was supported by a CIHR clinician-scientist salary award (D.S.L.), career investigator awards from the Heart and Stroke Foundation of Ontario (P.C.A., J.V.T.), and a Canada Research Chair in health services research (J.V.T.). The authors are solely responsible for the design and conduct of this study, all study analyses, the drafting and editing of the paper and its final contents. None of the authors have any conflicts of interest to declare.
Results
Patient characteristics
Of 8,594 patients discharged alive from the hospital, 4,324 were excluded, leaving 4,270 patients for analyses of dynamic furosemide dose (Figure 1). Baseline characteristics according to discharge furosemide dose categories are shown in Table I. The HD group had more symptom-directed therapy with high rates of metolazone and digoxin use but had lower use of evidence-guided therapy with lower utilization of angiotensin-converting enzyme inhibitors (ACEI) and β-adrenoreceptor antagonists.
Figure 1.
Patients included and excluded from the analysis.
Table I. Baseline characteristics
| Characteristic | LD | MD | HD | P |
|---|---|---|---|---|
| n = 2299 | n = 1407 | n = 700 | ||
| Age, mean (SD) | 78.7 ± 7.1 | 78.4 ± 6.7 | 77.3 ± 6.9 | <.001 |
| Male, n (%) | 1110 (48.3) | 722 (51.3) | 391 (55.9) | .002 |
| IHD, n (%) | 1172 (51.0) | 876 (62.3) | 473 (67.6) | <.001 |
| Previous MI, n (%) | 811 (35.3) | 598 (42.5) | 320 (45.7) | <.001 |
| Aortic/Mitral valve disease, n (%) | 337 (14.7) | 288 (20.5) | 149 (21.3) | <.001 |
| Diabetes mellitus, n (%) | 638 (27.8) | 523 (37.2) | 328 (46.9) | <.001 |
| Atrial fibrillation, n (%) | 682 (29.7) | 432 (30.7) | 303 (43.3) | <.001 |
| COPD, n (%) | 312 (13.6) | 259 (18.4) | 144 (20.6) | <.001 |
| ACEI/ARB, n (%) | 1651 (71.8) | 1034 (73.5) | 469 (67.0) | .007 |
| β-Blockers, n (%) | 709 (30.8) | 394 (28.0) | 182 (26.0) | .024 |
| Digoxin, n (%) | 867 (37.7) | 661 (47.0) | 392 (56.0) | <.001 |
| Metolazone, n (%) | 26 (1.1) | 62 (4.4) | 70 (10.0) | <.001 |
| Spironolactone, n (%) | 339 (14.7) | 292 (20.8) | 197 (28.1) | <.001 |
| Other diuretics, n (%) | 59 (2.6) | 31 (2.2) | 30 (4.3) | .017 |
| Potassium supplements, n (%) | 614 (26.7) | 457 (32.5) | 279 (39.9) | <.001 |
| Admission SBP (mm Hg) | ||||
 <90, n (%) | 21 (0.9) | 27 (1.9) | 10 (1.4) | .032 |
| 90-139, n (%) | 731 (31.8) | 525 (37.3) | 320 (45.7) | <.001 |
| ≥140, n (%) | 1529 (66.5) | 847 (60.2) | 367 (52.4) | <.001 |
| Discharge SBP (mm Hg) | ||||
| <90, n (%) | 36 (1.6) | 33 (2.3) | 28 (4.0) | <.001 |
| 90-139, n (%) | 1492 (64.9) | 972 (69.1) | 485 (69.3) | .011 |
| ≥140, n (%) | 637 (27.7) | 322 (22.9) | 152 (21.7) | <.001 |
| Tachycardia, n (%) | 1079 (38.6) | 588 (33.7) | 268 (28.7) | <.001 |
| Tachypnea, n (%) | 1816 (65.0) | 1187 (68.0) | 597 (63.9) | .048 |
| Hyponatremia, n (%) | 391 (17.0) | 265 (18.8) | 168 (24.0) | <.001 |
| Hyperkalemia, n (%) | 220 (7.9) | 164 (9.4) | 96 (10.3) | .043 |
| Serum hemoglobin, g/dL (SD) | 12.62 ± 1.93 | 12.38 ± 2.03 | 12.19 ± 1.96 | <.001 |
| Serum creatinine, mg/dL (SD) | 1.28 ± 0.67 | 1.43 ± 0.69 | 1.63 ± 0.94 | <.001 |
| Serum creatinine, μmol/L (SD) | 113.0 ± 58.9 | 126.1 ± 61.0 | 144.3 ± 83.5 | <.001 |
| Serum urea, mg/dL (SD) | 24.3 ± 12.6 | 28.9 ± 16.9 | 35.0 ± 21.8 | <.001 |
| Serum urea, mmol/L (SD) | 8.67 ± 4.49 | 10.32 ± 6.02 | 12.51 ± 7.79 | <.001 |
| Cardiomegaly, n (%) | 987 (35.3) | 669 (38.3) | 367 (39.3) | .035 |
| LVEF ≤ 40%, n (%) | 614 (22.0) | 360 (20.6) | 174 (18.6) | .083 |
| LVEF >40%, n (%) | 507 (18.1) | 262 (15.0) | 109 (11.7) | <.001 |
| LVEF unknown, n (%) | 1673 (59.9) | 1124 (64.4) | 652 (69.7) | <.001 |
| Preadmission furosemide use, n (%) | 859 (37.4) | 951 (67.6) | 576 (82.3) | <.001 |
| Mean furosemide dose, mg (SD) | 35.4 ± 8.5 | 77.4 ± 7.8 | 160.3 ± 55.7 | <.001 |
Dynamic furosemide doses after discharge
Figure 2 displays dynamic furosemide doses over the follow-up period. Figure 2, A shows furosemide exposures for patients initially discharged from the hospital on LD. Figure 2, B shows furosemide exposures over time for patients initially discharged from the hospital on MD, whereas Figure 2, C shows furosemide exposures over time for patients discharged initially on HD. Some patients were exposed only to LD (“low only”), MD (“mid only”), or HD (“high only”) furosemide throughout follow-up, whereas some patients never received LD (“never low”), MD (“never mid”), or HD (“never high”) furosemide at all. Patients exposed to combinations of furosemide categories are shown separately (eg, “low + mid” indicates patients exposed to LD and MD). Similar labels are used to indicate other combinations of dose exposures. Among patients discharged on any of the 3 initial dose categories, there were changes in drug dosages over time. Furosemide dose categories changed in 46% of patients in the first year, and 63% changed dose categories over the entire follow-up period. Patients discharged with LD furosemide had 5,270 person-years of follow-up available, and only 78% of these person-years were dynamically at LD throughout follow-up. Similarly, of the 3,017 and 1,754 person-years of follow-up available in those discharged at MD or HD, only 51% and 41% of the total person-years were dynamically at MD or HD throughout follow-up, respectively.
Figure 2.
A-C, Furosemide dose exposures as per outpatient pharmacare claims during the 5-year follow-up of patients stratified according to furosemide dose at discharge.
Mortality
A total of 12,751 person-years of follow-up were examined in survival analysis. During this time, 554 patients discharged on HD furosemide (79%) died, compared to 966 (69%) and 1,421 (62%) of patients discharged on MD and LD furosemide, respectively. Median survival was 1.9, 2.8, and 3.6 years respectively, and the corresponding adjusted survival curves stratified by the furosemide dose category are shown in Figure 3. The crude event rates based on discharge dose and adjusted hazard ratios (HRs) for mortality are shown in Table II. There was a dose-dependent increase in adjusted risk of in-hospital, out-of-hospital, and all deaths in the MD and HD groups relative to LD exposure.
Figure 3.
Adjusted Kaplan-Meier curves showing survival after hospital discharge, stratified by discharge dose category (log-rank, P < .001).
Table II. Event rates and adjusted HRs for mortality
| LD | MD | HD | |
|---|---|---|---|
| Discharge furosemide dose—event rate per 100 person-years | |||
| All mortality | 19.64 | 24.59 | 34.89 |
| In-hospital death | 12.43 | 15.27 | 20.15 |
| Out-of-hospital death | 7.22 | 9.32 | 14.74 |
| Dynamic time-varying furosemide dose—adjusted HRs (95% CI)⁎ | |||
| All mortality | Referent | 1.96 (1.79-2.15)† | 3.00 (2.72-3.31)† |
| In-hospital death | Referent | 2.00 (1.78-2.24)† | 3.12 (2.76-3.53)† |
| Out-of-hospital death | Referent | 1.91 (1.65-2.20)† | 2.81 (2.40-3.29)† |
⁎Hazard ratios adjusted for age, sex, IHD, prior MI, aortic/mitral valve disease, hypertension, diabetes, atrial fibrillation, cancer, cerebrovascular disease, COPD, dementia, smoking, NYHA class, LVEF, hypo/normo/hypernatremia, hypo/normo/hyperkalemia, creatinine, urea, cardiomegaly, heart rate, respiratory rate, preadmission use of furosemide, presenting and discharge systolic blood pressure, and other discharge medications. |
†P < .001 vs LD. |
Hospitalization
A total of 623 (89%), 1,228 (87%), and 2,015 (88%) of HD, MD, and LD furosemide patients, respectively, were hospitalized during follow-up; 361 (52%), 688 (49%), and 943 (41%) were admitted for HF, whereas 505 (72%), 966 (69%), and 1,492 (65%) were admitted for CV disease. Adjusted Kaplan-Meier survival curves for time to first HF hospitalization are shown in Figure 4. Table III shows increasing hospitalization rates with higher furosemide dose at discharge, whereas analysis of dynamic furosemide dose showed that higher furosemide dose exposure was independently associated with higher hospitalization risk in a dose-dependent fashion.
Figure 4.
Adjusted Kaplan-Meier curves showing survival free from HF hospitalizations after discharge, stratified by discharge dose category (log-rank, P < .001).
Table III. Event rates and adjusted HRs for hospitalization
| LD | MD | HD | |
|---|---|---|---|
| Discharge furosemide dose—event rate per 100 person-years | |||
| HF hospitalizations | 13.04 | 17.51 | 22.73 |
| CV hospitalizations | 20.62 | 24.59 | 31.80 |
| All hospitalizations | 27.85 | 31.25 | 39.23 |
| Dynamic time-varying furosemide dose—adjusted HRs (95% CI)⁎ | |||
| HF hospitalizations | Referent | 1.24 (1.12-1.38)† | 1.43 (1.26-1.63)† |
| CV hospitalizations | Referent | 1.12 (1.02-1.22)‡ | 1.29 (1.15-1.44)† |
| All hospitalizations | Referent | 1.06 (0.98-1.14) | 1.22 (1.10-1.35)† |
†P < .001 vs LD. |
‡P < .05 vs LD. |
Renal dysfunction, sudden death, and fractures
During the follow-up period, 44% of patients discharged on HD developed renal dysfunction, compared to 36% and 29% of patients discharged on MD and LD furosemide. We observed a prominent dose-dependent increase in the risk of renal dysfunction and arrhythmias with furosemide exposure, but there was no difference in fracture risk according to furosemide dose. The event rates and adjusted HRs for renal disease, arrhythmias, and fractures are listed in Table IV.
Table IV. Event rates and adjusted HRs for renal dysfunction, arrhythmias and fractures
| LD | MD | HD | |
|---|---|---|---|
| Discharge furosemide dose—event rate per 100 person-years | |||
| Renal dysfunction | 9.12 | 13.06 | 19.21 |
| Arrhythmias | 12.26 | 13.72 | 19.46 |
| Fractures | 3.18 | 3.00 | 3.53 |
| Dynamic time-varying furosemide dose—adjusted HRs (95% CI)⁎ | |||
| Renal dysfunction | Referent | 1.56 (1.38-1.76)† | 2.16 (1.88-2.49)† |
| Arrhythmias | Referent | 1.15 (1.03-1.30)‡ | 1.45 (1.27-1.66)† |
| Fractures | Referent | 1.13 (0.90-1.44) | 1.01 (0.75-1.37) |
†P < .001 vs LD. |
‡P < .05 vs LD. |
Sensitivity analyses
We examined the risk of death based on average daily furosemide exposure, determined as the cumulative dose of furosemide over time divided by the duration of prescription. High, medium, and low average daily dose during treatment were defined as ≥120, 60-119, and <60 mg/d, respectively. The results were comparable to the primary results shown in Table II, Table III, Table IV (online Table A). We also performed an analysis of patients that survived 1 year (n = 3,269), examining furosemide prescriptions from discharge until the end of year 1. Irrespective of initial discharge furosemide dose, patients were reclassified as HD1yr, MD1yr, or LD1yr if the majority of subsequent furosemide prescriptions were mostly high, medium, or low, respectively (online Table B). Thus, patients who were initially discharged on LD but filled mostly HD furosemide prescriptions within 1 year were considered to be HD1yr. These results also demonstrated that patients on higher furosemide were at increased risk irrespective of the initial discharge dose. Relative to those who were LD1yr, the fully adjusted HRs for mortality to 2-year follow-up were 1.19 (95% CI 1.07-1.33, P = .002) for MD1yr and 1.46 (95% CI 1.28-1.65, P < .001) for HD1yr.
Discussion
In this study, we examined the dynamic change in furosemide exposure among patients with HF. This has not been previously explored in the context of HF outcomes at the population level. Among elderly patients with HF, exposure to increasing furosemide dose after discharge was associated with increased risk of an array of adverse morbid and fatal outcomes including hospitalizations for HF, CV disease, renal dysfunction, arrhythmias, and both in-hospital and out-of-hospital death. Over time, there was growing discordance between discharge dose and patients' actual furosemide exposure, justifying our approach of analyzing furosemide exposure as a time-varying covariate. At baseline, patients discharged on HD had worse renal function and greater prevalence of ischemic and valvular heart disease. They were also more likely to receive concomitant symptom-targeted therapy such as metalozone and digoxin and had higher rates of prehospital furosemide use, hyponatremia, hypotension, and potassium supplementation. Medium-dose and high-dose dynamic furosemide exposures were associated with 96% and 200% increases, respectively, in adjusted mortality risk. There were increases in both in-hospital and out-of-hospital death, suggesting an increased risk of both pump failure and sudden death. There was also a dose-dependent increase in hospitalization risk that was strongest for HF events, suggesting that the adverse outcomes are related to HF progression.
In acute HF, diuretic dose has been correlated with mortality in-hospital,19 but the long-term prognostic implications were not examined. Retrospective analyses of data from clinical trials or tertiary referral centers have demonstrated that loop diuretic use is associated with increased mortality risk in chronic HF.7, 8, 20, 21 These studies did not examine the wide range of potential adverse events that may mediate the increased mortality risk, mostly included patients with advanced systolic dysfunction, and provided limited adjustment for noncardiac conditions. These studies also did not account for changes in furosemide dosing over time, which we found to be frequent in ambulatory HF management.
Our analysis examined long-term effects of dynamic furosemide exposure, accounting for dose fluctuations in a multicenter, population-based setting. Our data also extend the literature by examining the relationship between furosemide use and other possible adverse effects, such as renal dysfunction, arrhythmia, and fracture risk. Prior studies reported that loop diuretic use is associated with increased parathyroid hormone activity,22 decreased bone mineral density,23 and higher fracture risk.13 However, as demonstrated in our study, there was no significant fracture risk related to furosemide exposure in HF patients.
Given the strength of the association, our findings suggest that furosemide dose may represent a “pharmamarker” of HF disease severity. The value of furosemide dose as a pharmacologic biomarker of HF severity is enhanced by its dynamic nature, which may indicate progression or improvement in symptom control over time. Higher furosemide doses may indicate patients who have a tenuous cardiorenal axis, are at increased risk of renal dysfunction, and are more susceptible to decompensation.24 Dependence on HD furosemide also poses a challenge for optimal pharmacologic management. The use of ACEI and β-adrenoreceptor antagonists was reduced, whereas spironolactone use was higher in patients on HD furosemide, suggesting that medication prescription represents a somewhat complex balance between evidence-based therapy, drug intolerance, and symptom management.
There have also been suggestions that furosemide may accelerate HF progression25 by increasing activity of the renin-angiotensin-aldosterone system26, 27 and norepinephrine levels.27, 28 Furosemide may also induce hypokalemia,29 thus promoting cardiac dysrhythmias and sudden death.14 Furthermore, it has been postulated that neurohormonal activation and reduction of effective circulating volume by furosemide might directly compromise renal function.10 However, such concerns cannot be conclusively addressed outside the setting of a randomized clinical trial. Greater vigilance and monitoring of HF patients may be beneficial when HD furosemide is required because of their poor prognosis. Significant impacts could result if novel therapeutic alternatives can afford enhanced volume control with fewer adverse events than furosemide.
Our study had some notable limitations. Our methodology does not allow us to determine how much of the increased risk of adverse outcomes was due to confounding by indication for treatment because increasing furosemide dose often occurs with worsening clinical status. However, addressing this question in a randomized controlled trial may be challenging. We only identified 1 trial of furosemide use in HF,4 with prespecified primary outcomes of symptom improvement and change in creatinine (not mortality or hospitalization), an enrollment target of 300 patients, and a controlled furosemide exposure duration of 72 hours. Our study was able to evaluate long-term utilization and changes in dose of furosemide over time relative to a wide array of fatal and morbid outcomes. Nevertheless, we did not capture subtle changes in furosemide dose, which could have occurred by halving or doubling of a current dose, and thus, the true effect of drug dose may be even greater than we have identified. We also did not examine changes in laboratory markers or clinical measurements (eg, blood pressure) over time. Other limitations include the absence of complete data on LVEF and biomarkers, namely, brain natriuretic peptide. However, HF is a clinical syndrome for which furosemide can be used irrespective of the ejection fraction and throughout the range of brain natriuretic peptide values. Therefore, these factors do not detract from the value of furosemide dose as a dynamic marker of prognosis. Our analysis of morbidity evaluated hospitalizations and did not include less severe conditions that did not require admission to hospital or serial measurements of creatinine. Administrative database coding for renal dysfunction has previously been shown to have limited sensitivity, but high specificity, and similar considerations likely also apply for arrhythmias.30 These factors would have resulted in attenuated risk estimates. The accuracy for fractures is likely to be high in administrative databases as demonstrated in prior studies.13
Furosemide is commonly prescribed after HF hospitalization, with frequent changes in dose during subsequent management. High furosemide doses were prescribed to younger patients who more commonly had renal dysfunction and characteristics consistent with reduced ejection fraction. High furosemide dose was associated with greater risk of death, hospitalization for CV disease, renal dysfunction, and arrhythmias, but fracture risk was not increased, after multiple covariate adjustment. Our results suggest that furosemide dose can serve as a powerful, dynamic, and easily used marker of prognosis in HF. The marked increase in risk suggests the need for vigilance in patients who are dependent on HD furosemide and supports the need for clinical trials testing different strategies of furosemide dosing.
Disclosures
The Institute for Clinical Evaluative Sciences is supported in part by a grant from the Ontario Ministry of Health and Long Term Care. The opinions, results, and conclusions are those of the authors, and no endorsement by the Ministry of Health and Long-Term Care or by the Institute for Clinical Evaluative Sciences is intended or should be inferred. Supported by a Canadian Institutes of Health Research (CIHR) grant and a CIHR Team Grant in Cardiovascular Outcomes Research. Supported by a clinician-scientist award from the CIHR (D.S.L.), career investigator awards from the Heart and Stroke Foundation of Ontario (P.C.A. and J.V.T.), and a Canada Research Chair in health services research (J.V.T.).
Appendix A. Coding of diagnoses using ICD coding system
| Diagnoses | ICD-9 codes | ICD-10 codes |
|---|---|---|
| CV disease | ||
| HF | 428 | I50 |
| IHD | 410-414 | I20-I25 |
| Arrhythmia | 426,427 | I44-I49 |
| Myo-/peri-/endocardial disease | 420-423, 425 | I30-I33, I40-I43, I51.4 |
| Cerebrovascular disease | 430-438 | I60-I69 |
| Hypertensive disease | 401-405 | I10-I13, I15 |
| Pulmonary vascular disease | 415-417 | I26-I28 |
| Peripheral vascular disease | 440-448, 451-453, 785.4 | I70-I74, I77-I82, R02 |
| Rheumatic heart disease | 390-398 | I00-I02, I05-I09 |
| Shock | 785.5 | R57 |
| Syncope, sudden death | 780.2, 798.1 | R55, R96.0 |
| Valvular heart disease | 424 | I34-I39 |
| Other cardiac | 429, 458 | I51, I52, I95, I97 |
| Renal dysfunction | ||
| Acute renal failure | 584, 997.5, 639.3, 669.3, 572.4 | N17, N99.0, K76.7, O90.4, O08.4 |
| Chronic renal failure | 585 | N18 |
| Other renal failure | 586, 788.9 | N16, N19, E10.22, E11.22, E13.22, E14.22 |
| Disorder of kidney and ureter | 593.9 | N28.9 |
| Care involving dialysis | V56 | Z49 |
| Dependence on renal dialysis | V45.1 | Z99.2 |
| Volume depletion | 276.5 | E86 |
| Other fluid, electrolyte, acid-base | 276.0-276.9, 253.6 | E87 |
| Arrhythmias | ||
| Cardiac arrest | 427.5 | I46 |
| Paroxysmal tachycardia | 427.1, 427.2 | I47 |
| Atrial fibrillation and flutter | 427.3 | I48 |
| Ventricular fibrillation and flutter | 427.4 | I49 |
| Other sudden death | 798 | R96 |
| Fractures | 800-816 and 818-828 including subclassifications, 873.6, 873.7 | S02, S32, S42, S52, S62, S72, S82, S92, T02 |
Appendix B. Online Table A—Sensitivity analysis of adjusted risk of death relative to cumulative furosemide exposure
| Outcome | HR for LD | HR for MD (95% CI) | HR for HD (95% CI) |
|---|---|---|---|
| Death | Referent | 1.31 (1.20-1.42)⁎ | 1.72 (1.54-1.92)⁎ |
| HF hospitalizations | Referent | 1.70 (1.53-1.89)⁎ | 2.09 (1.83-2.40)⁎ |
| CV hospitalizations | Referent | 1.36 (1.25-1.48)⁎ | 1.62 (1.45-1.81)⁎ |
| Any hospitalization | Referent | 1.22 (1.13-1.31)⁎ | 1.41 (1.28-1.56)⁎ |
| Renal dysfunction | Referent | 1.57 (1.39-1.78)⁎ | 2.27 (1.95-2.65)⁎ |
| Arrhythmias | Referent | 1.32 (1.18-1.47)⁎ | 1.50 (1.29-1.74)⁎ |
| Fractures | Referent | 1.07 (0.85-1.34)† | 1.23 (0.90-1.69)† |
⁎P < .001 vs LD. |
†P = NS vs LD. |
Appendix C. Online Table B—Reclassification based on furosemide dose at end of year 1 for sensitivity analysis for death
| Discharge dose | Dose category at end of year 1 | No. of patients (%) |
|---|---|---|
| High | HD1yr | 287 (8.8) |
| High | MD1yr | 102 (3.1) |
| High | LD1yr | 51 (1.6) |
| Medium | HD1yr | 216 (6.6) |
| Medium | MD1yr | 510 (15.6) |
| Medium | LD1yr | 300 (9.2) |
| Low | HD1yr | 136 (4.2) |
| Low | MD1yr | 335 (10.2) |
| Low | LD1yr | 1332 (40.7) |
References
- 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
- The EuroHeart Failure Survey programme—a survey on the quality of care among patients with heart failure in Europe. Part 2: treatment. Eur Heart J. 2003;24:464–474
- . Dose-dependent association between use of loop diuretics and mortality in advanced systolic heart failure. Am J Cardiol. 2006;98:1416–1417
- . Loop diuretics in acute decompensated heart failure: necessary? Evil? A necessary evil?. Circ Heart Fail. 2009;2:56–62
- . Diuretic usage in heart failure: a continuing conundrum in 2005. Eur Heart J. 2005;26:644–649
- . Furosemide in the long-term management of heart failure: the good, the bad, and the uncertain. J Am Coll Cardiol. 2004;44:1308–1310
- A propensity-matched study of the effects of chronic diuretic therapy on mortality and hospitalization in older adults with heart failure. Int J Cardiol. 2008;125:246–253
- . Relation of loop diuretic dose to mortality in advanced heart failure. Am J Cardiol. 2006;97:1759–1764
- Relationship between heart failure treatment and development of worsening renal function among hospitalized patients. Am Heart J. 2004;147:331–338
- Diuretics, mortality, and nonrecovery of renal function in acute renal failure. JAMA. 2002;288:2547–2553
- . Risk factors for acute renal insufficiency induced by diuretics in patients with congestive heart failure. Am J Kidney Dis. 2006;47:798–808
- . Diuretic drug use and the risk for hip fracture. Ann Intern Med. 1991;115:1–6
- . Fracture risk in patients treated with loop diuretics. J Intern Med. 2006;259:117–124
- Diuretics and risk of arrhythmic death in patients with left ventricular dysfunction. Circulation. 1999;100:1311–1315
- Effectiveness of public report cards for improving the quality of cardiac care: the EFFECT study: a randomized trial. JAMA. 2009;302:2330–2337
- The natural history of congestive heart failure. NEJM. 1971;285:1441–1446
- “Dose-dependent” impact of recurrent cardiac events on mortality in patients with heart failure. Am J Med. 2009;122:162–169
- Coding accuracy of administrative drug claims in the Ontario Drug Benefit database. Can J Clin Pharmacol. 2003;10:67–71
- Relation between dose of loop diuretics and outcomes in a heart failure population: results of the ESCAPE trial. Eur J Heart Fail. 2007;9:1064–1069
- Diuretic use, progressive heart failure, and death in patients in the Studies Of Left Ventricular Dysfunction (SOLVD). J Am Coll Cardiol. 2003;42:705–708
- Diuretic resistance predicts mortality in patients with advanced heart failure. Am Heart J. 2002;144:31–38
- Loop diuretics alter the diurnal rhythm of endogenous parathyroid hormone secretion. A randomized-controlled study on the effects of loop- and thiazide-diuretics on the diurnal rhythms of calcitropic hormones and biochemical bone markers in postmenopausal women. Eur J Clin Invest. 2001;31:764–772
- Loop diuretics increase bone turnover and decrease BMD in osteopenic postmenopausal women: results from a randomized controlled study with bumetanide. J Bone Miner Res. 2006;21:163–170
- Renal function as a predictor of outcome in a broad spectrum of patients with heart failure. Circulation. 2006;113:671–678
- Furosemide and the progression of left ventricular dysfunction in experimental heart failure. J Am Coll Cardiol. 2004;44:1301–1307
- Untreated heart failure: clinical and neuroendocrine effects of introducing diuretics. Br Heart J. 1987;57:17–22
- Acute vasoconstrictor response to intravenous furosemide in patients with chronic congestive heart failure. Activation of the neurohumoral axis. Ann Intern Med. 1985;103:1–6
- Neurohormonal effects of furosemide withdrawal in elderly heart failure patients with normal systolic function. Eur J Heart Fail. 2003;5:47–53
- Frequency of hypokalemia after successfully resuscitated out-of-hospital cardiac arrest compared with that in transmural acute myocardial infarction. Am J Cardiol. 1987;59:84–88
- Validity of International Classification of Diseases, Ninth Revision, Clinical Modification Codes for Acute Renal Failure. J Am Soc Nephrol. 2006;17:1688–1694
PII: S0002-8703(10)00438-2
doi:10.1016/j.ahj.2010.05.032
© 2010 Mosby, Inc. All rights reserved.
