Prescriptions of β-blockers over time expressed as the percentage of new patients seen at each period who started β-blocker therapy. MDC indicates Metoprolol in Dilated Cardiomyopathy Trial; Carvedilol RCTs, the US and Australia/New Zealand Carvedilol Trials; CIBIS, Cardiac Insufficiency Bisoprolol Study; and MERIT-HF, Metoprolol CR/XL Randomized Intervention Trial in Congestive Heart Failure.
Survival of patients with heart failure receiving or not receiving β-blockers. Median duration of follow-up was 32 months. Four hundred seventy-five patients received β-blockers, and 566 patients did not receive β-blockers. Results of Cox multivariate regression analysis are given in the "Survival Analyses" subsection of the "Results" section.
Dosages of β-blockers expressed as the percentage of patients achieving a maximum prescribed dosage in 1 of 4 strata (defined relative to the mean dosages attained in the trials proving the efficacy of β-blockers).
Tandon P, McAlister FA, Tsuyuki RT, Hervas-Malo M, Dupuit R, Ezekowitz J, Cujec B, Armstrong PW. The Use of β-Blockers in a Tertiary Care Heart Failure ClinicDosing, Tolerance, and Outcomes. Arch Intern Med. 2004;164(7):769-774. doi:10.1001/archinte.164.7.769
Little is known about the dosing, tolerability, and impact of β-blockers in nontrial participants. This study was conducted to evaluate the use and outcomes of β-blockers in a tertiary care heart failure clinic.
Analysis of prospectively collected data from a cohort of 1041 patients with heart failure seen at the University of Alberta Heart Function Clinic, Edmonton, from September 1, 1989, through July 1, 2001, with objective measurement of ejection fraction at baseline and prospective collection of data at all subsequent clinic visits.
Median age at baseline was 69 years; 65% were male; 75% had systolic dysfunction; mean ejection fraction was 33%; and 51% had New York Heart Association class III or IV symptoms. Median duration of follow-up was 32 months (interquartile range, 13-62 months). Overall, 46% of patients received β-blockers, but only 18% of these were ultimately prescribed the dosages achieved in the trials (mean maximum dosages achieved, 27 mg/d for carvedilol and 81 mg/d for metoprolol tartrate). Of those patients prescribed β-blockers, 74% continued to receive them during follow-up. Blood pressure, heart rate, and failure symptomatology did not change appreciably before and after β-blockers were prescribed, or during the upward titration of the dosage. Although our patients were prescribed lower dosages than those used in trials, Cox multivariate regression revealed that β-blockers were associated with improved survival, even after adjusting for potential confounders including New York Heart Association class, year of prescription, and concomitant medication use (relative risk, 0.63; 95% confidence interval, 0.50-0.81).
The benefits of β-blockers seen in randomized trials extend to nontrial participants treated in a tertiary care clinic specializing in heart failure. In our cohort of elderly patients with multiple comorbidities, β-blockers were well tolerated.
The development of heart failure leads to a compensatory and ultimately maladaptive activation of the adrenergic system. Agents that modify neurohormonal activity (such as angiotensin-converting enzyme [ACE] inhibitors and β-blockers) have been shown to improve outcomes in patients with systolic dysfunction.1,2 Although many trials have conclusively demonstrated the benefits of β-blockers in heart failure, a number of questions remain unanswered, as trial participants tend to be younger, healthier, more compliant, and more closely followed up than patients in everyday practice.3 In particular, we do not know the proportion of heart failure patients who would be eligible for and able to tolerate β-blocker therapy, whether the dosages used in trials are prescribed to real-world patients, the factors that have an impact on the use and dosage of β-blockers, and the effectiveness of β-blockers in nontrial participants.
A study from our group has previously shown that the use of β-blockers is associated with improved survival in nontrial patients in a population-based cohort study of almost 12 000 elderly patients with incident heart failure (consistent with the degree of benefit seen in trial participants).4 That study revealed that only 10% of heart failure patients cared for by community physicians were prescribed β-blockers from 1994 to 1999; however, as we used administrative databases in that study, we could not fully examine (or adjust for) potential confounding factors.
The present study was undertaken to evaluate the use of β-blockers in patients followed up at a tertiary care heart failure clinic, with particular focus on the proportion of patients who were prescribed and tolerated β-blockers, those demographic and clinical factors that influenced their use and affected their dosage titration, and the survival of patients treated with β-blockers.
Data were collected prospectively on 1151 adult patients referred to the University of Alberta Hospital Heart Function clinic, Edmonton, for evaluation and management of heart failure between September 1, 1989, and July 1, 2001. The clinic is multidisciplinary and staffed by a team of specialty-trained physicians, pharmacists, and nurse practitioners. Each patient was examined by a Heart Function Clinic physician at the first visit, and the 1041 confirmed to have congestive heart failure (on the basis of the Framingham Heart Study criteria5) were included in this study. Demographic, clinical, laboratory, imaging, and pharmacological data were collected prospectively at the first visit and each subsequent visit and entered into a computer database.
Objective assessments (by means of echocardiography, radionuclide imaging, or contrast ventriculography) of ejection fraction were obtained in all patients within 3 months of initial presentation to the clinic. Although the precision of these methods differs, each method has been shown to confer similar prognostic information.6 The demographic and clinical variables collected are outlined in Table 1. Patients were defined as having systolic dysfunction if their ejection fraction was less than 45%. If patients had a baseline ejection fraction of 45% to 50%, they were classified as having systolic dysfunction unless the echocardiogram indicated that the predominant problem was in ventricular diastolic relaxation (in which case they were classified as having diastolic dysfunction). If patients had a baseline ejection fraction of greater than 50%, they were classified as having diastolic dysfunction. The causes of heart failure were classified as ischemic (history of myocardial infarction, coronary revascularization, angiographically documented stenosis of >75% in ≥1 coronary artery, or focal areas of akinetic myocardium detected on echocardiography, radionuclide stress testing, or contrast ventriculography) or nonischemic by the Heart Function Clinic physician. Atrial arrhythmias included atrial fibrillation, atrial flutter, and multifocal atrial tachycardia. Ventricular arrhythmias included a history of cardiac arrest or documented ventricular tachycardia/fibrillation. The diagnosis of left ventricular hypertrophy was determined on the basis of the electrocardiogram using the Romhilt-Estes criteria.
For each patient, information about concurrent cardiac medications and dosage was obtained. Mortality data were obtained from the Alberta Vital Statistics Registry, death certificates, or review of medical records. Approval of the Research Ethics Board of the University of Alberta was obtained.
All variables were entered into a computer database, and data collation, quality assurance, and analysis were performed at the Epidemiology Coordinating and Research Center, Division of Cardiology, University of Alberta.
The baseline characteristics of those who were and were not prescribed β-blockers were compared using the χ2 test for dichotomous variables and unpaired, 2-tailed t test for continuous variables. Since multiple comparisons were planned a priori, a Bonferroni correction was used, and statistical significance was assumed at P≤.005. Analyses of baseline heart rate and blood pressure were performed excluding those patients who were receiving β-blockers at the time they were first seen in the clinic. Titration of β-blocker dosage was calculated for each β-blocker separately. To adjust for differences in baseline clinical characteristics and concomitant conditions or medications in determining the predictors of β-blocker use (and dosage), multiple logistic regression analysis with a backward stepwise technique and entry of all prespecified factors with P<.25 on bivariate analyses was performed. Age was run as a continuous variable in the analysis and was dichotomized into younger than 69 years or 69 years or older when age was significant in the first model (69 years was chosen as the cut point because it represented the median age in our cohort). For the multiple logistic regression analyses, statistical significance was accepted at P<.05. We generated survival curves by the Kaplan-Meier method and compared them using the Mantel log-rank test. Multiple logistic regression analyses were run to determine which factors were associated with mortality. Cox proportional hazards analyses were performed to determine the association of β-blocker use with mortality adjusted for age, sex, New York Heart Association (NYHA) class, year in which the patient was seen, concomitant medication use, and any other variables that significantly differed between β-blocker users and nonusers or predicted mortality in the multivariate models. All analyses were performed with SAS software (SAS Institute Inc, Cary, NC).
Our study population consisted of 1041 patients with confirmed congestive heart failure. Median age was 69 years; 65% were male; 75% had systolic dysfunction; 66% had heart failure with an ischemic cause; mean ejection fraction was 33%; and 89% had NYHA classes II through IV symptoms. Median duration of follow-up was 32 months (interquartile range, 13-62 months). Contraindications to β-blocker use were present in 112 (15%) of the 765 patients not receiving β-blockers at their first visit to the Heart Function Clinic (many of which were relative, but were defined broadly to include history of obstructive airway disease, hypotension, high degree of atrioventricular block, or advanced peripheral vascular disease).
During this study, β-blockers were prescribed for 46% of patients; there was a significant trend over time, with two thirds of patients seen for the first time after 2000 prescribed β-blockers (P<.001; Figure 1). The baseline characteristics of those prescribed and those not prescribed β-blockers are shown in Table 1. On multivariate logistic regression, adjusting for all variables in Table 1, the use of β-blockers was independently associated with NYHA class I or II symptoms (odds ratio [OR], 2.0; 95% confidence interval [CI], 1.5-2.6), being younger than 69 years (OR, 1.4; 95% CI, 1.05-1.8), having a history of dyslipidemia (OR, 2.8; 95% CI, 1.7-4.5), and also being prescribed ACE inhibitors (OR, 2.7; 95% CI, 1.8-4.2), aspirin (OR, 1.7; 95% CI, 1.3-2.2) or dihydropyridine calcium channel blockers (OR, 3.3; 95% CI, 2.2-4.9). Patients with a history of obstructive airway disease (OR, 0.3; 95% CI, 0.2-0.6) were less likely to be prescribed β-blockers.
During follow-up, 383 patients died. Survival analyses revealed 1-, 2-, and 5-year survival rates of 96%, 92%, and 74%, respectively, for patients with NYHA class I symptoms at baseline; 93%, 82%, and 66%, respectively, for those with NYHA class II symptoms; 85%, 71%, and 54%, respectively, for those with NYHA class III symptoms; and 69%, 54%, and 36%, respectively, for those with NYHA class IV symptoms. Factors independently associated with 1-year mortality risk on multiple logistic regression are outlined in the following tabulation:
Kaplan-Meier analysis demonstrated that β-blocker users had substantially better survival than nonusers (Figure 2). On Cox proportional hazards analysis, use of β-blockers was associated with reduced mortality risk (hazard ratio, 0.63; 95% CI, 0.50-0.81), even after adjusting for age, sex, ejection fraction, NYHA class, causes, type of dysfunction (systolic vs diastolic), year of prescription, and concomitant use of ACE inhibitors, digoxin, and/or amiodarone. On multivariate logistic regression, the benefits of β-blockers were particularly pronounced in those patients with systolic dysfunction (OR, 0.42 [95% CI, 0.27-0.67]) compared with those with diastolic dysfunction (OR, 0.66 [95% CI, 0.30-1.47]).
The dosages attained in the clinical trials establishing the efficacy of β-blockers were rarely achieved in these older, more frail patients (Figure 3).2 In fact, only 18% of our patients prescribed β-blockers achieved the target dosages in the β-blocker trials. Although a significantly higher proportion of those prescribed carvedilol achieved the target dosage compared with those prescribed metoprolol tartrate (31% vs 17%; P = .006) or other β-blockers (31% vs 9%; P<.001), these differences disappeared after adjustment for year of prescription, because the upward titration of β-blocker dosage was more aggressive over time (with 24% of patients achieving the target dosage in 1998-2001 vs 10% before 1998; P<.001). The dosage titration curves were relatively flat, with a mean increase of only 29% in β-blocker dosage in the first 12 months after the initial prescription. The mean maximum dosage in our patients treated with β-blockers was only 47% greater than the mean initial prescribed dosage. Mean maximum dosages were 27 mg/d for carvedilol and 81 mg/d for metoprolol tartrate.
Although there were no statistically significant differences between the groups, there were strong trends toward use of higher dosages of β-blockers in those patients who were younger, had less severe symptoms (NYHA classes I and II), had higher ejection fractions, had higher systolic blood pressure, and were also prescribed ACE inhibitors or dihydropyridine calcium channel blockers. On multiple logistic regression, only concurrent use of dihydropyridine calcium channel blockers (OR, 1.7; 95% CI, 1.0-2.9) was significantly associated with prescription of higher dosages.
Of the 475 patients prescribed β-blockers in our clinic, 351 (74%) continued to receive these agents throughout follow-up. Aside from a small decline in mean heart rate when β-blocker therapy was first initiated, there was little change in blood pressure, heart rate, or NYHA class as β-blocker dosages were increased (Table 2). None of the clinical or laboratory data we collected predicted the tolerability of β-blockade.
Use and dosage of β-blockers increased in our clinic as the evidence evolved; however, relative to the clinical trial evidence, β-blockers were still underused and dosages were low in our patients. Furthermore, our data demonstrate that heart failure patients seen in clinical practice who are prescribed β-blockers have better outcomes than those who are not provided this therapy, even after adjusting for age, sex, ejection fraction, cause of heart failure, type of dysfunction, year of prescription, and concomitant use of other medications shown to influence prognosis (such as ACE inhibitors). This reinforces the results of the published trials and extends the evidence base for β-blockers from efficacy (established by the trials) to effectiveness (established by outcomes research). We have also demonstrated that β-blockers (at least in the dosages we used) are generally well tolerated by nontrial heart failure patients.
Although some of this underuse is accounted for by patient factors such as obstructive airway disease, hypotension, or bradycardia at baseline, and although some is attributable to the evolving evidence base (β-blockers were only recently shown to be beneficial in NYHA class IV failure),7 patients in our cohort were less likely to be prescribed a β-blocker if they were older, even after adjustment for covariates and concomitant medication use. In addition, patients were more likely to be prescribed β-blockers if they were also prescribed other therapies such as ACE inhibitors, aspirin, or dihydropyridine calcium channel blockers. The underuse of proven efficacious therapies such as β-blockers in the elderly is a common theme in outcomes research and is not supportable on the basis of the heart failure trial evidence (where older patients appear to derive similar relative risk reductions from β-blockade as younger patients).8,9 Although it may be hypothesized that older patients are sicker or have a higher burden of disease, the relative underuse in these patients was seen even after adjustment for comorbidities, concomitant medication use, symptoms, signs, and ejection fraction. As older patients are at higher absolute risk for poor outcomes, they are likely to benefit even more than younger patients from the use of efficacious therapies.
The use of β-blockers in our clinic is substantially higher than that observed in a cohort of heart failure patients not treated in our specialized clinic but drawn contemporaneously from the same geographic locale (46% vs 10%).4 Furthermore, our β-blocker use was substantially higher than that previously reported in audits of outpatient and inpatient care in a variety of countries,10- 14 and is similar to the data recently reported from a heart failure disease management program in the United States and heart failure clinics in Germany and the United States.15- 17 This provides further proof that a specialized clinic devoted to the management of heart failure can optimize the use of proven efficacious therapies in these patients.18 We believe our recent experience provides guidance as to the proportion of heart failure patients who are likely to be eligible for β-blocker therapy (more than two thirds of our patients seen in the past 3 years).
It is not surprising (given the older and more frail patients in our cohort compared with the β-blocker trial populations) that the dosages attained in clinical trials were infrequently achieved in our cohort. This mirrors the recently reported results from a German heart failure clinic in which patients prescribed β-blockers achieved only approximately 40% of trial dosages.16 Although no trials have been completed comparing higher vs lower dosages of β-blockers in heart failure, some preliminary evidence suggests that higher dosages may be better. For example, in 345 patients assigned various dosages of carvedilol, those patients assigned higher dosages had better ejection fractions during follow-up.19 Furthermore, the cohort study evaluating outcomes in heart failure patients who were not treated in a specialized clinic suggested greater survival benefits with higher-dosage therapy.4 Regardless, the issue of whether there is a dosage-dependent effect with β-blockers will remain unresolved until a trial is performed to compare high- vs low-dosage β-blockade. Until such a trial is carried out, our data suggest that even low-dosage β-blockers are associated with improved survival in nontrial heart failure patients.
Similar to an analysis of 808 Australian heart failure patients treated with carvedilol, we failed to find any baseline clinical variables that predicted tolerability to β-blockade.20 In our clinic, β-blockers were not associated with hypotension, bradycardia, or worsening failure symptomatology. This echoes the results from a German heart failure clinic study16 in which patients treated with β-blockers did not differ appreciably from nontreated patients in NYHA class, blood pressure, peak oxygen consumption, or N-terminal probrain natriuretic peptide level during follow-up. This finding suggests that unmeasured variables (eg, fatigue, impotence, or financial cost) may have an impact on β-blocker tolerability and limit upward titration of the dosage. Indeed, a recently published systematic review21 of 15 trials (35 000 subjects) reported increased risks for fatigue (OR, 2.6; 95% CI, 1.2-5.9) and sexual dysfunction (OR, 4.9; 95% CI, 3.0-8.0) in patients treated with β-blockers; however, the absolute risks were small (annual rates of 4/1000 for fatigue and 2/1000 for sexual dysfunction) in these trial participants (who were younger and healthier than our study patients).
The present study is observational and, as such, we cannot make definitive statements on whether low-dosage β-blockers in older patients confer survival benefits. However, the rigorous manner in which we prospectively collected data on all consecutive patients with heart failure, our ability to adjust for relevant covariates (including symptoms, signs, comorbidities, ejection fraction, time periods, and concomitant medications), and the similarity of our treatment effect estimate to that reported in the clinical trials of β-blockade provide reasonable comfort in our findings.22- 24 Indeed, an observational study such as ours is a useful adjunct to the β-blocker randomized trials in showing that the efficacy of β-blockers established in trials with relatively young and healthy participants extends to the older and more complicated patients usually seen in clinical practice.
Although concerns may be raised about the potential for selection bias in our study (since sicker or more complicated patients are most likely to be referred to a specialized heart failure clinic), this is less important for the study questions we wanted to examine. Although our study sample is not a random sample of community-dwelling patients with heart failure, and patients in our clinic receive educational materials and access to resources generally not available to nonclinic patients, they are a consecutive series of patients with documented heart failure and objective assessment of ejection fractions followed up prospectively in a clinic that emphasizes the maximal application of proven efficacious therapies. Thus, our study sample is ideal for investigating the tolerability of β-blockers (and their dosages) in heart failure.
We have shown that the benefits of β-blockers extend beyond the young and relatively uncomplicated patients enrolled in trials to encompass older patients with multiple comorbidities. In general, β-blockers were well tolerated in our clinic patients. We have also demonstrated that the care of patients with heart failure changed over time in a specialized clinic, in step with the evolving evidence from clinical trials, and comparison with the rates of β-blocker use in contemporaneous nonclinic patients4 confirms that uptake of this evidence was greater in the specialized clinic. However, even in our specialized clinic, few patients were prescribed trial dosages of β-blockers, and further prospective studies are required to determine why. Although our data suggest that even low-dosage β-blockade is associated with a survival benefit, a trial is needed to compare the effects of low- vs high-dosage β-blockers on morbidity and mortality end points in patients with heart failure. Until such a trial is performed, we believe that heart failure patients treated with β-blockers should have their dosages titrated to the trial target dosages if tolerated.
Corresponding author: Finlay A. McAlister, MD, MSc, 2E3.24 WMC, University of Alberta Hospital, 8440 112 St, Edmonton, Alberta, Canada T6G 2R7 (e-mail: Finlay.McAlister@ualberta.ca).
Accepted for publication May 29, 2003.
This study was supported by the Alberta Heritage Foundation for Medical Research, Edmonton, and the Canadian Institutes of Health Research, Ottawa, Ontario (Dr McAlister).
This study was presented in part at the 51st annual meeting of the American College of Cardiology; March 18, 2002; Atlanta, Ga.
We thank K. Teo, MD, PhD, T. Montague, MD, D. Humen, MD, R. Gutierrez, I. Scott, L. Sanderson, L. Haykowsky, K. Dewart, and T. Walker for their expert assistance in the University of Alberta Heart Function Clinic, Edmonton; S. Kimber, MD, for insights into an earlier version of this study; J. Sieben for maintaining the computer database; and W. Chang, PhD, N. Franjic, Fradette, Dolan, and Priest for assistance with data collection/coordination.