- Email: [email protected]
Trends in heart failure outcomes and pharmacotherapy: 1992 to 2000
CLINICAL STUDIES
Trends in Heart Failure Outcomes and Pharmacotherapy: 1992 to 2000 Douglas S. Lee, MD, Muhammad M. Mamdani, PharmD, MA, MPH, Peter C. Austin, PhD, Yanyan Gong, MSc, Peter P. Liu, MD, Jean L. Rouleau, MD, Jack V. Tu, MD, PhD PURPOSE: To review trends in drug therapy and concomitant outcomes of elderly heart failure patients in Ontario, Canada. METHODS: Utilization of drug therapies, mortality, and rehospitalization rates from April 1992 to March 2000 were determined in 77,421 elderly (aged ⱖ65 years), community-based heart failure patients using linked administrative databases. Treatment effects were identified from published meta-analyses and randomized trials. The effect of drug trends on mortality and morbidity were assessed based on their absolute treatment effects. RESULTS: From 1992 to 2000, angiotensin-converting enzyme inhibitor or angiotensin receptor blocker use increased from 58% to 62% (P ⫽ 0.001) while beta-blocker use increased from 6% to 22% (P ⬍0.001). There was a decrease in the use of treatments for which no survival advantage had been demonstrated in randomized trials, such as digitalis (49% to 35%, P
⬍0.001), Vaughan-Williams class I antiarrhythmic agents (3.5% to 1.4%, P ⬍0.001), and first-generation calcium antagonists (21.3% to 9.6%, P ⬍0.001). The trends in drug therapy were associated with a 2.8% reduction in age-, sex-, and comorbidity-adjusted 1-year mortality and a 4.1% reduction in 1-year hospitalization rates. The observed trends in therapy over time explained 37% of the decrease in mortality and 30% of the decrease in rehospitalization rates. The treatment effect from beta-blockers was most pronounced, explaining 30% of the decrease in mortality and 10% of the decrease in rehospitalization rates. CONCLUSION: During 1992 to 2000, mortality and morbidity improved among elderly patients with heart failure, with increased utilization of beta-blockers contributing most to the beneficial trends in outcomes. Am J Med. 2004;116:581–589. ©2004 by Excerpta Medica Inc.
H
apy (2). Since the publication of several landmark randomized trials of angiotensin-converting enzyme (ACE) inhibitors beginning in the late 1980s, several studies have demonstrated the effects of drug intervention on mortality and morbidity in heart failure patients (3,4). However, changes in drug treatments for community-based heart failure patients and their effects on clinical outcomes have not been described. In this study, we reviewed trends in drug therapy and outcomes in a large sample of heart failure patients in Ontario, Canada. We estimated the effect of treatments initiated after a hospitalization episode on the trends in mortality and rehospitalization rates from 1992 to 2000. We also explored the effect of published randomized heart failure trials on temporal changes in drug utilization patterns.
eart failure affects approximately 6 million persons in North America, where it is a leading cause of hospitalization (1). Over the past two decades, hospital discharges have increased as have the number of deaths attributed to heart failure (1). Although early treatment was directed primarily towards the relief of symptoms, the cornerstone of current heart failure management is based on modification of the natural history of the condition using pharmacologic therFrom the Institute for Clinical Evaluative Sciences (DSL, MMM, PCA, YG, JVT); and the Department of Health Policy, Management and Evaluation/Clinical Epidemiology (DSL, JVT), the Faculty of Pharmacy (MMM), the Department of Public Health Sciences (PCA), the Heart and Stroke/Richard Lewar Centre of Excellence (PPL), the Division of Cardiology, University Health Network, Toronto General Hospital (PPL, JLR), and the Division of General Internal Medicine, Sunnybrook and Women’s College Health Science Centre (JVT), University of Toronto, Toronto, Canada. This study was funded by grants from the Ontario Ministry of Health (Ontario Program for Optimal Therapeutics), and to the Canadian Cardiovascular Outcomes Research Team from the Canadian Institutes of Health Research and Heart and Stroke Foundation. Dr. Lee is supported by a research fellowship from the Heart and Stroke Foundation of Canada/Canadian Institutes of Health Research. Dr. Tu is supported by a Canada Research Chair in Health Services Research. Requests for reprints should be addressed to Jack V. Tu, MD, PhD, Institute for Clinical Evaluative Sciences, Room G-106, 2075 Bayview Avenue, Toronto, Ontario M4N 3M5, Canada, or [email protected]. Manuscript submitted May 13, 2003, and accepted in revised form November 26, 2003. © 2004 by Excerpta Medica Inc. All rights reserved.
METHODS Sources of Data We linked data elements from a number of administrative data sources to identify heart failure patients and medications utilized over time. From the Canadian Institute of Health Information hospital discharge abstracts, which contain information on all hospital separations (discharges, transfers, and deaths) in Ontario, data per0002-9343/04/$–see front matter 581 doi:10.1016/j.amjmed.2003.11.025
Trends in Heart Failure Outcomes and Therapy/Lee et al
taining to the index admission, demographic characteristics, and subsequent heart failure hospitalizations were identified. Prescription drug information and dispensation dates for patients aged ⱖ65 years were identified from the Ontario Drug Benefit database. Data on mortality were gathered from the Registered Persons Vital Statistics database. The cohort consisted of all patients ⱖ65 years of age, admitted to a hospital in Ontario with a primary/most responsible diagnosis of heart failure (code 428) according to the International Classification of Diseases, Ninth Revision, Clinical Modification (5). Audits of coding have shown a high degree of accuracy in the database for heart failure diagnosis (6 – 8). The cohort comprised newly admitted patients, as previously defined (9,10). Other exclusion criteria were no admittance to an acute care hospital, transfer from another acute care facility, age ⱖ105 years, not a resident of Ontario, invalid health card number, and heart failure coded as an in-hospital complication (6). The study was approved by the ethics review board of Sunnybrook and Women’s College Health Sciences Centre.
Trends in Incidence, Mortality, and Morbidity Incident heart failure cases were identified from April 1992 to March 2000, and followed for at least 1 year until March 2001, the last date of follow-up. Each index admission was categorized in chronological order according to the month and year of hospital discharge. We identified the vital status of all patients after admission and whether patients were rehospitalized for heart failure within 1 year of hospital discharge. Mortality rates were adjusted for age, sex, and comorbid conditions using the Deyo-Charlson Comorbidity Index. Linear trends were assessed using adjusted 30-day and 1-year mortality rates as the dependent variable and month of the series as the independent variable (11). Separate mortality analyses were conducted in all patients and hospital survivors. We used similar methods to evaluate unadjusted and adjusted 1-year rehospitalization rates.
Trends in Heart Failure Drug Therapy The drug classes evaluated were beta-adrenergic blockers, ACE inhibitors with angiotensin receptor blockers as an alternative, first-generation calcium channel blockers (e.g., verapamil, diltiazem, nifedipine), class I antiarrhythmic drugs, and digitalis. Drugs within the ACE inhibitor class included benazepril, captopril, cilazapril, enalapril, fosinopril, lisinopril, perindopril, quinapril, and ramipril. Beta-blockers included acebutolol, atenolol, carvedilol, labetolol, metoprolol, nadolol, oxprenolol, pindolol, propranolol, and timolol. Class I antiarrhythmic drugs included disopyramide, flecainide, mexiletine, procainamide, propafenone, quinidine, and tocainide (12). We also considered whether patients received at least one prescription of an ACE inhibitor or 582
May 1, 2004
THE AMERICAN JOURNAL OF MEDICINE威
Volume 116
angiotensin receptor blocker to account for ACE inhibitor intolerance. Examination of drug utilization trends was conducted by comparing utilization rates in the initial and final years of the series. Binary outcome variables were assigned for the drug classes based on prescription status within 30 days of discharge. The average monthly rate in the first year was compared with the rate in the last year of the series. Descriptive analysis of trends was by linear regression, with utilization rate as the dependent variable and month of the series as the independent variable. When a randomized controlled trial was published during the study years, the rates of utilization were described before and after the trial. Three groups of potentially relevant trials that reported mortality outcomes (and their release dates) were identified (12–15): Evaluation of Losartan in the Elderly (ELITE)-I study (March 1997), the Australia– New Zealand (February 1997) and U.S. Carvedilol (May 1996) trials, and the Digitalis Investigation Group (DIG) trial (February 1997).
Contribution of Treatments to Changes in Mortality The contributions of each drug class to the reduction in mortality and hospitalization rates were estimated from the absolute mortality benefits reported from published meta-analyses or randomized trials (if the former were not available). We estimated the contribution of each treatment over time using previously described methods (16). Briefly, mortality benefit was calculated as ABj(t2) * Usej(t2) – ABj(t1) * Usej(t1), where ABj is the absolute benefit of drug j in the first (t1) and last (t2) years. Usej is the utilization rate of drug j at t1 and t2. The mortality effect of drug classes that were found to have adverse effects was signified by a negative value of ABj. The mortality benefits of drug therapies were assumed to apply only to heart failure patients with abnormal left ventricular systolic function, and the proportion of patients with this condition was estimated from published literature. We then calculated the proportion of the reduction in number of deaths from 1992/1993 to 1999/2000 that could be attributed to the change in drug therapy over time.
Effect of Randomized Trials on Trends in the Utilization of Heart Failure Drugs We conducted an exploratory analysis of the effect of randomized trials published during the study period using time series analysis, a collection of techniques for modeling autocorrelation in temporally sequenced data (17). We used multivariate Box-Jenkins autoregressive integrated moving average intervention models to assess early effects and simple autoregressive models to examine sustained effects of randomized trials (18,19). The presence of white noise was assessed by examining the auto-
Trends in Heart Failure Outcomes and Therapy/Lee et al
Table 1. Trends in New Hospitalizations for Heart Failure and Outcomes in Ontario, Canada (April 1992 to March 2000)
Year
Number of Heart Failure Admissions
Number Discharged Alive
Number of In-Hospital Deaths
Number of 1-Year Deaths (%)
Number Readmitted at 1 Year (%*)
1992 1993 1994 1995 1996 1997 1998 1999
11,084 11,432 11,177 10,891 11,049 11,265 10,858 10,684
9689 9983 9758 9560 9774 9889 9402 9366
1395 1449 1419 1331 1275 1376 1456 1318
4044 (36.5) 4056 (35.5) 3983 (35.6) 3932 (36.1) 3905 (35.3) 4033 (35.8) 3868 (35.6) 3759 (35.2)
2818 (29.1) 2682 (26.9) 2658 (27.2) 2563 (26.8) 2522 (25.8) 2584 (26.1) 2372 (25.2) 2355 (25.1)
* Denominator ⫽ those discharged alive.
correlations at various lags using the Ljung-Box chisquared statistic (20). Analyses were adjusted for age, sex, and the Deyo-Charlson Comorbidity score (11).
Statistical Analysis Continuous variables were reported as means (⫾ SD) and compared using the Student t test. All years are reported as fiscal year, beginning on April 1 and ending on March 31 of the following year. Gaussian smoothing of temporal trends was conducted using S-Plus, version 6.1 (Seattle, Washington). Analyses were conducted using SAS statistical software, version 8.0 (SAS Institute, Cary, North Carolina).
RESULTS Trends in Hospitalizations, Demographics, and Mortality Between April 1992 and March 2000, 88,440 patients aged ⱖ65 years were hospitalized for heart failure in Ontario, of whom 77,421 were discharged alive (12.5% inhospital mortality). The numbers of yearly admissions were stable (Table 1). Although crude in-hospital mortality did not change appreciably, 1-year mortality decreased from the initial year to 35.2% in the last year of analysis. During this time, there was also a reduction in the number of 1-year rehospitalizations for heart failure. The mean age of the entire cohort at the time of index hospitalization was 79 years (women: 80 years; men: 77 years). Women comprised 53% of the cohort, and the sex distribution was similar in the first and last years of the study (P ⫽ 0.10). Mean (⫾ SD) age increased slightly from 78.4 ⫾ 0.2 years in the initial study year to 78.9 ⫾ 0.4 years in the last year (P ⬍0.001). Comorbid conditions also increased over time, with an increase in the average Charlson score from 0.87 ⫾ 0.04 in the first year to 1.01 ⫾ 0.06 in the last year (P ⬍0.001).
Trends in Mortality and Morbidity The decline in mortality and the concomitant increase in age and comorbidity burden of heart failure patients were
reflected in greater reductions in adjusted 1-year mortality rates than unadjusted rates. Risk-adjusted 1-year mortality decreased by 2.8%, and the crude rates decreased by 1.3%. Thirty-day mortality did not decrease among all patients (P ⫽ 0.9 for linear trend) or in hospital survivors (P ⫽ 0.10; Figure 1). There was, however, a significant decline in 1-year mortality over time. Adjusted 1-year mortality decreased by 0.024% per month among hospital survivors (P ⬍0.001), and by 0.029% per month in all patients including in-hospital deaths (P ⬍0.001). During this time, adjusted 1-year rehospitalization rates also decreased from 29.1% to 25.0%, an absolute 4.1% reduction in adjusted and 4.0% reduction in unadjusted rehospitalization rates.
Trends in Heart Failure Drug Utilization The use of ACE inhibitors or angiotensin receptor blockers increased significantly (P ⫽ 0.001) in the last year of the study period relative to the initial year (Table 2; Figure 2); however, the largest increase in utilization was with beta-blockers, which increased by 15.6% (P ⬍0.001; Table 2; Figure 3).
Effect of Drug Treatments on Mortality and Morbidity Rates Of the published reports of community-based heart failure patients with data on left ventricular systolic function, the largest study (n ⫽ 19,710), which had a demographically similar study sample as ours, reported that 66% of patients had impaired systolic function (21). Review of five smaller studies yielded a weighted average of 53% with low ejection fraction (22–26). Given the similarity between our study sample and the former study, we assumed that 66% of heart failure admissions would have low ejection fraction in the primary analysis and a 53% rate in the secondary analysis. Therapies that have been demonstrated to have a clear mortality benefit in heart failure include ACE inhibitors and beta-blockers (Table 3). The suggested absolute reductions in mortality from the meta-analyses were 4% with ACE inhibitors and 3% with beta-blockers (27,28). May 1, 2004
THE AMERICAN JOURNAL OF MEDICINE威
Volume 116 583
Trends in Heart Failure Outcomes and Therapy/Lee et al
Figure 1. Thirty-day and 1-year mortality among hospital survivors and all patients from April 1992 to March 2000. Thirty-day mortality rates (among hospital survivors; solid triangles), adjusted for age, sex, and Deyo-Charlson Comorbidity score, demonstrated no linear trend over time (P ⫽ 0.10). Adjusted 30-day mortality rates including in-hospital deaths (open triangles) also exhibited no trend over time (P ⫽ 0.9). However, there was a decreasing trend in 1-year mortality among patients who survived to hospital discharge (solid squares) and among all patients including in-hospital deaths (open squares) (both P ⬍0.001).
A comparative meta-analysis of angiotensin receptor blockers versus ACE inhibitors found no mortality differences (32), therefore we assumed that mortality effects would be similar for both drug classes. Based on the absolute effects from published meta-analyses and assuming that 66% of hospitalized heart failure patients aged ⱖ65 years have left ventricular systolic dysfunction, 30%
of the improvement in mortality rates can be explained by increased beta-blocker use over time. In addition, 5.9% of the improvement in mortality can be explained by the increase in ACE inhibitor/angiotensin receptor blocker utilization. Assuming that the distribution of patients was toward fewer patients with low ejection fraction, then 24% of the observed reduction in mortality can be ex-
Table 2. Utilization Rates of Drug Therapies in Elderly Heart Failure Patients Over Time Drug Class
Fiscal 1992/1993
Fiscal 1999/2000
P Value
Mean Utilization Rate (95% Confidence Interval) Beta adrenergic blockers Calcium channel blockers* Digitalis preparations Antiarrhythmic agents† ACE inhibitors ACE inhibitor or angiotensin receptor blocker
5.9 (3.9–7.9) 21.3 (19.7–22.9) 48.6 (45.3–51.9) 3.5 (2.1–4.9) 57.8 (52.3–63.3) 57.8 (52.3–63.3)
* First-generation calcium channel blockers. † Vaughan-Williams class I antiarrhythmic agents. ACE ⫽ angiotensin-converting enzyme. 584
May 1, 2004
THE AMERICAN JOURNAL OF MEDICINE威
Volume 116
21.5 (17.2–25.8) 9.6 (7.8–11.4) 34.5 (29.8–39.2) 1.4 (0.6–2.2) 59.1 (55.0–63.2) 61.7 (56.8–66.6)
⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 0.2 0.001
Trends in Heart Failure Outcomes and Therapy/Lee et al
Figure 2. Utilization of ACE inhibitors or angiotensin receptor blockers (A), digitalis preparations (B), class I antiarrhythmic drugs (C), and first-generation calcium channel blockers (D) in heart failure patients from April 1992 to March 2000 (thin lines). Thick lines represent smoothed estimates for each drug class. The vertical axes represent the percentage of all heart failure patients prescribed and obtaining the drug within 30 days of hospital discharge. ACE ⫽ angiotensin-converting enzyme.
plained by increased use of beta-blockers and 4.8% of the mortality reduction by increased use of ACE inhibitors/ angiotensin receptor blockers. Meta-analyses of class I antiarrhythmic agents have suggested that these drugs increase mortality by 0.7% (29). The decrease in use of this drug class over time may explain 0.7% of the decline in mortality rates during the study period. A meta-analysis suggested no effect on mortality with digitalis (30), and two randomized trials of first-generation calcium channel blockers did not find notable detrimental effects on mortality (31,33). Several studies have demonstrated an effect on morbidity in heart failure patients (Table 3). The absolute reductions in hospitalization with ACE inhibitors and beta-blockers suggested by meta-analyses were 4% for both drug classes (27,28). A meta-analysis of angiotensin receptor blockers versus ACE inhibitors suggested no significant difference in hospitalization, which suggests that their effects would be similar (32). Based on the absolute effects above, ACE inhibitors/angiotensin receptor blockers can explain 2.6% of the decline in hospitalization rates over time, whereas beta-blockers can explain
10.4% of the decline. Assuming that 53% of patients had low ejection fraction, then the morbidity reductions explained by therapy decrease slightly to 2.1% with ACE inhibitors/angiotensin receptor blockers and to 8.5% with beta-blockers. Digitalis was found to decrease the risk of rehospitalization by 8% (30), and the decrease in utilization of this class of drugs may have increased the number of hospitalizations (up to 70 additional events). In contrast, reduction in the use of diltiazem, verapamil, and nifedipine could decrease rehospitalization. Patients with reduced ejection fraction in the Multicenter Diltiazem Postinfarction Trial had an 8.6% increase in heart failure (31). Therefore, decreased utilization of this latter group can explain 17% of the decline in hospitalization rates over time, and 14% of the decline if a lower proportion of elderly patients had systolic dysfunction.
Effect of Randomized Trials on Drug Utilization Rates Publication of the results of randomized trials did not have a detectable immediate effect on the use of digitalis May 1, 2004
THE AMERICAN JOURNAL OF MEDICINE威
Volume 116 585
Trends in Heart Failure Outcomes and Therapy/Lee et al
Figure 3. Thirty-day utilization rate of beta-blockers in heart failure patients over time. The inset shows smoothed estimates of drug utilization (inset) over time. The dashed vertical bar represents the time of sequential release of the U.S. Carvedilol and Australia– New Zealand trials. The observed data (points shown as stars) before the intervention were used to generate a predicted utilization curve and upper and lower 95% percent confidence intervals (CI). The observed utilization of beta-blockers after the trials (shown as stars) increased significantly more than predicted (shown as crosses), suggesting a sustained effect (P ⫽ 0.001). RCT ⫽ randomized controlled trial.
(DIG), ACE inhibitors or angiotensin receptor blockers (ELITE-I), and beta-blockers (U.S. Carvedilol and Australia–New Zealand trials, all P ⬎0.20). A sustained analysis demonstrated that after publication of the DIG results, there was little effect on the rate of digitalis utilization (P ⫽ 0.6). Indeed, the rate of digitalis utilization over time was ⫺1.7 per 1000 discharges per month (95% confidence interval [CI]: ⫺2.3 to ⫺1.0 per 1000 discharges per month) after the DIG trial. After publication of the ELITE-I results, the slope of the age-/sex-adjusted utilization plots for ACE inhibitors or angiotensin receptor blockers was ⫹0.1 per 1000 discharges per month (95% CI: ⫺0.6 to ⫹0.8 per 1000 discharges), and was not increased significantly when compared with the period prior to the trial release. However, a sustained positive effect on beta-blocker use was observed after publication of the randomized trials of beta-blockers (P ⫽ 0.001; Figure 3). The slope of age-/sex-adjusted beta-blocker utilization after the randomized trials
586
May 1, 2004
THE AMERICAN JOURNAL OF MEDICINE威
Volume 116
was ⫹3.1 prescriptions per 1000 discharges per month (95% CI: ⫹2.6 to ⫹3.6 prescriptions per 1000 discharges per month). Assuming that the effect of the randomized trials occurred at the time of presentation at scientific meetings or at the time of regulatory approval did not alter results.
DISCUSSION In this study, we found that trends in drug therapy were concordant with the weight of scientific evidence from randomized trials. The use of first-generation calcium channel blockers decreased by 11.7% while use of class I antiarrhythmics decreased by 2.1%. There was a 3.9% increase in ACE inhibitor or angiotensin antagonist use over time, reaching steady state. However, the most striking increase occurred with beta-blocker use, which increased by 15.6%. At the same time, there was no significant change in 30-day mortality rates, and thus the
Trends in Heart Failure Outcomes and Therapy/Lee et al
Table 3. Published Reviews Summarizing Treatment Effects in Heart Failure Patients Drug Class
Total Number of Patients
Mortality
Hospitalization
First Author (Reference)
Odds Ratio (95% Confidence Interval) ACE inhibitors Beta-blockers Antiarrhythmics (class I) Digitalis Calcium channel antagonists
7105 10,135 23,229 7744 623
0.77 (0.67–0.88) 0.65 (0.53–0.80) 1.14 (1.01–1.28) 0.99 (0.89–1.09) —
0.65 (0.57–0.74) 0.64 (0.53–0.79) — 0.68 (0.61–0.75) 1.90*
Garg (27) Brophy (28) Teo (29) Hood (30) Goldstein (31)
* 95% confidence interval not reported. ACE ⫽ angiotensin-converting enzyme.
trends in drug utilization over time were not attributable to survivor selection bias (34). During the study period, there was improvement in outcomes of mortality and rehospitalization in elderly patients admitted for their first heart failure episode. The decreased unadjusted mortality rate and concomitant increase in age and comorbidity burden resulted in a decrease of 2.8% in the 1-year risk-adjusted mortality rate. We also found improvement in morbidity, with a 4.1% reduction in 1-year hospitalization rates. Our findings suggest that increased use of angiotensin receptor blockers and ACE inhibitors contributed to the observed reductions in mortality and morbidity during this time. However, the greatest contributor to decreased mortality and hospitalization rates was treatment with beta-blockers, which increased temporally after the publication of randomized trials showing benefit in heart failure patients. Other studies have reported similar trends in heart failure outcomes over time. Levy et al (35) reported improved trends in heart failure survival from 1950 to 1999 in the Framingham study. Using administrative data similar to the data source used in this study from 1986 to 1995, the absolute reduction in case fatality rates was also found to have improved over time in Scotland among patients of all ages (9). Two studies of hospital-associated mortality conducted during a similar time frame as the above also found that despite increased heart failure severity, outcomes improved over time (36,37). However, these prior studies did not examine the effect of temporal changes in the utilization of drug therapies. Our study extends upon prior research by examining the effect of drug therapies on mortality and rehospitalization outcomes in a large sample of heart failure patients. We observed increased use of effective treatments, and decreased use of drugs of no mortality benefit, over time. When compared with population-based trends of adult mortality in the United States (38), which have demonstrated far smaller reductions from 1992 to 1999, our findings suggest that changes in
medical treatment have contributed to the overall reduction in mortality and morbidity in patients with heart failure. Although beta-blocker use increased markedly in the latter part of the study, improvement in adjusted mortality rates were observed throughout the entire time period. This may have been due to increased use of ACE inhibitors and other nonpharmacologic approaches to the management of heart failure patients. Importantly, our findings also suggest that despite improvements in drug treatments, there are still large numbers of elderly patients with heart failure who are not receiving the benefits of angiotensin pathway antagonism or betablockade, and therefore the ultimate effect of drug therapy on outcomes could potentially be even greater than we have observed. Our study had a number of limitations. The study was limited to elderly patients because of age restrictions in the drug reimbursement formulary. However, heart failure is primarily a disease of the elderly, with over 80% of admissions for heart failure involving elderly patients (39). A presumption inherent in our analyses was that drug utilization was not polarized among the low or preserved ejection fraction groups, and thus an even distribution was assumed. Owing to the paucity of randomized trials involving patients with heart failure and preserved ejection fraction, we conservatively assumed that the drug classes studied did not have benefits in these patients. However, if the benefits extend to all heart failure patients regardless of ejection fraction, the contributions of these drugs to improvements in outcomes over time would be underestimated. We were unable to address the issue of appropriateness of drug treatment owing to the absence of relevant clinical data on contraindications to drug therapy. Our study could not determine whether the observed trends were due to improved prognosis among those with low or preserved ejection fraction (or both). Finally, the adjusted analyses did not account for severity of illness, which may have increased over the years with trends toward more outpatient care. If this was the case, then our estimates of changes in mortality rates over time may be underestimated.
May 1, 2004
THE AMERICAN JOURNAL OF MEDICINE威
Volume 116 587
Trends in Heart Failure Outcomes and Therapy/Lee et al
In summary, from 1992 to 2000 in Ontario, there was an increase in the use of therapies that had demonstrated clinical efficacy in randomized trials of heart failure. Conversely, there was decreased utilization of drug treatments that had no demonstrated mortality benefit. These trends in drug therapy contributed to the favorable trends in mortality and rehospitalizations observed among elderly heart failure patients. Still, despite the improvement in heart failure outcomes, the maximal potential benefits of existing effective therapies have not yet been attained.
REFERENCES 1. American Heart Association. Heart Disease and Stroke Statistics—2003 Update. Dallas, Texas: American Heart Association; 2002. 2. Braunwald E. Shattuck lecture— cardiovascular medicine at the turn of the millennium: triumphs, concerns, and opportunities. N Engl J Med. 1997;337:1360 –1369. 3. CONSENSUS Trial Study Group. Effects of enalapril on mortality in severe congestive heart failure: results of the Cooperative North Scandinavian Enalapril Survival Study (CONSENSUS). N Engl J Med. 1987;316:1429 –1435. 4. SOLVD Investigators. Effect of enalapril on survival in patients with reduced left ventricular ejection fractions and congestive heart failure. N Engl J Med. 1991;325:293–302. 5. International Classification of Diseases, Ninth Revision, Clinical Modification. Washington, D.C.: Public Health Service, U.S. Department of Health and Human Services; 1988. 6. Tu JV, Zhang H. Congestive heart failure outcomes in Ontario. In: Naylor C, Slaughter P, eds. Cardiovascular Health and Services: Technical and Methods Appendices. Toronto, Canada: Institute for Clinical Evaluative Sciences; 1999:25–30. 7. Weil E, Tu JV. Quality of congestive heart failure treatment at a Canadian teaching hospital. CMAJ. 2001;165:284 –287. 8. Jong P, Gong Y, Liu PP, et al. Care and outcomes of patients newly hospitalized for heart failure in the community treated by cardiologists compared with other specialists. Circulation. 2003;108:184 – 191. 9. MacIntyre K, Capewell S, Stewart S, et al. Evidence of improving prognosis in heart failure: trends in case fatality in 66 547 patients hospitalized between 1986 and 1995. Circulation. 2000;102:1126 – 1131. 10. Jong P, Vowinckel E, Liu PP, et al. Prognosis and determinants of survival in patients newly hospitalized for heart failure: a population-based study. Arch Intern Med. 2002;162:1689 –1694. 11. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45:613–619. 12. Australia/New Zealand Heart Failure Research Collaborative Group. Randomised, placebo-controlled trial of carvedilol in patients with congestive heart failure due to ischaemic heart disease. Lancet. 1997;349:375–380. 13. Pitt B, Segal R, Martinez FA, et al. Randomised trial of losartan versus captopril in patients over 65 with heart failure (Evaluation of Losartan in the Elderly Study, ELITE). Lancet. 1997;349:747–752. 14. Packer M, Bristow MR, Cohn JN, et al. The effect of carvedilol on morbidity and mortality in patients with chronic heart failure: U.S. Carvedilol Heart Failure Study Group. N Engl J Med. 1996;334: 1349 –1355. 588
May 1, 2004
THE AMERICAN JOURNAL OF MEDICINE威
Volume 116
15. Digitalis Investigation Group. The effect of digoxin on mortality and morbidity in patients with heart failure. N Engl J Med. 1997; 336:525–533. 16. Heidenreich PA, McClellan M. Trends in treatment and outcomes for acute myocardial infarction: 1975-1995. Am J Med. 2001;110: 165–174. 17. Pindyck RS, Rubinfeld DL. Econometric Models and Economic Forecasts. 4th ed. New York, New York: Irwin McGraw-Hill; 1998. 18. Diggle PJ. Time Series: A Biostatistical Introduction. New York, New York: Oxford University Press; 1990. 19. Yaffee R. Introduction to Time Series Analysis and Forecasting. San Diego, California: Academic Press; 2000. 20. Ljung GM, Box GEP. On a measure of lack of fit in time series models. Biometrika. 1978;65:297–303. 21. Masoudi FA, Havranek EP, Smith G, et al. Gender, age, and heart failure with preserved left ventricular systolic function. J Am Coll Cardiol. 2003;41:217–223. 22. Krumholz HM, Wang Y, Parent EM, et al. Quality of care for elderly patients hospitalized with heart failure. Arch Intern Med. 1997;157: 2242–2247. 23. Philbin EF, Rocco TA Jr, Lindenmuth NW, et al. Systolic versus diastolic heart failure in community practice: clinical features, outcomes, and the use of angiotensin-converting enzyme inhibitors. Am J Med. 2000;109:605–613. 24. Redfield MM, Jacobsen SJ, Burnett JC Jr, et al. Burden of systolic and diastolic ventricular dysfunction in the community: appreciating the scope of the heart failure epidemic. JAMA. 2003;289:194 – 202. 25. Senni M, Tribouilloy CM, Rodeheffer RJ, et al. Congestive heart failure in the community: a study of all incident cases in Olmsted County, Minnesota, in 1991. Circulation. 1998;98:2282– 2289. 26. Vasan RS, Larson MG, Benjamin EJ, et al. Congestive heart failure in subjects with normal versus reduced left ventricular ejection fraction: prevalence and mortality in a population- based cohort. J Am Coll Cardiol. 1999;33:1948 –1955. 27. Garg R, Yusuf S. Overview of randomized trials of angiotensinconverting enzyme inhibitors on mortality and morbidity in patients with heart failure: collaborative group on ACE inhibitor trials. JAMA. 1995;273:1450 –1456. 28. Brophy JM, Joseph L, Rouleau JL. Beta-blockers in congestive heart failure. A Bayesian meta-analysis. Ann Intern Med. 2001;134:550 – 560. 29. Teo KK, Yusuf S, Furberg CD. Effects of prophylactic antiarrhythmic drug therapy in acute myocardial infarction. An overview of results from randomized controlled trials. JAMA. 1993;270:1589 – 1595. 30. Hood WB Jr, Dans AL, Guyatt GH, Jaeschke R, McMurray JJV. Digitalis for treatment of congestive heart failure in patients in sinus rhythm (Cochrane Review). In: The Cochrane Library. Issue 3. Oxford, United Kingdom: Update Software; 2003. 31. Goldstein RE, Boccuzzi SJ, Cruess D, et al. Diltiazem increases lateonset congestive heart failure in postinfarction patients with early reduction in ejection fraction. Circulation. 1991;83:52–60. 32. Jong P, Demers C, McKelvie RS, Liu PP. Angiotensin receptor blockers in heart failure: meta-analysis of randomized controlled trials. J Am Coll Cardiol. 2002;39:463–470. 33. Elkayam U, Amin J, Mehra A, et al. A prospective, randomized, double-blind, crossover study to compare the efficacy and safety of chronic nifedipine therapy with that of isosorbide dinitrate and their combination in the treatment of chronic congestive heart failure. Circulation. 1990;82:1954 –1961.
Trends in Heart Failure Outcomes and Therapy/Lee et al 34. Glesby MJ, Hoover DR. Survivor treatment selection bias in observational studies: examples from the AIDS literature. Ann Intern Med. 1996;124:999 –1005. 35. Levy D, Kenchaiah S, Larson MG, et al. Long-term trends in the incidence of and survival with heart failure. N Engl J Med. 2002;347: 1397–1402. 36. Ni H, Nauman DJ, Hershberger RE. Analysis of trends in hospitalizations for heart failure. J Card Fail. 1999;5:79 –84. 37. Polanczyk CA, Rohde LE, Dec GW, DiSalvo T. Ten-year trends
in hospital care for congestive heart failure: improved outcomes and increased use of resources. Arch Intern Med. 2000;160:325– 332. 38. Hoyert DL, Arias E, Smith BL, et al. Deaths: final data for 1999. Natl Vital Stat Rep. 2001;49:1–113. 39. Tu JV, Zhang H. Congestive heart failure outcomes in Ontario. In: Naylor CD, Slaughter PM, eds. Cardiovascular Health and Services in Ontario: An ICES Atlas. Toronto, Canada: Institute for Clinical Evaluative Sciences; 1999:111–122.
May 1, 2004
THE AMERICAN JOURNAL OF MEDICINE威
Volume 116 589
Trends in Heart Failure Outcomes and Pharmacotherapy: 1992 to 2000 Douglas S. Lee, MD, Muhammad M. Mamdani, PharmD, MA, MPH, Peter C. Austin, PhD, Yanyan Gong, MSc, Peter P. Liu, MD, Jean L. Rouleau, MD, Jack V. Tu, MD, PhD PURPOSE: To review trends in drug therapy and concomitant outcomes of elderly heart failure patients in Ontario, Canada. METHODS: Utilization of drug therapies, mortality, and rehospitalization rates from April 1992 to March 2000 were determined in 77,421 elderly (aged ⱖ65 years), community-based heart failure patients using linked administrative databases. Treatment effects were identified from published meta-analyses and randomized trials. The effect of drug trends on mortality and morbidity were assessed based on their absolute treatment effects. RESULTS: From 1992 to 2000, angiotensin-converting enzyme inhibitor or angiotensin receptor blocker use increased from 58% to 62% (P ⫽ 0.001) while beta-blocker use increased from 6% to 22% (P ⬍0.001). There was a decrease in the use of treatments for which no survival advantage had been demonstrated in randomized trials, such as digitalis (49% to 35%, P
⬍0.001), Vaughan-Williams class I antiarrhythmic agents (3.5% to 1.4%, P ⬍0.001), and first-generation calcium antagonists (21.3% to 9.6%, P ⬍0.001). The trends in drug therapy were associated with a 2.8% reduction in age-, sex-, and comorbidity-adjusted 1-year mortality and a 4.1% reduction in 1-year hospitalization rates. The observed trends in therapy over time explained 37% of the decrease in mortality and 30% of the decrease in rehospitalization rates. The treatment effect from beta-blockers was most pronounced, explaining 30% of the decrease in mortality and 10% of the decrease in rehospitalization rates. CONCLUSION: During 1992 to 2000, mortality and morbidity improved among elderly patients with heart failure, with increased utilization of beta-blockers contributing most to the beneficial trends in outcomes. Am J Med. 2004;116:581–589. ©2004 by Excerpta Medica Inc.
H
apy (2). Since the publication of several landmark randomized trials of angiotensin-converting enzyme (ACE) inhibitors beginning in the late 1980s, several studies have demonstrated the effects of drug intervention on mortality and morbidity in heart failure patients (3,4). However, changes in drug treatments for community-based heart failure patients and their effects on clinical outcomes have not been described. In this study, we reviewed trends in drug therapy and outcomes in a large sample of heart failure patients in Ontario, Canada. We estimated the effect of treatments initiated after a hospitalization episode on the trends in mortality and rehospitalization rates from 1992 to 2000. We also explored the effect of published randomized heart failure trials on temporal changes in drug utilization patterns.
eart failure affects approximately 6 million persons in North America, where it is a leading cause of hospitalization (1). Over the past two decades, hospital discharges have increased as have the number of deaths attributed to heart failure (1). Although early treatment was directed primarily towards the relief of symptoms, the cornerstone of current heart failure management is based on modification of the natural history of the condition using pharmacologic therFrom the Institute for Clinical Evaluative Sciences (DSL, MMM, PCA, YG, JVT); and the Department of Health Policy, Management and Evaluation/Clinical Epidemiology (DSL, JVT), the Faculty of Pharmacy (MMM), the Department of Public Health Sciences (PCA), the Heart and Stroke/Richard Lewar Centre of Excellence (PPL), the Division of Cardiology, University Health Network, Toronto General Hospital (PPL, JLR), and the Division of General Internal Medicine, Sunnybrook and Women’s College Health Science Centre (JVT), University of Toronto, Toronto, Canada. This study was funded by grants from the Ontario Ministry of Health (Ontario Program for Optimal Therapeutics), and to the Canadian Cardiovascular Outcomes Research Team from the Canadian Institutes of Health Research and Heart and Stroke Foundation. Dr. Lee is supported by a research fellowship from the Heart and Stroke Foundation of Canada/Canadian Institutes of Health Research. Dr. Tu is supported by a Canada Research Chair in Health Services Research. Requests for reprints should be addressed to Jack V. Tu, MD, PhD, Institute for Clinical Evaluative Sciences, Room G-106, 2075 Bayview Avenue, Toronto, Ontario M4N 3M5, Canada, or [email protected]. Manuscript submitted May 13, 2003, and accepted in revised form November 26, 2003. © 2004 by Excerpta Medica Inc. All rights reserved.
METHODS Sources of Data We linked data elements from a number of administrative data sources to identify heart failure patients and medications utilized over time. From the Canadian Institute of Health Information hospital discharge abstracts, which contain information on all hospital separations (discharges, transfers, and deaths) in Ontario, data per0002-9343/04/$–see front matter 581 doi:10.1016/j.amjmed.2003.11.025
Trends in Heart Failure Outcomes and Therapy/Lee et al
taining to the index admission, demographic characteristics, and subsequent heart failure hospitalizations were identified. Prescription drug information and dispensation dates for patients aged ⱖ65 years were identified from the Ontario Drug Benefit database. Data on mortality were gathered from the Registered Persons Vital Statistics database. The cohort consisted of all patients ⱖ65 years of age, admitted to a hospital in Ontario with a primary/most responsible diagnosis of heart failure (code 428) according to the International Classification of Diseases, Ninth Revision, Clinical Modification (5). Audits of coding have shown a high degree of accuracy in the database for heart failure diagnosis (6 – 8). The cohort comprised newly admitted patients, as previously defined (9,10). Other exclusion criteria were no admittance to an acute care hospital, transfer from another acute care facility, age ⱖ105 years, not a resident of Ontario, invalid health card number, and heart failure coded as an in-hospital complication (6). The study was approved by the ethics review board of Sunnybrook and Women’s College Health Sciences Centre.
Trends in Incidence, Mortality, and Morbidity Incident heart failure cases were identified from April 1992 to March 2000, and followed for at least 1 year until March 2001, the last date of follow-up. Each index admission was categorized in chronological order according to the month and year of hospital discharge. We identified the vital status of all patients after admission and whether patients were rehospitalized for heart failure within 1 year of hospital discharge. Mortality rates were adjusted for age, sex, and comorbid conditions using the Deyo-Charlson Comorbidity Index. Linear trends were assessed using adjusted 30-day and 1-year mortality rates as the dependent variable and month of the series as the independent variable (11). Separate mortality analyses were conducted in all patients and hospital survivors. We used similar methods to evaluate unadjusted and adjusted 1-year rehospitalization rates.
Trends in Heart Failure Drug Therapy The drug classes evaluated were beta-adrenergic blockers, ACE inhibitors with angiotensin receptor blockers as an alternative, first-generation calcium channel blockers (e.g., verapamil, diltiazem, nifedipine), class I antiarrhythmic drugs, and digitalis. Drugs within the ACE inhibitor class included benazepril, captopril, cilazapril, enalapril, fosinopril, lisinopril, perindopril, quinapril, and ramipril. Beta-blockers included acebutolol, atenolol, carvedilol, labetolol, metoprolol, nadolol, oxprenolol, pindolol, propranolol, and timolol. Class I antiarrhythmic drugs included disopyramide, flecainide, mexiletine, procainamide, propafenone, quinidine, and tocainide (12). We also considered whether patients received at least one prescription of an ACE inhibitor or 582
May 1, 2004
THE AMERICAN JOURNAL OF MEDICINE威
Volume 116
angiotensin receptor blocker to account for ACE inhibitor intolerance. Examination of drug utilization trends was conducted by comparing utilization rates in the initial and final years of the series. Binary outcome variables were assigned for the drug classes based on prescription status within 30 days of discharge. The average monthly rate in the first year was compared with the rate in the last year of the series. Descriptive analysis of trends was by linear regression, with utilization rate as the dependent variable and month of the series as the independent variable. When a randomized controlled trial was published during the study years, the rates of utilization were described before and after the trial. Three groups of potentially relevant trials that reported mortality outcomes (and their release dates) were identified (12–15): Evaluation of Losartan in the Elderly (ELITE)-I study (March 1997), the Australia– New Zealand (February 1997) and U.S. Carvedilol (May 1996) trials, and the Digitalis Investigation Group (DIG) trial (February 1997).
Contribution of Treatments to Changes in Mortality The contributions of each drug class to the reduction in mortality and hospitalization rates were estimated from the absolute mortality benefits reported from published meta-analyses or randomized trials (if the former were not available). We estimated the contribution of each treatment over time using previously described methods (16). Briefly, mortality benefit was calculated as ABj(t2) * Usej(t2) – ABj(t1) * Usej(t1), where ABj is the absolute benefit of drug j in the first (t1) and last (t2) years. Usej is the utilization rate of drug j at t1 and t2. The mortality effect of drug classes that were found to have adverse effects was signified by a negative value of ABj. The mortality benefits of drug therapies were assumed to apply only to heart failure patients with abnormal left ventricular systolic function, and the proportion of patients with this condition was estimated from published literature. We then calculated the proportion of the reduction in number of deaths from 1992/1993 to 1999/2000 that could be attributed to the change in drug therapy over time.
Effect of Randomized Trials on Trends in the Utilization of Heart Failure Drugs We conducted an exploratory analysis of the effect of randomized trials published during the study period using time series analysis, a collection of techniques for modeling autocorrelation in temporally sequenced data (17). We used multivariate Box-Jenkins autoregressive integrated moving average intervention models to assess early effects and simple autoregressive models to examine sustained effects of randomized trials (18,19). The presence of white noise was assessed by examining the auto-
Trends in Heart Failure Outcomes and Therapy/Lee et al
Table 1. Trends in New Hospitalizations for Heart Failure and Outcomes in Ontario, Canada (April 1992 to March 2000)
Year
Number of Heart Failure Admissions
Number Discharged Alive
Number of In-Hospital Deaths
Number of 1-Year Deaths (%)
Number Readmitted at 1 Year (%*)
1992 1993 1994 1995 1996 1997 1998 1999
11,084 11,432 11,177 10,891 11,049 11,265 10,858 10,684
9689 9983 9758 9560 9774 9889 9402 9366
1395 1449 1419 1331 1275 1376 1456 1318
4044 (36.5) 4056 (35.5) 3983 (35.6) 3932 (36.1) 3905 (35.3) 4033 (35.8) 3868 (35.6) 3759 (35.2)
2818 (29.1) 2682 (26.9) 2658 (27.2) 2563 (26.8) 2522 (25.8) 2584 (26.1) 2372 (25.2) 2355 (25.1)
* Denominator ⫽ those discharged alive.
correlations at various lags using the Ljung-Box chisquared statistic (20). Analyses were adjusted for age, sex, and the Deyo-Charlson Comorbidity score (11).
Statistical Analysis Continuous variables were reported as means (⫾ SD) and compared using the Student t test. All years are reported as fiscal year, beginning on April 1 and ending on March 31 of the following year. Gaussian smoothing of temporal trends was conducted using S-Plus, version 6.1 (Seattle, Washington). Analyses were conducted using SAS statistical software, version 8.0 (SAS Institute, Cary, North Carolina).
RESULTS Trends in Hospitalizations, Demographics, and Mortality Between April 1992 and March 2000, 88,440 patients aged ⱖ65 years were hospitalized for heart failure in Ontario, of whom 77,421 were discharged alive (12.5% inhospital mortality). The numbers of yearly admissions were stable (Table 1). Although crude in-hospital mortality did not change appreciably, 1-year mortality decreased from the initial year to 35.2% in the last year of analysis. During this time, there was also a reduction in the number of 1-year rehospitalizations for heart failure. The mean age of the entire cohort at the time of index hospitalization was 79 years (women: 80 years; men: 77 years). Women comprised 53% of the cohort, and the sex distribution was similar in the first and last years of the study (P ⫽ 0.10). Mean (⫾ SD) age increased slightly from 78.4 ⫾ 0.2 years in the initial study year to 78.9 ⫾ 0.4 years in the last year (P ⬍0.001). Comorbid conditions also increased over time, with an increase in the average Charlson score from 0.87 ⫾ 0.04 in the first year to 1.01 ⫾ 0.06 in the last year (P ⬍0.001).
Trends in Mortality and Morbidity The decline in mortality and the concomitant increase in age and comorbidity burden of heart failure patients were
reflected in greater reductions in adjusted 1-year mortality rates than unadjusted rates. Risk-adjusted 1-year mortality decreased by 2.8%, and the crude rates decreased by 1.3%. Thirty-day mortality did not decrease among all patients (P ⫽ 0.9 for linear trend) or in hospital survivors (P ⫽ 0.10; Figure 1). There was, however, a significant decline in 1-year mortality over time. Adjusted 1-year mortality decreased by 0.024% per month among hospital survivors (P ⬍0.001), and by 0.029% per month in all patients including in-hospital deaths (P ⬍0.001). During this time, adjusted 1-year rehospitalization rates also decreased from 29.1% to 25.0%, an absolute 4.1% reduction in adjusted and 4.0% reduction in unadjusted rehospitalization rates.
Trends in Heart Failure Drug Utilization The use of ACE inhibitors or angiotensin receptor blockers increased significantly (P ⫽ 0.001) in the last year of the study period relative to the initial year (Table 2; Figure 2); however, the largest increase in utilization was with beta-blockers, which increased by 15.6% (P ⬍0.001; Table 2; Figure 3).
Effect of Drug Treatments on Mortality and Morbidity Rates Of the published reports of community-based heart failure patients with data on left ventricular systolic function, the largest study (n ⫽ 19,710), which had a demographically similar study sample as ours, reported that 66% of patients had impaired systolic function (21). Review of five smaller studies yielded a weighted average of 53% with low ejection fraction (22–26). Given the similarity between our study sample and the former study, we assumed that 66% of heart failure admissions would have low ejection fraction in the primary analysis and a 53% rate in the secondary analysis. Therapies that have been demonstrated to have a clear mortality benefit in heart failure include ACE inhibitors and beta-blockers (Table 3). The suggested absolute reductions in mortality from the meta-analyses were 4% with ACE inhibitors and 3% with beta-blockers (27,28). May 1, 2004
THE AMERICAN JOURNAL OF MEDICINE威
Volume 116 583
Trends in Heart Failure Outcomes and Therapy/Lee et al
Figure 1. Thirty-day and 1-year mortality among hospital survivors and all patients from April 1992 to March 2000. Thirty-day mortality rates (among hospital survivors; solid triangles), adjusted for age, sex, and Deyo-Charlson Comorbidity score, demonstrated no linear trend over time (P ⫽ 0.10). Adjusted 30-day mortality rates including in-hospital deaths (open triangles) also exhibited no trend over time (P ⫽ 0.9). However, there was a decreasing trend in 1-year mortality among patients who survived to hospital discharge (solid squares) and among all patients including in-hospital deaths (open squares) (both P ⬍0.001).
A comparative meta-analysis of angiotensin receptor blockers versus ACE inhibitors found no mortality differences (32), therefore we assumed that mortality effects would be similar for both drug classes. Based on the absolute effects from published meta-analyses and assuming that 66% of hospitalized heart failure patients aged ⱖ65 years have left ventricular systolic dysfunction, 30%
of the improvement in mortality rates can be explained by increased beta-blocker use over time. In addition, 5.9% of the improvement in mortality can be explained by the increase in ACE inhibitor/angiotensin receptor blocker utilization. Assuming that the distribution of patients was toward fewer patients with low ejection fraction, then 24% of the observed reduction in mortality can be ex-
Table 2. Utilization Rates of Drug Therapies in Elderly Heart Failure Patients Over Time Drug Class
Fiscal 1992/1993
Fiscal 1999/2000
P Value
Mean Utilization Rate (95% Confidence Interval) Beta adrenergic blockers Calcium channel blockers* Digitalis preparations Antiarrhythmic agents† ACE inhibitors ACE inhibitor or angiotensin receptor blocker
5.9 (3.9–7.9) 21.3 (19.7–22.9) 48.6 (45.3–51.9) 3.5 (2.1–4.9) 57.8 (52.3–63.3) 57.8 (52.3–63.3)
* First-generation calcium channel blockers. † Vaughan-Williams class I antiarrhythmic agents. ACE ⫽ angiotensin-converting enzyme. 584
May 1, 2004
THE AMERICAN JOURNAL OF MEDICINE威
Volume 116
21.5 (17.2–25.8) 9.6 (7.8–11.4) 34.5 (29.8–39.2) 1.4 (0.6–2.2) 59.1 (55.0–63.2) 61.7 (56.8–66.6)
⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 0.2 0.001
Trends in Heart Failure Outcomes and Therapy/Lee et al
Figure 2. Utilization of ACE inhibitors or angiotensin receptor blockers (A), digitalis preparations (B), class I antiarrhythmic drugs (C), and first-generation calcium channel blockers (D) in heart failure patients from April 1992 to March 2000 (thin lines). Thick lines represent smoothed estimates for each drug class. The vertical axes represent the percentage of all heart failure patients prescribed and obtaining the drug within 30 days of hospital discharge. ACE ⫽ angiotensin-converting enzyme.
plained by increased use of beta-blockers and 4.8% of the mortality reduction by increased use of ACE inhibitors/ angiotensin receptor blockers. Meta-analyses of class I antiarrhythmic agents have suggested that these drugs increase mortality by 0.7% (29). The decrease in use of this drug class over time may explain 0.7% of the decline in mortality rates during the study period. A meta-analysis suggested no effect on mortality with digitalis (30), and two randomized trials of first-generation calcium channel blockers did not find notable detrimental effects on mortality (31,33). Several studies have demonstrated an effect on morbidity in heart failure patients (Table 3). The absolute reductions in hospitalization with ACE inhibitors and beta-blockers suggested by meta-analyses were 4% for both drug classes (27,28). A meta-analysis of angiotensin receptor blockers versus ACE inhibitors suggested no significant difference in hospitalization, which suggests that their effects would be similar (32). Based on the absolute effects above, ACE inhibitors/angiotensin receptor blockers can explain 2.6% of the decline in hospitalization rates over time, whereas beta-blockers can explain
10.4% of the decline. Assuming that 53% of patients had low ejection fraction, then the morbidity reductions explained by therapy decrease slightly to 2.1% with ACE inhibitors/angiotensin receptor blockers and to 8.5% with beta-blockers. Digitalis was found to decrease the risk of rehospitalization by 8% (30), and the decrease in utilization of this class of drugs may have increased the number of hospitalizations (up to 70 additional events). In contrast, reduction in the use of diltiazem, verapamil, and nifedipine could decrease rehospitalization. Patients with reduced ejection fraction in the Multicenter Diltiazem Postinfarction Trial had an 8.6% increase in heart failure (31). Therefore, decreased utilization of this latter group can explain 17% of the decline in hospitalization rates over time, and 14% of the decline if a lower proportion of elderly patients had systolic dysfunction.
Effect of Randomized Trials on Drug Utilization Rates Publication of the results of randomized trials did not have a detectable immediate effect on the use of digitalis May 1, 2004
THE AMERICAN JOURNAL OF MEDICINE威
Volume 116 585
Trends in Heart Failure Outcomes and Therapy/Lee et al
Figure 3. Thirty-day utilization rate of beta-blockers in heart failure patients over time. The inset shows smoothed estimates of drug utilization (inset) over time. The dashed vertical bar represents the time of sequential release of the U.S. Carvedilol and Australia– New Zealand trials. The observed data (points shown as stars) before the intervention were used to generate a predicted utilization curve and upper and lower 95% percent confidence intervals (CI). The observed utilization of beta-blockers after the trials (shown as stars) increased significantly more than predicted (shown as crosses), suggesting a sustained effect (P ⫽ 0.001). RCT ⫽ randomized controlled trial.
(DIG), ACE inhibitors or angiotensin receptor blockers (ELITE-I), and beta-blockers (U.S. Carvedilol and Australia–New Zealand trials, all P ⬎0.20). A sustained analysis demonstrated that after publication of the DIG results, there was little effect on the rate of digitalis utilization (P ⫽ 0.6). Indeed, the rate of digitalis utilization over time was ⫺1.7 per 1000 discharges per month (95% confidence interval [CI]: ⫺2.3 to ⫺1.0 per 1000 discharges per month) after the DIG trial. After publication of the ELITE-I results, the slope of the age-/sex-adjusted utilization plots for ACE inhibitors or angiotensin receptor blockers was ⫹0.1 per 1000 discharges per month (95% CI: ⫺0.6 to ⫹0.8 per 1000 discharges), and was not increased significantly when compared with the period prior to the trial release. However, a sustained positive effect on beta-blocker use was observed after publication of the randomized trials of beta-blockers (P ⫽ 0.001; Figure 3). The slope of age-/sex-adjusted beta-blocker utilization after the randomized trials
586
May 1, 2004
THE AMERICAN JOURNAL OF MEDICINE威
Volume 116
was ⫹3.1 prescriptions per 1000 discharges per month (95% CI: ⫹2.6 to ⫹3.6 prescriptions per 1000 discharges per month). Assuming that the effect of the randomized trials occurred at the time of presentation at scientific meetings or at the time of regulatory approval did not alter results.
DISCUSSION In this study, we found that trends in drug therapy were concordant with the weight of scientific evidence from randomized trials. The use of first-generation calcium channel blockers decreased by 11.7% while use of class I antiarrhythmics decreased by 2.1%. There was a 3.9% increase in ACE inhibitor or angiotensin antagonist use over time, reaching steady state. However, the most striking increase occurred with beta-blocker use, which increased by 15.6%. At the same time, there was no significant change in 30-day mortality rates, and thus the
Trends in Heart Failure Outcomes and Therapy/Lee et al
Table 3. Published Reviews Summarizing Treatment Effects in Heart Failure Patients Drug Class
Total Number of Patients
Mortality
Hospitalization
First Author (Reference)
Odds Ratio (95% Confidence Interval) ACE inhibitors Beta-blockers Antiarrhythmics (class I) Digitalis Calcium channel antagonists
7105 10,135 23,229 7744 623
0.77 (0.67–0.88) 0.65 (0.53–0.80) 1.14 (1.01–1.28) 0.99 (0.89–1.09) —
0.65 (0.57–0.74) 0.64 (0.53–0.79) — 0.68 (0.61–0.75) 1.90*
Garg (27) Brophy (28) Teo (29) Hood (30) Goldstein (31)
* 95% confidence interval not reported. ACE ⫽ angiotensin-converting enzyme.
trends in drug utilization over time were not attributable to survivor selection bias (34). During the study period, there was improvement in outcomes of mortality and rehospitalization in elderly patients admitted for their first heart failure episode. The decreased unadjusted mortality rate and concomitant increase in age and comorbidity burden resulted in a decrease of 2.8% in the 1-year risk-adjusted mortality rate. We also found improvement in morbidity, with a 4.1% reduction in 1-year hospitalization rates. Our findings suggest that increased use of angiotensin receptor blockers and ACE inhibitors contributed to the observed reductions in mortality and morbidity during this time. However, the greatest contributor to decreased mortality and hospitalization rates was treatment with beta-blockers, which increased temporally after the publication of randomized trials showing benefit in heart failure patients. Other studies have reported similar trends in heart failure outcomes over time. Levy et al (35) reported improved trends in heart failure survival from 1950 to 1999 in the Framingham study. Using administrative data similar to the data source used in this study from 1986 to 1995, the absolute reduction in case fatality rates was also found to have improved over time in Scotland among patients of all ages (9). Two studies of hospital-associated mortality conducted during a similar time frame as the above also found that despite increased heart failure severity, outcomes improved over time (36,37). However, these prior studies did not examine the effect of temporal changes in the utilization of drug therapies. Our study extends upon prior research by examining the effect of drug therapies on mortality and rehospitalization outcomes in a large sample of heart failure patients. We observed increased use of effective treatments, and decreased use of drugs of no mortality benefit, over time. When compared with population-based trends of adult mortality in the United States (38), which have demonstrated far smaller reductions from 1992 to 1999, our findings suggest that changes in
medical treatment have contributed to the overall reduction in mortality and morbidity in patients with heart failure. Although beta-blocker use increased markedly in the latter part of the study, improvement in adjusted mortality rates were observed throughout the entire time period. This may have been due to increased use of ACE inhibitors and other nonpharmacologic approaches to the management of heart failure patients. Importantly, our findings also suggest that despite improvements in drug treatments, there are still large numbers of elderly patients with heart failure who are not receiving the benefits of angiotensin pathway antagonism or betablockade, and therefore the ultimate effect of drug therapy on outcomes could potentially be even greater than we have observed. Our study had a number of limitations. The study was limited to elderly patients because of age restrictions in the drug reimbursement formulary. However, heart failure is primarily a disease of the elderly, with over 80% of admissions for heart failure involving elderly patients (39). A presumption inherent in our analyses was that drug utilization was not polarized among the low or preserved ejection fraction groups, and thus an even distribution was assumed. Owing to the paucity of randomized trials involving patients with heart failure and preserved ejection fraction, we conservatively assumed that the drug classes studied did not have benefits in these patients. However, if the benefits extend to all heart failure patients regardless of ejection fraction, the contributions of these drugs to improvements in outcomes over time would be underestimated. We were unable to address the issue of appropriateness of drug treatment owing to the absence of relevant clinical data on contraindications to drug therapy. Our study could not determine whether the observed trends were due to improved prognosis among those with low or preserved ejection fraction (or both). Finally, the adjusted analyses did not account for severity of illness, which may have increased over the years with trends toward more outpatient care. If this was the case, then our estimates of changes in mortality rates over time may be underestimated.
May 1, 2004
THE AMERICAN JOURNAL OF MEDICINE威
Volume 116 587
Trends in Heart Failure Outcomes and Therapy/Lee et al
In summary, from 1992 to 2000 in Ontario, there was an increase in the use of therapies that had demonstrated clinical efficacy in randomized trials of heart failure. Conversely, there was decreased utilization of drug treatments that had no demonstrated mortality benefit. These trends in drug therapy contributed to the favorable trends in mortality and rehospitalizations observed among elderly heart failure patients. Still, despite the improvement in heart failure outcomes, the maximal potential benefits of existing effective therapies have not yet been attained.
REFERENCES 1. American Heart Association. Heart Disease and Stroke Statistics—2003 Update. Dallas, Texas: American Heart Association; 2002. 2. Braunwald E. Shattuck lecture— cardiovascular medicine at the turn of the millennium: triumphs, concerns, and opportunities. N Engl J Med. 1997;337:1360 –1369. 3. CONSENSUS Trial Study Group. Effects of enalapril on mortality in severe congestive heart failure: results of the Cooperative North Scandinavian Enalapril Survival Study (CONSENSUS). N Engl J Med. 1987;316:1429 –1435. 4. SOLVD Investigators. Effect of enalapril on survival in patients with reduced left ventricular ejection fractions and congestive heart failure. N Engl J Med. 1991;325:293–302. 5. International Classification of Diseases, Ninth Revision, Clinical Modification. Washington, D.C.: Public Health Service, U.S. Department of Health and Human Services; 1988. 6. Tu JV, Zhang H. Congestive heart failure outcomes in Ontario. In: Naylor C, Slaughter P, eds. Cardiovascular Health and Services: Technical and Methods Appendices. Toronto, Canada: Institute for Clinical Evaluative Sciences; 1999:25–30. 7. Weil E, Tu JV. Quality of congestive heart failure treatment at a Canadian teaching hospital. CMAJ. 2001;165:284 –287. 8. Jong P, Gong Y, Liu PP, et al. Care and outcomes of patients newly hospitalized for heart failure in the community treated by cardiologists compared with other specialists. Circulation. 2003;108:184 – 191. 9. MacIntyre K, Capewell S, Stewart S, et al. Evidence of improving prognosis in heart failure: trends in case fatality in 66 547 patients hospitalized between 1986 and 1995. Circulation. 2000;102:1126 – 1131. 10. Jong P, Vowinckel E, Liu PP, et al. Prognosis and determinants of survival in patients newly hospitalized for heart failure: a population-based study. Arch Intern Med. 2002;162:1689 –1694. 11. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45:613–619. 12. Australia/New Zealand Heart Failure Research Collaborative Group. Randomised, placebo-controlled trial of carvedilol in patients with congestive heart failure due to ischaemic heart disease. Lancet. 1997;349:375–380. 13. Pitt B, Segal R, Martinez FA, et al. Randomised trial of losartan versus captopril in patients over 65 with heart failure (Evaluation of Losartan in the Elderly Study, ELITE). Lancet. 1997;349:747–752. 14. Packer M, Bristow MR, Cohn JN, et al. The effect of carvedilol on morbidity and mortality in patients with chronic heart failure: U.S. Carvedilol Heart Failure Study Group. N Engl J Med. 1996;334: 1349 –1355. 588
May 1, 2004
THE AMERICAN JOURNAL OF MEDICINE威
Volume 116
15. Digitalis Investigation Group. The effect of digoxin on mortality and morbidity in patients with heart failure. N Engl J Med. 1997; 336:525–533. 16. Heidenreich PA, McClellan M. Trends in treatment and outcomes for acute myocardial infarction: 1975-1995. Am J Med. 2001;110: 165–174. 17. Pindyck RS, Rubinfeld DL. Econometric Models and Economic Forecasts. 4th ed. New York, New York: Irwin McGraw-Hill; 1998. 18. Diggle PJ. Time Series: A Biostatistical Introduction. New York, New York: Oxford University Press; 1990. 19. Yaffee R. Introduction to Time Series Analysis and Forecasting. San Diego, California: Academic Press; 2000. 20. Ljung GM, Box GEP. On a measure of lack of fit in time series models. Biometrika. 1978;65:297–303. 21. Masoudi FA, Havranek EP, Smith G, et al. Gender, age, and heart failure with preserved left ventricular systolic function. J Am Coll Cardiol. 2003;41:217–223. 22. Krumholz HM, Wang Y, Parent EM, et al. Quality of care for elderly patients hospitalized with heart failure. Arch Intern Med. 1997;157: 2242–2247. 23. Philbin EF, Rocco TA Jr, Lindenmuth NW, et al. Systolic versus diastolic heart failure in community practice: clinical features, outcomes, and the use of angiotensin-converting enzyme inhibitors. Am J Med. 2000;109:605–613. 24. Redfield MM, Jacobsen SJ, Burnett JC Jr, et al. Burden of systolic and diastolic ventricular dysfunction in the community: appreciating the scope of the heart failure epidemic. JAMA. 2003;289:194 – 202. 25. Senni M, Tribouilloy CM, Rodeheffer RJ, et al. Congestive heart failure in the community: a study of all incident cases in Olmsted County, Minnesota, in 1991. Circulation. 1998;98:2282– 2289. 26. Vasan RS, Larson MG, Benjamin EJ, et al. Congestive heart failure in subjects with normal versus reduced left ventricular ejection fraction: prevalence and mortality in a population- based cohort. J Am Coll Cardiol. 1999;33:1948 –1955. 27. Garg R, Yusuf S. Overview of randomized trials of angiotensinconverting enzyme inhibitors on mortality and morbidity in patients with heart failure: collaborative group on ACE inhibitor trials. JAMA. 1995;273:1450 –1456. 28. Brophy JM, Joseph L, Rouleau JL. Beta-blockers in congestive heart failure. A Bayesian meta-analysis. Ann Intern Med. 2001;134:550 – 560. 29. Teo KK, Yusuf S, Furberg CD. Effects of prophylactic antiarrhythmic drug therapy in acute myocardial infarction. An overview of results from randomized controlled trials. JAMA. 1993;270:1589 – 1595. 30. Hood WB Jr, Dans AL, Guyatt GH, Jaeschke R, McMurray JJV. Digitalis for treatment of congestive heart failure in patients in sinus rhythm (Cochrane Review). In: The Cochrane Library. Issue 3. Oxford, United Kingdom: Update Software; 2003. 31. Goldstein RE, Boccuzzi SJ, Cruess D, et al. Diltiazem increases lateonset congestive heart failure in postinfarction patients with early reduction in ejection fraction. Circulation. 1991;83:52–60. 32. Jong P, Demers C, McKelvie RS, Liu PP. Angiotensin receptor blockers in heart failure: meta-analysis of randomized controlled trials. J Am Coll Cardiol. 2002;39:463–470. 33. Elkayam U, Amin J, Mehra A, et al. A prospective, randomized, double-blind, crossover study to compare the efficacy and safety of chronic nifedipine therapy with that of isosorbide dinitrate and their combination in the treatment of chronic congestive heart failure. Circulation. 1990;82:1954 –1961.
Trends in Heart Failure Outcomes and Therapy/Lee et al 34. Glesby MJ, Hoover DR. Survivor treatment selection bias in observational studies: examples from the AIDS literature. Ann Intern Med. 1996;124:999 –1005. 35. Levy D, Kenchaiah S, Larson MG, et al. Long-term trends in the incidence of and survival with heart failure. N Engl J Med. 2002;347: 1397–1402. 36. Ni H, Nauman DJ, Hershberger RE. Analysis of trends in hospitalizations for heart failure. J Card Fail. 1999;5:79 –84. 37. Polanczyk CA, Rohde LE, Dec GW, DiSalvo T. Ten-year trends
in hospital care for congestive heart failure: improved outcomes and increased use of resources. Arch Intern Med. 2000;160:325– 332. 38. Hoyert DL, Arias E, Smith BL, et al. Deaths: final data for 1999. Natl Vital Stat Rep. 2001;49:1–113. 39. Tu JV, Zhang H. Congestive heart failure outcomes in Ontario. In: Naylor CD, Slaughter PM, eds. Cardiovascular Health and Services in Ontario: An ICES Atlas. Toronto, Canada: Institute for Clinical Evaluative Sciences; 1999:111–122.
May 1, 2004
THE AMERICAN JOURNAL OF MEDICINE威
Volume 116 589