Author + information
- Received May 28, 2013
- Revision received July 17, 2013
- Accepted July 25, 2013
- Published online December 1, 2013.
- Padma Kaul, PhD∗∗ (, )
- Shelby D. Reed, PhD†,‡,
- Adrian F. Hernandez, MD, MPH‡,
- Jonathan G. Howlett, MD§,
- Justin A. Ezekowitz, MBBCh, MSc∗,
- Yanhong Li, MD†,‡,
- Yinggan Zheng, MA, MEd∗,
- Jean L. Rouleau, MD‖,
- Randall C. Starling, MD¶,
- Christopher M. O'Connor, MD‡,
- Robert M. Califf, MD# and
- Paul W. Armstrong, MD∗
- ∗Division of Cardiology, Department of Medicine, University of Alberta, and Mazankowski Alberta Heart Institute, Edmonton, Alberta, Canada
- †Center for Clinical and Genetic Economics, Duke Clinical Research Institute, Durham, North Carolina
- ‡Duke Clinical Research Institute, Durham, North Carolina
- §Department of Medicine, University of Calgary, Calgary, Alberta, Canada
- ‖Montreal Heart Institute, Montreal, Quebec, Canada
- ¶Department of Cardiovascular Medicine, Cleveland Clinic, Cleveland, Ohio
- #Duke Translational Medicine Institute, Duke University School of Medicine, Durham, North Carolina
- ↵∗Reprint requests and correspondence:
Dr. Padma Kaul, Department of Medicine, University of Alberta, 2-132 Li Ka Shing Centre for Health Research Innovation, Edmonton AB T6G 2E1, Canada.
Objectives The aim of this study was to compare clinical outcomes, resource utilization, and health-related quality of life between Canadian and U.S. patients enrolled in ASCEND-HF (Acute Study of Clinical Effectiveness of Nesiritide in Decompensated Heart Failure). A further aim was to supplement the within-trial analysis with a contemporaneous population-based comparison of all patients hospitalized with primary diagnoses of heart failure (HF) in the 2 countries.
Background Little is known about intercountry differences in outcomes of patients with HF in Canada and the United States.
Methods Trial patients consisted of 465 Canadian and 2,684 U.S. patients enrolled in ASCEND-HF. Population-level cohorts consisted of 1.9 million U.S. and 81,016 Canadians hospitalized for HF in 2007 and 2008.
Results Canadian patients in ASCEND-HF were older, were more likely to be white, and had lower body weights and blood pressures than U.S. patients. Canadians also had lower baseline-adjusted odds of 30-day mortality (odds ratio: 0.46; 95% confidence interval: 0.23 to 0.92) and better health-related quality of life than U.S. patients. In both countries, trial patients differed significantly from population-level cohorts. In contrast to ASCEND-HF, unadjusted in-hospital mortality at the population level was significantly lower in the United States (3.4%) compared with Canada (11.1%) (p < 0.01).
Conclusions Intercountry differences in outcomes of patients hospitalized with HF differed significantly between trial and population cohorts. Further study on how cardiac care is delivered in the 2 countries and how it influences the results of clinical trials and population-level outcomes, especially in the long term, is warranted. (A Study Testing the Effectiveness of Nesiritide in Patients With Acute Decompensated Heart Failure; NCT00475852)
Differences in the management of patients with acute myocardial infarction in Canada and the United States are well documented (1–6). However, less is known about intercountry differences in the treatment and outcomes of patients hospitalized with acute decompensated heart failure (HF). Ko et al. (7) compared short-term and long-term mortality among elderly U.S. Medicare beneficiaries hospitalized with HF between 1998 and 2001 and a similarly aged cohort of patients in Ontario, Canada. They found that 30-day mortality rates were lower in the United States, but by 1 year, the mortality differences between the 2 countries had dissipated. Whether these findings persist in contemporary practice and extend to patients under the age of 65 years is not known. Also, no previous study has compared health-related quality of life (HRQOL) among patients with HF in the 2 countries.
Accordingly, we compared clinical outcomes, resource utilization, and HRQOL between Canadian and U.S. patients enrolled in ASCEND-HF (Acute Study of Clinical Effectiveness of Nesiritide in Decompensated Heart Failure) (8). Although patients enrolled in clinical trials have the advantage of being relatively homogenous, they may not be representative of the overall populations with the disease in the 2 countries. We therefore supplemented our within-trial analysis with a contemporaneous population-based comparison of all patients hospitalized with primary diagnoses of HF in the 2 countries.
ASCEND-HF trial and study population
The study design and results of ASCEND-HF have been previously published (8,9). Briefly, ASCEND-HF evaluated nesiritide versus placebo in patients with acute decompensated HF enrolled within 24 h of first intravenous HF-related treatment. Participants were required to have each of the following at the time of randomization: dyspnea at rest or with minimal activity, ≥1 accompanying sign (respiratory rate ≥20 breaths/min or pulmonary congestion or edema with rales ≥1/3 baseline), and ≥1 objective measure of HF (evidence of congestion or edema on chest x-ray, B-type natriuretic peptide ≥400 pg/ml or N-terminal pro–B-type natriuretic peptide ≥1,000 pg/ml, pulmonary capillary wedge pressure >20 mm Hg, or left ventricular ejection fraction <40% in the previous 12 months).
Between May 2007 and August 2010, 7,007 patients were enrolled in ASCEND-HF and received a study drug. All 2,684 patients enrolled in the United States and 465 enrolled in Canada were included in the present analysis.
HRQOL in ASCEND-HF was measured using the EQ-5D descriptive system. The EQ-5D descriptive system is a self-administered instrument consisting of 5 items, each representing 1 dimension, and a visual analogue scale used to record an individual's rating of his or her current health state. The 5 dimensions are mobility, self-care, usual activities, pain and discomfort, and anxiety and depression. The visual analogue scale ranges from 0 to 100, with 0 representing worst imaginable health and 100 representing the best imaginable health. The different health states attainable from the EQ-5D profile can each be assigned an EQ-5D utility index, where 1 corresponds to perfect health and 0 to death. Utility levels below zero can occur, indicating a health state worse than death. The single index scores of the states are based on the U.S. time trade-off tariff, derived from a sample of the U.S. population who were asked to value EQ-5D states in terms of equal preference for the amount of time in perfect health relative to the amount of time in each health state. The EQ-5D descriptive system has been extensively used within the cardiovascular field to assess patient utility in trials of new treatments and has demonstrated both high validity and reliability (10).
Statistical analysis of ASCEND-HF data
Continuous variables are presented both as mean ± SD and as median (interquartile range) and categorical variables as percents. Intercountry differences in continuous variables were assessed using t tests (for means) and Kruskal-Wallis tests (for medians) and using chi-square tests for categorical variables.
We examined intercountry differences in baseline characteristics, the use of adjunct medical therapies (at baseline and at discharge), and resource utilization during the index hospitalization. With respect to outcomes, we examined differences in in-hospital and 30-day mortality, as well as the combined end point of 30-day all-cause rehospitalization and mortality. We also calculated the number of days alive and out of the hospital at 30 days as a measure of the total burden of HF in the 2 countries. To examine whether country of enrollment was independently associated with 30-day mortality and combined 30-day all-cause rehospitalization and mortality after baseline adjustment, we used previously developed multivariate logistic regression models based on the entire ASCEND-HF trial population. Variables included in the models were as follows: for 30-day mortality, age, blood urea nitrogen, baseline sodium, dyspnea, and systolic blood pressure; and for 30-day all-cause rehospitalization and mortality, age, blood urea nitrogen, baseline sodium, dyspnea, systolic blood pressure, creatinine, depression, history of cerebrovascular disease, elevated jugular venous pressure, HF hospitalization before admission, and history of chronic obstructive pulmonary disease.
In ASCEND-HF, the EQ-5D data were collected at randomization, 24 h, hospital discharge, and 30 days. Average EQ-5D index scores at baseline, hospital discharge, and 30 days among patients enrolled in Canada and the United States were compared using t tests. Patients who died before each time point were assigned an EQ-5D index score value of zero. Linear regression analysis was used to examine whether EQ-5D index scores at 30 days were different between the 2 countries after adjusting for age, sex, and baseline EQ-5D score.
For the population-level comparison of HF outcomes in the 2 countries, we identified all hospitalizations with primary diagnoses of HF between January 1, 2007, and December 31, 2008, for patients aged 40 years and older in the 2 countries. Data on U.S. hospitalizations were from the Healthcare Cost and Utilization Project Nationwide Inpatient Sample. These inpatient data, made publicly available by the Agency for Healthcare Research and Quality, are representative of all Healthcare Cost and Utilization Project state inpatient databases (http://www.hcup-us.ahrq.gov). Data on Canadian hospitalizations were from the Canadian Institutes of Health Information Discharge Abstract Database (http://www.cihi.ca). This database contains data on all acute care hospital discharges in all Canadian provinces (except Quebec). Both databases include data on admission and discharge dates, patient demographic variables, information on comorbidities and diagnostic and therapeutic procedures, and discharge disposition. The unit of analysis of the population data is hospitalizations; therefore, patients with multiple HF hospitalizations were included more than once.
Previously established International Classification of Diseases-Ninth Revision, and International Classification of Diseases-Tenth Revision, code-based definitions were used to identify comorbidities (11,12). Both the Healthcare Cost and Utilization Project and Canadian Institutes of Health Information databases can include up to 25 diagnostic codes. However, the average number of diagnoses coded was lower in Canada (5.7) than in the United States (10.4). To control for intercountry differences in comorbidity rates due to completeness of coding, we compared U.S. rates with those in the Canadian province of Alberta, which had the highest average number of diagnoses coded (7.6) and where administrative data coding has been validated against chart audit (11).
Statistical analysis of population-level data
Policies related to transport and release of data prohibited us from combining the 2 population-level databases. Therefore, all statistical calculations were performed separately for each country. Continuous variables are presented both as mean ± standard deviation and as median (interquartile range) and categorical variables as percents. Chi-square tests were used to examine differences in summary rates of baseline demographic and clinical characteristics, in-hospital mortality, and discharge destinations in the 2 countries. In addition to the overall cohorts, we examined in-hospital mortality in specific age and sex subgroups.
All analyses were conducted using SAS version 9.2 (SAS Institute Inc., Cary, North Carolina).
Differences in U.S. and Canadian patients enrolled in ASCEND-HF
There were significant differences in baseline characteristics of U.S. and Canadian patients in ASCEND-HF (Table 1). U.S. patients were younger, were less likely to be white, had higher body weights and blood pressures, and were more likely to be hospitalized for HF in the year before enrollment. Given that the trial included patients with de novo HF as well as patients with ejection fractions over 40%, the overall use of implantable devices was relatively low. However, U.S. patients had higher rates of implantable cardioverter-defibrillator and combined biventricular pacemaker and implantable cardioverter-defibrillator use than Canadians. Although Canadians were more likely to have histories of myocardial infarction and atrial fibrillation, they were less likely to have depression and chronic respiratory disease. In general, the use of adjunct medical therapies at randomization and at discharge was similar in the 2 countries; although the proportion of Canadian patients on angiotensin-converting enzyme inhibitors was slightly higher and that on beta-blockers slightly lower than among the U.S. patients (Fig. 1).
In ASCEND-HF, mean and median hospital length of stay were significantly shorter in the United States compared with Canada (Table 2). Rates of diagnostic and invasive procedures were generally low; however, the rates of cardiac magnetic resonance imagining, computed tomography, transthoracic echocardiography, nuclear imaging, and cardiac catheterization were significantly higher in Canada compared with the United States.
There was no significant difference in unadjusted in-hospital mortality or 30-day mortality rates between the 2 countries (Fig. 2). However, after adjustment for baseline characteristics shown to be predictive of 30-day mortality on the basis of the entire ASCEND-HF population, patients enrolled in Canada had significantly lower odds of 30-day mortality (odds ratio: 0.46; 95% confidence interval: 0.23 to 0.92) relative to U.S. patients. The rate of combined 30-day all-cause readmission or death was significantly lower in Canada (15.1%) compared with the United States (20.2%) (p = 0.010), and this difference remained after baseline adjustment (odds ratio: 0.72; 95% confidence interval: 0.54 to 0.96; p = 0.025). However, the higher rate of readmission in the United States did not translate into increased disease burden: at 30 days, the total number of days alive and out of the hospital was significantly higher among U.S. patients (median 24 days; interquartile range: 20 to 26 days) compared with Canadian patients (median 21 days; interquartile range: 14 to 23 days) (p < 0.001).
Overall HRQOL, as measured by the EQ-5D utility index, was similar at baseline among patients enrolled in Canada and the United States (Fig. 3). At baseline, a higher proportion of Canadians reported extreme problems with mobility and self-care, while a higher proportion of U.S. patients reported extreme problems with pain and anxiety or depression (Table 3). By discharge, HRQOL improved in both countries and was higher among Canadian patients compared with U.S. patients (Fig. 3). At 30 days, Canadians reported significantly better HRQOL than U.S. patients, which remained true even after adjustments for age, sex, and baseline HRQOL. Intercountry differences in the proportion of patients reporting no problems with pain at baseline (37.5% in the United States vs. 50.9% in Canada, p < 0.001) were attenuated by 30 days (56% in the United States vs. 56.9% in Canada, p = 0.651). Relative to baseline, at 30 days, the proportion of patients reporting no problems in the 5 dimensions increased in both countries (Table 3).
Population-level differences in U.S. and Canadian patients hospitalized with HF
Baseline characteristics and hospital outcomes of approximately 1.9 million U.S. and 81,016 Canadian HF hospitalizations between January 2007 and December 2008 are presented in Table 4. In both countries, patients in the population-based cohorts were older, more likely to be women, and differed significantly with respect to rates of comorbidities compared with patients enrolled in ASCEND-HF. Canadian patients were older, while U.S. patients had higher rates of other risk factors such as ischemic heart disease, hypertension, renal disease, and diabetes. These differences in comorbidity rates were maintained, although slightly attenuated, when the U.S. rates were compared with rates among patients hospitalized in Alberta. Average length of stay was more than twice as long in Canada (12 days) compared with the United States (5 days). Although the proportion of patients discharged to home care was similar in both countries, the percent of patients transferred to long-term care or skilled nursing facilities was significantly higher in the United States than in Canada. Overall, unadjusted in-hospital mortality was 3.4% in the United States compared with 11.1% in Canada (p < 0.001). These mortality differences were maintained among specific patient subgroups.
This is the largest study to examine Canadian-American differences in HF outcomes among patients enrolled in a clinical trial as well as at a population level. The recent international ASCEND-HF trial of nesiritide in acute decompensated HF offered us a unique opportunity to examine intercountry differences in a contemporary cohort of patients hospitalized for HF in the 2 countries. Canadian patients enrolled in ASCEND-HF had similar unadjusted in-hospital mortality compared with patients enrolled in the United States. However, baseline-adjusted odds of 30-day mortality and combined 30-day all-cause rehospitalization and mortality were significantly lower among Canadian relative to U.S. patients. The lower readmission rates in Canada may be due partly to the significantly longer hospital stays (13). The longer stays may also explain why despite higher readmission rates, disease burden, as measured by the number of days alive and out of the hospital, was lower among U.S. patients.
No previous study has examined differences in HRQOL outcomes between U.S. and Canadian patients with HF. Canadian patients enrolled in ASCEND-HF reported better age-adjusted, sex-adjusted, and baseline-adjusted quality-of-life outcomes at 30 days. As previously hypothesized, a possible explanation for these findings is that U.S. and Canadian patients have important sociocultural differences in how they perceive quality of life (5). However, patients in both countries reported similar HRQOL at baseline. Other factors that were not measured in our study may explain some of the difference in HRQOL. These include differences in the incomes and education of enrolled patients or differences in follow-up care.
We supplemented this intercountry comparison of patients enrolled in a clinical trial with a population-based analysis of HF-related hospitalizations in the 2 countries. As in the clinical trial population, lengths of stay were significantly longer in Canada relative to the United States. However, in contrast to our finding in ASCEND-HF, unadjusted in-hospital mortality was significantly lower in the United States compared with Canada.
There are several potential reasons for the discrepancy between our findings based on patients enrolled in the clinical trial and the population-level analysis. First, some of the mortality difference observed at the population level may be explained by differences in demographic and clinical characteristics between U.S. and Canadian patients. On average, Canadian patients were 4 years older but had lower rates of ischemic heart disease and hypertension compared with U.S. patients. Unfortunately, because of privacy and data-sharing constraints, we were unable to combine the population-level data to calculate baseline risk-adjusted mortality rates.
Second, it is important to view the intercountry differences in population-level mortality rates in the context of the striking differences in lengths of stay in the 2 countries. We have previously documented significant differences in the location of death among elderly patients with HF in Canada and the United States (14,15). A higher percent of Canadians die in the hospital relative to U.S. patients, probably as a result of limited availability of alternative venues of care. In our population-based cohorts, the percent of patients discharged to long-term care or skilled nursing facilities was twice as high in the United States (20%) compared with Canada (11%). Given these differences in practice patterns, a more appropriate metric to compare intercountry differences in outcomes may be 30-day or 1-year mortality rates. Unfortunately, the lack of follow-up data for the population-level cohorts in our study did not allow us to examine longer term outcomes. In a previous study of elderly Medicare patients in the United States and a matched cohort in Ontario, Ko et al. (7) reported lower 30-day mortality (8.9% vs. 12.2%, respectively, p < 0.01) but similar mortality rates at 1 year. In ASCEND-HF, Canadian patients had significantly lower baseline-adjusted odds of 30-day mortality. These inconsistent findings suggest a need for a more contemporary assessment of population-level short and long-term mortality outcomes among patients with HF in the 2 countries.
Third, and perhaps the most dominant reason for the difference in outcomes among clinical trial versus population-level cohort patients, is selection bias. Patients enrolled in the trial do not appear to be representative of the entire population of patients with the disease in the 2 countries. In both countries, patients enrolled in the clinical trial were significantly younger and more likely to be men than the population-level cohorts. These differences may explain, to a certain extent, the intracountry difference in in-hospital mortality in the trial versus population cohorts. In the United States, patients enrolled in ASCEND-HF had a hospital mortality rate of 1.4%, compared with 3.7% in the population-level cohort. The difference in trial versus population cohorts in terms of in-hospital mortality was even more pronounced in Canada: 1.3% among patients enrolled in ASCEND-HF and 11.1% among all HF hospitalizations. Surprisingly, the rates of traditional risk factors, such as hypertension and diabetes, appeared to be higher in the trial populations. However, the extent to which this reflects either better capture of clinical information in the setting of a controlled clinical trial and less comprehensive capture by medical records personnel, or a truly higher risk population selected for enrollment in the clinical trial, cannot be assessed.
Our study had several strengths and some limitations. Although the comparison of intercountry differences based on ASCEND-HF offers unique data on quality-of-life outcomes and baseline-adjusted mortality rates in a relatively homogenous, contemporary patient population, our examination of HF hospitalizations for a similar time period provides evidence on intercountry differences at the population level. Although the results are somewhat contradictory, we believe that supplementing the clinical trial data with population-level data provided a more complete picture of intercountry differences in HF outcomes. First, unlike patient-level data in ASCEND-HF, the population-level data in the 2 countries were for HF hospitalizations in which a patient may be represented more than once. Therefore, some of the observed differences in baseline characteristics in the 2 cohorts may be explained by differences in readmission rates in the 2 countries. Second, another factor that could affect comorbidity rates is the accuracy and completeness of coding in the 2 countries. In the United States, hospital payment is tied to the presence of comorbidities, whereas in Canada, only those comorbidities that significantly affect treatment, require treatment beyond maintenance of the preexisting condition, or increase a hospital stay by 24 h are mandatory to capture (16). We found significant variability in the mean number of diagnoses coded across the Canadian provinces. However, intercountry differences in comorbidity rates remained significant in our comparison of U.S. patients and patients hospitalized in the province of Alberta, which had the highest mean number of coded diagnoses. Third, comorbidity data are based on International Classification of Disease, Ninth Revision, codes in the US and on International Classification of Disease, Tenth Revision, codes in Canada; however, a previously validated cross-walk was used in the present analysis (12). Fourth, data on measures such as patient height, weight, ejection fraction, cardiac markers such as B-type natriuretic peptide, and medications were not available for the population-level cohorts. Finally, HRQOL was assessed using the generic EQ-5D instrument. Although this instrument has the advantage of being completed with relative ease by patients, incorporated into case-report forms, and validated in many languages, it may be less sensitive than a disease-specific quality-of-life instrument. However, this should not have influenced the observed between-country difference in HRQOL, which was significant at 30 days.
Evidence on intercountry differences in outcomes of patients hospitalized with HF in Canada and the United States remains equivocal. In a contemporary clinical trial, patients enrolled at sites in Canada had longer lengths of stay, better quality of life, and lower baseline-adjusted mortality at 30 days compared with patients enrolled at U.S. sites. However, these findings are not generalizable: a population-level examination of HF hospitalizations in the 2 countries found a significantly higher hospital mortality rate in Canada compared with the United States. This dichotomy in findings from clinical trial and population-level data highlights the important role that each plays in observational studies. Further study on how cardiac care is delivered in the 2 countries and how it influences the results of clinical trials and population-level outcomes, especially in the long term, is warranted.
The authors thank Wei Liu, MS, Canadian VIGOUR Center, University of Alberta (Edmonton, AB, Canada), for statistical assistance (he did not receive compensation for his assistance apart from his employment at the institution at which the study was conducted).
ASCEND-HF was funded by Scios Inc. Individual financial disclosures pertaining to the original ASCEND-HF study have been previously published online. Dr. Hernandez has relationships with Janssen, Corthera, and Amgen. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Javed Butler, MD, served as Guest Editor for this paper.
- Abbreviations and Acronyms
- heart failure
- health-related quality of life
- Received May 28, 2013.
- Revision received July 17, 2013.
- Accepted July 25, 2013.
- American College of Cardiology Foundation
- Ko D.T.,
- Krumholz H.M.,
- Wang Y.,
- et al.
- Kaul P.,
- Armstrong P.W.,
- Chang W.-C.,
- et al.
- Fu Y.,
- Chang W.C.,
- Mark D.,
- et al.,
- for the Global Use of Strategies to Open Occluded Coronary Arteries (GUSTO) II Investigators
- Eapen Z.J.,
- Reed S.D.,
- Li Y.,
- et al.