Evidence of Improving Prognosis in Heart Failure
Trends in Case Fatality in 66 547 Patients Hospitalized Between 1986 and 1995
Background—Contemporary survival in unselected patients with heart failure and the population impact of newer therapies have not been widely studied. Therefore, we have documented case-fatality rates (CFRs) over a recent 10-year period.
Methods and Results—In Scotland, all hospitalizations and deaths are captured on a single database. We have studied case fatality in all patients admitted with a principal diagnosis of heart failure from 1986 to 1995. A total of 66 547 patients (47% male) were studied. Median age was 72 years in men and 78 years in women. Crude CFRs at 30 days and at 1, 5, and 10 years were 19.9%, 44.5%, 76.5%, and 87.6%, respectively. Median survival was 1.47 years in men and 1.39 years in women (2.47 and 2.36 years, respectively, in those surviving 30 days). Age had a powerful effect on survival, and sex, comorbidity, and deprivation had modest effects. One-year CF was 24.2% in those aged <55 years and 58.1% in those aged >84 years. After adjustment, 30-day CFRs fell between 1986 and 1995, by 26% (95% CI 15 to 35, P<0.0001) in men and 17% (95% CI 6 to 26, P<0.0001) in women. Longer term CFRs fell by 18% (95% CI 13 to 24, P<0.0001) in men and 15% (95% CI 10 to 20, P<0.0001) in women. Median survival increased from 1.23 to 1.64 years.
Conclusions—Heart failure CF is much higher in the general population than in clinical trials, especially in the elderly. Although survival has increased significantly over the last decade, there is still much room for improvement.
Clinical trials have shown that the case fatality related to heart failure is high but can be significantly reduced with therapies such as angiotensin-converting enzyme (ACE) inhibitors and, more recently, β-blockers.1 2 3 4 However, these studies have mainly enrolled middle-aged men. Thus, the trials are unrepresentative of the general population of patients with heart failure who tend to be elderly and as frequently female as male.5 6 7 The prognosis of these older patients has been less well studied. Furthermore, the population impact of newer treatments such as ACE inhibitors, now in widespread use for more than a decade, has not been investigated. We have used the unique Scottish National Health Service Linked Patient Database to study short- and long-term case-fatality rates in all patients admitted to hospital for the first time with a primary diagnosis of heart failure. We have also examined trends in these rates over the period between 1986 and 1995, during which ACE inhibitor therapy was introduced into routine clinical practice.
The Information and Statistics Division of the National Health Service in Scotland collects and collates data on all National Health Service hospital admissions by use of the Scottish Morbidity Record scheme.5 8 This database is linked, with the use of probability matching, to information held by the General Registrar’s Office for Scotland on in-hospital and out-of-hospital deaths. This linkage permits analysis of each individual patient’s outcome.8 Although some minor losses to follow-up might be expected, emigration of people of “pensionable age” from Scotland was <2% per decade. All deaths occurring within the United Kingdom are captured by the linked database.
Time Period of Analysis
A Scottish-wide retrospective cohort study was undertaken with use of this database to identify all admissions attributed to heart failure (International Classification of Diseases, 9th Revision, codes 425.4, 425.5, 425.9, 428.0, 428.1, and 428.9) to Scottish Hospitals between 1986 and 1995 and subsequent deaths related to these admissions in the same period. Only the first episode of heart failure leading to a hospital admission per patient was analyzed (see below).
Each patient’s record provided information on age, sex, postal code of residence, date of admission, and death, if it occurred. The postal codes of residence were used to attribute a Carstairs Deprivation category (from 1 to 5) to each individual. These categories are derived from 1991 census data on the proportion of residents who are unemployed, live in overcrowded accommodations, do not have access to a car, or belong to a low occupational social class. Category 1 represents the least deprived section of the population; category 5, the most deprived.8 This identified those patients who had been admitted to hospital for any other reason within 5 years before their first admission for heart failure. To consistently obtain a 5-year history of prior hospital admission for each patient, the principal analyses in the present study were confined to patients admitted between January 1986 and December 1995. This allowed patients to be followed up for a minimum of 1 year to the end of the study (December 31, 1996).
Definition of First Admission
A “first admission” was defined as no previous admission with heart failure in the past 5 years. Patients with a hospitalization related to heart failure in the previous 5 years were excluded from this analysis.
The linked database allowed analysis of survival data for all identified patients until December 31, 1996. All surviving patients were censored at this time point to provide between 1 and 10 years of follow-up depending on the year of the index admission. If death from any cause occurred, survival time was calculated as the time from date of index admission to the date of death from any cause. Crude case-fatality rates were calculated for follow-up periods from 30 days to 10 years by use of the actuarial life-table method. This takes account of admission dates and periods of follow-up, which differ between patients. Crude case-fatality rates were stratified by age, sex, deprivation category, prior admission (comorbidity), and year of first admission for heart failure. Kaplan-Meier analyses were used to determine median survival. For patients admitted to hospital with heart failure, mortality at 30 days was modeled by use of logistic regression to analyze the independent effects of these factors. Because changes in case fatality for men and women appeared to differ over the short term depending on age, the sexes were considered separately in the multivariate analyses. All variables were entered simultaneously into the models. Each model was subject to the Hosmer-Lemeshow goodness-of-fit test, and all were statistically nonsignificant. To examine the independent effect of these factors on survival thereafter, data from patients who survived ≥30 days were entered into the Cox proportional hazards models. Once again, models were performed separately for men and women, and all variables were entered simultaneously into the model. The assumption of constant hazard was met for these models. For both multiple logistic regression and Cox proportional hazards models, age was recoded and entered in ascending order as follows: <55, 55 to 64, 65 to 74, 75 to 84, and >84 years. Deprivation data were reentered as the 5 categories described above. The 2350 patients not assigned a deprivation category were excluded from these analyses. Prior admission categories were entered as either present or absent. The year of admission was coded chronologically from 1 to 10 (1986 to 1995). For each variable entered into a model, the lowest class was set at unity. Adjusted odds and hazards ratios for the remaining 1 to 9 classes for each variable are therefore relative to that of the lowest class. Significance was accepted at P<0.05. All analyses were undertaken with use of the Statistical Package for Social Scientists (SPSS Inc).
Sex and Age
A total of 66 547 patients were admitted to hospital in Scotland for the first time with heart failure between 1986 and 1995. Men (31 040) accounted for less than half (46.6%) of the patients admitted. The median age at admission was 72 years in men and 78 years in women. Only 18.7% of patients were aged <65 years, whereas 53% were aged >75 years. The median age of patients increased over the period of study. For women, the median age increased from 76.0 years in 1986 to 79.0 years in 1995. In men, the increase was from 70.7 to 73.0 years. These trends were highly significant (P<0.0001).
Almost half (44%) of the cases came from the lowest 2 deprivation quintiles. The admission rate was 56% higher in the most deprived quintile compared with the most affluent quintile (P<0.001).
Prior Hospital Admissions
One third of the patients had a history of other admissions to hospital within the previous 5 years. Coronary heart disease accounted for the greatest proportion of these: 10 074 (15.1%) patients with an acute myocardial infarction and 7408 (11.1%) with other forms of coronary heart disease. Other vascular disease (cerebral 3677 and peripheral 2288 patients), diabetes mellitus (1760 patients), and hypertension (752 patients) were also frequently coded. Other conditions commonly coded were respiratory disease, cancer, and atrial fibrillation (Table 1⇓). There was a trend over time for patients to have more prior admissions. In 1986, 42% of men had at least one prior hospitalization, whereas this proportion had increased to 52% in 1995. The respective proportions for women were 37% and 46%.
Crude Case-Fatality Rates (Univariate Analysis)
The overall crude case-fatality rate at 1 month, 1 year, 5 years, and 10 years was 19.9%, 44.5%, 76.5%, and 87.6%, respectively. These respective rates were 19.4%, 44.0%, 75.0%, and 87.2% in men and 20.3%, 44.9%, 76.2%, and 89.3% in women. Age had a powerful effect with the 1-month case-fatality rate, increasing from 10.4% in those aged <55 years to 25.9% in those aged >84 years. The respective rates at 1 year were 24.2% and 58.1%. Sex and deprivation had a minimal effect on crude case-fatality rates (Table 1⇑).
Median unadjusted survival over the period of study was 1.47 years in men and 1.39 years in women. For men surviving 30 days, median survival was 2.47 years; in women, it was 2.36 years.
Adjusted Case-Fatality Rates (Multivariate Analysis)
Multivariate analysis confirmed the effect of age. The effect of age on long-term case fatality (30 days to end of follow-up period), as expressed by the hazards ratio per decade of age, was 1.42 for men and 1.38 for women. However, there was also an effect of sex and social deprivation in this analysis (Table 2⇓).
The effect of sex was modest and complex. There was a highly significant interaction between age and sex, but only for 30-day case fatality. Women <64 years fared less well than did men at 30 days, although few were in this age group (≈12%). In contrast, women aged ≥65 years had a better outcome than did men. Because this was the majority of female patients, women (as a whole) fared better than did men in the short term. In the longer term (>30 days), no age-sex interaction was detected, and women had a lower case fatality than did men. The hazards ratio for women was 0.87 (95% CI 0.85 to 0.89), with that for men set at 1.
Deprivation principally increased the short-term case-fatality rate (by 26% in men and 11% in women), affecting the longer term case-fatality rate by only 10% in men and 6% in women (Table 2⇑).
A variety of prior admissions increased the short-term case-fatality rate, including cancer (by 44% in men and 47% in women) and peripheral vascular disease (by 36% in men and 19% in women). Similar trends were seen with renal and respiratory disease and stroke (Table 2⇑). In general, the same prior admissions also increased long-term mortality. A prior admission with atrial fibrillation reduced the short-term case-fatality rate (by 32% in men and 24% in women). Prior admissions with myocardial infarction and other coronary heart disease had the same effect (Table 2⇑).
Trends in Case Fatality Over Time
Crude case fatality at 1 month, 1 year, and 5 years showed a modest improvement between 1986 and 1995. Median survival increased from 1.23 to 1.64 years over this period (Table 3⇓). After adjustment for age, deprivation, and prior admission, short-term (30-day) case-fatality rates fell by 26% in men (95% CI 15 to 35, P<0.0001) and by 17% in women (95% CI 6 to 26, P<0.0001). Longer term case-fatality rates fell by ≈18.0% in men (95% CI 13 to 24, P<0.0001) and 15.0% in women (95% CI 10 to 20, P<0.0001) (Table 3⇓, Figures 1⇓ and 2⇓).
The present study confirms that unselected patients in the community with heart failure differ from those enrolled in clinical trials in that they are elderly and, in more than half the cases, female.1 2 3 4 Our data also show that the prognosis for patients admitted to hospital with heart failure is even worse than indicated by clinical trials. One fifth of the patients die within 1 month of admission, nearly half die by 1 year, and three fourths die by 5 years. Median survival is only ≈1.5 years.
We know of no similar data from the whole population of a single country. However, our findings can be compared with the 14.8-year follow-up of the Framingham study. During this time, 652 subjects developed heart failure (subjects were screened between 1948 and 1988).9 10 The mean age of those diagnosed in the 1980s in the Framingham study was 76.4 years. The average follow-up of the 652 subjects identified was 3.9 years after the onset of heart failure. Median survival in the Framingham population was 1.66 years in men and 3.17 years in women. One-, 2-, 5-, and 10-year mortality rates were 43%, 54%, 75%, and 89% in men. Women in the Framingham study had a better crude and adjusted survival (crude survival 36%, 44%, 62%, and 79% for the corresponding periods of follow-up). Patients in the Framingham study with heart failure caused by valvular heart disease or other/unknown causes had a worse survival than did those with underlying coronary heart disease. Women fared better than men. Our findings are remarkably similar to those of the Framingham investigators, including the better survival in patients with a presumed coronary etiology (although, interestingly, this finding contrasts with some clinical trials).
The Rochester Epidemiology project has also described the prognosis in 107 patients presenting to associated hospitals with new-onset heart failure in 1981 and in 141 patients presenting in 1991.11 The median follow-up in these cohorts was 1061 and 1233 days, respectively. The mean age of the 1981 patients was 75 years, rising to 77 years in 1991. We also found that the average age of patients hospitalized with heart failure has increased over time. Crude and adjusted survival did not improve over the period of study.
Respective 1-year and 5-year mortality rates were 28% and 66% in the 1981 cohort and 23% and 67% in the 1991 cohort. In other words, although the same diagnostic criteria used in the Framingham study were used in the Rochester project, the prognosis was somewhat better in the latter. The only other large representative epidemiological study reporting long-term outcome in patients with heart failure is the National Health and Nutrition Examination Survey (NHANES-I).12 The initial program evaluated 14 407 adults aged 25 and 74 years in the United States between 1971 and 1975. Follow-up studies were carried out in 1982 to 1984 and again in 1986 (for those aged ≥55 years and alive during the 1982 to 1984 review). The estimated 10-year mortality in subjects aged 25 to 74 years with self-reported heart failure was 42.8% (49.8% in men and 36% in women). Mortality in those aged 65 to 74 years was 65.4% (71.8% and 59.5% in men and women, respectively). These mortality rates are considerably lower than those observed in Scotland and Framingham. In the present study, the 10-year case-fatality rates in men and women aged 65 to 74 years were 89% and 86%, respectively. The explanation for this difference is not clear. The patients in NHANES-I were not institutionalized, and their heart failure was self-reported. Follow-up was incomplete. NHANES-I was also carried out in a more recent time period than the Framingham study, and the prognosis in patients may have improved by this time (see below).
The second major finding in the present study is that case fatality in patients admitted to hospital with heart failure has been falling over the last decade. Adjusted short-term case fatality has fallen by ≈20% to 25%, and longer term case fatality has fallen by 15% to 20%. Thus, median life expectancy has increased by almost half a year (or by a third). This is a quite different finding from that reported by the Framingham investigators in 1993,9 who studied patients developing heart failure in the period 1948 to 1988, and by the Rochester investigators, who studied patients in the period 1981 to 1991.11 In both of these studies, no temporal change in prognosis was identified. Therefore, it is clearly tempting to suggest that our more encouraging observations, from an era when ACE inhibitors have become more widely used, reflect a true improvement in survival, consequent on better treatment. Clearly, this observation must be speculative. Although ACE inhibitors are used widely among patients admitted to Scottish hospitals with heart failure,13 14 15 it is also possible that other factors could account for the apparent reduction in short- and long-term case-fatality rates. One obvious explanation is a reduction in admission threshold and the consequent creation of a cohort of patients with milder disease. We know of no evidence to support (or refute) this possibility. However, our observations are supported by 2 recent reports of decreasing age-adjusted population mortality rates for heart failure.16 17 Furthermore, the decline in case fatality observed between 1986 and 1995 is consistent with the best estimation of the population impact of ACE inhibitor treatment, having allowed for relatively low treatment uptake (40%) and imperfect compliance (70%).18
One further important finding in the multivariate analysis is the effect of atrial fibrillation to reduce case-fatality rates. Some prior studies agree with this, whereas others do not.19 20 It is likely that in the present study, a proportion of patients with atrial fibrillation may have heart failure but preserved left ventricular function, and this may account for their better prognosis.
There are obvious limitations to any study of the type that we have conducted. We have had to rely on discharge coding to identify cases. Although these have been found to be quite accurate in Scotland, we do not know whether patients with a diagnosis of heart failure had left ventricular systolic dysfunction, preserved systolic function, or some other cause of their syndrome. Although there is some debate, it seems that the prognosis in patients with normal systolic function is better than that in those with depressed function.21 22 We have clearly described the outcome in a mixture of such patients, as have the Framingham study, the Rochester study, and, almost certainly, NHANES-I. Nevertheless, these are the patients who have heart failure in the community. We have also only studied patients admitted to the hospital. Arguably, these are patients at the more severe end of the spectrum of heart failure. Community surveys, however, show that most patients with heart failure have been admitted to hospital, the majority within 2 years of identification.23 Comorbidity was identified only from prior hospital admissions, clearly representing only the more severe cases.
In summary, the prognosis of patients admitted to hospital for the first time with a diagnosis of heart failure is very poor indeed. Although there has, at last, been some modest increase in survival, there is much room for further improvement.
This study was funded by the British Heart Foundation.
- Received January 21, 2000.
- Revision received March 29, 2000.
- Accepted April 4, 2000.
- Copyright © 2000 by American Heart Association
McMurray J, McDonagh T, Morrison CE, et al. Trends in hospitalization for heart failure in Scotland 1980–1990. Eur Heart J. 1993;14:1158–1162.
Capewell S, Kendrick S, Boyd J, et al. Measuring outcomes: one month survival after acute myocardial infarction in Scotland. Heart. 1996;76:70–75.
Ho KKL, Anderson KM, Kannel WB, et al. Survival after the onset of congestive heart failure in Framingham heart study subjects. Circulation. 1993;88:107–115.
Ho KKL, Pinsky JL, Kannel WB, et al. The epidemiology of heart failure: the Framingham study. Circulation. 1994;89:506–507.
Hillis GS, Trent RJ, Winton P, et al. Angiotensin-converting-enzyme inhibitors in the management of cardiac failure: are we ignoring the evidence? QJM.. 1996;89:145–150.
Hillis GS, AlMohammad A, Wood M, et al. Changing patterns of investigation and treatment of cardiac failure in hospital. Heart. 1996;76:427–429.
Davie AP, McMurray JJ. ACE inhibitors and heart failure in hospital: any difference between cardiologists and general physicians. Postgrad Med J. 1999;75:219–222.
Murdoch DR, Love MP, Robb SD, et al. Importance of heart failure as a cause of death: changing contribution to overall mortality and coronary heart disease mortality in Scotland 1979–1992. Eur Heart J. 1998;19:1829–1835.
Capewell S, Morrison C, McMurray J. Contribution of modern cardiovascular treatment and risk factor changes to the decline in coronary heart disease mortality in Scotland between 1975–1994. Heart. 1999;81:380–386.
Carson PE, Johnson GR, Dunkman WB, et al. The influence of atrial fibrillation on prognosis in mild to moderate heart failure: the V-Heft studies. Circulation. 1993;87(suppl VI):VI-102–VI-110.
Dries DL, Exner DV, Gersh BJ, et al. Atrial fibrillation is associated with an increased risk for mortality and heart failure progression in patients with asymptomatic and symptomatic left ventricular systolic dysfunction: a retrospective analysis of the SOLVD trials. J Am Coll Cardiol. 1998;32:695–703.
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.
Clarke KW, Gray D, Hampton JR. Evidence of inadequate investigation and treatment of patients with heart failure. Br Heart J. 1994;71:584–587.