HOW DO HEART FAILURE PATIENTS DIE?

European Heart Journal Supplements (2002) 4 (Supplement D), D59-D65

How do heart failure patients die? S. Orn and K. Dickstein University of Bergen and Cardiology Division, Central Hospital in Rogaland, Stavanger, Norway

Introduction The heterogeneity of the heart failure population is reflected in the different ways in which these patients die. Some deteriorate progressively, whereas others die after acute episodes of decompensation. Others die suddenly and unexpectedly, and some (relatively few) die from noncardiac causes. Before the angiotensin-converting enzyme (ACE) inhibitor era, it was estimated that 90% of the total deaths in heart failure patients were from cardiovascular causes, 49% were related to worsening heart failure, 22% to arrhythmias and 11% to acute myocardial infarction[S].

Classification of cause and mode of death It is conventional to categorise death according to mode and cause o f death. Cause o f death addresses the mechanisms by which death occurs, such as arrhythmia, acute myocardial infarction or progressive heart failure (Table 1). Mode of death is perhaps easier to categorise,

Correspondence:Dr Stein Orn, University of Bergen and Cardiology Division, Central Hospital in Rogaland, Stavanger, Norway. 1520-765X/02/0D0059 + 07 $35.00/0

treatment strategies for heart failure prolong life expectancy, but have relatively little impact on the proportion of heart failure patients who die from cardiovascular causes. The ultimate goal of intervention is to shift the balance toward more deaths from non-cardiovascular causes.

(Eur Heart J Supplements 2002; 4 (Suppl D): D59-D65) © 2002 The European Society of Cardiology

Key Words: Autopsy, cause of death, classification of death, heart failure, mode of death.

reflects how patients die and is described as sudden or nonsudden. Mode and cause of death are not the same, although they are often used interchangeably. Sudden death has various underlying causes, such as arrhythmia, acute myocardial infarction, pulmonary embolism, myocardial or aortic rupture, and stroke. Sudden cardiac death is defined as natural death due to cardiac causes, heralded by abrupt loss of consciousness within 1 h of the onset of acute symptoms[2]. In order to avoid confusion in terminology, some clinical trials subclassify death without using the term 'cause of death' and end-point committees focus instead on mode and place of death (Table 1)[31. However, although it is more difficult to classify cause of death than mode of death, it is nevertheless productive to examine the causes of death among heart failure patients. The cause of death reflects the underlying pathophysiology of the disease, and helps us to understand the mechanisms responsible for its progression. Unravelling the mechanisms that lead to death is clinically relevant and may reveal potential new treatment targets. Effective treatment may alter the cause of death, and should ideally shift the operative mechanism from cardiovascular to noncardiovascular. Most of our loaowledge of the cause and mode of death in heart failure comes from the large randomized mortality trials and from official death registries. However, both of these sources of information have their problems. © 2002 The European Society of Cardiology

Downloaded from https://academic.oup.com/eurheartjsupp/article-abstract/4/suppl_D/D59/593247 by guest on 16 September 2018

Approximately 90% of heart failure patients die from cardiovascular causes. Fifty per cent die from progressive heart failure, and the remainder die suddenly from arrhythmias and ischaemic events. Autopsy reveals the presence of an acute ischaemic event in approximately 50% of sudden deaths and in 35% of all deaths among patients with ischaemic heart failure. An accurate description of the cause and mode of death is important if we are to elucidate the mechanisms that are operative in the heart failure population. At present, the most accurate data on mode of death are obtained from large randomized heart failure trials. They indicate that current

D60

S. Orn and K. Dickstein

Table 1 A simplified classification o f h e a r t f a i l u r e deaths

•

SOLVD treatment ACE inhibitor

[] MERITplacebo Cardiovascular

Non-cardiovascular

100 90 80 70 60 .g 50 40 30 20

Death registries The advantage of official death registries lies in the large amount of data available on a non-selected population exposed to various treatment strategies. However, even though data on total mortality is fairly reliable, official death registries make it difficult to analyse the cause of death. Data are often scarce or biased, and the clinical basis of the diagnosis may be absent or inadequate. In most countries, official death certificates discourage terms such as 'sudden death' and 'worsening heart failure'. Aetiological diagnoses, such as arrhythmia and myocardial infarction, are encouraged. However, most of these diagnoses are based on the judgement of the physician signing the death form, which may be arbitrary. Autopsy data has demonstrated the limited reliability of death certificates, medical records and interviews as sources of mortality statistics in cardiovascular epidemiology[41. Moreover, most death registries use World Health Organization International Classification of Diseases coding. The current trend toward linking payment for health services to International Classification of Diseases codes threatens to create bias and renders the use of these codes unreliable in epidemiological research.

Randomized trials The most accurate data on cause of death is obtained from large randomized trials. However, several issues complicate their interpretation. Classification of cause of death appears to differ between trials. Comparison of two large randomized trials - the ACE inhibitor treated arm of the Studies of Left Ventricular Dysfunction (SOLVD) treatment trial[ 1] and the Metoprolol CR/XL Randomized Intervention Trial in Congestive Heart Failure (MERIT-HF)[5] - reveals wide variations in cause of death between the trials, despite nearly identical selection criteria (Fig. 1). Furthermore, classification problems may arise when death is not witnessed or information is scarce. In the Cardiac Insufficiency Bisoprolol Study (CIBIS) II[6], for example, 30% of deaths in the placebo group and 20% in the bisoprolol group were classified as of un!mown cause. Eur Heart J Supplements,VoI.4 (Suppi D) April 2002

10 Total Worsening Sudden/ cardiovascular heart failure arrhythmic deaths

Other

Figure 1 Modes and causes of death in patients treated with angiotensin-converting enzyme inhibitors in the Studies of Left Ventricular Dysfunction (SOLVD) treatment trial[tl and patients treated with placebo in Metoprolol CR/XL Randomized Intervention Trial in Congestive Heart Failure (MERIT-HF)[S]. In SOLVD the term 'arrhythmic death' was used to describe presumed arrhythmic death in patients dying suddenly, whereas a broader definition of sudden death was used in MERIT-HE The inconsistency in interpretation of death narratives among centres is another major problem. Ziesche et al.[7] reported agreement only in 50% of cases when 21 SOLVD investigators independently classified 10 death narratives from the Valsartan Heart Failure Trial (V-HeFT) trial. Much of the inconsistency between centres can be reduced, however, if deaths are evaluated by end-point committees according to well-defined guidelines. However, many large trials, such as SOLVD, the Digitalis Investigative Group (DIG) trial[S], the U.S. carvedilol trials[9] and the Survival and Ventricular Enlargement (SAVE) trial[l°l, did not use end-point committees. In most trials, records are provided for cardiovascular death, sudden death, myocardial infarction and death caused by worsening heart failure. As with official death registries, the classification of death according to clinical records shows considerable discordance with autopsy data. In the Assessment of Treatment with Lisinopril and Survival (ATLAS) trial[111, autopsy revealed acute coronary findings (myocardial infarction) in 54% of heart failure patients with known coronary artery disease who died suddenly. That trial found myocardial infarction at autopsy in 27-5% of all heart failure patients who died during the study. The high incidence of myocardial infarction demonstrated by autopsy in heart failure patients who die suddenly is in sharp contrast to death records based on clinical data alone; on the basis of clinical judgement, the ATLAS trial recorded the occurrence of fatal myocardial


Cardiac Myocardial infarction Progressive heart failure Other cardiac Sudden death Non-cardiac Stroke Other Procedure-related

How do heart failure patients die?

D61

extrapolate it to the general heart failure population, we need to know more about the impact of treatment: its relationship with the severity of heart failure, its effects on cause of death and mode of death over time, and its interactions with sex and the underlying aetiology of heart failure.

• Diagnosisbased on clinical information [] Diagnosisbased on autopsy 100 90 8O

Impact of treatment

60

50 ~3

4O 30 20 10 Total Worsening Sudden death cardiovascular heartfailure deaths

Other

Figure 2 Data from a cohort of 171 patients included in the A s s e s s m e n t of Treatment with Lisinopril and Survival (ATLAS) triali221, comparing cause and mode of death based on clinical information alone with that based on autopsy results. The difference between the two groups is almost entirely due to an underestimation of acute myocardial infarction by clinical judgement alone (represented by an increase in the 'other' column for autopsy results).

infarction as only 5.8%. Thus, clinical records appear to underestimate considerably the incidence of acute myocardial infarction as a cause of death (Fig. 2). Notably, none of the patients classified in the ATLAS trial as having died of non-cardiovascular causes had acute coronary findings at autopsy. It is clear, therefore, that comparison of the cause or mode of death between trials is not meaningful. The steering or end-point committee of each individual trial defines its own rules, making it difficult to assess differences between trials. There is no universally accepted system that takes into account the problems in classifying cause of death in clinical trials. In order to overcome this problem, working groups have been established that are attempting to standardize classification. The Task Force on Sudden Death of the European Society of CardiologyE21 recently published its recommendations for the definition of sudden cardiac death. Within trials, however, definitions of the cause or mode of death are consistent, albeit subject to systematic error. Within-trial data may therefore be used to evaluate the effect of intervention. However, it is important to recognize that clinical trials are not conducted in populations that are representative of heart failure patients in the community. The typical trial population is highly selected. It consists mainly of men who are, on average, younger than the typical heart failure patient, and without serious comorbidities. Women, the elderly and patients with other chronic diseases are not adequately studied. To evaluate the existing data and

Improvements in treatment have significantly increased the life expectancy of heart failure patients, but have done little to reduce the proportion of patients who die from cardiovascular causes. The percentage of patients who die from cardiovascular causes was 88% in the SOLVD treatment trial[i], 88% in the metoprolol-treated patients in MERIT-HF[51 and 87% in the spironolactone-treated patients in the Randomized Aldactone Evaluation Study (RALES)[12]. It is true that, in MERIT-HF, significant reductions in both worsening heart failure deaths and sudden deaths in metoprolol-treated patients accounted for a significant reduction in overall cardiovascular deaths, as compared with placebo. However, in RALES there was no reduction in the proportion of patients who died from cardiovascular causes. Rather, the highly significant reduction in total mortality was mostly attributable to a reduction in non-cardiovascular deaths. It might be argued that modes of death compete, and that a treatment that reduces one mode or cause of death will increase the frequency of another. 'Therapeutic conversion' may therefore be taken as evidence of benefit. The most reliable measure of the real efficacy of cardiovascular treatments may be a reduction in the proportion of cardiovascular deaths within the context of an overall reduction in mortality. A failure to reduce cardiovascular deaths may be recorded due to fundamental problems of classification, or may occur because the underlying pathophysiological processes are slowed but not terminated.

Severity of heartfailure Patients with heart failure are readily classified according to the severity of their symptoms, and most clinicians are experienced in the use of the New York Heart Association (NYHA) classification. The SOLVD treatment[11 and prevention[ 131 trials allow a comparison of causes and modes of death in patients with left ventricular dysfunction (with and without overt symptoms of heart failure). In the SOLVD prevention trial 67% of patients were classified as NYHA class I and 33% as NYHA class II. In the SOLVD treatment trial 11% were classified as NYHA class I, 57% as NYHA class II, 30% as NYHA class III and 2% as NYHA class IV[~4]. There were other significant differences in the baseline characteristics o f the patients; more prevention trial patients had a history of myocardial infarction and revascularization procedures, but they were less likely to have a history of diabetes mellitus, systemic hypertension and chronic lung disease. Eur Heart J Supplements,Voh 4 (Suppl D) April 2002


70

D62

S. Orn and K. Dickstein

• TreatmentACEinhibitor [] PreventionACE inhibitor 10090" 80 ~3

70. 50 40 30 20 10 0

Total Worsening Arrythmia Myocardial cardiovascular heart failure infarction deaths

Gender

Figure 3

Relationship between cause of death and clinical status in the Studies of Left Ventricular Dysfunction (SOLVD) trial: patients with asymptomatic left ventricular dysfunction (SOLVD prevention)J131 compared with patients with symptomatic left ventricular dysfunction (SOLVD treatment)Iq. In the symptomatic group, more patients died from worsening heart failure. In the asymptomatic group, more patients died as a result of arrhythmias. ACE=angiotensin converting enzyme.

The mortality data showed that fewer SOLVD treatment trial patients died from acute myocardial infarction (9% versus 15%) and arrhythmias (23% versus 31%) than in the SOLVD prevention group, but more died from worsening

Experimental and clinical studies have suggested the presence of gender-specific differences in left ventricular remodellingE~5,161. There are also data that suggest genderrelated differences in the cause of death in the heart failure population. The Framingham Heart Study[ 171 reported that women with heart failure were 36% less likely to die than were men. A post-hoc analysis of the CIBIS II trial[ 18] attempted to evaluate gender-related differences in survival by analysing data from the 515 women in the trial (n = 2647). A total of 53 women died during follow-up. In women as compared with men, the relative risk for cardiovascular death was 0.64 ( P = 0.013) and for noncardiovascular causes it was 0.11 (P = 0.005). Male patients

[] NYHAII [] NYHAIII

[] NYHAIV 70 60 50 40 30

~-

2O

Figure 4 Severity of heart failure

10

Worsening heart failure

Sudden death

Eur Heart J Supplements,Vol. 4 (Suppl D) April 2002

Other

and mode of death in Metoprolol CR/XL Randomized Intervention Trial in Congestive Heart Failure (MERIT-HF)ISl. NYHA=New York Heart Association.


60

heart failure (46% versus 31%; Fig. 3). The same tendency was noted in post-hoc analyses of data from the MERIT-HF trial[51, which assessed total mortality and mode of death in relation to NYHA class at randomization. The proportion of sudden deaths generally decreased with increasing severity of heart failure according to NYHA functional class. Conversely, the proportion of patients who died from worsening heart failure increased with increasing severity of heart failure (Fig. 4). These intriguing results from SOLVD and MERIT-HF provide data that are central to our understanding of the mechanisms of death and progression of heart failure. One may speculate that many patients who die suddenly may do so because of underlying ischaemic disease before the process of remodelling has caused deterioration in NYHA class. These questions may be answered by looking at the mode and cause o f death in patients whose N Y H A classification improves after treatment.

How do heart failure patients die?

•

D63

Ischaemiccardiomyopathy

[] Non-ischaemiccardiomyopathy 35 30

~z 2

20

~

15

Sudden

Myocardial failure

Myocardial infarction

Other

Noncardiovascular

Figure 5 Assessment of Treatment with Lisinopril and Survival (ATLAS) triali221: comparison of mode and cause of death in autopsied patients diagnosed as having ischaemic or non-ischaemic dilated cardiomyopathy. were also more likely to die suddenly. These results are consistent with the findings of the Framingham Heart Study[~7], the first National Health and Nutrition Examination Survey (NHANES-1)[191 and the Flolan International Randomized Survival Trial (FIRST)E20]. Those studies included only a limited number of women, however, and further studies are needed in populations that include larger numbers of female patients.

Aetiology of heart failure In the Metoprolol in Dilated Cardiomyopathy (MDC) trial[a1], 383 patients with idiopathic dilated cardiomyopathy and symptomatic heart failure were studied. Only 42 patients died during the trial. As compared with patients receiving metoprolol, more patients receiving placebo underwent cardiac transplantation (19 versus 2; P = 0-0001). Sudden death occurred in 78% of patients receiving metoprolol and in 63% of patients receiving placebo. Death from progressive heart failure occurred in 22% of patients receiving metoprolol versus 26% of those receiving placebo. In the ATLAS trial[22], a specific analysis was made of mode of death in patients with ischaemic and non-ischaemic cardiomyopathy. The total number of deaths analysed was substantially larger than in the MDC trial (950 deaths in the ischaemic and 394 deaths in the non-ischaemic group). Cardiovascular mortality was generally higher in the ischaemic subgroup (41.9% versus 29.5% in patients with non-ischaemic heart disease). However, there was no

difference in the relative proportions of sudden deaths, which accounted for 44% of deaths in both groups. In the same trial, death from worsening heart failure occurred in 31% of patients with ischaemic cardiomyopathy and in 29% of those with non-ischaemic cardiomyopathy. There were more deaths from non-cardiovascular causes among patients with non-ischaemic heart disease. Not surprisingly, there was a difference between the groups in deaths clinically reported as related to myocardial infarction (8% in the ischaemic subgroup versus 5% in the nonischaemic subgroup). Interestingly, however, autopsy revealed the number of fatal myocardial infarctions in both subgroups to be much higher than estimated on clinical grounds (31.7% ischaemic and 17-6% non-ischaemic; Fig. 5). This clearly indicates that there may be underlying ischaemic disease even in heart failure patients diagnosed as having non-ischaemic cardiomyopathy.

Time and place of death In recent years there has been a trend toward fewer patients dying from cardiac causes outside the hospital setting[231. This trend also applies to the heart failure population. In a Scottish study based on a national database including 12,640 patients and 4877 deaths during a 3-year period[24], 73% of all heart failure deaths occurred in hospital. Median time from hospitalization until death was less than 10 days, and 45% of patients died within 90 days of discharge. Most deaths following discharge from hospital occurred within Eur Heart J Supplements, Vol. 4 (Suppl D) Apr/1 2002


25

D64

S. O r n a n d K. D i c k s t e i n

Pump failure I death

I

Figure 6 Interaction between underlying mechanisms and cause of death in heart failure.

the first year. Patients hospitalized with an acute myocardial infarction subsequent to a heart failure admission had a particularly poor prognosis. Once a patient had been admitted with an acute ischaemic event, this remained the dominant reason for later readmission.

Conclusion A n intricate interplay of underlying pathophysiological m e c h a n i s m s determines the progression and clinical presentation of heart failure (Fig. 6). The process of remodelling and neurohumoral activation, through central and peripheral mechanisms, increases susceptibility to arrhythmias and exacerbates the progression of ischaemic disease. Ischaemia induces arrhythmias and myocardial infarction, which in turn may cause increased neurohumoral activation and cardiac remodelling. Tailored heart failure therapy requires the identification and treatment of the detrimental pathophysiological factors in each individual patient. The goal of therapy in this large and heterogeneous population should be both to prolong survival and to increase the percentage of patients dying from noncardiovascular causes. While non-cardiovascular death is a robust parameter, subclassification of cardiovascular death is highly complicated because of the complex variation in clinical presentation of death in the heart failure population. We do know, however, that there is a strong relationship between mode of death and severity of heart failure, with patients in lower N Y H A classes being more likely to die suddenly. This observation may well be due to the underestimated effects of underlying ischaemic disease. In fact, according to autopsy data, ischaemia is the single most important cause of death in patients with heart failure. Ischaemia may cause arrythmias, worsening heart failure and sudden death. However, the role of stunning and hibernation is difficult to quantify. Aggressive treatment of underlying coronary Eur Heart J Supplements, Vol. 4 (Suppl D) April 2002

artery disease as well as supraventricular arrhythmias is therefore essential in the heart failure population. In conclusion, it is not possible to separate cause from simple association accurately in the pathophysiological maze of heart failure. In spite of major advances in the treatment of heart failure, the distribution between the modes of death remains largely unchanged. The understanding of how heart failure patients die requires an accurate description of the mode and cause of death. A simple and clinically relevant classification system is required if current controversies are to be resolved.

References [1] The 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. [2] Priori SG, Aliot E, Blomstrom-Lundqvist C et al. Task Force on Sudden Cardiac Death of the European Society of Cardiology. Eur Heart J 2001; 22: 1374-450. [3] Dickstein K, Kjekshus J. Comparison of the effects of losartan and captopril on mortality in patients after acute myocardial infarction: the OPTIMAAL trial design. Optimal Therapy in Myocardial Infarction with the Angiotensin II Antagonist Losartan. Am J Cardiol 1999; 83: 477-81. [4] Smith CJ, Scott SM, Wagner BM. The necessary role of the autopsy in cardiovascular epidemiology. Hum Pathol 1999; 29: 1469-79. [5] MERIT-HFStudy Group. Effect of metoprolol CR/XL in chronic heart failure: Metoprolol CR/XL Randomized Intervention Trial in Congestive Heart Failure (MERIT-HF).Lancet 1999; 353: 2001-9. [6] CIBIS-II Investigators and Committees. The Cardiac Insufficiency Bisoprolol Study II (CIBIS-II): a randomised trial. Lancet 1999; 353:9 13. [7] Ziesche S, Rector TS, Cohn JN. Interobserver discordance in the classification of mechanisms of death in studies of heart failure. J CardFail 1995; 1:127 32. [8] Digitalis Investigative Group. The effect of digoxin on mortality and morbidity in patients with heart failure. N Engl J Med 1997; 336: 525-33. [9] Packer M, Bristow MR, Cohn JN et al., on behalf of the US Carvedilol Heart Failure Study Group. The effect of carvedilol on morbidity and mortality in patients with chronic heart failure. N Engl J Med i996; 334: 1349-55. [10] Pfeffer MA, Brannwald E, Moye LA et al. Effect of captopril on mortality and morbidlity in patients with left ventricular dysfunction after myocardial infarction. N Engl J Med 1992; 327: 669-77. [11] Uretslcy BF, Thygesen K, Armstrong PW et al. Acute coronary findings at autopsy in CHF patients with sudden death. Results from the ATLAS trial. Circulation 2000; 102:611-6. [12] Pitt B, Zannad F, Remme WJ et al., on behalf of the Randomized Aldactone Evaluation Study Investigators. The effect of spironolactone on morbidity and mortality in patients with severe heart failure. N Engl J Med 1999; 341: 709-17. [13] SOLVD Investigators. Effect of enalapril on mortality and the development of heart failure in asymptomatic patients with reduced left ventricular ejection fractions. N Engl J Med 1992; 327: 68541. [14] Johnstone D, Limacher M, Rousseau M e t al., for the SOLVD investigators. Clinical characteristics of patients in Studies of Left Ventricular Dysfunction (SOLVD). Am J Cardiol 1992; 70: 894-900. [15] Litwin SE, Katz SE, Litwin CM et al. Gender differences in postinfarction left ventricular remodelling. Cardiology 1999; 91: 173-83. [16] Caroll JD, Caroll EP, Feldman T et al. Sex-associated differences in left ventricular function in aortic stenosis of the elderly. Circulation 1992; 86:1099 107.


I Arrhythmic death

Ischaemic death

H o w do heart f a i l u r e p a t i e n t s die?

metoprolol in idiopathic dilated cardiomyopathy. Metoprolol in Dilated Cardiomyopathy (MDC) Trial Study Group. Lancet 1993; 342: 1441-6. [22] Cleland JG, Thygesen K, Uretsky BF et al., on behalf of the ATLAS investigators. Cardiovascular critical event pathways for the progression of heart failure; a report from the ATLAS study. Eur Heart J 2001; 22:1601 12. [23] Capewell S, MacIntyre K, Stewart S e t al. Age, sex and social trends in out-of-hospital cardiac deaeths in Scotland 1986-95: a retrospective cohort study. Lancet 2001; 358: 1213-7. [24] Khand AU, Gemmell I, Rankin AC, CMand JGF. Clinical events leading to the progression of heart failure: new insights from a national database of hospital discharges. Eur Heart J 2001; 22: 153-64.

Eur Heart J Supplements, Vol. 4 (Suppl D) April 2002


[17] Ho KK, Anderson KM, Kannel WB et al. Survival after the onset of congestive heart failure in Framingham Heart Study subjects. Circulation 1993; 88:107 15. [18] Simon T, Mary-Krause M, Funck-Brentano C, Jaillon P, on behalf of the CIBIS II investigators. Sex differences in the prognosis of congestive heart failure: results from the Cardiac Insufficiency Bisoprolol Study (CIBIS II). Circulation 2001; 103: 375-80. [19] Schocken DD, Arriata M, Laver PEet al. Prevalence and mortality rate of congestive heart failure in the United States. J Am Coll Cardiol 1992; 20:301 6. [20] Adams KF, Sueta CA, Gheorgiade M et al. Gender differences in survival in advanced heart failure: insights from the FIRST study. Circulation 1999; 99: 1816-21. [21] Waagstein F, Bristow MR, Swedberg K et al. Beneficial effects of

D65

HOW DO HEART FAILURE PATIENTS DIE?

Recommend Documents