25 Prognostication

Recent advances in medical therapy, notably angiotensin converting enzyme (ACE) inhibitors, β-adrenoreceptor antagonists, and aldosterone antagonists, as well as device therapy have improved that prognosis such that even in patients with chronic heart failure (CHF) who are in New York Heart Association (NYHA) class IV, the annual mortality can be as low as 9.7%.¹ Nevertheless, many patients do not respond to medical therapy and remain symptomatically very limited with a poor prognosis. It is for such patients that cardiac transplantation and ventricular assist devices are options. However, identifying such patients is one of the great challenges of HF management.

The 1-year mortality following cardiac transplantation is approximately 19%.² The selection of candidates for cardiac transplantation is therefore heavily involved with identifying those patients whose annual mortality from HF exceeds this rate and who might therefore benefit prognostically from a transplant.

There are over 300 prognostic markers described in patients with HF, the most significant of which are shown in Table 25.1. Many studies have been carried out looking at clinical, haemodynamic, and neurohormonal variables to assist with risk stratification, although it is important to look at such data in the context of the latest disease-modifying therapy. The traditional markers, including left ventricular ejection fraction (LVEF) and the peak oxygen uptake (pVo₂), consistently offer useful prognostic information, although scoring systems involving a combination of markers have also been developed. More recently, neurohormones have been shown to demonstrate the greatest prognostic potential in identifying patients at the greatest risk of an adverse outcome.

The Framingham Heart Study showed that mortality increased with advancing age, with a 27% increase in mortality per decade in men and 61% increase per decade in women.³ This is confirmed by other studies which have shown that increasing age is an independent risk factor for all-cause mortality and HF hospitalization.^4,–9 However, despite the increased risk elderly HF patients currently face, their fate is substantially better than that it would have been 50 years ago. Further data from the Framingham study demonstrated a 5-year mortality of 54% in men and 40% in women in the period 1990–1999, compared to 65% and 66% in 1950–1969, in patients surviving at least 30 days after the onset of HF.¹⁰

Although there are occasional conflicting reports, women with HF are generally regarded to have a better outlook than men.^11,12 The Framingham study showed that between 1948 and 1988, women had a median survival time after diagnosis of 3.2 years compared to 1.7 years for men; after 5 years, 38% of women and 25% of men with CHF were alive.¹¹ In the period from 1990 to 1999, the 5-year survival had improved to 60% in women compared to 46% in men.¹⁰

A reduced risk in women has also been seen in modern therapeutic trials.^13,–17 However, in the Italian Network on Congestive HF Registry,¹⁸ no difference in outcome was found between men and women, and in the Studies of Left Ventricular Dysfunction (SOLVD) trial¹⁹, where an ischaemic aetiology was more common, women were found to have a poorer prognosis than men.

In a study by Adams et al.,²⁰ women with nonischaemic HF had a significantly better outcome than men (male RR = 3.08). In contrast, ischaemic HF had a similar outcome in both men and women. A similar finding was also shown in a pooled analysis of over of 11,000 patients from five modern therapeutic randomized trials in patients with reduced LVEF (MERIT HF, PRAISE, PRAISE II, PROMISE, and VEST).¹² In multivariable analysis, male sex was associated with significantly worse prognosis, particularly for those with a nonischaemic aetiology of HF.

There were similar findings in a comparison of outcomes in 2400 women and 5199 men in the CHARM trial, which included patients with both reduced and normal LVEF.¹⁵ Women had lower risks of most fatal and nonfatal outcomes; these differences were not explained by LVEF or the cause of HF.

In a prospective study of patients admitted with decompensated HF, African Americans had a similar mortality but greater functional decline and were around 8 years younger on presentation than white Americans.²¹ In the SOLVD registry,²² black patients with CHF were also at a greater risk of death and worsening HF, but had a higher prevalence of diabetes, prior stroke, and left ventricular dysfunction of a nonischaemic aetiology, making interpretation difficult. Physiological differences in the renin–angiotensin–aldosterone and neuroendocrine systems could account for any difference in outcome, with ACE inhibitors thought to be less effective at halting disease progression in black compared to white patients.

There is some evidence to suggest a significant seasonal variation in hospitalization and death from HF, with winter peaks most notable in the elderly population.^23,24 In one report at least a fifth of cases could be attributable to the associated seasonal increase in respiratory disease.²³

Chronic HF of ischaemic aetiology carries an greater risk of morbidity and mortality than that of a nonischaemic aetiology.¹³ Exceptions to this rule are infiltrative causes of myocardial disease, such as amyloidosis and haemochromatosis.²⁵ Although an ischaemic aetiology is an independent predictor of mortality in patients with a reduced ejection fraction, patients with mild coronary artery disease appear to have a similar 5-year survival to those with a non–ischaemic cardiomyopathy.⁸ There is some evidence that revascularization of ischaemic myocardium may improve prognosis,²⁶ although results from large prospective randomized trials are awaited.²⁷

Renal impairment is often associated with HF due to renal hypoperfusion, and the use of diuretics, ACE inhibitors, angiotensin receptor antagonists, aldosterone antagonist, and other concomitant medication. Serum creatinine concentration, which is often quoted as a barometer of renal impairment, is actually a poor indicator of renal function.²⁸ An estimation of the glomerular filtration rate (GFR) is better for the accurate assessment of renal function,²⁸ and the Modification of Diet in Renal Disease (MDRD) equations²⁹ have recently been validated in patients with severe CHF (see Fig. 18.3, p. 180).³⁰ A GFR below 60 mL/min/1.73 m² is associated with complications of renal disease.²⁸ Moreover, a reduced GFR is independently predictive of all-cause mortality in asymptomatic³¹ and symptomatic^31,–34 left ventricular systolic dysfunction. In advanced HF, however, NT-proBNP appears to be a superior marker of prognosis.³⁵

Diabetes mellitus is a strong independent predictor of mortality in patients with HF^36,37 and this increased risk was later shown to be in patients with an ischaemic cardiomyopathy, rather than those with a dilated cardiomyopathy.³⁸

Excessive intake of alcohol is also a strong independent predictor of mortality.³⁹ Importantly, with total abstinence from alcohol, patients with an alcoholic cardiomyopathy can demonstrate a significant improvement in LVEF and functional status.⁴⁰ However, the prognosis for those who continue to consume excess alcohol is poor.⁴¹

HF patients with major depression are at an increased risk of death as well as hospitalization.⁴² Social isolation is also a significant predictor of mortality.⁴³ Marital quality, as assessed by the marital satisfaction scale, predicts 4-year survival independent of NYHA class.⁴⁴

A higher NYHA class has frequently been shown to be an independent predictor of mortality.^8,17,18,45

In a cohort of mild to moderate CHF, the Minnesota Living with Heart Failure Questionnaire (MLHFQ) was found to be an independent predictor of 1-year mortality or worsening HF.⁴⁶ However, there is no correlation between MLHFQ and traditional prognostic indicators, such as LVEF and pVo₂.⁴⁷ Another health-related quality of life (HRQL) questionnaire has also been shown to predict mortality and CHF-related hospitalization.⁴⁸

Syncope in CHF, whether cardiac in origin or not, is independently predictive of sudden death.⁴⁹

A heart rate above 86/min and systolic blood pressure below 119 mmHg are independently associated with a poorer outcome in CHF.^17,18,50 The prognostic importance of elevated jugular venous pressure and a third heart sound in patients with HF was evaluated in a retrospective analysis of the SOLVD treatment trial.⁵¹ Both signs were associated with a significantly poorer NYHA class, but each was independently associated with an adverse outcome. A third heart sound was also an independent predictor of 1-year mortality in the Italian Network on Congestive HF Registry.¹⁸

In a prospective analysis of 452 patients, subjects were classified by clinical assessment into four profiles: dry–warm, wet–warm, wet–cold, and dry–cold, on the basis of the absence/presence of signs of congestion, and evidence suggesting adequate or inadequate perfusion. Patients who were either wet–cold or wet–warm had an increased risk of death plus urgent transplantation on multivariate analysis.⁵²

Obesity, as defined by a body mass index (BMI) greater than 31, is not associated with increased mortality in patients with advanced CHF after 5 years of follow-up.⁵³ Paradoxically, an elevated BMI was shown to be an independent predictor of improved survival. This could be partly explained by higher blood pressure in the overweight and obese groups, allowing a significantly greater use of disease-modifying therapy. Low body weight and significant weight loss predict increased mortality, however, possibly reflecting a higher degree of cytokine activation.^54,55

Although a meta-analysis of small-scale trials of rehabilitation suggested that there may have been a survival benefit from exercise training,⁵⁶ the near-definitive HF-ACTION trial showed no effect of a formal training programme on survival. Training was associated with better quality of life.^57,58

There is compelling evidence that ACE inhibitors,^59,–63, β-blockers^64,–66 and aldosterone antagonists⁶⁷ are associated with better survival. The relation between some other drugs and survival is less clear-cut.

In a retrospective study, high doses of diuretic (〉80 mg furosemide or equivalent per day) were independently associated with total mortality, sudden death, and pump failure death.⁶⁸ Another study corroborates this finding that diuretic dose relates to mortality.⁶⁹

Statin therapy is beneficial for the primary and secondary prevention of ischaemic heart disease^70,71. However, two recent studies of rosuvastatin have shown that this statin does not alter prognosis in patients with HF, whether or not it is due to coronary heart disease.^72,73

Although it appears to be a relatively safe antiarrhythmic agent in CHF, there is conflicting information about the effect of amiodarone on mortality. In the GESICA study⁷⁴, there was a 28% relative risk reduction in mortality, but the mortality reduction was not verified in the larger, placebo-controlled CHF-STAT study.⁷⁵

Although digoxin improves symptoms and reduces hospitalization in CHF, it has a neutral effect on mortality^76,77. However, a post-hoc analysis of the DIG trial suggests that male subjects with higher serum digoxin concentrations(〉1.2 ng/mL) have a higher mortality than patients receiving placebo.⁷⁸

The combination of hydralazine and isosorbide dinitrate is associated with a lower mortality than placebo,⁷⁹ but is less effective than enalapril.⁸⁰ However, the A-HeFT study was stopped early because it showed that the addition of hydralazine and isosorbide dinitrate to standard care (including 69% on ACE inhibitors and 74% on β-blockers) in black patients was superior to placebo (43% RRR in all-cause mortality).⁸¹

Several studies have shown hyponatraemia to be an independent predictor of mortality^82,–86 and hypokalaemia to be an independent predictor of sudden cardiac death⁸². Serum magnesium is not an independent risk factor for death in patients with moderate to severe CHF.⁸⁷

Over the past decade, an increasing number of studies has demonstrated that a significant proportion of patients with CHF (10–49%)have detectable troponin (Tn).⁸⁸ Persistently elevated Tn concentrations are associated with an adverse prognosis regardless of the aetiology of HF.⁸⁹ A raised Tn level is also associated with an adverse outcome in acute, and acute decompensated, HF.^88,90

Serum uric acid is strongly related to circulating markers of inflammation in patients with HF.⁹¹ Several studies have shown that a raised uric acid concentration is independently predictive of an impaired prognosis.^69,92 Interestingly, a retrospective study has suggested that long-term use of high-dose allopurinol could be associated with a reduced mortality, possibly by negating the adverse effect of an elevated urate concentration.⁹³

Abnormalities in liver function tests adversely affect prognosis in CHF, most notably AST and bilirubin.⁶⁹

Inflammatory markers such as C-reactive protein (CRP), as well as the interleukins IL-4 and IL-6, increase during episodes of acute decompensation, returning to baseline once patients become compensated.⁹⁴ Patients admitted with decompensated HF who subsequently died or required readmission following discharge had a higher baseline CRP concentration than those who remained event free.⁹⁵ In the Val-HeFT trial, the cumulative likelihood of death or a first morbid event increased progressively with quartiles of serum CRP.⁹⁶

Anaemia is an independent predictor of mortality in patients with new onset HF,⁹⁷ mild to moderate CHF³⁵ and advanced CHF.¹⁰⁵ Indeed, in the latter study, patients in the lowest haemoglobin (Hb) quartile were 86% more likely to die at 1 year than those in the highest Hb quartile. Hb is a significant predictor of progressive pump failure but not sudden death. Treatment of anaemia in CHF with subcutaneous erythropoietin and intravenous iron improves some aspects of the condition,⁹⁸ but large randomized controlled trials of erythropoietin in anaemic HF patients are awaited.

Red cell distribution width is a readily available measure of the variation in erythrocyte volume. As well as offering prognostic information in CHF,⁹⁹ it is also a marker of adverse outcome in patients with acute HF, regardless of anaemia status.¹⁰⁰ It has prognostic value additional to that of B-type natriuretic peptide.¹⁰¹

In a retrospective analysis of the SOLVD study, a white cell count (WCC) greater than 7000/mm³ was an independent predictor of all-cause and cardiovascular mortality in patients with LVSD of ischaemic aetiology, but not in those with a dilated cardiomyopathy.¹⁰²

Although platelet activity is increased in 22% of patients with stable CHF (vs 7% in normal controls), the degree of activation was similar in CHF of ischaemic and nonischaemic aetiologies and platelet activation was not related to either NYHA class or subsequent outcome.¹⁰³

An erythrocyte sedimentation rate (ESR) above the median value (14 mm/h) is associated with a poor survival, independent of age, NYHA class, LVEF, and peak Vo₂.^104,105

Atrial fibrillation (AF) is much more common in patients with CHF than in normal individuals, with prevalence ranging from 10% to 50%. Data assessing the outcome of AF in patients with CHF have been conflicting, with most showing no impact on survival.

In the V-HeFT study, atrial fibrillation did not increase major morbidity or mortality in mild to moderate HF.¹⁰⁶ A follow-up of patients in the SOLVD trials with asymptomatic left ventricular dysfunction or NYHA class II–III HF found that AF (present in only 6.4%) was a significant predictor of all-cause mortality,¹⁰⁷ primarily due to pump failure, as there was no increase in mortality from arrhythmia. Around 18% of patients in the CHARM series had AF at baseline which was independently linked to mortality, both in patients with either low or preserved LVEF.¹⁰⁸

One study noted an improvement in the prognosis of AF in patients with CHF with the advent of ACE inhibitor therapy, amiodarone, and avoidance of class Ia antiarrhythmic drugs.¹⁰⁹

In a prospective study, a standard deviation of R-R interval of less than 100 ms identified patients at increased risk of death due to progressive pump failure but not sudden cardiac death.⁸² However, conversely, a retrospective analysis of data from the Veterans Affairs’ Survival Trial of Anti-arrhythmic Therapy in Congestive Heart Failure, the lowest quartile of the standard deviation of R-R intervals was an independent predictor of sudden death, as well as total mortality.¹¹⁰

QRS prolongation (〉120 ms) is an independent predictor of both total mortality and sudden death in patients with severe CHF (LVEF〈30%).¹¹¹ In moderate CHF (LVEF 30–40%), however, QRS duration is associated only with increased mortality but not sudden death. Right bundle branch block is not associated with excess mortality.

In the Italian Network Network on Congestive registry,¹¹² left bundle breach block (LBBB)—diagnosed with a QRS duration in excess of 140 ms—was found in 25.2% of individuals with CHF. LBBB was more common in women and in patients with dilated cardiomyopathy, and was an independent predictor of all-cause death and sudden death in patients with CHF. LBBB was also an independent predictor of mortality in the CIBIS-2 study¹⁷. In two subsequent studies, accelerated QRS-interval widening was independently associated with deterioration of cardiac function,¹¹³ and death or need for urgent cardiac transplantation.¹¹⁴

In a substudy of Diamond-CHF,¹¹⁵ QT dispersion was not a predictor of outcome and in another study QT dispersion and maximum QT interval were found to be univariate but not independent predictors of all-cause mortality and sudden death.¹¹⁶

Patients with moderate to severe CHF who have evidence of nonsustained ventricular tachycardia (NSVT) on 24-h Holter monitoring have an increased risk of total mortality and sudden death.^82,117

A higher cardiothoracic ratio (CTR) is predictive of the risk of worsening symptoms, hospitalization, and mortality,¹¹⁸ particularly in the patient with CHF and a low LVEF. In addition, UK-HEART (United Kingdom-Heart Failure Evaluation and Assessment of Risk Trial)—a multicentre prospective study designed to identify noninvasive markers of death and mode of death in patients with CHF—showed that a greater CTR was also predictive of sudden death.¹¹⁹

Left ventricular dimensions (end-systolic and end-diastolic) independently predict all-cause mortality and sudden cardiac death.^82,120 Furthermore, HF patients with severe mitral or tricuspid regurgitation on echocardiography are also at increased risk of death.¹²¹

In a study using dobutamine echocardiography, patients with moderate to severe left ventricular dysfunction and viable or ischaemic myocardium had an adverse prognosis, independent of age and LVEF.¹²²

There are few prognostic data from studies of cardiac magnetic resonance imaging (CMR) in patients with CHF. However, in a study of 279 patients with poor-quality echocardiograms, the presence of a reduced left ventricular function (〈40%) on CMR stress testing was independently associated with all-cause mortality.¹²³

A lower ejection fraction is associated with a poorer outcome in patients with CHF.^17,118,124 A improvement in LVEF during exercise radionuclide ventriculography in mild CHF is a strong independent predictor of survival.¹²⁵

In a small study by Di Salvo et al.,¹²⁶ a right ventricular ejection fraction of 35% or more at exercise was an independent predictor of survival and more potent than peak Vo₂ or percentage of predicted Vo₂.

Peak oxygen consumption (pVo₂) of less than 10 mL/kg/min is associated with a 1-year mortality of 77% compared with 21% for those patients who achieved a pVo₂ of 10–18 mL/kg per min.¹²⁷ pVo₂ has become widely accepted as a marker of prognosis,^{7,118,128,–131} as well as a marker for the timing of transplantation.¹³² Initially, those patients with a pVo₂ of less than 14 mL/kg/min were identified as a high-risk cohort of patients, with a 1-year mortality of 30% compared with those with a value greater than 14 mL/kg/min who had a 1-year mortality of 6%. However, with widespread use of β-blockade, the cut-off has fallen to 12 mL/kg/min.¹³³

Many studies have since attempted to ‘fine-tune’ the predictive power of pVo₂. As pVo₂ is affected by age, sex, body composition, and body conditioning, the percentage of predicted pVo₂ might be a better marker, but added minimal precision over pVo₂ alone.¹³⁴ As oxygen consumption is corrected for total body weight, and body fat consumes very little oxygen, body fat adjusted pVo₂ (pVo₂ lean) may provide greater prognostic precision, especially in female and obese subgroups.¹³⁵

The six-minute walk test is simple and noninvasive, and independently predicts morbidity and mortality in patients with mild–moderate¹³⁶ as well as advanced¹³⁷ CHF. However, in other studies, the six-minute walk test was only able to predict mortality in univariate analysis, and was not an independent predictor of survival.^138,139

There are inconsistent reports of the predictive power of right heart catheter data in CHF. No single variable consistently predicts outcome, although many studies have found the pulmonary capillary wedge pressure (PCWP) to be independently predictive of mortality.^140,141 Other studies have found other right heart pressure measurements to be similarly predictive of outcome.^142,143 Most studies were conducted prior to the routine use of modern disease-modifying therapy and so their relevance to today’s practice is uncertain. There is a role for right heart catheterization in patient selection for cardiac transplantation, however, since an increased pulmonary vascular resistance has consistently been shown to increase the risk of early graft failure.^144,145

This 52-amino-acid peptide is almost ubiquitously expressed throughout the human cardiovascular system. It has potent vasodilating and natriuretic effects, and is raised in HF in proportion to the severity of the disease.¹⁴⁶ Although adrenomedullin is an independent predictor of death or urgent cardiac transplantation in patients with mild–moderate CHF,¹⁴⁷ it is not as powerful a predictor of prognosis as endothelin-1 or NT-proBNP.^148,149 However, as plasma levels of adrenomedullin can be inaccurate because of its short half-life, an assay has been developed for the more stable midregional portion of the pro-peptide (MR-proADM) which may offer further prognostic information.¹⁵⁰

A high plasma noradrenaline (NA) level is an independent predictor of morbidity and mortality in LVSD patients with NYHA class I–II.¹⁵¹ In a substudy of the Val-HeFT trial, a NA level above the median at baseline was an independent predictor of mortality, although not as powerful as B-type natriuretic peptide.¹⁵² Interestingly, subjects with the greatest increase or greatest decrease in NE over the first 4 months of the study had the highest mortality risk, which complements preliminary data from the BEST study.¹⁵²

Endothelin-1 is a 21-amino-acid structure with potent and long-lasting vasoconstricting properties that is elevated in patients with CHF. It independently predicts mortality, clinical deterioration, and the need for cardiac transplantation in patients with CHF.^153,–157 However, one study showed that there was no difference in endothelin-1 concentrations between patients with mild CHF and healthy controls.¹⁵⁸ Endothelin also correlates positively with the degree of pulmonary hypertension in CHF,¹⁵⁹ which in itself has been shown to alter prognosis unfavourably.¹⁶⁰

Atrial and B-type natriuretic peptides (ANP and BNP) are polypeptides that are produced in response to cardiac stretch. They stimulate natriuresis, induce vasodilatation, and are antiproliferative. They also inhibit the renin–angiotensin–aldosterone and sympathetic nervous systems.

ANP and NT-proANP are raised in patients with CHF,^161,162 correlating closely with the severity of CHF, and are associated with increased mortality.¹⁶³ The midregional segment of the pro-atrial natriuretic peptide molecule (MR-proANP) is more stable in plasma than either proANP or mature ANP and is emerging as a promising biomarker. MR-proANP is an independent predictor of mortality in acutely decompensated HF,¹⁶⁴ and may be a better marker than BNP and its variants in patients with CHF.^165,166

BNP and its N-terminal inactive fragment (NT-proBNP), are increased in both symptomatic and asymptomatic LVSD,¹⁶¹ increasing in proportion to the severity of CHF.¹⁶⁷ Currently, they appear to be the most potent prognostic markers available in HF, and are predictive of morbidity and mortality in asymptomatic or minimally symptomatic LVSD,¹⁵¹ in mild–moderate CHF,¹³⁸ and in patients with advanced HF referred for consideration of cardiac transplantation.¹⁶⁸

Tumour necrosis factor α (TNFα) is a proinflammatory cytokine and is increased in patients with severe CHF, particularly in those with cachexia.^169,170 Although TNF-α has not been shown to be an independent marker of prognosis, tumour necrosis factor soluble receptor-1 has.¹⁷¹

IL-6 is a proinflammatory and vasodepressor cytokine that mediates both inflammatory and immune responses: like TNFα, it is increased in patients with HF. In a study of NYHA class III patients, IL-6 was an independent predictor of mortality at 1 year, and was as least as predictive as LVEF.¹⁷² In contrast, there were no significant differences in the plasma concentrations of IL-1, IL-10, IL-12, and TNFα between survivors and non–survivors. Plasma concentrations of IL-6 follow a circadian rhythm, peaking at midnight.¹⁸⁵

Several emerging biomarkers have shown promise as predictors in HF, and are currently being investigated (Table 25.2).¹⁷³ Many of them arise from our greater understanding of the pathophysiological processes in HF, in particular the components of novel neurohormonal pathways with possible roles—both protective and deleterious.

Retrospective analyses have been used to develop composite scoring systems as predictive models that generate a more accurate and individual estimate of prognosis than single variables.^174,–176 As with the single-variable studies, however, the scoring systems quickly become out of date with the development of new therapies, and the models are often derived and validated in different populations of patients.

The EFFECT model was retrospectively derived and tested, and was intended for use in patients hospitalized for HF.¹⁷⁶ The derivation cohort included patients from the EFFECT study who presented with HF between 1999 and 2001. The model was then validated in a separate cohort presenting between 1997 and 1999. Multiple clinical characteristics (age, respiratory rate, systolic pressure, blood urea nitrogen, and serum sodium concentration) and comorbidities were included and the resultant scores correlated with 30-day and 1-year mortality. However, no information was given on background therapy, and the model was created prior to the routine use of natriuretic peptides. An online calculator is available atwww.ccort.ca/CHFriskmodel.aspx.

One commonly used scoring system is the Heart Failure Survival Score (HFSS),⁸⁵ which was developed and validated in patients with advanced HF (NYHA III and IV). The score stratifies patients by risk—low, medium, and high (equating to a 1-year survival rate of 88%, 60%, and 35%, respectively)—of death or urgent transplantation and incorporates seven variables (heart rate, mean blood pressure, serum sodium, ejection fraction, pVo₂, presence of ischaemic heart disease, and conduction delay on electrocardiography; Fig 25.1). In an invasive version of the HFSS, PCWP was included as an eighth variable. Unfortunately, due to the timing of the original model, only a small percentage of patients involved in the initial HFSS study were established on current standards of medical or device therapy. In particular, the use of β-blockers and spironolactone was very low, and both can alter some of the variables used in the scoring system, as well as the prognosis of CHF. However, a later study validated the HFSS in patients on current therapy.¹⁷⁷ A conflicting study found that a simplified risk stratification model with only two variables—LVEF and Vo₂ or six-minute walk test—was superior to the HFSS.¹⁷⁸

The Seattle Heart Failure Model (SHFM) was derived from the PRAISE-1 database of 1125 HF patients with the use of a multivariable Cox model. It was subsequently prospectively validated in five additional cohorts: ELITE-2, Val-HeFT, UW, RENAISSANCE, and INCHF involving 9942 HF patients and 17 307 person-years of follow-up.¹⁷⁴ Although no patients in the derivation cohort were on β-blockers, up to 72% of the validation population were. Importantly, the validation cohorts also included patients with a wide range of ages (14–100 years), ejection fractions (1–75%), and HF symptoms (NYHA class I–IV).

The SHFM accurately predicts survival of HF patients (Fig. 25.2) with the use of commonly obtained clinical characteristics (NYHA class, ischaemic aetiology, diuretic dose, LVEF, systolic BP, serum sodium, haemoglobin, percentage lymphocytes, uric acid, and serum cholesterol). It has a distinct advantage over the HFSS which relies on pVo₂ to calculate a score. The Seattle model also provides information about the likely mode of death. In an analysis of 10 538 ambulatory patients with predominantly systolic HF (NYHA class II–IV), the score was predictive of the risk of sudden death and of pump failure.¹⁷⁹

Interestingly, renal function was not an independent predictor in the SHFM, and two extremely powerful prognostic markers—VO₂ and BNP/NT-proBNP—were not included in the development of the model, as the data were available in fewer that 1% of the patients in the six data sets. An online calculator is available at www.SeattleHeartFailureModel.org.

In clinical trials, the 1-year mortality of patients with severe CHF can be as low as 9.7%,¹⁸⁰ but perhaps the real challenge is identifying those patients at greatest risk of death, and therefore those who would benefit most from advanced therapies such as ventricular assist devices and cardiac transplantation.

Many variables have prognostic power in CHF but the markers vary in their success for predicting outcome because of the heterogeneous nature of HF and the populations in which the variables were studied. Many of the described variables are inter-related and although they may be strong univariate markers of prognosis, they can be competitively removed in any multivariate model.

The ideal prognostic tool would be cost-effective, readily available, easy to measure, reproducible, minimally invasive, and both sensitive and specific. Until recently, the best single marker of prognosis was pVo₂, but many patients with CHF fail to achieve a true pVo₂, it is difficult to perform, and is not widely available. Many of the studies proposing the use of specific markers were carried out before the widespread use of β-blockers, ACE inhibitors, and spironolactone.

The HF survival score is currently used by many transplant centres to help identify patients who would benefit most from cardiac transplantation.¹⁷⁵ However, the score relies on seven variables being available for each patient, including pVo₂, and without one variable the score is useless.

The variables with perhaps the most promise as prognostic tools in CHF are the B-type natriuretic peptides. They are relatively cheap, are stable in whole blood for 3 days and can be measured on most modern autoanalysers.¹⁸¹

However, in view of the multifactorial nature of CHF, it seems likely that a combination of variables would be useful in the prediction of prognosis in HF. Large prospective studies are therefore required in order to further clarify the subject.