The natural history of heart failure shows a general decline in function and quality of life, marked by ‘troughs’ representing decompensation into acute exacerbations of heart failure (Figure 9). It should be noted that a sudden decompensation due to a new insult is possible at any point in established heart failure. However the general pattern is that following an episode of acute decompensation, some function will be regained. Over time, these exacerbations are likely to become more frequent and cause more severe symptoms, as the baseline becomes lower (Figure 9) (Allen et al., 2012; Chaudhry & Stewart, 2016).
The New York Heart Association (NYHA) has a classification system for the stages of heart failure, which is discussed in more detail in the diagnosis of advanced heart failure section. Although it was created in 1928, it remains one of the most important functional scores in the ‘staging’ of heart failure.
In general, the prognosis of an individual is difficult to estimate due to the variable nature of heart failure. Allen et al., (2012) suggest that over 100 variables have been found to be independently associated with mortality and admission to hospital. The most important of these include:
Scoring systems have been established, looking both at outpatient prognosis and outcomes following admission for acute decompensation (Allen et al., 2011). Scoring systems that stratify risk could aid in clinical decisions surrounding admission and the timing of discharge, as some studies have reported that patients considered high-risk are discharged to home and may subsequently die despite being deemed safe to discharge (Lee et al., 2010). At the other end of the spectrum, some patients considered low-risk are being admitted to hospital, which can increase exposure to hospital-related adverse events and lead to inefficiently assigned resources (Pang & Schuur, 2014). Results were recently published from the Acute Congestive Heart Failure Urgent Care Evaluation (ACUTE) study that prospectively validated the Emergency Heart Failure Mortality Risk Grade (EHMRG7), a tool to predict 7-day mortality risk, and an extended model of EHMRG7 that predicts 30-day mortality (EHMRG30-ST). The EHMRG7 and EHMRG30-ST tools stratified patients into very low, low, intermediate, high, and very high risk of mortality within 7 or 30 days. As predicted, patients within the higher risk categories had higher mortality rates at both 7 and 30 days (Figure 10) (Lee et al., 2018).
In addition to validating the EHMRG tools, the investigators also compared the risk stratification results to individual clinical judgement. Before the results were displayed, the treating physician entered their prediction of 7-day mortality and their proposed management plan. Interestingly, physicians estimated higher mortality rates for all categories except for the final very high-risk category in comparison to EHMRG7 predictions (Figure 11). Comparing these predictions to the reported mortality rates seen in Figure 10 found EHMRG7 to be superior to physician-estimated risk of 7-day mortality (c-statistic 0.81, 95% CI 0.75–0.87 vs. c-statistic 0.71, 95% CI 0.64–0.78). Using a combination of EHMRG7 and physician-estimated risk was also superior to physician-estimated risk alone (p=0.003). These tools could predict mortality more accurately and guide future clinical decisions as study-reported hospital admissions could reduce by 9.8% if admissions and discharges were based on the EHMRG7 risk stratification (Lee et al., 2018).
As there is such a variable course and prognosis, it is difficult to apply generalities to the individual patient. However, heart failure in general terms has a worse prognosis than most forms of cancer, with 5-year mortality following diagnosis consistently over 75% (McMurray & Stewart, 2000; Avellino et al., 2011). A study published in 2011 found some improvement in outcomes following admission with heart failure between 1995 and 2000, but still noted that of 2,587 patients admitted in 2000, the 3-month, 1-year and 5-year mortality rates were 18.8%, 39.7%, and 82.9%, respectively (Park et al., 2011). Other, older studies have had similar findings; the Rochester Epidemiology Project found the 1-year and 5-year mortality rates were 23% and 67% in a relatively small population of 141 patients.
A larger prospective study (n = 14,407) evaluated non-hospitalised people who self-reported symptoms, finding a 10-year mortality of 42.8% in the general population, much lower than other studies based on diagnosis following hospital admission; the 10-year mortality in subjects aged 65–74 was more reflective of previously established statistics with a mortality of 65.4% (Allen et al., 2011). Meanwhile, data from the Atherosclerosis Risk in Communities (ARIC) Study Heart Failure Community Surveillance observed a 1-year mortality rate of 34% in patients with acute decompensated heart failure with preserved ejection fraction. Predictors of mortality included a higher heart rate, being underweight (BMI <18.5) and higher natriuretic peptide levels (Thorvaldsen et al., 2017).
Early changes such as these may also identify patients at increased risk of poor outcomes and prolonged hospitalisation. One prospective study identified 103 patients with acute decompensated heart failure requiring intravenous diuretic therapy and analysed patient outcomes based on urine sodium concentration measurement. Patients with a low urine sodium concentration after the first diuretic dose were more likely to reach the composite primary endpoint of mechanical circulatory support device placement, discharge from hospital on inotropes, or death at admission or during the follow-up period (HR 2.4, 95% CI 1.02–5.66; p=0.045). Mortality was highest among those with a low urine sodium concentration (≤60 mmol/L) compared to a high concentration (>60 mmol/L) (22.6% vs. 8.3%, p=0.0575), along with more frequent inotrope use at discharge (12.9% vs. 1.4%, p=0.03). The results from this study suggest that early spot urine sodium concentration measurement after the initial dose of intravenous diuretics may aid in triage to identify the appropriate intensity of therapy needed to relieve congestion (Luk et al., 2018). However, there are limitations to the practicality of this test in clinical practice.
The clinical category that a patient presents with may also influence the patient’s outcomes. The 1-year mortality rate differed substantially between classifications (Chioncel et al., 2017) (Table 2):
Table 2. 1-year mortality rate associated with different acute heart failure classification categories (Chioncel et al., 2017).
A recent small-scale study sought to determine what other factors were predictive of mortality in patients admitted with acute heart failure. Of the 322 patients investigated, they noted three characteristics that were the most indicative of mortality rate: age >85 years (OR=1.5; 95% CI 0.8–2.7; p=0.01), creatinine clearance <30 mL/min (OR=2.6; 95% CI 1.5–5.0; p<0.001) and Nt-proBNP (N-terminal pro-B-type natriuretic peptide) >5000 pg/mL (OR=2.2; 95% CI 1.2–4.0; p<0.001) (Marchetti et al., 2017). A similar study in 287 patients with acute heart failure identified Nt-proBNP, serum sodium and soluble ST2 as potential predictors of 1-year mortality (Jin et al., 2017). Further evidence for the predictive value of low urinary sodium concentrations (UNa) has been observed in 669 Japanese patients with acute heart failure. In this prospective registry study, patients with lower UNa at admission were more likely to have a history of prior heart failure admission, β-blockers and diuretics use, and lower blood pressure and serum sodium levels, but higher blood urea nitrogen, estimated glomerular filtration rate, blood glucose and troponin T levels, than patients with higher UNa. Furthermore, lower UNa was associated with worse long-term outcomes, including significantly higher incidence of worsening renal function (Honda et al., 2018). Meanwhile, hyperkalaemia may also be an indicator for increased 90-day mortality with β-blockers potentially having a protective effect in these patients (Legrand et al., 2018).
An assessment combination of hepatic dysfunction, as measured by the Fibrosis-4 (FIB4) index, and renal dysfunction may improve the predictability of prognosis in patients with acute decompensated heart failure, according to a retrospective single centre study of 758 patients admitted for advanced heart failure (Shibata N et al., 2019).
Non-alcoholic fatty liver disease (NAFLD) has also been shown to increase the risk of all-cause mortality in elderly patients with acute heart failure. In these patients, this increasingly prevalent disease has previously been shown to be associated with an increased risk in 1-year all-cause and cardiac rehospitalisations (Valbusa et al., 2017). Meanwhile, the association between NAFLD and all-cause mortality was revealed in an Italian study of 264 patients admitted between 2013 and 2015 with a diagnosis of acute heart failure; 57.9% of patients had NAFLD and 60% of these patients died either in-hospital or post-discharge during a maximum follow-up period of 58 months (mean 23.3 months). Univariate regression analysis revealed NAFLD was associated with a ~70% increase in risk of all-cause mortality; the significance of this association was also confirmed with multivariate Cox regression analysis (adjusted HR 1.82: 95% CI 1.22–2.81; p<0.005). In addition, the association appeared to increase among patients with advanced NAFLD fibrosis (Valbusa et al., 2018).
Malnutrition is common among patients hospitalised for acute heart failure and in a study of 145 patients, proBNP levels were directly correlated with nutritional status. In this patient population, malnutrition was observed to be a mediator of disease progression and offers a poor prognosis (Agra Bermejo et al., 2017).
A study examining 15,828 patients with acute heart failure determined that the precipitating factor also plays a role in determining 90-day mortality, and more than half (8,784) had an identifiable precipitating cause. Investigators suspected that part of the reason behind persistently high mortality in acute heart failure is that the condition is clinically heterogeneous. To that end they tested whether the precipitating pathology independently predicted 90-day mortality. They determined that acute heart failure precipitated by acute coronary syndrome (ACS) or infection was associated with a higher mortality versus those with undetermined precipitating factors (ACS: HR=1.69; 95% CI 1.44–1.97; p<0.001), (infection: HR=1.51; 95% CI 1.18–1.92; p<0.001). Heart failure precipitated by atrial fibrillation (AF) and uncontrolled hypertension was associated with a significantly lower 90-day mortality (hypertension: HR=0.72; 95% CI 0.60–0.86, AF: HR=0.56; 95% CI 0.42–0.75) (Figure 12) (Arrigo et al., 2017). While this is perhaps an unsurprising finding, it does emphasise the importance of not viewing heart failure as a single clinical entity.
Similarly, the presence of functional mitral regurgitation (FMR) in patients with acute heart failure and reduced left ventricular ejection fraction has been assessed as a predictor of adverse outcomes. In a prospective study of 938 consecutive patients, FMR was assessed at 120 ± 24 hours post-admission with 56.8% having none or mild (grade 0 or 1), 26.9% having moderate (grade 2) and 16.2% having severe (grade 3 or 4) FMR. After multivariable adjustment, patients with moderate or severe FMR had a significantly increased risk of short-term adverse clinical outcomes (De la Espriella et al., 2017).
In patients with acute heart failure, an inverse relationship has been described between frequency of cardiovascular events and body mass index (BMI). In 808 acute heart failure patients, it was recently shown that severely thin (BMI <16 kg/m2) patients were more likely to have valvular disease and be at an increased risk of 910-day mortality (HR=3.372; 95% CI 1.362–8.351) than normal/underweight (BMI ≥16–<25 kg/m2) patients. Meanwhile, overweight patients (BMI ≥25–<30 kg/m2) were younger and more likely to have hypertensive heart disease than the normal/underweight group. They also had a significantly better prognosis than severely thin, normal/underweight and obese (BMI ≥30 kg/m2) patients with a 910-day mortality HR of 0.615 (95% CI 0.391–0.966) patients that were of normal weight or underweight (Matsushita et al., 2017). A recent study explored whether this effect persisted in patients with co-morbid type-II diabetes. A multicentre observational study was carried out, examining 2,492 patients with acute heart failure and type-II diabetes – they found that underweight patients (BMI <20.0 kg/m2) had a higher 3-month mortality risk (OR=2.04; 95% CI 1.02–4.08), while obese patients (BMI 30.0–34.9 kg/m2) had a lower mortality over the same time period (OR=0.53; 95% CI 0.34–0.83). There was no significant effect for those of normal and overweight patients (Abi Khalil et al., 2017).
While several studies have investigated factors that are predictive of mortality, a small study of 71 hospitalised acute heart failure patients recently demonstrated that assessing a 6-minute walking distance prior to discharge could predict 30-day readmission. It was shown that for every additional 100 feet (30.48 meters) walked, the odds of readmission within 30 days were reduced by 16% (McCabe et al., 2017). Similar findings were seen in a recent Japanese study of 1,253 patients with heart failure analysing decline in activities of daily living (ADL). Patients were categorised into four groups based on their mobility: independent outdoor walking, independent indoor walking, indoor walking with assistance, or unable to walk both before admission and at discharge. Decline in ADL was defined as a downgrade in the four categories before discharge compared to admission. Within the independent outdoor walking group, patients with heart failure who showed a decline in ADL had a higher risk of hospitalisation (HR 2.38; 95% CI 1.56–3.49; p<0.001) and a higher risk of all-cause mortality (HR 3.24: 95% CI, 1.70–5.83; p<0.001) compared to the non-decline group. These findings could aid in decision making for early-phase cardiac rehabilitation (Takabayashi et al., 2018).
An important distinction to be aware of is the need to manage any intervention against the corresponding expected change in quality of life. This is discussed in greater detail in the advanced heart failure prognosis and quality of life section.