The recent guidelines from the American Society of Echocardiography and European Society of Echocardiography have defined an abnormal ejection fraction (EF) of the left ventricle, as one that is <55%.1 Owing to the fact that it is a continuous biological variable, there is inevitable debate over what constitutes mild, moderate and severe left ventricular (LV) dysfunction across the ranges of EF. Up to now, the lower limit of normal in clinical practice has usually been set at 40%.2

Despite the recent guidelines,1 there has been little debate or evidence to suggest altering the lower limit of normality to include patients with EFs of 50–54%. These guidelines therefore represent a step change in the definition of (echocardiographic) LV systolic dysfunction and will include many more patients into the category of “impaired LV function” with EFs >50%. The main limitation of this approach is how best to define risk. The cut-off points suggested for a single parameter can vary broadly for the risk of death, myocardial infarction, congestive heart failure and atrial fibrillation. Also, much of the literature applies to specific populations (eg, after myocardial infarction, the elderly) and not to overall cardiovascular risk.1

The recommended new cut-off EF value of 54% is based on expert opinion, interpreting data from the limited number of studies that have reported that an EF <54% is associated with moderate adverse outcomes.1 For example, Roman et al followed up 486 patients undergoing echocardiography to determine the prevalence of carotid atherosclerosis and to examine its relationship to LV hypertrophy.3 Two hundred and seventy-seven subjects were normotensive and 209 subjects had untreated hypertension, and all were asymptomatic with no overt manifestations of cerebrovascular disease. The investigators found that 392 patients were essentially without any evidence of carotid plaques and 366 patients had no evidence of carotid plaques or LV hypertrophy.3 The lower limit of the 95% confidence interval of the EF of the 366 normal patients was 54% and this value was therefore used as the cut-off point for LV systolic dysfunction in two subsequent studies—the Strong Heart study4 5 and HyperGEN study.6 It is important to note that the method used to calculate the EF in the Roman study is not reported, nor why 366 patients were classified as “normal” although only 277 patients were normotensive.

In 2001, the Strong Heart study documented the prevalence and correlates of LV systolic dysfunction in 3184 American Indians.4 5 LV systolic dysfunction was divided into severe and mild, based on EF calculated by echocardiography using M-mode and two-dimensional measures by means of the Teicholz method (EF of 40–54% was defined as mild and EF <40% as severe). In this study, 86% (n = 2739) of patients had normal LV function, 11.1% (n = 352) had mildly depressed EF (40–54%) and 2.9% (n = 92) had severely depressed EF of <40%. A stepwise increase in prevalence in overt coronary heart failure at baseline was observed (2% vs 6% vs 28%) from participants with normal EF to those with mild and severe LV systolic dysfunction, respectively. Importantly, at baseline, 52% (n = 1641) of these patients were diabetic and 32% (n = 1007) patients were receiving antihypertensive drugs. A further 8.1% (n = 257) patients had ischaemic heart disease and 3.0% (n = 95) had congestive heart failure.5 These patients were followed up for 37 months, and cardiovascular death occurred in 2%, 5% and 12% of patients with normal, mildly reduced and severely reduced EF.5 All-cause mortality was 6%, 10% and 32%, respectively (both p<0.01).5 In analyses adjusting for previously described covariates, the relative risk of cardiovascular and all-cause mortality were 2.9- and 1.4-fold higher in the 10% of Strong Heart study participants with mildly reduced EF and 6.9 and 4.8 higher in the 2% with severely reduced EF.4

The HyperGEN study was published in 20016 using 40–54% as the cut-off point for “mild” LV systolic dysfunction. This study was to determine the prevalence and correlates of LV systolic dysfunction in hypertensive patients in a biracial population-based sample. In this study, 2086 patients were identified and 86% had a normal EF, while 10% had mildly reduced EFs and 4% had severely reduced ventricular systolic function( EF<40%). The prevalence of coronary heart failure was 17% in the group with severe LV dysfunction, compared with 4.7% and 0.5% in the groups with mild and normal LV function, respectively. Coronary heart disease was significantly more prevalent in the group with mild and severe LV systolic dysfunction, at 23% and 63%, respectively, compared with the group with normal function (12.6%).6

Defining a cut-off value of 54% based on the evidence above will result in the prevalence of LV systolic dysfunction in the community being around 13–14% in a predominantly middle-aged population (mean age 55–61), as seen in the Strong Heart and HyperGEN studies (table 1).4 6

The Echocardiographic Heart Of England Screening (ECHOES) study, a population-based screening study of 3960 patients for prevalence of LV systolic dysfunction and heart failure was conducted on a large representative adult population aged >45 years (mean age 61), in central England in the late 1990s. An objectively measured LVEF was obtained in 3141 of these subjects; 53 of these (1.7%) had an EF <40%; visual assessment showed that 19/819 of those without a measurable EF also had significant systolic dysfunction, giving an overall prevalence of 1.8%. Analysing only those with a measurable EF, if the recent ASE/EAE guidelines are applied, the percentage of LV systolic dysfunction in the ECHOES population would be 27.2%—over a quarter the population aged over 45. The mean (SD) EF of the cohort was 57.9 (6.7)% with a median EF of 58% (table 2).7 In the MONICA study conducted in Glasgow, a cross-sectional survey of 1467 patients between the ages of 25 and 74 years randomly surveyed for the prevalence of heart failure, the prevalence of LV systolic dysfunction was 7.7% (based on a definition of EF <35%). The prevalence of patients with an EF <55% with no cardiovascular disease would be more than 50% in this population cohort.8

One major problem with the diagnostic criteria is that no systolic heart failure or LV systolic dysfunction treatment study has included patients with an EF between 50% and 54%. ACE inhibitors were shown to improve outcomes in asymptomatic LV systolic dysfunction, in both the SOLVD9 and SAVE10 clinical trials but the definitions of LV systolic dysfunction used was an EF of <35% and <40%, respectively.

Even the studies investigating “preserved” LV systolic function have used a cut-off EF of ⩽50%. For instance, the recent CHARM-Preserved study, used a cut-off point of an EF >45%, and showed there was no significant mortality benefit in those patients treated with candesartan, and only a modest reduction in hospitalisation.11 Similar findings were noted in the PEP-CHF study, recently published which used perindopril in an elderly population with a preserved EF defined as an EF >40%.12

The CHARM investigators also reported that in their entire cohort of 7599 patients, the LVEF was a powerful predictor of cardiovascular outcome in patients with heart failure across a broad spectrum of ventricular function. Once raised above 45%, EF did not contribute further to cardiovascular risk assessment in patients with heart failure.13

The two main reasons for these findings are that patients in these studies were classified as having heart failure without having other causes of breathlessness excluded. The pathophysiology of heart failure preserved EF also probably differs from systolic heart failure and therefore these patients may not respond to the same treatment as seen in Charm Preserved and PEP-CHF.

In conclusion, the ASE/EAE guidelines represent a significant stepwise change in the criteria to define systolic dysfunction. However, these are incompatible with definitions used in all major heart failure trials and in particular the two major trials exploring heart failure and preserved EF. If these new criteria are accepted then a substantial proportion of patients over the age of 45 years will be classified as having LV systolic dysfunction, including large numbers of apparently “healthy” subjects. Yet we have only limited data that an EF of 50–54% correlates with worse outcomes and no evidence to guide us as to whether such patients should receive treatment (ACE/angiotensin II receptor channel blockers/β-blockers). The resource implications are huge. Furthermore this change in definition will have an impact on the diagnosis of diastolic heart failure or heart failure with a normal EF. One of the main diagnostic criteria is based on a cut-off EF >50%, which is assumed to be normal or “preserved” systolic function. Will those patients with an EF between 50% and 54% currently classified as having “diastolic heart failure” be reclassified as having “systolic heart failure”?14

Based on current evidence, we suggest the following classification for EF: mild dysfunction 41–49%, moderate 35–40% and <35% as severe. Before classifying those with an EF between 50% and 55% as abnormal, outcome studies are required where other illnesses and risk factors such as diabetes, obesity, smoking and hypertension, which might influence outcome and cause breathlessness, are clearly documented before study.