Chronic heart failure: a look through the rear view mirror.

Heart failure is now the most common condition that leads to hospital admission in industrialized nations. Although the overall prognosis for patients with chronic heart failure is still gloomy, and similar to that of many of the most common forms of cancer, a combination of pre-clinical and clinical research conducted over the past 25 years has led to five significant advances, each of which has received a IA recommendation in the European Society of Cardiology Guidelines for the treatment of heart failure with reduced ejection fraction. These advances, well known to cardiologists, are: (i) blockers of the renin–angiotensin system [angiotensin-converting enzyme (ACE) inhibitors or angiotensin II type I receptor blockers (ARBs)]; (ii) beta-adrenergic blockers; (iii) aldosterone antagonists; (iv) implantation of cardioverterdefibrillators; and (v) the application of cardiac resynchronization therapy in patients with QRS prolongation. Each has been shown to reduce mortality in patients with heart failure. As a consequence, the cumulative benefits of these therapies has improved the prognosis in these patients quite substantially. The development of a management plan for individual patients with heart failure (as is the case with many conditions) requires an assessment of prognosis. This has been a challenging task in patients with chronic heart failure. A number of risk scores have been established since 2003 to aid physicians in this task. These scores have been of varying size and clinical value. In a very ambitious and far-reaching effort, Pocock and colleagues have formed the Meta-analysis Global Group in Chronic Heart Failure (MAGICC). They now report an analysis of individual data on almost 40 000 patients with chronic heart failure enrolled into 30 cohort studies, six of which were clinical trials and the remainder registries. These patients were followed for a median of 2.5 years, providing almost 100 000 patient-years of observation, with almost 16 000 deaths! Certainly, this represents the largest number of patients and deaths ever investigated in heart failure. Among these 30 cohorts, the seven largest cohorts contributed 78% of the patients (and deaths) to this meta-analysis (see supplementary table S1 in Pocock et al.). The authors identified 13 highly significant individual predictors of mortality. Not surprisingly, age, ejection fraction, New York Heart Association class, serum creatinine, diabetes, male gender, and chronic obstructive pulmonary disease were confirmed as independent predictors of mortality. Quite appropriately, the authors separated patients into the two haemodynamically major subgroups of heart failure by the level of ejection fraction. Interestingly, lower systolic pressure was more predictive of mortality in chronic heart failure patients with reduced ejection fraction than it was in those with preserved ejection fraction; the opposite was the case with age. Surprisingly, atrial fibrillation did not emerge as an independent predictor of risk. Pocock et al. then created a model which combined all 13 independent predictors of mortality into an integer score, which, in turn, was linked to the probability of dying within 1 year and 3 years. The stated goal was to ‘create a generalizable, easily used risk score for mortality in heart failure’. The authors accomplished this and should be congratulated on their prodigious efforts. However, the score has five limitations. First, the inclusion of, as two of the 13 independent predictors of mortality, the lack of administration of an ACE inhibitor/ARB and of a beta-blocker is puzzling. These two classes of drugs, as already pointed out, are now routine in the treatment of heart failure and it is difficult to understand why they should be included in a risk score that is to be used to evaluate future patients with this condition. The second limitation is that the concentration of natriuretic peptides, which have been shown to be useful predictors of mortality in heart failure, were not considered as potential risk predictors. The third limitation is the considerable amount of missing data. Only two of the 13 variables that were entered into the model— age and gender—were available in all patients. Five variables (body mass index, systolic pressure, chronic obstructive pulmonary disease, creatinine, and heart failure duration) were each missing in . 10 000 patients; and three others were not available in between 6500 and 10 000 patients. The fourth is the authors’ statement that there were ‘substantial between-study differences