One hundred years of atrial fibrillation

We have known about atrial fibrillation in man (originally called auricular fibrillation) for about a hundred years, and 1905 was a pivotal year. Although Jean Baptiste Sénac (1693–1770) connected 'rebellious palpitation' and mitral stenosis in 1783 [1,2], and although the arrhythmia was subsequently described in animals [3], the cause of the palpitation that atrial fibrillation can produce was not definitively recognized until the start of the twentieth century. Of course, physicians had described different types of irregularities of the pulse ever since they started taking it, and terms such as 'ataxia of the pulse'[4], 'delirium cordis'[5], and 'pulsus irregularis perpetuus'[6] had been used to describe the pulse that we now recognize to be associated with atrial fibrillation [7]. But the pathophysiology was not properly delineated until James Mackenzie (1853–1925) described the loss of the 'a' wave in the venous pulse, using the 'clinical polygraph' that he had invented to demonstrate the form of the pulse wave [8,9]. Willem Einthoven (1860–1927) published the first electrocardiogram showing atrial fibrillation in 1906 [10], and the connection between irregularity of the pulse and atrial fibrillation was established by the pharmacologist Arthur Cushny [11,12], the cardiologist Thomas Lewis [13], and two Viennese physicians, Rothberger and Winterberg [14]. However, despite this long experience of the condition, we still do not know how best to manage it. What, for example, is the ideal ventricular rate in a patient with permanent atrial fibrillation? Cardiac output is maximal in sinus rhythm at a resting ventricular rate of about 70 per minute, and there is evidence that a rate that is 25% higher is most effective in atrial fibrillation [15]. However, it has been argued that maintaining a rate of 85–90 per minute is deleterious for the heart in the long term, and there is certainly an inverse relation between life span and heart rate, the average allotment being about 7 × 108 beats per lifetime in all mammals bar man, in whom it is about 3 × 109[16]; a resting rate of 70–85 per minute in atrial fibrillation has therefore been recommended [17]. There is also uncertainty about the ideal ventricular rate during exercise. It may, for example, be different in patients with good or poor left ventricular function [18], and a rate below 100 per minute during normal daily activities has been recommended [17]. Now it turns out that we do not know whether it is better, in patients with atrial fibrillation, to control the ventricular rate (rate control) or to convert the rhythm to sinus rhythm and try to maintain it (rhythm control). This dilemma has arisen because of the availability of pharmacological and nonpharmacological methods of converting atrial fibrillation to sinus rhythm, in the condition known as persistent atrial fibrillation, which would previously have been regarded as permanent. The advantages and disadvantages of both strategies have been discussed [19]. The advantages of rhythm control are the absence of symptoms of the arrhythmia, improved haemodynamic function, better exercise tolerance, and less need for anticoagulation; the disadvantages are that it may be difficult to achieve and maintain sinus rhythm, that the antiarrhythmic drugs used can cause adverse effects, and that recurrence of atrial fibrillation is common, increasing hospital admissions. The main advantage of rate control is that it is easier to achieve, while the disadvantages are that atrial fibrillation causes symptoms, particularly during exercise, which may be limited, that anticoagulation is required, and that cardiac structural changes occur during long-term atrial fibrillation. So much for theory. What is the evidence? In this issue of the Journal, Kumana et al. report the results of a meta-analysis of randomized controlled comparisons of rate control and rhythm control [20]. Their analysis included five trials in a total of 5239 patients with paroxysmal or persistent atrial fibrillation, and they calculated relative risk reductions and numbers needed to treat (NNTs) per year. In the longest trial, which was also the largest, follow-up was for a mean of 3.5 years. The results overall favoured rate control: one additional patient was hospitalized for every 35 assigned to rhythm control; there were no significant differences in deaths, ischaemic strokes, or bleeding from sites other than the nervous system; quality of life assessments did not differ; adverse drug reactions were significantly more common in the patients who were assigned to rhythm control; thromboembolism was associated with withdrawal of anticoagulation or subtherapeutic anticoagulation. These results suggest that on average it is better to opt for the counterintuitive option: pharmacological rate control (rather than rhythm control) and optimization of anticoagulation. However, the balance of benefit to harm in the individual patient may not always be in that direction. There will be some patients in whom the opposite strategy will be preferable, particularly those in whom there are intolerable adverse drug effects or in whom anticoagulation is undesirable. Nevertheless, after about 100 years of study, and despite the introduction of various nonpharmacological methods, it looks as if we shall have to continue to use the pharmacological approach to manage persistent and permanent atrial fibrillation for some time to come, and moreover to choose drugs that will not convert the rhythm to sinus rhythm. And in each patient we shall have to continue to consider the relative benefit to harm balance of doing so and of using drugs such as beta-blockers, calcium channel blockers (such as diltiazem and verapamil), and of course cardiac glycosides, which have been used to slow the ventricular rate in atrial fibrillation for almost as long as it has been known about [21].