In principle medical research is supposed to result in unequivocal guidance to doctors as to how to treat their patients. As often as not, however, the waters are muddied as much as cleared. Two papers in a recent edition of the New England Journal of Medicine about atrial fibrillation and the cause of stroke illustrate this. It has long been known that people with a clinically-detected chaotic heart rhythm called atrial fibrillation (AF) have an increased incidence of stroke by embolism; and likewise that no cause of such stroke can be found in up to 40 percent of patients who suffer from one. Their strokes are called cryptogenic. The two papers addressed the question whether, if you monitor patients with cryptogenic stroke for long enough, some or many of them will turn out to suffer from AF. This is important, because it is generally agreed that, in patients with clinically detected and symptomatic AF, anti-coagulation reduces the subsequent risk of stroke. AF, however, is not an all or none phenomenon. Some people suffer it continuously, but others only occasionally and for only a few seconds at a time. The additional risk of stroke in the latter is unknown, but is an important question because the anticoagulation designed to reduce the risk of stroke is not itself without risk, including that of another kind of stroke, the haemorrhagic kind. In other words, the risk caused by treatment could outweigh its benefits.
In the first paper, one group of patients with cryptogenic stroke heart rhythm was monitored for up to 30 days, while another was monitored for only 1 or 2 days according to the current, standard protocol. At three month follow-up, AF had been detected in 16.1 percent of the former and only 3.2 percent of the latter. In the second paper, the heart rhythm of patients with cryptogenic stroke were either monitored for 12 months or they were treated according to the standard protocol. At six months, 8.9 percent of monitored patients had been found to have suffered from AF lasting at least 30 seconds, against 1.4 per cent of those not so monitored; at 12 months, the figures were 12.4 percent and 2 percent, respectively. On a superficial reading, the case for monitoring might seem to have been made, along with that for consequent anticoagulation. Certainly that is what I should have concluded if I had simply skimmed the papers, which is how I (and I suspect most doctors) normally read papers. I think it likely that as a result of these papers many patients are going to be monitored and treated who would not have been monitored and treated before. But actually the case is not made out. It is always important to remember that the object of all medical activity is the benefit of the patient. If monitoring is carried out, it is essential to know a) the clinical significance of what is found an b) the ability to improve the outcome. Neither is the case here. For example, in over half the patients in which AF was found, it lasted for less than 6 minutes per day. As a friend of mine pointed out, a drink of alcohol will do this. Yet these patients were anti-coagulated. There was no significant difference in the outcomes of the monitored and non-monitored groups, both of whom were anti-coagulated if AF was found among them (and many more were found in the monitored group). In other words, a number of people were quite possibly anti-coagulated for no benefit to themselves, at the expense of time, anxiety and money. Five patients died, three in the monitored group, two in the controls. They were excluded from the analysis of the results. But if the former died of hemorrhage consequent upon anticoagulation, while the latter died of incidental causes, the whole analysis of the results would change. The paper is silent on this point. But monitors will be sold and patients anti-coagulated. **** images via shutterstock / Alila Medical Media / ievgen sosnytskyi / Puwadol Jaturawutthichai