In Reply:


  • Miguel S. Cabalag BMedSci,

  • David McDonald Taylor MD, MPH, DRCOG, FACEM,

  • Jonathan C. Knott MBBS, GDipClinEpiBiostats, FACEM, PhD,

  • Paul Buntine MBBS, FACEM,

  • DeVilliers Smit MB, ChB, FACEM,

  • Alastair Meyer MBBS, BSc, FACEM

We thank Drs. Karydes and Bryant for their comments on our article1 and agree that our study design had flaws that limit the strength of the conclusions. These limitations are well described in the article.

It is perplexing to note that Karydes and Bryant have entitled their letter “Adenosine and Caffeine-induced Paroxysmal Supraventricular Tachycardia.” In fact, our study did not examine, in any way, the management of supraventricular tachycardia or other tachyarrhythmia specifically induced by caffeine. Hence, their comments regarding the tachyarrhythmia and deaths induced by theophylline are irrelevant. Their reports of the useful effects of benzodiazepines and lack of effect of adenosine on tachyarrhythmia management are also irrelevant. Adenosine is only effective in terminating arrhythmias where the AV node is an inherent part of the mechanism,2,3 and we only examined patients with supraventricular tachycardia for which adenosine is a specific treatment.

We do agree with Karydes and Bryant regarding the effect of chronic exposure to caffeine. Such exposure does result in up-regulation of adenosine receptors4 and is clearly a potential confounder in our study, as we reported. Much larger studies are required with samples sizes sufficient to allow adequate stratification of patients into both acute and chronic caffeine exposure levels.

We noted clearly that recall bias is likely to have affected our results. Karydes and Bryant are correct in stating that the presence of disease can affect patient perception of causality. However, all our patients had the disease (supraventricular tachycardia), all were administered adenosine, all were asked about their caffeine intake, and none was aware of our hypothesis. Hence, the direction and magnitude of the recall bias should have been similar for all our patients. We agree that accurate timing of the last caffeine intake is important. However, how does one design a method of determining this apart from patient recall? No one logs the time when they have a caffeinated beverage, and no one expects the onset of supraventricular tachycardia. Finally, if the timing of caffeine intake is ultimately considered in clinical decision-making, it is highly likely that patient recall will be employed. Hence, it is most appropriate to use this method in investigative studies.

Our study was a simple observational exercise with findings consistent with what could be expected from adenosine receptor blockade by caffeine. We hope that it will stimulate further research in the area that can overcome the limitations that both we and Karydes and Bryant describe.