Author’s response to the Letter to the Editor of Tfelt-Hansen

Abstract

Sir, We have read with care the letter of Dr. Tfelt-Hansen regarding our article published in the European Journal of Clinical Pharmacology in which we assessed the placebo effect in migraine using meta-analytic techniques [1]. While we consider that his comments will help to clarify some of the issues raised in our article, we would like to address a number of specific points brought up by Dr. Tfelt-Hansen with the aim of improving the reader’s understanding of our contribution. Dr. Tfelt-Hansen expresses his concern regarding the definition of headache relief, stating the need to accept that headache relief should be defined as the decrease in its intensity from moderate (2) or severe (3) to none (0) or mild (1). In fact, in our article we write that ‘pain relief means a pain reduction higher than 50%’. In the case of severe pain, a pain reduction of 50% is a decrease in headache intensity from severe (3) to mild (1) or none (0). For moderate pain, a decrease in headache intensity to mild (1) or none (0) is considered to be a success. Therefore, we consider that our definition of ‘higher than 50%’ is not different from that suggested by Dr. Tfelt-Hansen. However, we do accept that his definition is clearer than the one used by us in our article, which is also in accordance with the Guideline of EMEA for the investigation of medicinal products for the treatment of migraine [2] and the International Headache Society Clinical Trials Subcommittee [3]. However, for practical purposes, the two definitions are almost equivalent. Dr. Tfelt-Hansen also suggests that pain free at 2 h is increasingly used as the primary efficacy measure. We fully agree with that point of view. In fact, all of the tables in our article [1] include both pain-free and pain-relief variables, which enable the reader to make a comparison. The first can be considered as a hard outcome and the second as a soft outcome. However, it is indisputable that the main objective of treatment should be to leave patients pain free, even though pain relief can obviously not be considered to be of minor importance in clinical practice. As expected, the magnitude of the placebo effect was higher when pain relief was used as a variable and decreased when the pain free was considered as a variable. Unfortunately, although pain free is desirable, it is not always achieved in clinical practice or in clinical trials. For example, in a recent study in acute migraine [4], the pain freedom at 2 h ranged from 17.2 to 31.3% with active treatments and was 9.6% with placebo. Values of pain relief at 2 h ranged from 49.8 to 56.4% but it was only 27.7% with placebo. These values are very similar to those obtained in our meta-analysis, i.e., 28.6% of pain relief and 8.8% of pain freedom. The main objective of our article was to analyze the significance of placebo effect differences according to clinical trial methodology independently of all other variables (disease, population, and even outcomes). If we assume that methodology may modify placebo results, then we may also assume that some aspects associated to clinical A. Macedo :M. Farre Department of Pharmacology, Therapeutics and Toxicology, Universitat Autonoma de Barcelona, Bellaterra, Spain