Visual and auditory cortical evoked potentials in interictal episodic migraine: An audit on 624 patients from three centres. Response to the letter by Omland et al.

Omland et al. argue that we overestimated the diagnostic usefulness of visual (VEP) and auditory evoked potentials (IDAP), suggesting that we did not take into account pre-test probability (1). As described in the “methods” section, however, the area under the ROC (AUC) was based on a pre-test probability of 15%, which is the widely accepted average of migraine prevalence. The positive predictive value of 94.1% and diagnostic accuracy of 81.1% were thus calculated taking into account this 15% pre-test probability (see ROC analysis in (2)). They are correct in mentioning that we were “overoptimistic” in some of the terms we used in the discussion. In particular, in the sentence “Diagnostic accuracy is moderate to fair for VEP habituation (65.3% and 69%) and IDAP (54.3%) taken separately”, we agree that “poor” could replace “fair”. Concordantly, we clearly stated in the “discussion” and “clinical implications” sections that neither VEP nor IDAP are useful for diagnosis, but that an abnormality of at least one of them is highly predictive of a diagnosis of migraine in subjects in whom both evoked potentials are recorded. We agree that this may be useful only in a small subset of patients and should be confirmed in an adequate trial, although tension-type headache might not be an ideal comparator because of its pathophysiological heterogeneity (3).

The methodological limitations cited by Omland et al. would be appropriate had we performed a randomized controlled trial, which was not the case (1). Our objective was merely an audit of a large, multicentre database of VEP and IDAP recordings in migraine and healthy subjects. From this perspective, we consider that the methodological variations rather reinforce our present and previously published conclusions: VEP habituation is significantly reduced and IDAP increased in migraine patients, which holds true even with different recording protocols or devices. Concerning the role of blinding, we have addressed this on several occasions in our previous publications, and recently published a formal study showing that blinding does not significantly bias the results (4).

Finally, the initial objective of our study was to verify that our previously published findings of evoked potential abnormalities in small patient groups are reproducible in the largest sample of recordings hitherto analysed in migraine. As mentioned in the introduction, the idea arose thanks to Omland et al.’s studies, in which they were unable to reproduce our results and attributed this mainly to methodological and blinding problems (1,5). The time has come to abandon sterile discussions about methods and to explore pathophysiological, genetic and environmental factors that might explain why the differences in evoked potential profiles between migraineurs and non-migraineurs are not reproducible above 60° latitude north. The regional difference in solar radiation could be one of several interesting tracks to pursue, as suggested by our recent study (6).