Reply to the letter by Omland et al.: “Visual and auditory cortical evoked potentials in interictal episodic migraine: The predictive value of a proposed diagnostic test should always be corrected for prevalence”

We agree with Omland et al, (1) that “it is important to take disease-prevalence within the target population (pre-test probability) correctly into account” for calculating Positive Predictive Value (PPV). However, this applies to a single screening test in the general population. Our study differs in that we assessed a combination of two tests in a very selected population of patients consulting tertiary headache clinics with a very high prevalence of migraine, and thus the proposed correction is not applicable (2). PPV mainly reflects sensitivity, which is necessary in screening studies. When a test is used for diagnostic confirmation, it should be highly specific. In fact, if the sample sizes in the positive (disease present) and the negative (disease absent) groups do not reflect the real prevalence of the disease, then positive and negative predictive values cannot be estimated and should be ignored (3).

Moreover, we assessed a combined VEP-IDAP test, considered to be positive if there is a supra-threshold value in at least one of two tests. We proposed this combination of tests as a potential aid for diagnosing migraine in clinical samples of patients with uncertain diagnosis, and not for screening purposes in the general population. We underscored that this proposal was not made on the basis of the PPV but of accuracy, which combines sensitivity and specificity values. An abnormality of either VEP or IDAP (or of both) may thus confirm the diagnosis of migraine in patients not fulfilling all diagnostic criteria in the setting of a headache clinic.