Objective: To verify the presence of different age at onset (AAO) subgroups of patients in a sample of patients with migraine without aura (MWA) and compare clinical correlates among them.
Background: MWA is a long-lasting disease whose prognosis has not yet been fully investigated. Patients may present complete remission, partial clinical remission, persistence and progression (migraine attack frequency and disability may increase over time leading to chronic migraine). Limited evidence exists regarding the identification of risk factors or predictors which might influence migraine prognosis. AAO has been proven a useful tool in the investigation of the clinical, biological, and genetic characteristics able to influence the prognosis of a number of neuropsychiatric disorders. AAO distribution was studied using mixture analysis, a statistical approach that breaks down the empirical AAO distribution observed into a mixture of normal components.
Methods: A sample of 334 outpatients affected by MWA, recruited in a clinical genetic study at our Headache Center from 2004 to 2008, was enrolled for this study. Diagnosis was made according to International Headache Society criteria 2004. AAO distribution in patients was studied using mixture analysis. Chi-square test was used to compare clinical correlates among identified subgroups. Logistic regression was performed in order to correct for effect of possible confounders.
Results: Mixture analysis broke up the observed distribution of AAO into 3 normal theoretical distributions. Informational criteria clearly showed a better 3-component model rather than the 2-component one. An early-onset (≤ 7 years of age), an intermediate-onset (≥ 8 and ≤ 22), and a late-onset group (≥ 23) were identified. Comparison of clinical correlates among subgroups by means of chi-square test showed a statistically significant result for migraine frequency (χ(2) = 7.41, P = .02). Considering the frequency of migraine attacks as a main outcome, the regression model showed a higher AAO is associated with low frequency (odds ratio =0.95; P= .02).
Conclusions: The significant association between AAO and attack frequency found in our study supports the hypothesis that AAO could act as a predictor factor able to influence prognosis. AAO could represent a phenotype suitable for identifying MWA susceptibility genes.