A Cox's hazard model for new development of medication overuse headache in pure chronic migraine patients under migraine prophylaxis

Background: Medication Overuse Headache (MOH) occurrence during migraine preventive therapy is poorly known. The present study aims at identifying the rate and factors influencing MOH development in patients with Chronic Migraine (CM). Methods: Patients with a diagnosis of CM without MOH were consecutively recruited and followed at least three times for 1 year under preventive therapy. The number of headache days and the use of acute medication were recorded; the occurrence of MOH was considered as the main event. Kaplan-Meier curves were used to determine the time of the event. After the third follow-up patients were censored. Proportional hazards Cox regression was used to identify predictors of MOH among demographic variables (e,g. age), main comorbidities (e.g. sleep disorders, epilepsy, thyroid dysfunctions), migraine features (e.g. aura, number of previous preventive therapies, headache days and number of medications). Results: We recruited 260 first-visit patients with CM. 46 patients (17.70%) developed a new onset of MOH over 1 year, with 35 at the first follow-up. No difference in MOH occurrence was found among different pharmacological groups (Chi-square = 2.99, p = 0.70). About 1/3 of non-responders developed MOH. Monthly headache days, number of acute medications, thyroid dysfunction, epilepsy, and MIDAS score were associated with MOH (Chi-square = 14.08; p = 0.01). Headache days (Wald test: 6.0; p = 0.014) and MIDAS score (Wald test: 4.07, p = 0.04) were the only stand-alone significant factors. Conclusions: The rate of new occurrence of MOH in patients without a previous history is 18%. Monthly number of headache days and MIDAS score are the most important predictors.