Potential Characteristics for Prediction of the Efficacy of Non-Steroid Anti-Inflammatory Drugs for Migraine Treatment: A Retrospective Cohort Study

Abstract Background: Although several special medications for migraine are currently available, non-steroid anti-inflammatory drugs (NSAIDs) are still the first-line pharmacological option. A considerable proportion of migraineurs still be unresponsive to NSAIDs for reasons that remain unknown. The study aimed to develop significant characteristics to predict the efficacy of NSAIDs for patients with migraine. Methods: In this retrospective study, 567 patients suffering migraine were included and divided into an effective NSAIDs (M-eNSAIDs) group and a noneffective NSAIDs (M-neNSAIDs) group according to the analgesic efficacy at 2 hours after taking NSAIDs. Clinical and neuropsychiatric characteristics were collected and used to build a logistic regression model. And a receiver operating characteristic (ROC) curve was drawn to represent the prediction capability.Results: Five predictors including education, attack duration, headache impact intensity, anxiety and depression scores were identified to build the logistic regression modal. The area under the curve (AUC) values of each predictor were 0.706, 0.639, 0.560, 0.683 and 0.632, respectively, failing to predict the efficacy of NSAIDs. All five predictors-combined method achieved an acceptable AUC value of 0.834 and a sensitivity of 90.9%.Conclusions: Despite the insufficient predictive capability of these predictors when analyzed individually, this study developed an effective and convenient method to accurately predict the efficacy of NSAIDs, which would be helpful for developing individualized therapeutic strategies for treatment of migraine.