The effect of gender disparity on migraine pharmacotherapy: a propensity score-matched cohort study

Objectives: Migraine is a debilitating disease that affects 28 million adults in the United States. Gender differences in migraine treatment have been studied; however, results are often overlooked due to lack of controlling for selection bias. This study aimed to investigate gender differences in migraine treatment, using propensity score-matched analyses. Methods: A retrospective cohort study was conducted by analysing the 2012 National Ambulatory Medical Care Survey. All patient visits with a diagnosis of migraine were included. Weighted descriptive analyses were used to estimate the prevalence of medications recommended in the American Neurology Association guidelines. Multivariate logistic regression and Greedy 8-to-1 digit match were used to create propensity scores. A propensity score-weighted logistic regression model was fitted to compare gender differences on migraine treatment while controlling for covariates. All analyses utilized SAS PROCSURVEY applications and incorporated sample weights and standard errors. Key findings: Among the 6 408 461 outpatient visits with migraine diagnosis that took place in 2012, only 54.1% had received at least one abortive or prophylactic migraine prescription. Prior to propensity score matching, there was no significant association between gender and migraine pharmacotherapy (odds ratios (OR): 1.0, 95% CI: 0.510–1.961). After matching, women received significantly more drug treatment than men (OR: 3.2, 95% CI: 1.137–8.932). Conclusions: Assessment of gender differences is challenging because of potential biases that require adjustment. This study successfully provides empirical evidence that the problem of selection bias must be properly considered during analysis in observational studies. An understanding of this potential pitfall is important to critically assess relevant research.