Comorbidity of Migraine and Psychiatric Disorders Among Substance-Dependent Inpatients


  • Funding: This study was supported by National Institutes of Health grants [R21 DA022383] awarded to Dr. Tull and [R01 NS06525701] awarded to Dr. Houle. The data were obtained from [R21 DA022383] awarded to Dr. Tull.
  • Conflicts of Interest: Dr. Houle receives research support from Merck and GlaxoSmithKline. Dr. Smitherman receives research support from Merck.



Limited and conflicting data exist regarding the prevalence of psychiatric disorders, particularly substance use disorders (SUDs), among migraineurs in inpatient clinical settings.


As part of a larger cross-sectional study, 181 substance-dependent inpatients completed a structured psychiatric interview and measures of psychiatric symptoms and migraine. Standardized mean differences were used to quantify differences between inpatients with and without migraine across 4 domains of predictors (demographic variables, non-SUD psychiatric diagnoses, specific SUDs, and self-reported psychiatric symptoms). The predictors within each domain that best discriminated between the migraine and no-migraine groups were identified using a classification tree approach with Bonferroni corrections. These candidate predictors were subsequently entered into a multivariate logistic regression to predict migraine status, which was then replicated using bootstrapping of 500 samples. Associations between migraine status and SUD treatment dropout were also examined.


Forty-four of the 181 (24.3%) participants met criteria for migraine. Migraineurs were more likely to be female (34.8% vs 18.3%) and reported higher levels of current anxiety symptoms (mean [standard deviation]: 19.7 [11.0] vs 11.3 [10.3]). Having a lifetime diagnosis of generalized anxiety disorder (56.8% vs 27%, odds ratio 3.47, 95% confidence interval [CI] 1.39-10.58) or a current diagnosis of alcohol dependence (45.5% vs 24.1%, odds ratio 3.79, 95% CI 1.63-13.62) was associated with more than a 3-fold risk of migraine. These 4 variables in combination were forced into the final multivariate model, which differentiated well between those with and without migraine (area under the receiver operating characteristic curve = 0.81; 95% CI 0.73-0.88). Migraine was not differentially associated with increased risk for SUD treatment dropout (13.6% vs 16.1% dropout among those without migraine).


A history of generalized anxiety disorder, high levels of current anxiety symptoms, and current alcohol dependence are the strongest psychiatric predictors of migraine status among substance-dependent inpatients. However, migraine status is not associated with SUD treatment dropout.