Sleep Disturbance and Affective Comorbidity Among Episodic Migraineurs


  • Conflict of Interest: Ms. Walters and Mr. Hamer report no conflicts of interest.
  • Dr. Smitherman receives research support from Merck.

Address all correspondence to T.A. Smitherman, PhD, 207 Peabody, Department of Psychology, University of Mississippi, University, MS 38677, USA, email:



Disturbances in sleep are common among migraineurs, particularly those with frequent (ie, chronic) migraine. Examination of specific types of sleep disturbance and behaviors among episodic migraineurs, however, has not been sufficiently explored. Further, few studies have investigated whether sleep disturbance is attributable to comorbid affective symptomatology.


The present case-control study sought to (1) assess sleep quality, daytime sleepiness, and sleep hygiene among a large sample of episodic migraineurs; (2) quantify relations between sleep disturbance and headache-related variables; and (3) determine if these relations remain after accounting for comorbid depression and anxiety.


Two hundred ninety-two undergraduate students (69.9% female, mean age = 19.19, standard deviation [SD] = 3.21 years) completed measures of sleep quality, daytime sleepiness, and sleep hygiene along with well-validated measures of depression and anxiety symptomatology. Those screening positive for migraine were subsequently administered a structured diagnostic interview to verify diagnosis of migraine consistent with the International Classification of Headache Disorders, 2nd edition. Episodic migraineurs and non-migraine controls were compared on the sleep disturbance variables, and among those with migraine, relations with headache frequency, severity, and disability were quantified with linear regression analyses.


Seventy-eight (26.7%) participants met International Classification of Headache Disorders, 2nd edition criteria for episodic migraine. Compared with participants without migraine, episodic migraineurs reported poorer sleep quality (mean = 8.90 [SD = 3.39] vs 6.63 [SD = 3.02], P < .0001), with 85.9% reporting clinically significant poor sleep quality (vs 62.0% of controls). Poor sleep quality was significantly associated with headache frequency and headache-related disability, accounting for proportions of variance (14.8% in frequency and 18.2% in disability, both P ≤ .001) similar to those attributable to depression and anxiety. These relationships remained significant after controlling for these affective symptoms, in which sleep quality accounted for 5.3% and 5.8% of unique variance in frequency and disability, respectively (P < .05). By comparison, daytime sleepiness and poor sleep hygiene were not consistently associated with migraine or migraine-related variables.


Consistent with prior studies on chronic migraine, poor sleep quality is uniquely associated with episodic migraine, and this relationship is not solely attributable to comorbid psychiatric symptomatology. Sleep quality should be preferentially assessed (vs sleepiness and sleep hygiene) when subjective self-report measures of insomnia are used in clinical headache settings. Future studies should supplement these findings by evaluating the efficacy of interventions that specifically target sleep quality and insomnia (eg, stimulus control, sleep restriction) among episodic migraineurs.

Migraine affects 12% of Americans annually and is ranked by the World Health Organization as one of the top 20 causes of disability worldwide. Large-scale population studies indicate that migraineurs are 2-5 times more likely to suffer from depression and anxiety disorders than individuals without migraine.1-3 These affective comorbidities are of interest clinically because they further compound the negative impact of migraine by increasing health care costs and utilization,[4] exacerbating disability and poor quality-of-life,[5, 6] and promoting persistence and chronification of headache over time.[7]

In addition to affective comorbidities, disturbances in sleep are also highly prevalent among migraine sufferers.[8, 9] Migraine often precedes the onset of sleep disturbance,[10, 11] and sleep disturbance also functions as one of the most common “triggers” for migraine.[12] Data from clinical samples confirm that insomnia is the most prevalent sleep-related disorder among migraineurs, occurring in 1/2 to 2/3 of individuals who present to community headache clinics (vs 10.8% of the general population).[9, 13]

Beyond the objective measurement of sleep via polysomnography, which is not feasible in many headache practice settings, sleep disturbance has multiple subjective components, the most central of which are poor sleep quality, resulting daytime sleepiness, and poor sleep hygiene. Poor sleep quality and daytime sleepiness represent orthogonal constructs of sleep disturbance and are associated with poor health, significant functional and cognitive impairment, and psychiatric comorbidity.[14, 15] Inadequate sleep hygiene (ISH) involves engaging in behaviors or maintaining a sleep environment not conducive to sleep (eg, frequent daytime napping, variable sleep-wake times, participating in stimulating activities before bed). Poor sleep hygiene appears to be a prominent contributor to sleep disturbance among patients with chronic migraine (CM),[16] and interventions to modify poor sleep hygiene have shown promise in reducing headache frequency among adults with CM[17] and children/adolescents with migraine.[18]

Despite the high comorbidity of sleep disturbance with migraine, examining the relative importance of sleep quality, daytime sleepiness, and sleep hygiene among community samples with episodic migraine is of importance to more accurately characterize the nature of their sleep disturbance. Episodic migraineurs are of interest because the majority of migraineurs (86.1%) have less than 5 attacks per month and do not seek medical treatment for their condition,[19] and because their headache presentation is not yet complicated by chronification or medication overuse. Relations between sleep disturbance and migraine variables of clinical interest (ie, severity, frequency, disability) also merit exploration. Finally, because affective comorbidities and sleep disturbance commonly co-occur among migraineurs, additional research is needed to determine whether sleep problems persist after controlling for affective symptomatology, as was reported in a recent study.[20] Thus, the aims of the present study were to (1) assess sleep quality, daytime sleepiness, and sleep hygiene among episodic migraineurs; (2) assess relations between these variables and migraine frequency, severity, and disability; and (3) determine if these relations remain after accounting for comorbid depression and anxiety.



An a priori power analysis indicated that a total sample size of 236 participants was required for the present study, assuming a small effect size (f2 = 0.08), a power level of 0.80, and an alpha level of 0.05. Three hundred and twenty-three college students completed a variety of measures assessing headache symptoms and headache-related disability, psychiatric comorbidity, and sleep disturbance. Both individuals with and without migraine were recruited and were blind to specific hypotheses of the present study. Thirty participants (9.3%) with missing data on the measure of sleep quality were excluded from analyses as well as 1 multivariate outlier, as described later.


ID MigraineTM[21].—

The ID Migraine is a widely-used 3-item migraine screening instrument that assesses diagnostic features of nausea, photophobia, and interference with activity. The endorsement of 2 or more items is considered a positive screen for migraine and has a sensitivity of 0.81 and a specificity of 0.75 (positive predictive value = 0.93).

Structured Diagnostic Interview for Headache-Revised.—

Participants screening positive on the ID Migraine were individually administered the Structured Diagnostic Interview for Headache[22] that was modified to comport with current diagnostic criteria of the International Classification of Headache Disorders, 2nd edition (ICHD-II).[23] This measure has strong validity for identifying primary headache disorders[22] and was used for establishing ICHD-II diagnoses of migraine. Data on headache frequency and severity were also obtained from the Structured Diagnostic Interview for Headache-Revised (SDIH-R).

Migraine Disability Assessment Questionnaire[24].—

The Migraine Disability Assessment Questionnaire (MIDAS) is a 5-item measure of migraine-related disability that inquires about the number of days during the past 3 months that migraine has limited the respondent's ability to function at school/work, home, and in social activities.[25, 26] Scores from 0 to 5 indicate little or no disability, 6-10 mild disability, 11-20 moderate disability, and ≥21 severe disability.

Pittsburgh Sleep Quality Index[27].—

The Pittsburgh Sleep Quality Index (PSQI) is an 18-item measure of sleep quality that is often used as a means of identifying insomnia.[14] The validity of using PSQI scores >5 to differentiate “poor” from “good” sleepers has been established in multiple studies with both clinical and non-clinical samples.27-29

Epworth Sleepiness Scale[30].—

The Epworth Sleepiness Scale (ESS) is an 8-item measure of daytime sleepiness that quantifies the likelihood of dozing during various activities. Summed scores 10-15 are indicative of moderate sleepiness and 16-24 of severe sleepiness. The ESS is one of the most widely used paper-pencil measures of daytime sleepiness in adults,[9] has established reliability,[31] and has been validated against polysomnographic recordings.[30]

Sleep Hygiene Index[32].—

The Sleep Hygiene Index (SHI) is a 13-item measure assessing how frequently the respondent engages in behaviors comprising the International Classification of Sleep Disorders-Revised criteria[33] for ISH. Higher scores are indicative of more frequent maladaptive sleep behaviors (ie, poorer sleep hygiene). Although relatively new, test-retest reliability and internal consistency of the SHI appear superior to those of previously published sleep hygiene instruments.[32, 34]

Patient Health Questionnaire-9[35].—

The Patient Health Questionnaire-9 (PHQ-9) is a 9-item measure of depression commonly used in medical settings. Scores of 10 or higher are indicative of significant depressive symptomatology. Validity has been established by comparing the PHQ-9 to self-report measures of general health and symptom-related disability,[36] as well as to structured interview diagnoses of major depression.[37]

Generalized Anxiety Disorder-7[38].—

The Generalized Anxiety Disorder-7 (GAD-7) is a commonly used self-report measure of GAD symptomatology. Scores of 10 or higher are indicative of significant anxiety symptomatology and demonstrate good sensitivity (89%) and specificity (82%) for identifying GAD.[38] Most recently, the GAD-7 has also been shown to be effective at detecting other anxiety disorders such as panic disorder, social phobia, and post-traumatic stress disorder.[39] As such, the GAD-7 was used as a general measure of anxiety symptomatology.


The institutional review board at the University of Mississippi approved this study. Participants were undergraduate students recruited using an online research program. They self-selected for participation, provided written informed consent, and completed the aforementioned measures in small groups in exchange for modest course credit. Those who screened positive for migraine on the ID Migraine screening measure were administered the SDIH-R in a face-to-face interview with trained graduate students to confirm headache diagnoses. Those who met ICHD-II diagnostic criteria for episodic migraine comprised the migraine group, and those screening negative for migraine comprised the control group. Individuals with CM were excluded.


Statistical analyses were conducted using SPSS 20.0 (IBM, Armonk, NY, USA). Histograms, Q-Q plots, and descriptive statistics data (ie, skewness, kurtosis) were used to assess data analytic assumptions for the total scores of interest and found satisfactory. Levene's test confirmed that homogeneity of variance was met for all major criterion variables. Episodic migraineurs and control participants were then assessed for multivariate outliers by group using Mahalanobis distance; 1 multivariate outlier (control participant) was identified using a conservative P < .001 chi-square cut-off and thus removed prior to analyses.

Independent t-tests were used to compare migraineurs and non-migraine controls on the 3 variables of sleep disturbance (ie, sleep quality, daytime sleepiness, sleep hygiene), depression, and anxiety. Chi-square analyses were then used to assess clinically significant group differences on measures with established clinical cut-offs. A multivariate analysis of variance (MANOVA) was conducted to compare the groups on specific sleep hygiene behaviors as indexed by the 13 SHI items while controlling for family-wise Type I error rates. A series of linear regression analyses was used to assess relations between the 3 sleep disturbance variables and migraine frequency, severity, and disability both before and after adjusting for symptoms depression and anxiety (entered simultaneously).


Migraine Prevalence.—

Of the 292 participants with sleep quality data, 204 (69.9%) were female. The mean age was 19.19 years (standard deviation [SD] = 3.21); 38.0% were of ethnic minority status. Of these, 78 (26.7%) met ICHD-II criteria for episodic migraine (74.4% female). Six participants met diagnostic criteria for CM and thus were excluded from subsequent analyses; the remaining 208 without migraine served as controls (67.3% female). Percentage of female participants did not differ significantly between groups. Among migraineurs, average migraine frequency was 5.26 days per month (SD = 3.90; range = 1-14 days/month; 39.2% with >5 days/month) and average severity was 6.70 (SD = 1.51) on a scale of 1-10. Over half (55.8%) of the episodic migraineurs reported moderate or severe headache-related disability on the MIDAS (27.2% moderate; 28.6% severe).

Sleep Disturbance and Migraine.—

The 3 measures of sleep disturbance correlated significantly (P < .001) with one another (PSQI and ESS: r = 0.31; PSQI and SHI: r = 0.44; ESS and SHI: r = 0.29) but not close to a level at which multicollinearity would be of concern. Table 1 presents group differences on the 3 variables of sleep disturbance. Episodic migraineurs obtained significantly higher mean total scores on both the PSQI (indicative of poorer sleep quality) and SHI (indicative of poorer sleep hygiene) but did not differ from controls on mean daytime sleepiness. Group differences in sleep quality were replicated with comparisons of clinically significant scores on the PSQI (scores >5; 85.9% of migraineurs vs 62.0% of controls, P = .0001). Similarly, a higher proportion of migraineurs endorsed excessive daytime sleepiness on the ESS (scores ≥10; 54.6% vs. 40.7%, P = .037). An optimal cut-off for the SHI has not been established, although the aforementioned mean difference of 1.64 points between migraineurs and controls is unlikely to be of clinical significance given that the SHI has a range of 52 (13-65).

Table 1. Group Scores on Sleep and Affective Variables
MeasureMigraine (n = 78) 74.4% Female M (SD)Control (n = 208) 67.3% Female M (SD)P-value% Migraineurs With Clinically Significant Scores% Controls With Clinically Significant ScoresP-value of χ2
  1. — = validated clinical cutoffs have not been established for the SHI; ESS = Epworth Sleepiness Scale; GAD-7 = Generalized Anxiety Disorder-7; M = mean; PHQ-9 = Patient Health Questionnaire-9; PSQI = Pittsburgh Sleep Quality Index; SD = standard deviation; SHI = Sleep Hygiene Index.
PSQI8.90 (3.39)6.63 (3.02)<.000185.9062.02.0001
SHI38.27 (6.62)36.63 (6.01).046
ESS10.09 (3.50)9.19 (3.83).07254.5540.69.037
PHQ-98.67 (5.08)6.37 (4.55)<.00139.7420.19.001
GAD-77.95 (4.92)5.73 (4.68).00134.6218.36.004

Regarding specific sleep hygiene behaviors, over half of the entire sample endorsed frequently or always using their bed for something other than sleep or sex (62.0%), doing something that may wake them up before bedtime (61.3%), going to bed at different times each day (56.8%), doing important work before bed (56.5%), and thinking, planning, or worrying in bed (55.8%). However, the MANOVA failed to reveal any significant between-group differences across the 13 specific sleep behaviors, Wilks' lambda = 0.949, F(13,269) = 1.116, P = not significant.

Psychiatric Symptoms.—

As shown in Table 1, migraineurs reported higher levels of both depression and anxiety than controls. These mean differences were replicated in comparisons of group proportions meeting clinical cut-offs for moderate or greater symptomatology on both the PHQ-9 and GAD-7 (scores ≥10). Specifically, 39.7% of migraineurs vs 20.2% of controls reported clinically significant depression (P = .001), and 34.6% of migraineurs vs 18.4% of controls reported clinically significant anxiety (P = .004). As such, depression and anxiety scores were entered as covariates in the subsequent regression analyses.

Sleep Disturbance and Migraine-Related Variables.—

As depicted in Table 2, sleep quality, depression symptomatology, and anxiety symptomatology were all significantly predictive of migraine frequency in the univariate analyses. Daytime sleepiness and sleep hygiene were not predictive of headache frequency and were thus not analyzed in the adjusted analyses with covariates. After first adjusting for depression and anxiety, the association between sleep quality and migraine frequency remained significant (Block 2 ΔR2 = 5.3%, P = .04).

Table 2. Associations With Headache Frequency
Beta95% CI for BetaR2P-ValueBeta95% CI for BetaR2ΔR2P-value of ΔR2P-value of Model
  1. CI = confidence interval; ESS = Epworth Sleepiness Scale; GAD-7 = Generalized Anxiety Disorder-7; NS = not significant; PHQ-9 = Patient Health Questionnaire-9; PSQI = Pittsburgh Sleep Quality Index; SHI = Sleep Hygiene Index. Boldface indicates unique variance accounted for by poor sleep quality after controlling for depression and anxiety.
PSQI0.46(0.20, 0.71)14.8%.001      
ESS0.04(−0.23, 0.32)0.1%NS      
SHI0.07(−0.07, 0.21)1.5%NS      
PHQ-90.28(0.11, 0.45)13.0%.002      
GAD-70.23(0.05, 0.41)8.2%.013      
Block 1      13.0%  .007
PHQ-9    0.29(0.00, 0.58)    
GAD-7    −0.01(−0.31, 0.29)    
Block 2      18.2%5.3%.037.003
PSQI    0.32(0.02, 0.62)    

None of the sleep disturbance or psychiatric variables were significantly associated with headache severity. As shown in Table 3, both sleep quality and sleep hygiene were associated with headache-related disability, as were symptoms of depression and anxiety. In the adjusted analyses, depression and anxiety were first entered as covariates, and a stepwise entry procedure was employed with sleep quality and sleep hygiene in the second block (P < .05 required for entry into the model) in order to assess their relative contributions to disability. The stepwise procedure selected only sleep quality into the covariate-adjusted model, which accounted for 5.8% of unique variance in headache-related disability after controlling for depression and anxiety (P = .02).

Table 3. Associations With Headache-Related Disability (MIDAS)
Beta95% CI for BetaR2P-ValueBeta95% CI for BetaR2ΔR2P-value of ΔR2P-value of Model
  1. *Block 2 employed stepwise entry of the PSQI and SHI (requiring P < .05 for entry).
  2. CI = confidence interval; ESS = Epworth Sleepiness Scale; GAD-7 = Generalized Anxiety Disorder-7; MIDAS = Migraine Disability Assessment Questionnaire; NS = not significant; PHQ-9 = Patient Health Questionnaire-9; PSQI = Pittsburgh Sleep Quality Index; SHI = Sleep Hygiene Index. Boldface indicates unique variance accounted for by poor sleep quality after controlling for depression and anxiety.
PSQI2.46(1.26, 3.66)18.2%.0001      
ESS0.07(−1.23, 1.37)0.0%NS      
SHI0.97(0.34, 1.61)11.0%.003      
PHQ-91.62(0.81, 2.42)17.6%.0001      
GAD-71.43(0.57, 2.29)12.8%.001      
Block 1      17.7%  .001
PHQ-9    1.47(0.07, 2.87)    
GAD-7    0.19(−1.27, 1.64)    
Block 2*      23.5%5.8%.021<.001
PSQI    1.64(0.25, 3.02)    


The current study examined the relative importance of 3 distinct sleep disturbance variables (ie, sleep quality, daytime sleepiness, sleep hygiene) pertinent to insomnia among young adult episodic migraineurs, a population of interest because of their high rates of migraine,[40, 41] disturbed sleep,[42, 43] and psychiatric comorbidities.[41, 44] Of additional interest was delineating any potential relationship between sleep disturbance and affective symptomatology. Episodic migraine was differentially associated with greater sleep disturbance, and sleep quality in particular emerged as the most robust and consistent differentiator of migraine. Poor sleep quality was strongly associated with variance in headache frequency (14.8%) and headache-related disability (18.2%), accounting for similar proportions as depression and anxiety. Furthermore, poor sleep quality remained associated with migraine frequency and disability even after controlling for symptoms of both depression and anxiety. The percentages of unique variance in frequency (5.3%) and disability (5.8%) accounted for by sleep quality were modest but non-trivial. Notably, when depression and anxiety scores were entered into the headache frequency and disability regressions subsequent (instead of prior) to the PSQI, the affective symptoms did not significantly improve either model beyond sleep quality alone. As such, the effects of poor sleep quality on migraine-related variables appear independent of, and potentially even more relevant than, comorbid affective symptoms.

By comparison, daytime sleepiness and sleep hygiene produced inconsistent results and smaller effects. Similar to the study by Calhoun et al among women with CM,[16] episodic migraineurs in the current study consistently engaged in a variety of poor sleep hygiene behaviors. However, specific sleep hygiene behaviors did not differentiate them from individuals without migraine, and sleep hygiene only was associated with headache disability before controlling for affective symptomatology. In the latter instance, sleep hygiene did not improve the relation beyond that already afforded by poor sleep quality. A significantly larger proportion of migraineurs than controls reported clinically significant daytime sleepiness, paralleling the findings of Barbanti and colleagues,[45] but daytime sleepiness was not associated with headache frequency, severity, or disability.

Sleep quality is an important factor uniquely associated with migraine disability,[20, 46] and the current findings indicate that sleep quality is of clinical importance even among young, non-clinical episodic migraineurs. As such, sleep quality should be assessed preferentially to other sleep disturbance variables when subjective self-report measures of insomnia are used, particularly among those reporting high headache frequency and associated disability, and even in the absence of significant depression or anxiety. In light of the present findings, conclusions from systematic reviews,[47] and practice guidelines,[48, 49] the most potent means of improving sleep quality and insomnia among migraineurs is likely a treatment package that incorporates stimulus control and/or sleep restriction in addition to basic sleep hygiene education and management of comorbid psychiatric symptoms.[50, 51]

Despite the high rates of migraine-related disability and sleep disturbance among our sample,[52, 53] use of well-validated measures, and reliance on a face-to-face structured headache interview to establish migraine diagnoses, several limitations exist. Alcohol use/abuse was not assessed but is often comorbid with and influences sleep problems[54] and is disproportionately prevalent among young adult populations.[55] Future research should consider whether potential group differences in substance use affect the roles of sleep and affective comorbidities in migraine. Incorporating daily sleep diary data would further strengthen the present design by allowing prospective examination of the sleep disturbance variables with new-onset migraine, although examining sleep as a trigger of individual headache attacks was not a goal of this study. Given that this was not a treatment-seeking sample, we did not assess frequency of medication use for headache or insomnia, although future studies should consider incorporating these variables into similar analyses. Finally, given our broad interest in comparing aspects of sleep disturbance, we did not attempt to isolate the specific contributors to poor sleep quality in particular, such as delayed sleep onset latency or shortened sleep duration, although their relation with headache-related variables merits future exploration. In light of our findings and the stark paucity of data regarding the effects on migraine of treating comorbid psychiatric disorders, a strong need remains for treatment studies that assess the effects on migraine of comprehensive strategies to treat sleep disturbance and psychiatric comorbidities.

Statement of Authorship

Category 1

  • (a)Conception and DesignA. Brooke Walters, Todd A. Smitherman
  • (b)Acquisition of DataA. Brooke Walters, Joshua D. Hamer, Todd A. Smitherman
  • (c)Analysis and Interpretation of DataA. Brooke Walters, Joshua D. Hamer, Todd A. Smitherman

Category 2

  • (a)Drafting the ManuscriptA. Brooke Walters, Joshua D. Hamer, Todd A. Smitherman
  • (b)Revising It for Intellectual ContentA. Brooke Walters, Joshua D. Hamer, Todd A. Smitherman

Category 3

  • (a)Final Approval of the Completed ManuscriptA. Brooke Walters, Joshua D. Hamer, Todd A. Smitherman