The association between multidimensional sleep health and migraine burden among patients with episodic migraine
ABSTRACT
Study Objectives:
Using the Sleep Regularity, Satisfaction, Alertness, Timing, Satisfaction, and Duration (Ru-SATED) sleep health framework, we examined the association between multidimensional sleep health and headache burden in a cohort of 98 adults with episodic migraine.
Methods:
Participants wore wrist actigraphs and completed twice-daily electronic diaries regarding sleep, headaches, and other health habits for 6 weeks. We calculated separate composite sleep health scores from diary and actigraphy assessed measures using the Ru-SATED framework. We used adjusted multivariable linear regression models to examine the association between composite sleep health scores and headache frequency, duration, and pain intensity.
Results:
Among 98 participants (mean age: 35 ± 12 years; 87.8% female), 83 had healthy ranges in ≥ 3 sleep dimensions. In models adjusted for age, sex, menopausal status, physical activity and alcohol intake, good sleep health was associated with fewer headache days/month (actigraphy: 3.1 fewer days; 95% confidence interval: 0.9, 5.7; diary: 4.0 fewer days; 95% confidence interval: 1.1, 6.9). Results did not change substantively with further adjustment for stress and depressive symptoms. We did not observe an association between sleep health and headache duration or intensity, respectively.
Conclusions:
Among patients with episodic migraine, good multidimensional sleep health, but not the majority of singular dimensions of sleep, is associated with approximately 3–4 fewer headache days/month. In addition, there was no association with headache duration or intensity. These findings highlight the importance of assessing multiple dimensions of sleep and suggest that improving sleep health may be a potential clinical strategy to reduce headache frequency.
Citation:
Yoo A, Vgontzas A, Chung J, et al. The association between multidimensional sleep health and migraine burden among patients with episodic migraine. J Clin Sleep Med. 2023;19(2):309–317.
BRIEF SUMMARY
Current Knowledge/Study Rationale: Poor sleep is frequently reported among patients with migraine and described as a trigger for headache. Examination of the relationship between sleep disturbance and migraine has largely been limited to retrospective and self-reported sleep characteristics, or mainly focused on single indices, for which there have been discrepant findings. A multidimensional assessment of sleep using both self-reported and objective sleep measures over time may improve the characterization of sleep and its potential impact on migraine burden.
Study Impact: Good sleep health, as measured by a composite multidimensional sleep health assessment, is associated with fewer headache days per month. Multidimensional and longitudinal sleep measures in aggregate may be more helpful in capturing the relationship between sleep and migraine, compared with single indices of sleep.
INTRODUCTION
Migraine is a chronic neurologic disorder characterized by recurrent moderate to severe headaches associated with nausea, photophobia, or phonophobia. It affects an estimated 1 billion adults worldwide and is the leading cause of disability under age 50.1–3 Compared with individuals without migraine, patients with migraine endorse higher rates of insomnia symptoms,4,5 lower-quality sleep,6,7 and shorter self-reported sleep duration,8 and frequently describe poor sleep as a migraine trigger.9 However, despite the high burden of sleep problems reported among patients with migraine, our understanding of the relationship between sleep and migraine remains incompletely understood, which limits our ability to devise sleep interventions to reduce migraine burden.
Examination of the relationship between sleep disturbance and migraine has largely been limited to the use of retrospective and self-reported sleep characteristics, or mainly focused on short sleep duration and sleep quality, for which there have been discrepant findings. Although cross-sectional studies have linked several individual sleep indices (ie, < 6 hours/night, later sleep midpoint, poorer sleep quality) with higher migraine frequency in patients with episodic and chronic migraine,8,10–12 prospective studies examining nightly individual sleep dimensions and migraine frequency have been mixed.12,13 For example, a 4-week observational study in 55 patients with chronic migraine or tension-type headache reported that < 4 hours of self-reported sleep duration on 2 consecutive nights was associated with next-day headache.13 Conversely, a 6-week prospective study among 98 patients with episodic migraine observed no association between nightly (or 2 consecutive nights of) self-reported or objective sleep duration < 6.5 hours or poor-quality sleep and headache occurrence.12 These inconsistencies may be due to the reliance on single sleep indices, limited use of objective estimates of sleep, and/or the different periods of sleep assessments (eg, nightly vs monthly), or different populations (episodic vs chronic migraine). Given the uncertainty regarding the influence of individual sleep indices on migraine, the time course by which sleep may impact migraine (and vice versa), and limited objective data on habitual sleep in patients with migraine, examining a more global assessment of sleep14 using both self-reported and objective sleep measures over time may improve the characterization of multidimensional habitual sleep and its impact on migraine. Assessing multidimensional sleep health as a composite metric provides a global assessment that may reflect the complex neurophysiologic processes of sleep compared with individual sleep dimensions. Additionally, using a composite metric may also be advantageous in detecting relationships between sleep and health outcomes when single metrics may interact with one another or may be underpowered.15,16 For example, prior studies have reported associations between multidimensional sleep health and all-cause mortality17 and heart disease,18 whereas there was no association or weaker associations with single dimensions of sleep. This strength can be leveraged in migraine and sleep research, which has produced some conflicting findings thus far.
Therefore, we evaluated the association between multidimensional sleep health and headache frequency in 98 individuals with episodic migraine followed over 6 weeks. Component dimensions of the Ru-SATED framework14 (sleep regularity, satisfaction, alertness, timing, efficiency, and duration) were operationalized using longitudinal actigraphy and diary data over 6 weeks to develop separate composite scores. We hypothesized that better sleep health would be associated with fewer headache days and, secondarily, shorter average headache duration and lower average maximal pain intensity over the 6-week observation period.
METHODS
Study setting and overview
Full details of the study methods have been previously described.12 In brief, from March 2016 to October 2017, potential participants meeting the criteria for episodic migraine with or without aura according to the International Classification of Headache Disorders, third edition were recruited from 3 academic medical centers in Boston, Massachusetts (Beth Israel Deaconess Medical Center, Massachusetts General Hospital, Brigham and Women’s Hospital) as well as at local colleges. A total of 126 participants expressed interest in participation, 110 were screened, and 101 met inclusion criteria. We excluded 3 participants with fewer than 21 days of daily data, resulting in a final analytic sample of 98 participants.
At enrollment, participants completed standardized questionnaires. Participants wore a wrist actigraph throughout the 6-week study period and completed daily morning and evening diaries. Questionnaire and diary data were collected using REDCap (Research Electronic Data Capture) hosted at Beth Israel Deaconess Medical Center. We conducted all visits at Beth Israel Deaconess Medical Center. The Beth Israel Deaconess Medical Center Committee on Clinical Investigations approved the study, and all participants provided written informed consent.
Study population
Enrolled participants met criteria for migraine with or without aura according to the International Classification of Headache Disorders, third edition based on an in-person physician interview at baseline. Additionally, participants were required to be ≥18 years of age, have at least a 3-year history of migraine, have at least 2 headaches/month during the previous 3 months, communicate in English, and give informed consent. Exclusion criteria included a self-report of ≥15 headache days/month over the past 3 months, presence of a chronic pain condition, current opioid use, high risk of obstructive sleep apnea based on the Berlin Questionnaire19 or reported history of untreated obstructive sleep apnea of any severity, pregnancy, current uncontrolled medical problems that would preclude participation, or failure to complete 4 out of the 7 days of diary data during an initial diary run-in period before enrollment.
Headache assessments
Participants completed daily morning and evening diaries that assessed the presence of headaches. If a headache occurred, participants provided details on the time of onset and offset, maximal pain intensity, and medications used to acutely treat headache each day. When headache onset was on the same calendar day of a prior headache’s resolution, we considered it a relapse (same headache event) as has been done previously in this cohort.20 We calculated monthly headache days by multiplying the total number of headache days by 30 and dividing by the number of total diary days, consistent with an operationalization from a prior study.21 We also measured additional outcomes, including headache duration and maximum pain intensity (measured by a 100-point visual analog scale). For headaches classified as a relapse (ie, recurrence of headache within the same calendar day), the total duration was measured from the start of the first headache to the end of the second headache, and pain intensity was recorded as the higher of the 2 headaches. Descriptive information on overall headache-related disability was collected using the Headache Impact Test (HIT-6) completed at baseline.
Sleep measures
Participants completed a daily morning sleep diary (Consensus Sleep Diary)22 that assessed sleep quality (very poor, poor, fair, good, very good), sleep and bed timings, awakenings, and use of sedative-hypnotic medications over 6 weeks. Additionally, each participant wore an actigraph (Actiwatch Spectrum; Philips Respironics, Murraysville, PA) on their nondominant wrist 24 hours per day for 42 consecutive days. The actigraph recorded data on activity and light in 30-second intervals and indicated any “off-wrist” time. Study staff then downloaded and transmitted the data to the Brigham Women’s and Health Sleep Reading Center. A trained technician blinded to daily headache status scored all the studies using a standardized protocol described previously.12 Participants also completed the Patient-Reported Outcomes Measurement Information System Sleep-Related Impairment (PROMIS SRI) item bank,23 the Pittsburgh Sleep Quality Index (PSQI),24 and the Insomnia Severity Index,25 and they provided information about their use of hypnotic medications at baseline.
Sleep health score
We calculated a sleep health composite score comprising the 6 sleep dimensions of the Ru-SATED framework14: regularity, satisfaction, alertness, timing, efficiency, and duration. These domains were operationalized using Sleep Regularity Index for regularity,26 average sleep quality per participants’ daily diary ratings for satisfaction, and PROMIS SRI T-score for alertness, sleep midpoint for timing, sleep efficiency for efficiency, and total sleep time for duration. Sleep Regularity Index, sleep midpoint, sleep efficiency, and total sleep time were operationalized using separate values from actigraphy and diary for the regularity, timing, efficiency, and duration dimensions, respectively. We dichotomized each dimension based on cutoffs recommended in guidelines (for sleep duration and PROMIS SRI scores),23,27 estimates of normative values or cutoffs established in prior studies (for Sleep Regularity Index, sleep efficiency, and sleep midpoint),26,28 or percentiles of the sample distribution (for sleep satisfaction), consistent with previous approaches to defining multidimensional sleep health in epidemiologic studies.29,30Table 1 summarizes the operationalization of the individual dimensions and the justification for threshold values that we used to dichotomize each dimension into poor or good sleep health.
| Sleep Health Dimension | Operationalization | Dichotomization Rationale | Individuals Rated as “Healthy: by Actigraphy (%) | Individuals Rated as “Healthy” by Diary/ Self-Report (%) |
|---|---|---|---|---|
| Regularity | Sleep Regularity Index ≥ 68.0 (actigraphy)a; ≥ 76.3 (diary)a | Lower quartile of Sleep Regularity Index. Comparable cutoff to prior studies26 | 75 (76.5%) | 73 (74.5%) |
| Satisfaction | Average self-reported sleep quality of 3.5b or greater (diary only) | Represents greater good quality days vs bad; also represents near median split in the current sample | Not Applicable | 52 (53.1%) |
| Alertness | PROMIS SRI T-Score ≤ 50 | Normalized cutoff for PROMIS measures (population mean = 50) | Not Applicable | 44 (44.9%) |
| Timing | Average sleep midpoint between 02:00 and 04:00 am | Consistent with dichotomization rules in prior studies51 | 65 (66.3%) | 71 (72.5%) |
| Efficiency | Average sleep efficiency ≥ 88.6 (actigraphy)a; ≥ 90.8 (diary)a | Lower quartile of efficiency within our distribution. Consistent with upper bounds of polysomnography sleep efficiency in healthy adults of similar age28 | 74 (75.5%) | 74 (75.5%) |
| Duration | Average duration between 420 and 540 minutes (7–9 hours) | ATS guidelines for young adults and adults27 | 63 (64.3%) | 80 (81.6%) |
Regularity was defined using the average Sleep Regularity Index,26 which estimates the probability of an individual being awake or asleep at any 2 time points, 24 hours apart, averaged over the 6-week study (range: 0–100). We also derived average values for nightly sleep efficiency (proportion of total sleep time/duration of the rest period, duration, and midpoint across the 6 weeks). For both the diary- and actigraphy-derived sleep health scores, we defined satisfaction using the mean nightly self-assessed sleep quality across the study period, rated on a scale of 1 to 5, where 1 = “very poor” and 5 = “very good.” Alertness was assessed using the 8-item PROMIS SRI short form and was the only dimension based on data collected at baseline. Based on these categories, each dimension was scored as 0 (poor) or 1 (good) for a maximum composite score of 6, with a higher score reflecting better sleep health. We derived composite sleep health scores for diary and actigraphy data separately. Based on the empiric shape of the distributions, we defined good overall sleep health as a score of ≥ 3, corresponding to about the 25th percentile for both actigraphy and diary derivations.
Covariates
At baseline, we collected information on demographics, medical history, medication use, health habits, stress, and depressive symptoms using standardized questionnaires. A priori, we specified age, sex, menopausal status, alcohol use, physical activity, stress, and depressive symptoms as potential confounders. Participants reported menopausal status as “pre-menopausal,” “menopausal,” and “I don’t know.” We categorized those reporting “I don’t know” as “pre-menopausal” or “menopausal” based on their age compared with the median age of menopause in the United States (52 years). Alcohol use was classified based on reporting any current use (yes/no). We also calculated weekly drinks of alcohol per week from daily diaries for sensitivity analysis. We categorized physical activity as no or low activity vs regular activity, defined as at least 3 days/week of moderate to strenuous physical activity.20 We used the 10-item Perceived Stress Scale31 (PSS-10) to categorize stress symptom burden (high/low) based on a cutoff ≥ 12 on the PSS-10.32 We used the 20-item Center for Epidemiologic Studies–Depression scale33 (CES-D) to assess depressive symptom burden, excluding the item, “My sleep was restless,” from the total summed CES-D scores given that our exposure of interest was sleep health.
Statistical analysis
We used multivariable linear regression to estimate the associations between actigraphy and diary-derived sleep health with monthly headache days separately. We also estimated the associations between sleep health and average headache duration and average maximal headache intensity separately. In our primary models, we adjusted for age, sex, menopausal status, alcohol use, and physical exercise. Since both high stress and depressive symptoms can bidirectionally relate to some sleep dimensions, we further adjusted for stress (high vs low) and depressive symptoms (total CES-D scores) in a subsequent model. In sensitivity analyses, we further adjusted for prescription migraine prophylactic medication use (ie, daily use of metoprolol, propranolol, topiramate, amitriptyline, or venlafaxine to reduce headache frequency as a binary variable). In a separate model, we adjusted for weekly alcohol use as a continuous variable. For both sensitivity analyses adjusting for prophylactic medication use and prospective weekly alcohol use, identical models were used to evaluate the associations of sleep health and average headache duration with intensity. In exploratory analyses, we assessed the associations between individual dichotomized sleep health dimensions and migraine outcomes separately using multivariable linear regression models adjusting for age, sex, menstrual status, alcohol use, and physical exercise. In post hoc analyses, we also explored the association between good sleep quality (cutoff PSQI < 5) and monthly headache days using the same modeling approach. Two-sided P values < .05 were considered statistically significant. All analyses were performed using SAS version 9.4 (SAS Institute, Inc, Cary, NC).
RESULTS
Participant characteristics
Table 2 presents the sample’s baseline characteristics categorized by good vs poor sleep health scores. Among the 98 participants, 74 were premenopausal women, 12 were menopausal women, and 12 were men. There was no significant difference in the distribution of sex and menopausal characteristics across groups. The 2 groups did not differ on baseline headache characteristics. Participants with good sleep health reported an average of 4.9 ± 3.5 headaches per month and a mean HIT-6 score of 60.6 ± 6.4, suggesting severe headache-related disability. Participants with good sleep health were less likely to experience moderate to high stress and had less severe depression symptoms. Additionally, they had lower total Insomnia Severity Index scores and total PSQI scores (Table 2).
| Sleep Health Score < 3 (n = 15) | Sleep Health Score ≥ 3 (n = 83) | P | |
|---|---|---|---|
| Sociodemographics | |||
| Age, y | 34.9 ± 11.6 | 35.1 ± 12.2 | .94 |
| Race | |||
| White | 12 (80.0) | 69 (83.1) | .56 |
| Black | 0 | 4 (4.8) | |
| Asian | 1 (6.7) | 3 (3.6) | |
| Multiple races/other | 2 (13.3) | 7 (8.4) | |
| Sex and menopausal status | |||
| Male | 1 (6.8) | 11 (13.3) | .71 |
| Premenopausal female | 13 (86.7) | 61 (73.5) | |
| Postmenopausal female | 1 (6.7) | 11 (13.3) | |
| Headache characteristics | |||
| Baseline headache days per month | 5.7 ± 3.7 | 4.9 ± 3.5 | .48 |
| Headache Impact Test score (range: 36–78) | 63.2 ± 4.8 | 60.6 ± 6.4 | .08 |
| Migraine with aura | 5 (33.3) | 27 (32.5) | .99 |
| On migraine prophylaxisa | 5 (33.3) | 21 (25.3) | .53 |
| Health habits and psychological health | |||
| Alcohol use | 7 (46.7) | 59 (72.0) | .07 |
| Smoking status | .34 | ||
| Never smoker | 12 (85.7) | 69 (84.1) | |
| Past or current smoker | 2 (14.3) | 13 (15.9) | |
| Regular physical activityb | 7 (50.0) | 52 (65.0) | .28 |
| PSS-10 ≥ 14 | 13 (86.7) | 38 (45.8) | .003 |
| CES-D (sleep question removed), total score | 14.6 ± 9.2 | 7.7 ± 6.5 | .01 |
| Using regular hypnotic medication | 5 (33.3) | 17 (20.5) | .30 |
| Insomnia Severity Index, total score | 11.7 ± 6.2 | 6.7 ± 4.9 | .01 |
| PSQI, total score | 7.5 ± 4.4 | 4.2 ± 2.4 | .01 |
Sample sleep health scores
Eighty-three participants were categorized as having good sleep health (Sleep Health Score ≥ 3) by actigraphy. Characteristics for actigraphy-derived sleep health status overall and by individual dimensions are described in Table 2. The distributions of participants among actigraphy-derived sleep health scores were as follows: 0 (n = 1), 1 (n = 4), 2 (n = 10), 3 (n = 22), 4 (n = 31), 5 (n = 21), and 6 (n = 9). The distribution among diary-derived sleep health scores was similar, although more individuals were classified as “healthier” sleep compared with actigraphy: 0 (n = 1), 1 (n = 2), 2 (n = 8), 3 (n = 18), 4 (n = 35), 5 (n = 22), and 6 (n = 12). Agreement between diary and actigraphy derivations of poor (Sleep Health Score < 3) vs good (Sleep Health Score ≥ 3) sleep health was high, with 89.8% of participants being identically classified. McNemar’s test revealed no significant difference between the 2 sleep health derivations (P = .21). Table 3 describes the average values of each dimension by sleep health status.
| Sleep Health Score < 3 (n = 15) | Sleep Health Score ≥ 3 (n = 83) | P | |
|---|---|---|---|
| Sleep dimensions by actigraphy | |||
| Regularity: Sleep Regularity Index | 63.5 ± 7.5 | 73.3 ± 6.0 | <.001 |
| Satisfaction: nightly sleep quality rating, range 1–5 | 3.1 ± 0.6 | 3.5 ± 0.4 | .009 |
| Alertness: PROMIS Sleep-Related Impairment T-score | 56.6 ± 4.0 | 48.3 ± 5.6 | <.001 |
| Timing: sleep midpoint, clock time ± min | 4:21 am ± 69.4 | 3:29 am ± 77.7 | .02 |
| Efficiency: time asleep/time in bed, % | 85.8 ± 4.1 | 90.2 ± 2.4 | .001 |
| Duration: total sleep time, min | 411.9 ± 41.3 | 441.2 ± 36.3 | .02 |
| Sleep Health Score < 3 (n = 11) | Sleep Health Score ≥ 3 (n = 87) | ||
| Sleep dimensions by diary | |||
| Regularity: Sleep Regularity Index | 71.6 ± 9.3 | 80.7 ± 6.3 | .008 |
| Satisfaction: nightly sleep quality rating, range 1–5 | 2.9 ± 0.5 | 3.5 ± 0.4 | <.001 |
| Alertness: PROMIS Sleep-Related Impairment T-score | 55.0 ± 3.9 | 48.9 ± 6.0 | <.001 |
| Timing: sleep midpoint, clock time ± min | 4:21 am ± 91.3 | 3:23 am ± 54.3 | .06 |
| Efficiency: time asleep/time in Bed, % | 85.5 ± 12.7 | 93.4 ± 3.7 | .07 |
| Duration: total sleep time, min | 445.7 ± 71.5 | 466.7 ± 39.7 | .36 |
Sleep health and headache outcomes
Participants with good actigraphy-derived sleep health had 3.4 (95% confidence interval [CI]: 1.1, 5.7) fewer headache days per month than those with poor sleep health after adjusting for age, sex, and menopausal status (Table 4). Results were similar with additional adjustment for alcohol use and physical activity (3.1 fewer headache days/month; 95% CI: 0.7, 5.5) as well as further adjustment for stress and depressive symptoms (2.7 fewer headache days/month; 95% CI: 0.2, 5.2). Good diary-based sleep health was associated with 4.0 (95% CI: 1.1, 6.9) fewer headache days/month in the primary model and was similar after additional adjustment for stress and depressive symptoms. Results did not differ substantively with further adjustment for use of a migraine prophylactic medication or when accounting for longitudinally calculated weekly alcohol use. We found no association between actigraphy- or diary-derived sleep health and headache duration or intensity. Results were similar with further adjustment.
| Model 1a | Model 2b | Model 3c | |
|---|---|---|---|
| Average headache frequency (days per month) | |||
| Actigraphy | –3.4 (–5.7, –1.1) | –3.1 (–5.5, –0.7) | –2.7 (–5.2, –0.2) |
| Diary | –3.5 (–6.2, –0.8) | –4.0 (–6.9, –1.1) | –3.5 (–6.6, –0.5) |
| Average headache durationd (hours) | |||
| Actigraphy | –4.4 (–9.3, 0.6) | –2.0 (–7.1, 2.9) | –2.7 (–7.9, 2.5) |
| Diary | –4.7 (–10.5, 1.1) | –1.5 (–7.6, 4.5) | –1.6 (–7.9, 4.7) |
| Average maximal headache intensityd,e | |||
| Actigraphy | –7.3 (–15.1, 0.5) | –5.8 (–14.0, 2.4) | –6.8 (–15.3, 1.6) |
| Diary | –6.6 (–15.7, 2.5) | –4.4 (–14.4, 5.6) | –4.2 (–14.5, 6.1) |
In exploratory analyses of the individual dimensions, those with actigraphy-estimated efficiency of ≥ 88.6%12 had 2.2 fewer headache days/month (95% CI: 0.2, 4.1) than those with “poor” sleep efficiency. However, good diary-estimated efficiency was not associated with fewer headache days (1.4 fewer days; 95% CI: –0.7, 4.5). In our primary model, no other actigraphy- or diary-estimated sleep individual dimensions were associated with headache days/month (Table 5).
| Model 1a | Model 2b | Model 3c | |
|---|---|---|---|
| Regularity | |||
| Actigraphy (Sleep Regularity Index ≥ 68.0) | –0.9 (–3.0, 1.1) | –0.6 (–2.8, 1.5) | –0.4 (–2.5, 1.7) |
| Diary (Sleep Regularity Index ≥ 76.3) | –0.2 (–2.1, 1.8) | –0.2 (–1.8, 2.2) | 0.4 (–1.6, 2.4) |
| Satisfaction (quality ≥ 3.5) | –0.9 (–2.7, 0.9) | –1.0 (–2.7, 0.8) | –0.6 (–2.4, 1.2) |
| Alertness (PROMIS SRI T-score ≤ 50) | –1.8 (–3.6, 0.1) | –1.6 (–3.4, 0.1) | –1.1 (–3.2, 1.0) |
| Timing (midpoint between 2:00 and 4:00 am) | |||
| Actigraphy | –1.0 (–2.9, 0.9) | –0.5 (–2.5, 1.4) | –0.4 (–2.3, 1.5) |
| Diary | –0.7 (–2.7, 1.3) | –0.2 (–2.2, 1.8) | –0.1 (–2.1, 1.9) |
| Efficiency | |||
| Actigraphy (efficiency ≥ 88.6) | –2.1 (–4.1, –0.2) | –2.2 (–4.1, –0.2) | –2.0 (–3.9, –0.01) |
| Diary (efficiency ≥ 90.8) | –1.7 (–3.7, 0.3) | –1.4 (–4.5, 0.7) | –1.3 (–3.4, 0.8) |
| Duration (duration between 420 and 540 minutes) | |||
| Actigraphy | 0.6 (–1.3, 2.4) | 0.4 (–1.5, 2.3) | –0.2 (–1.7, 2.1) |
| Diary | –0.7 (–3.0, 1.6) | –1.0 (–3.3, 1.4) | –0.8 (–3.2, 1.7) |
In post hoc analyses, a PSQI total score < 5 was associated with 1.7 fewer headache days (95% CI: –0.1, 3.4) per month in the model adjusted for age, sex, menopausal status, physical activity, and alcohol use (Table 6).
| Model 1a | Model 2b | Model 3c | |
|---|---|---|---|
| Average headache frequency (days per month) | –1.9 (–3.7, –0.2) | –1.7 (–3.4, 0.1) | –1.2 (–3.1, 0.7) |
| Average headache durationd (hours) | –0.7 (–4.4, 3.0) | 0.1 (–3.5, 3.7) | –0.1 (–4.0, 3.8) |
| Average headache intensityd,e | –0.7 (–6.5, 5.1) | –1.0 (–6.9, 4.9) | –1.2 (–7.6, 5.1) |
DISCUSSION
In this cohort of 98 adults with episodic migraine, good multidimensional sleep health was associated with 3–4 fewer headache days/month compared with poor sleep health, when derived from either actigraphy or diary data, although estimates were imprecise. We did not observe an association between multidimensional sleep health and average maximal headache intensity or duration for either actigraphy- or diary-derived composite scores. In exploratory analyses of single sleep dimensions, only good actigraphy-estimated sleep efficiency was significantly associated with fewer headache days.
These results build on previous work examining the relationship between sleep and migraine frequency by providing evidence that aggregate, multidimensional, and longitudinal sleep measures may be more helpful for capturing the relationship between sleep and migraine, compared with single indices of sleep. This is supported by our exploratory analysis of individual dimensions and migraine burden over 6 weeks, which resulted in more modest estimates and were largely null. The more robust relationship seen between our sleep health composite score and migraine frequency may reflect the cumulative effects of and/or interactions between individual dimensions that single indices cannot otherwise account for.15 Therefore, in adults with episodic migraine, the multidimensional sleep health framework may provide a more sensitive tool than single sleep indices in detecting levels of sleep health relevant to headache frequency.
Although the sleep health framework14 was recently described to comprehensively capture sleep dimensions salient to various measures of health, the PSQI is another widely used global assessment of sleep that captures multiple dimensions of sleep34 and has been commonly used in migraine research. A recent meta-analysis of 20 studies reported that PSQI scores are higher among people with migraine compared with healthy participants.35 Multiple cross-sectional studies have reported that poor-quality sleep and higher PSQI scores are associated with a higher frequency of headache attacks among patients with both episodic and chronic migraine.6,7,36 Additionally, in this cohort, we found that a PSQI ≥ 5 was associated with a 22% higher odds of headache recurrence over 6 weeks.32 Our current analyses found that a PSQI < 5 was statistically significantly associated with 1.7 fewer headache days/month after adjustment for potential confounders, although estimates were imprecise. The weaker association of PSQI compared with our sleep health composite score may reflect the limitations of a retrospective, single-time-point assessment of sleep quality in discriminating between good vs poor sleepers compared with data collected longitudinally.
The association between poor sleep health and migraine may be mediated by plausible common neurobiological mechanisms. For example, the hypothalamus is likely a key important structure due to its role in sleep-wake regulation and activation in the premonitory phase of acute migraine attacks.37 Although its exact function during an attack is unclear, hormones secreted by the hypothalamus (eg, gonadotropin-releasing hormone, which stimulates follicle-stimulating hormone/luteinizing hormone and ultimately estrogen release) and neurotransmitters under hypothalamic influence (eg, serotonin) are associated with the onset of migraine attacks.38,39 Limited evidence from animal studies suggests a link between cortical spreading depression, the physiological correlate of migraine aura, and sleep. For example, the threshold for inducing cortical spreading depression is lower in rats that are sleep deprived40 Cortical spreading depression itself may lead to temporary impairment of glymphatic flow,41 a waste clearance system primarily active during sleep.
Despite our findings, further work is needed to validate the utility of multidimensional sleep health among patients with episodic migraine. For example, studies using factor analysis42 or clustering techniques43 to study sleep health dimensions may determine the most representative actigraph and self-reported measures for each dimension related to migraine burden. Optimal cutoff thresholds for representative metrics without known normative values will need to be defined. Finally, the relative informativeness of different sleep dimensions on headache frequency should be explored. While we assumed all dimensions are equally informative in our global score, it is unclear if poor health in certain domains, such as sleep efficiency, may bear a more substantial impact in terms of headache risk than sleep timing.
Our study has multiple strengths, including 6 weeks of longitudinally collected data for both exposure and outcomes, with high adherence to diary completion (> 96%). This is particularly important since prior studies have reported a low agreement between estimates from retrospective and longitudinal measures of sleep and headache frequency.44–46 We also developed sleep health scores using data collected from diaries and actigraphs separately, unlike most other studies examining multidimensional sleep health.29,47,48 As a result, we were able to obtain similar estimates between self-report and objective derivations of sleep health. This finding is notable, as the agreement between self-reported and objective derivations of sleep health may allow the framework to be optimally adapted to both clinical and research settings. For example, it is free and practical to obtain self-reported measures, making it favorable in the clinical setting. In contrast, the resolution of data and the lower burden on participants provided by actigraphy make it a valuable research tool. Thus, the sleep health framework may have unique advantages in assessing sleep in adults with episodic migraine.
Despite these strengths, our study has several limitations. Estimates from our analyses were imprecise, which may reflect variability across individuals in the influence of habitual sleep disturbance on headache frequency. Although we collected data longitudinally, our analyses were cross-sectional to capture sleep health as a more stable “trait.” As a result, our results may be capturing reverse causation (ie, more frequent headaches causing poorer sleep health). However, this may be less likely given that we previously found only modest temporal associations between daily headache status and subsequent nightly sleep characteristics.20 Also, our use of a composite score may obscure the effects of individual dimensions, some of which may be substantially influenced by headache frequency and associated medications (eg, alertness). This particular issue may be less problematic in our cohort, as excessive daytime sleepiness occurs less frequently in adults with episodic migraine compared with those with chronic migraine49,50 Moreover, our sensitivity analysis adjusting for migraine prophylactic medication use (including tricyclics) did not substantively alter our results. Although our study contains the largest collection of objective sleep data in episodic migraine to date, it was not powered to assess for differences in headache frequency across the entire spectrum of sleep health (as a range of 0 to 6) or to conduct analyses to evaluate interactions among individual dimensions. Additionally, our cohort largely comprised healthy sleepers as evidenced by the relatively modest deviations from each dimension cutoff. This homogeneity may have limited our power to detect effects and potentially reduced the generalizability of our findings. However, our findings may suggest a role for sleep optimization in adults with migraine with relatively mildly reduced sleep health. Finally, this study excluded children and adolescents in whom sleep may also have an important relationship with headache; however, children and adolescents have biologically different sleep needs across developmental stages.
In conclusion, compared with poor sleep health, good multidimensional sleep health (healthy in at least 3 of 6 dimensions of sleep) was associated with 3 to 4 fewer headache days per month among adults with episodic migraine. Our data support the development of future, larger studies to replicate our findings and prospective studies examining the causal relationship between good sleep health and headache risk to inform novel interventions targeting sleep health to reduce migraine burden for patients with episodic migraine.
ABBREVIATIONS
| CES-D | Center for Epidemiologic Studies–Depression Scale |
| CI | confidence interval |
| PROMIS | Patient-Reported Outcomes Measurement Information System |
| PSQI | Pittsburgh Sleep Quality Index |
| SRI | Sleep-Related Impairment |
DISCLOSURE STATEMENT
All authors have seen and approved the manuscript. Work for this study was performed jointly in the Department of Neurology in the University of Rochester and the Department of Sleep and Circadian Disorders associated with the Brigham and Women’s Hospital. This study was funded by grants from the National Institute of Neurological Disorders and Stroke (R21-NS091627), the American Sleep Medicine Foundation, Harvard Catalyst/The Harvard Clinical and Translational Science Center (National Center for Advancing Translational Sciences, National Institutes of Health [UL 1TR002541, K01 AA027831, 5T32NS007338-32]), and financial contributions from Harvard University and its affiliated academic health care centers. The content is solely the responsibility of the authors and does not necessarily represent the official views of the University of Rochester, Harvard Catalyst, Harvard University, and its affiliated academic health care centers, or the National Institutes of Health or the official views of the funding organizations. Dr. Bertisch has served as a paid consultant to Idorsia, OptumCare, and ResMed. Over the past 3 years, Dr. Buysse has served as a paid consultant to National Cancer Institute, Pear Therapeutics, Sleep Number, Idorsia, Eisai, and Weight Watchers International. Dr. Buysse is an author of the Pittsburgh Sleep Quality Index, Pittsburgh Sleep Quality Index Addendum for PTSD (PSQI-A), Brief Pittsburgh Sleep Quality Index (B-PSQI), Daytime Insomnia Symptoms Scale, Pittsburgh Sleep Diary, Insomnia Symptom Questionnaire, and Ru-SATED (copyrights held by University of Pittsburgh). These instruments have been licensed to commercial entities for fees. He is also co-author of the Consensus Sleep Diary (copyright held by Ryerson University), which is licensed to commercial entities for a fee. Dr. Yoo, Dr. Vgontzas, Dr. Chung, Dr. Mostofsky, Dr. Li, Mr. Rueschman, and Dr. Mittleman report no conflicts of interest.
REFERENCES
1. . Migraine prevalence, disease burden, and the need for preventive therapy. Neurology. 2011;77(21):1905 .
2. . What is the actual prevalence of migraine? Brain Behav. 2018;8(6):e00950 .
3. GBD 2016 Disease and Injury Incidence and Prevalence Collaborators. Global, regional, and national incidence, prevalence, and years lived with disability for 328 diseases and injuries for 195 countries, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet. 2017;390(10100):1211–1259 .
4. . Insomnia in probable migraine: a population-based study. J Headache Pain. 2016;17(1):92 .
5. . The impact of headache and chronic musculoskeletal complaints on the risk of insomnia: longitudinal data from the Nord-Trøndelag Health Study. J Headache Pain. 2013;14(1):24 .
6. PAMINA Study Group. Quality of sleep, fatigue and daytime sleepiness in migraine—a controlled study. Cephalalgia. 2009;29(6):662–669 .
7. . Poor sleep quality in migraine and probable migraine: a population study. J Headache Pain. 2018;19(1):58 .
8. . Headache and sleep: examination of sleep patterns and complaints in a large clinical sample of migraineurs. Headache. 2005;45(7):904–910 .
9. . The triggers or precipitants of the acute migraine attack. Cephalalgia. 2007;27(5):394–402 .
10. . Sleep and headaches: relationships between migraine and non‐migraine headaches and sleep duration, sleep quality, chronotype, and obstructive sleep apnoea risk. Aust J Psychol. 2017;69(3):210–217 .
11. . Short sleep duration and poor sleep quality among migraineurs: a population-based study. Cephalalgia. 2018;38(5):855–864 .
12. . Nightly sleep duration, fragmentation, and quality and daily risk of migraine. Neurology. 2020;94(5):e489–e496 .
13. . Stress and sleep duration predict headache severity in chronic headache sufferers. Pain. 2012;153(12):2432–2440 .
14. . Sleep health: can we define it? Does it matter? Sleep. 2014; 37(1):9–17 .
15. . Multidimensional sleep health in a diverse, aging adult cohort: concepts, advances, and implications for research and intervention. Sleep Health. 2021;7(6):699–707 .
16. . Beyond single sleep measures: a composite measure of sleep health and its associations with psychological and physical well-being in adulthood. Soc Sci Med. 2021;274:113800 .
17. Osteoporotic Fractures in Men (MrOS) Study Research Group. Which sleep health characteristics predict all-cause mortality in older men? An application of flexible multivariable approaches. Sleep. 2018;41(1):zsx189 .
18. . Sleep health composites are associated with the risk of heart disease across sex and race. Sci Rep. 2022;12(1):2023 .
19. . Validity of the Berlin questionnaire in detecting obstructive sleep apnea: a systematic review and meta-analysis. Sleep Med Rev. 2017;36:116–124 .
20. . Associations between migraine attacks and nightly sleep characteristics among adults with episodic migraine: a prospective cohort study. Sleep. 2020;43(7):zsaa001 .
21. . Prospective cohort study of routine exercise and headache outcomes among adults with episodic migraine. Headache. 2021;61(3):493–499 .
22. . The consensus sleep diary: standardizing prospective sleep self-monitoring. Sleep. 2012;35(2):287–302 .
23. . Development and validation of patient-reported outcome measures for sleep disturbance and sleep-related impairments. Sleep. 2010;33(6):781–792 .
24. . The Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research. Psychiatry Res. 1989;28(2):193–213 .
25. . Validation of the Insomnia Severity Index as an outcome measure for insomnia research. Sleep Med. 2001;2(4):297–307 .
26. . Irregular sleep/wake patterns are associated with poorer academic performance and delayed circadian and sleep/wake timing. Sci Rep. 2017;7(1):3216 .
27. American Thoracic Society ad hoc Committee on Healthy Sleep. An official American Thoracic Society statement: the importance of healthy sleep. Recommendations and future priorities. Am J Respir Crit Care Med. 2015;191(12):1450–1458 .
28. . Normal polysomnography parameters in healthy adults: a systematic review and meta-analysis. Lancet Respir Med. 2019;7(6):533–543 .
29. . Multidimensional sleep health is not cross-sectionally or longitudinally associated with adiposity in the Study of Women’s Health Across the Nation (SWAN). Sleep Health. 2020;6(6):790–796 .
30. . The relationship between childhood trauma and poor sleep health in adulthood. Psychosom Med. 2018;80(2):200–207 .
31. . A global measure of perceived stress. J Health Soc Behav. 1983;24(4):385–396 .
32. . Baseline sleep quality, stress, and depressive symptoms, and subsequent headache occurrence in a six-week prospective cohort study of patients with episodic migraine. Headache. 2021;61(5):727–733 .
33. . The CES-D scale: a self-report depression scale for research in the general population. Appl Psychol Meas. 1977;1(3):385–401 .
34. . Validation of a 3-factor scoring model for the Pittsburgh Sleep Quality Index in older adults. Sleep. 2006;29(1):112–116 .
35. . Subjective sleep quality and sleep architecture in patients with migraine: a meta-analysis. Neurology. 2021;97(16):e1620–e1631 .
36. . Associations between sleep quality and migraine frequency: a cross-sectional case-control study. Medicine (Baltimore). 2016;95(17):e3554 .
37. . Hypothalamic activation in spontaneous migraine attacks. Headache. 2007;47(10):1418–1426 .
38. . Hypothalamic control of sleep. Sleep Med. 2007;8(4):291–301 .
39. . Migraine and the hypothalamus. Cephalalgia. 2009;29(8):809–817 .
40. . Acute sleep deprivation enhances susceptibility to the migraine substrate cortical spreading depolarization. J Headache Pain. 2020;21(1):86 .
41. . Cortical spreading depression closes paravascular space and impairs glymphatic flow: implications for migraine headache. J Neurosci. 2017;37(11):2904–2915 .
42. . Multidimensional sleep health domains in older men and women: an actigraphy factor analysis. Sleep. 2021;44(2):zsaa181 .
43. . Actigraphy-derived sleep health profiles and mortality in older men and women. Sleep. 2022;45(4):zsac015 .
44. . The occurrence of recall bias in pediatric headache: a comparison of questionnaire and diary data. Headache. 2001;41(1):11–20 .
45. . A comparison of prospective and retrospective assessments of sleep. J Clin Epidemiol. 1996;49(4):455–460 .
46. . Comparing prospective headache diary and retrospective four-week headache questionnaire over 20 weeks: secondary data analysis from a randomized controlled trial. Cephalalgia. 2020;40(13):1523–1531 .
47. . Multidimensional sleep health and physical functioning in older adults. Gerontol Geriatr Med. 2021;7:1–9 .
48. . Sleep health mediates the relationship between physical activity and depression symptoms. Sleep Breath. 2022;26(3):1341–1349 .
49. . A case-control study on excessive daytime sleepiness in episodic migraine. Cephalalgia. 2007;27(10):1115–1119 .
50. . A case-control study on excessive daytime sleepiness in chronic migraine. Sleep Med. 2013;14(3):278–281 .
51. . Racial-ethnic differences in actigraphy, questionnaire, and polysomnography indicators of healthy sleep: the Multi-Ethnic Study of Atherosclerosis [manuscript published online September 9, 2021]. Am J Epidemiol.
