Sleep quality and chronotype in primary headache disorders: a nationwide population-based study of 5,311 Polish adults

Abstract

Background

Primary headache disorders, particularly migraine, are closely linked to sleep disturbances. However, the relationship between headache phenotypes, sleep quality, and chronotype in large populations remains to be fully elucidated. This study aimed to investigate sleep quality and chronotype across different primary headache types compared to headache-free individuals.

Methods

This nationwide, cross-sectional study included 5,311 Polish adults recruited via an online research panel. Headaches were classified using the HARDSHIP questionnaire according to ICHD-3 criteria. Sleep quality was assessed using the Pittsburgh Sleep Quality Index (PSQI) and chronotype using the Morningness-Eveningness Questionnaire (MEQ). Sleep-related variables were then compared between specific headache types and individuals without headache.

Results

Participants with migraine (n = 1,523) reported the poorest sleep quality (median PSQI: 7 vs. 5 in controls) and the highest prevalence of poor sleep (64.1% vs. 38.7% in controls; p < 0.001). After adjustment, migraine (β = 2.09), unclassified headache (β = 1.57), and tension-type headache (TTH; β = 0.59) remained significantly associated with higher PSQI scores. Regarding chronotype, the migraine group showed a significant shift toward eveningness (adjusted β=-1.38; p < 0.001) and a lower representation of morning types (24.9% vs. 39.1% in controls). In the migraine subgroup, poorer sleep quality was moderately correlated with lower quality of life (r=-0.45) and higher perceived stress (r = 0.34), whereas MEQ scores showed only weak correlations with clinical outcomes.

Conclusions

Poor sleep quality and an evening chronotype are prominent features of primary headache disorders, especially migraine. Future prospective studies are needed to determine the causality of these associations. Meanwhile, sleep quality assessment and interventions aiming to improve sleep patterns should be considered in migraine patients.

Introduction

Primary headache disorders are among the most prevalent neurological conditions worldwide and represent a significant source of disease-related burden, especially in people of productive age [1]. According to the recent Global Burden of Disease study [2], headache disorders affected 2.9 billion people worldwide in 2023. The global burden attributable to these disorders was estimated at 541.9 years lived with disability (YLDs) per 100,000 population, the majority of which was caused by migraine [1, 2].

Sleep is commonly considered a factor strongly associated with primary headaches, particularly migraine [3]. Sleep disturbances may act as a trigger for headache attacks, occur because of recurrent pain, or exist as a comorbid disorder with a partially shared pathophysiological background with migraine and other primary headaches [3–5]. Proposed mechanisms linking migraine and sleep include the involvement of the hypothalamus and the dysregulation of several neurotransmitter molecules, such as orexins, melatonin, pituitary adenylate cyclase-activating polypeptide (PACAP), serotonin, dopamine, and adenosine. Furthermore, they share neural networks responsible for pain modulation and the regulation of the sleep-wake cycle [3, 4].

However, the direction of this relationship remains difficult to determine unequivocally.

Previous population-based and clinical studies indicate that poorer sleep quality is more common in people with migraine than in those without headache and may be associated with higher headache frequency and headache-related burden [6–8]. Studies incorporating objective sleep assessment methods provide additional support for the relationship between migraine and sleep disturbance. Polysomnographic studies indicate altered sleep architecture in migraine, including reduced rapid eye movement (REM) sleep and changes in sleep efficiency [9]. Actigraphy studies have also suggested changes in sleep-wake patterns, sleep fragmentation and abnormal nocturnal activity related to migraine, although results are heterogeneous [10, 11].

Sleep dysfunction is also observed in other primary headache disorders, including tension-type headache (TTH). This suggests that the association is not specific to migraine but may differ in severity and clinical correlates across different headache phenotypes [7, 12].

Beyond sleep quality, which reflects global sleep dysfunction, the chronotype - defined as an individual’s preference for sleep timing, activity, and peak performance during the day - is hypothesized to be an important dimension of the headache-sleep relationship [13]. Migraine has previously been considered in a chronobiological context, as attacks can exhibit circadian and seasonal rhythmicity [14]. Nevertheless, studies on this topic remain inconclusive.

Despite growing evidence linking headache disorders with sleep and circadian alterations, direct comparative data across different primary headache phenotypes remain limited, as most previous studies have focused primarily on migraine or compared migraine patients with headache-free subjects.

The aim of this study was to investigate subjective sleep quality and chronotype in migraine and to compare these parameters with those in other headache types and in individuals without headache. The secondary objective was to evaluate the association between sleep parameters, headache burden, and quality of life measures in the migraine group.

Methods

Study design

We conducted a cross-sectional analysis based on data collected as part of a nationwide epidemiological study investigating the prevalence of primary headache disorders in the adult population of Poland. Data were collected using the Computer-Assisted Web Interview (CAWI) technique between May and June 2025. Participants completed a structured questionnaire that included items regarding sociodemographic data, headache characteristics, and selected health- and sleep-related variables.

Population

The study population comprised adults (age ≥ 18 years) recruited from the Polish general population through the online Ariadna National Research Panel, which includes > 150,000 verified members. Quotas were applied to approximate the distribution of the Polish adult population regarding age, sex, and place of residence. Eligible panel members had an equal probability of being invited. Email invitations used neutral wording and did not mention headache or migraine to minimize selection bias. To ensure data quality, the following exclusion criteria were applied: extremely short completion time (defined as > 3 SD below the mean after log-transformation), attention-check failure, inconsistent repeated responses, or invalid open-text entries. Of 30,850 invited panel members, 5,992 accessed the survey, yielding a participation rate of 19.4%. After excluding 681 respondents during quality control, the final analytical sample comprised 5,311 valid responses.

In the original epidemiological study, the estimated sample size was 5,000 participants. Cochran’s formula indicated that approximately 2,300 participants were required to estimate a chronic headache prevalence of 3–4% with 0.7% precision (at a 95% confidence level). The sample size was increased to ensure adequate numbers of cases for subgroup analyses and to account for anticipated exclusions during data cleaning.

Headache assessment

Headache characteristics were assessed using the Headache-Attributed Restriction, Disability, Social Handicap and Impaired Participation (HARDSHIP) questionnaire [15]. HARDSHIP is a standardized modular questionnaire developed for population-based studies conducted within the Global Campaign against Headache [16], designed for estimating primary headache prevalence and assessing headache-attributed burden. The questionnaire comprises 101 items and includes modules on sociodemographic characteristics (including age, sex, relationship status, education level, and job type), headache characteristics, and frequency (including the number of headache days per month), and the use of acute pain medication. In addition, it incorporates the Migraine Disability Assessment (MIDAS) scale to assess headache-attributed burden (higher scores indicate higher burden), the eight-item World Health Organization Quality of Life scale (WHOQoL-8) (lower scores indicate lower quality), and items assessing subjective well-being, such as self-rated life satisfaction and perceived stress. Classification of primary headache types was based on the prespecified HARDSHIP algorithm [15], following a hierarchical order (definite migraine, definite TTH, probable migraine, probable TTH, and unclassified headache) and criteria consistent with the International Classification of Headache Disorders, 3rd edition (ICHD-3) [17]. Each respondent was assigned to a single category based on the headache type reported as the most bothersome during the preceding 12 months.

Sleep assessment

Pittsburgh sleep quality index

The Pittsburgh Sleep Quality Index (PSQI) [18] is a self-report, 19-item questionnaire evaluating sleep quality and sleep disturbances over the preceding month. The PSQI comprises seven component scores: sleep duration, sleep disturbance, sleep latency, daytime dysfunction due to sleepiness, sleep efficiency, overall sleep quality, and the use of sleep medication. The component scores are summed to obtain a total score (range: 0–21), with higher scores indicating poorer sleep quality. A total score > 5 is considered the cut-off value for poor sleep quality.

Morningness-eveningness questionnaire

Chronotype was assessed using the Morningness-Eveningness Questionnaire (MEQ) [19]. It consists of 19 items, with a total score ranging from 16 to 86; lower scores indicate eveningness and higher scores indicate morningness. Based on the global score, respondents can be further classified into five chronotype categories: definite evening type (16–30), moderate evening type (31–41), intermediate type (42–58), moderate morning type (59–69), and definite morning type (70–86). Morningness and eveningness reflect individual differences in the timing of circadian rhythmic expression. Morningness can be characterized by a tendency to go to bed and wake up early, with peak mental and physical performance occurring in the early part of the day, while eveningness refers to a tendency to go to bed and wake up late, with optimal performance occurring later in the day and during the evening hours [20].

Ethics approval and consent

The study was approved by the Ethics Committee at Wrocław Medical University (No. KB/39/2025). The study was conducted according to the principles set out in the Declaration of Helsinki. All participants provided informed consent prior to enrollment.

Statistical analysis

For the purpose of the analysis, definite and probable migraine, as well as definite and probable TTH, were combined into single categories. Hence, participants were classified into four mutually exclusive headache groups: migraine, tension-type headache (TTH), unclassified headache, and no headache in the preceding 12 months. Descriptive statistics were used to summarize sociodemographic characteristics, including sex, age, place of residence (urban or rural), relationship status (married, single, or divorced/separated/widowed and whether the participant is living with a partner), education level (primary, secondary, or higher), and job type (predominantly physical or intellectual). Continuous variables are presented as means with standard deviations (SD) or medians with interquartile ranges (IQR), depending on their distribution. Categorical variables are presented as counts and percentages.

Between-group comparisons were performed across the four headache categories. Because all continuous variables deviated from normality, they were compared using the Kruskal-Wallis test. Distributional assumptions were assessed by visual inspection of histograms and the Shapiro-Wilk test. When the global test was significant, post-hoc pairwise comparisons of mean ranks were performed. Categorical variables were compared using the χ² test or Fisher’s exact test (reported as χ²/Fisher’s test (degrees of freedom)), as appropriate.

The main sleep-related outcomes were sleep quality, assessed using the Pittsburgh Sleep Quality Index (PSQI) total score, and chronotype, assessed using the Morningness-Eveningness Questionnaire (MEQ) total score. Associations between headache groups and PSQI or MEQ total scores were examined using linear regression models. Headache group was treated as a categorical predictor, with the non-headache group used as the reference category; thus, migraine, tension-type headache, and unknown headache were compared with the non-headache group within the same model, adjusted for the specified covariates. Multivariable models were adjusted for predefined sociodemographic covariates, including age, sex, place of residence, education level, relationship status, and job type. In the multivariable analysis, sex (female vs. male) and relationship status (living with a partner: yes vs. no) were modeled as a binary covariates. Linear regression assumptions were assessed using residual diagnostics, including evaluation of linearity, homoscedasticity, residual normality, and influential observations. For the PSQI model, residual diagnostics showed mild deviations from normality, while no major heteroscedasticity, no clear nonlinear association with age, and no influential observations were identified for both models. Regression coefficients (β) with 95% confidence intervals (95% CIs) were reported as adjusted mean differences in PSQI or MEQ scores compared with the no-headache group.

Subsequently, a secondary analysis was performed, focusing solely on participants with migraine. In this subgroup, the associations between sleep quality and chronotype and sociodemographic, headache-related, or patient-reported variables were examined. Correlations were assessed using Spearman’s rank correlation coefficient. These analyses were considered exploratory and were interpreted as unadjusted descriptive associations. For correlation analyses, Spearman’s rho (R) coefficients with corresponding p-values were reported. All analyses were performed using STATISTICA software (version 14.1; StatSoft Inc., Tulsa, OK, USA). The significance level was established at α = 0.05.

Results

Characteristics of the study population

A total of 5,311 participants were included in the study: 1,523 with migraine, 2,366 with TTH, 209 with unclassified headache, and 1,213 reporting no headache in the preceding 12 months. Participants with migraine were significantly younger than those with TTH, unclassified headache, or no headache (median age 40 vs. 45, 44, and 56 years, respectively, Kruskal-Wallis test, p < 0.001). Additionally, the highest proportion of women was observed in the migraine group (67%; Pearson’s χ²(3) = 268.6, p < 0.001). While the proportion of participants living in urban areas was similar across all headache types (Pearson’s χ²(3) = 2.0, p = 0.57), higher education (Pearson’s χ²(6) = 47.5, p < 0.001) and predominantly intellectual work (Pearson’s χ²(6) = 207.7, p < 0.001) were more frequently reported in the migraine group compared to the headache-free group.

Headache frequency was higher in migraine and unclassified headache groups than in the TTH group (Kruskal-Wallis test, p < 0.001). Specifically, the median number of monthly headache days was 3 [IQR 2–5] for the migraine group, 2 [IQR 1–3] for the TTH group, and 3 [IQR 1–5] for the unclassified headache group. Compared to participants without headache, those reporting headache disorders had lower WHOQoL-8 scores (Kruskal-Wallis test, p < 0.001) and higher levels of perceived stress and anxiety (Kruskal-Wallis test, p < 0.001). The detailed characteristics of the study groups are presented in Table 1.

Table 1.

Sociodemographic characteristics of the study population by headache group

Variable	Total (n = 5311)	Migraine (n = 1523)	Tension-type headache (n = 2366)	Unclassified headache (n = 209)	No headache in preceding 12 months (n = 1213)	p-value
Age, years, median (IQR)	45 (34–59)	40 (31–51)	45 (34–58)	44 (33–56)	56 (38–68)	p < 0.001
Sex, n (%)						p < 0.001
Female	2790 (52.5%)	1020 (67%)	1196 (50.5%)	133 (63.6%)	441 (36.4%)
Male	2488 (46.9%)	493 (32.4%)	1160 (49%)	75 (35.9%)	760 (62.6%)
Other/not specified	33 (0.6%)	10 (0.6%)	10 (0.5%)	1 (0.5%)	12 (1%)
Place of residence, n (%)						p = 0.57
Urban	3464 (65.2%)	997 (65.5%)	1543 (65.2%)	127 (60.8%)	797 (65.7%)
Rural	1847 (34.8%)	526 (34.5%)	823 (34.8%)	82 (39.2%)	416 (34.3%)
Education level, n (%)						p < 0.001
Primary	507 (9.5%)	102 (6.7%)	228 (9.6%)	22 (10.5%)	155 (12.8%)
Secondary	2245 (42.3%)	629 (41.3%)	974 (41.2%)	87 (41.6%)	555 (45.7%)
Higher	2559 (48.2%)	792 (52%)	1164 (49.2%)	100 (47.9%)	503 (41.5%)
Relationship status, n (%)						p < 0.001
Married	2869 (54%)	786 (51.6%)	1298 (54.9%)	105 (50.2%)	680 (56.1%)
Not married	1673 (31.5%)	533 (35%)	752 (31.8%)	69 (33%)	319 (26.3%)
Widowed/divorced/separated	769 (14.5%)	204 (13.4%)	316 (13.3%)	35 (16.8%)	214 (17.6%)
Living with partner, n (%)						p < 0.001
Yes	3474 (65.4%)	1026 (67.4%)	1588 (67.1%)	121 (57.9%)	739 (60.9%)
No	1837 (34.6%)	497 (32.6%)	778 (32.9%)	88 (42.1%)	474 (39.1%)
Job type, n (%)						p < 0.001
Predominantly physical work	1079 (20.3%)	296 (19.4%)	501 (21.2%)	37 (17.7%)	245 (20.2%)
Predominantly mental work	2504 (47.2%)	878 (57.7%)	1116 (47.2%)	108 (51.7%)	402 (33.1%)
Not employed/student/retired	1728 (32.5%)	349 (29.1%)	749 (31.6%)	64 (30.6%)	566 (46.7%)
MHD		3 (2–5)	2 (1–3)	3 (1–5)	NA	p < 0.001
MIDAS total score		8 (2–20)	1 (0–6)	4 (0–13)	NA	p < 0.001
WHOQoL-8 score	29 (25–32)	28 (24–31)	29 (25–32)	27 (24–31)	30 (26–32)	p < 0.001
Current life satisfaction (0–10)	7 (5–8)	7 (5–8)	7 (5–8)	7 (5–8)	7 (6–8)	p < 0.001
Perceived stress/anxiety (0–10)	4 (2–6)	5 (3–7)	4 (1–6)	5 (2–6)	3 (1–5)	p < 0.001

Data are presented as median (interquartile range) for continuous variables and n (%) for categorical variables. Continuous variables were compared using a Kruskal-Wallis test and post-hoc comparisons of mean ranks were performed when global scores were significant. Categorical data was compared using χ² test or Fisher’s exact test

Abbreviations: MHD - monthly headache days; MIDAS - Migraine Disability Assessment; WHOQoL-8 - World Health Organization Quality of Life scale; NA - not applicable

Sleep quality and chronotype

Significant differences in sleep quality and chronotype were found between the study groups, as measured by the PSQI and MEQ, respectively. Participants with migraine had the highest PSQI total score, with a median of 7 [IQR 5–9], compared with 5 [IQR 3–8] in the TTH group, 6 [IQR 4–9] in the unclassified headache group, and 5 [IQR 3–7] among participants without headache (Kruskal-Wallis test, p < 0.001). The proportion of respondents with poor sleep quality (PSQI > 5) was highest in the migraine group, affecting 64.1% of participants, compared with 45.4% of those with TTH, 57.9% of those with unclassified headache, and 38.7% of respondents without headache (Pearson’s χ²(3) = 209.5, p < 0.001).

These differences were evident across all PSQI component scores. Post-hoc comparisons showed that participants with migraine had significantly different distributions of scores than those with TTH and those without headache in all PSQI domains: sleep duration, sleep disturbance, sleep latency, daytime dysfunction due to sleepiness, sleep efficiency, overall sleep quality, and use of sleep medication (Migraine vs. TTH: p < 0.001 and Migraine vs. No headache: p < 0.001 for all component scores).

Analysis of chronotype patterns showed that the migraine group had a lower MEQ total score than the TTH (p < 0.001) and no-headache (p < 0.001) groups, with median values of 53 [IQR 48–58], 55 [IQR 49–60], and 56 [IQR 51–62], respectively; this suggests a shift toward greater eveningness among participants with migraine. Both the TTH and unclassified headache groups also differed significantly from the no-headache group (p < 0.001 and p = 0.023, respectively). Furthermore, discrepancies were revealed in the distribution of MEQ chronotype categories across headache groups (Kruskal-Wallis test, p < 0.001). The intermediate chronotype was the most common category in all groups, while moderate and definite morning types were less frequent in the migraine group (22.2% and 2.7%, respectively) than in participants with TTH (29.0% and 3.0%) and those without headache (34.9% and 4.2%). PSQI and MEQ results across the study groups are presented in Table 2.

Table 2.

PSQI and MEQ results by headache group

Variable	Total (n = 5311)	Migraine (n = 1523)	Tension-type headache (n = 2366)	Unclassified headache (n = 209)	No headache in preceding 12 months (n = 1213)	p-value	Pairwise comparisons*
PSQI
1. sleep duration	0 (0–1)	0 (0–1)	0 (0–1)	0 (0–1)	0 (0–1)	< 0.001	M vs. TTH: p < 0.001 M vs. NH: p < 0.001
2. sleep disturbance	1 (1–2)	1 (1–2)	1 (1–1)	1 (1–2)	1 (1–1)	< 0.001	M vs. TTH: p < 0.001 M vs. NH: p < 0.001 TTH vs. NH: p = 0.008 UH vs. NH: p = 0.004
3. sleep latency	1 (1–2)	1 (1–2)	1 (0–2)	1 (1–2)	1 (0–2)	< 0.001	M vs. TTH: p < 0.001 M vs. NH: p < 0.001 UH vs. TTH: p = 0.012 UH vs. NH: p < 0.001
4. day dysfunction due to sleepiness	1 (0–2)	1 (1–2)	1 (0–1)	1 (1–2)	1 (0–1)	< 0.001	M vs. TTH: p < 0.001 M vs. NH: p < 0.001 M vs. UH: p = 0.027 UH vs. TTH: p = 0.035 UH vs. NH: p < 0.001 TTH vs. NH: p < 0.001
5. sleep efficiency	0 (0–1)	0 (0–1)	0 (0–1)	0 (0–1)	0 (0–1)	< 0.001	M vs. TTH: p = 0.033 M vs. NH: p = 0.001
6. overall sleep quality	1 (1–1)	1 (1–2)	1 (1–1)	1 (1–2)	1 (1–1)	< 0.001	M vs. TTH: p < 0.001 M vs. NH: p < 0.001 UH vs. TTH: p = 0.007 UH vs. NH: p < 0.001 TTH vs. NH: p < 0.001
7. use of sleep medication	0 (0–0)	0 (0–1)	0 (0–0)	0 (0–1)	0 (0–0)	< 0.001	M vs. TTH: p < 0.001 M vs. NH: p < 0.001 UH vs. NH: p < 0.02
PSQI total score	5 (4–8)	7 (5–9)	5 (3–8)	6 (4–9)	5 (3–7)	< 0.001	M vs. TTH: p < 0.001 M vs. NH: p < 0.001 UH vs. TTH: p < 0.001 UH vs. NH: p < 0.001 TTH vs. NH: p < 0.001
Poor sleep quality (PSQI > 5), %	2641 (49.7%)	977 (64.1%)	1074 (45.4%)	121 (57.9%)	469 (38.7%)	< 0.001
MEQ
MEQ total score	55 (49–60)	53 (48–58)	55 (49–60)	54 (48–61)	56 (51–62)	< 0.001	M vs. TTH: p < 0.001 M vs. NH: p < 0.001 UH vs. NH: P = 0.023 TTH vs. NH: p < 0.001
Chronotype category						< 0.001
Definitely evening type	39 (0.7%)	13 (0.9%)	19 (0.8%)	1 (0.5%)	6 (0.5%)
Moderately evening type	334 (6.3%)	114 (7.5%)	156 (6.6%)	11 (5.3%)	53 (4.3%)
Intermediate type	3250 (61.2%)	1016 (66.7%)	1424 (60.2%)	130 (62.2%)	680 (56.1%)
Moderately morning type	1517 (28.6%)	339 (22.2%)	696 (29.4%)	59 (28.2%)	423 (34.9%)
Definitely morning type	171 (3.2%)	41 (2.7%)	71 (3%)	8 (3.8%)	51 (4.2%)

Data are presented as median (interquartile range) for continuous variables and n (%) for categorical variables. Continuous variables were compared using a Kruskal-Wallis test and post-hoc comparisons of mean ranks were performed when global scores were significant. Categorical data was compared using χ² test or Fisher’s exact test

Abbreviations: MEQ - Morningness-Eveningness Questionnaire; PSQI- Pittsburgh Sleep Quality Index, M-migraine, TTH – tension-type headache, UH-unclassified headache, NH – no headache in preceding 12 months

* Only pairwise comparisons with statistically significant p-values are reported

In multivariable linear regression models adjusted for age, sex (male vs. female), place of residence, education level, relationship status, and job type, all headache groups had higher PSQI total scores than participants without headache in the preceding 12 months (Table 3). The largest adjusted mean difference was observed for the migraine group (β = 2.09; 95% CI 1.83 to 2.36; p < 0.001), followed by unclassified headache (β = 1.57; 95% CI 1.08 to 2.06; p < 0.001) and tension-type headache (β = 0.59; 95% CI 0.36 to 0.83; p < 0.001).

Table 3.

Adjusted mean differences in PSQI and MEQ total scores by headache type

Headache group	PSQI (total score), adjusted β (95% CI)	p-value	MEQ (total score), adjusted β (95% CI)	p-value
No headache in the preceding 12 months	Reference	—	Reference	—
Migraine	2.09 (1.83 to 2.36)	< 0.001	-1.38 (-2.05 to -0.71)	< 0.001
TTH	0.59 (0.36 to 0.83)	< 0.001	-0.90 (-1.49 to -0.31)	0.003
Unclassified headache	1.57 (1.08 to 2.06)	< 0.001	-0.60 (-1.83 to 0.63)	0.339

Values are presented as adjusted mean differences; the no-headache group was used as the reference category. Models were adjusted for age, sex, place of residence, education level, relationship status and job type. Abbreviations: CI - confidence interval; MEQ - Morningness-Eveningness Questionnaire; PSQI- Pittsburgh Sleep Quality Index; TTH- tension-type headache

Regarding the MEQ total score, participants with migraine had lower adjusted mean scores than those without headache (β = -1.38; 95% CI -2.05 to -0.71; p < 0.001), indicating greater eveningness. A similar but smaller difference was observed for tension-type headache (β = -0.90; 95% CI -1.49 to -0.31; p = 0.003), whereas the difference for unclassified headache was not statistically significant (β = -0.60; 95% CI -1.83 to 0.63; p = 0.339).

Exploratory analysis in participants with migraine

An additional analysis of sleep quality and chronotype measures, and their associations with headache burden and patient-reported outcomes, was performed in the migraine group. In this subpopulation, higher mean scores, reflecting more substantial disability, were found in the sleep disturbance, sleep latency, daytime dysfunction due to sleepiness, and overall sleep quality domains of the PSQI. These scores across the respective PSQI domains are displayed in Fig. 1.

Fig. 1.

Median scores across Pittsburgh Sleep Quality Index (PSQI) domains among participants with migraine. Points indicate median scores, and whiskers indicate interquartile ranges. Each PSQI domain is scored from 0 to 3, with higher scores indicating greater impairment in the respective domain

Among participants with migraine, poorer sleep quality (as assessed by the PSQI) was weakly associated with monthly headache days (r = 0.26, p < 0.001), higher MIDAS total score (r = 0.37, p < 0.001), lower self-rated life satisfaction (r = -0.35, p < 0.001), higher perceived stress and anxiety (r = 0.34, p < 0.001), and moderately correlated with lower WHOQoL-8 score (r = -0.45, p < 0.001). The MEQ total score showed very weak correlations with monthly headache days (r = -0.08, p = 0.002), MIDAS (r = -0.16, p < 0.001) and WHOQoL-8 scores (r = 0.17, p < 0.001), and life satisfaction (r = 0.17, p < 0.001), while no significant association was observed with stress/anxiety (r = -0.01, p = 0.635).

These findings are summarized in Table 4.

Table 4.

Correlations between PSQI and MEQ total scores and clinical or patient-reported variables among participants with migraine

Variable	PSQI (total score), R*	p-value	MEQ (total score), R*	p-value
Monthly headache days	0.26	< 0.001	-0.08	0.002
MIDAS total score	0.37	< 0.001	-0.16	< 0.001
WHOQoL-8 total score	-0.45	< 0.001	0.17	< 0.001
Current life satisfaction (0–10)	-0.35	< 0.001	0.17	< 0.001
Perceived stress/anxiety (0–10)	0.34	< 0.001	-0.01	0.635

Abbreviations: MEQ - Morningness-Eveningness Questionnaire; PSQI - Pittsburgh Sleep Quality Index; MIDAS - Migraine Disability Assessment; WHOQoL-8 - World Health Organization Quality of Life scale

*Spearman’s rank correlation coefficient

Discussion

Our findings confirm that impaired sleep quality is a hallmark of primary headache disorders, being most pronounced in migraine. While sleep dysfunction was observed across all headache phenotypes, the migraine group exhibited the highest global PSQI scores and the greatest proportion of “poor sleepers,” a relationship that remained robust even after adjusting for sociodemographic factors. This suggests that sleep impairment in migraine is an intrinsic clinical feature rather than a byproduct of background characteristics. While this analysis assesses subjective sleep quality and chronotype in migraine and tension-type headache, more rare primary headache disorders, including cluster headache and other trigeminal autonomic cephalalgias are also linked to sleep related measures. Cluster headache shows significant circadian and circannual rhythmicity, with recent evidence indicating that sleep may be negatively affected during cluster bouts and remission periods [21, 22]. Objective sleep studies in trigeminal autonomic cephalalgias remain limited, but available literature suggests that sleep alterations in this group of headache disorders are clinically relevant and should be further investigated [23].

In our study, the migraine group was particularly affected in domains of sleep latency, sleep disturbance, and daytime dysfunction. Prolonged sleep latency may be driven by sensory hypersensitivity, cognitive arousal, or anticipatory anxiety regarding future attacks [3, 24]. In turn, sleep disturbances likely reflect nocturnal awakenings caused by pain or associated symptoms like nausea. The resulting daytime dysfunction represents the cumulative interictal burden, where fatigue and reduced cognitive efficiency further diminish the patient’s quality of life [25]. Our observation that poorer sleep correlates with higher attack frequency and lower life satisfaction supports the view that sleep quality is a critical indicator of overall migraine burden.

These results align with existing literature. Population studies, such as those by Song et al. [26], have similarly identified worse PSQI component scores in migraine patients compared to both non-migraine headache and headache-free groups. Furthermore, the link between sleep quality and headache-related impact has been consistently reported across different populations [26, 27]. Interestingly, our data also highlights that sleep impairment - while most severe in migraine - is a significant issue in TTH, corroborating previous findings that sleep dysfunction is a common, albeit varying, feature across the headache spectrum [7, 28, 29].

A meta-analysis of 32 studies [9] confirmed that individuals with migraine exhibit significantly poorer subjective sleep quality and altered sleep architecture compared to healthy controls. While our study utilized the PSQI to capture subjective sleep quality, polysomnographic (PSG) data further support these findings, albeit with some inconsistencies. PSG studies have reported alterations in sleep macrostructure, such as lighter sleep, reduced slow-wave sleep, decreased sleep efficiency, and frequent awakenings; while microstructural analyses suggest NREM sleep instability and altered arousal profiles. Interestingly, sleep architecture may fluctuate across the migraine cycle; for instance, sleep onset latency has been found to be shorter in the preictal period compared to interictal phases [30]. Furthermore, specific REM sleep deficits have been linked to migraine chronification, with REM sleep duration below 23.1% associated with increased odds of chronic migraine [31].

The correlations we observed between PSQI scores and quality of life, anxiety, and stress are well-supported by clinical literature. Duan et al. [32] similarly found that poor subjective sleep quality, assessed using the PSQI scale, was independently associated with migraine-related burden, including pain severity, headache impact, and migraine-specific quality of life. These parallel findings reinforce the role of sleep quality as a clinically relevant marker of migraine severity.

Although the directionality of this association remains to be fully elucidated, poor sleep likely increases migraine susceptibility by impairing pain modulation. Experimental evidence suggests that sleep fragmentation and deprivation lower the threshold for nociceptive processing [33–35] and may facilitate central sensitization by impairing descending pain inhibition [36]. Neuroimaging has further demonstrated that sleep loss amplifies activity in somatosensory pain-processing regions while simultaneously reducing the response in brain networks responsible for endogenous analgesia [37].

Our results indicate a significant shift toward eveningness in the migraine group, evidenced by lower MEQ total scores compared to both headache-free individuals and those with TTH. When categorized, the migraine group exhibited a higher prevalence of evening chronotypes and a corresponding decrease in morning types. Although chronotype distribution differed significantly between participants with migraine, TTH and headache-free controls, the absolute differences between groups were not large. Therefore, the clinical relevance of this difference should be interpreted cautiously.

These findings add to a complex and sometimes inconsistent body of evidence regarding chronotypes in headache disorders. In a large-scale study by van Oosterhout et al. [38], migraine patients were more likely to fall into either extreme (morning or evening) rather than the intermediate chronotype. They also exhibited greater circadian rhythm rigidity and difficulty adapting to sleep-wake shifts. Similarly, Jung et al. [39] found a higher frequency of evening chronotypes in both migraine and TTH groups compared to controls.

Despite these distributional shifts, the clinical impact of chronotype remains debatable. Consistent with the findings of Jung et al. [39], we observed only weak correlations between MEQ scores and monthly headache days, quality of life, or life satisfaction. This suggests that while migraine may be associated with a later chronotype, the chronotype itself may not be a primary driver of headache frequency or burden. However, other clinical data offer a more nuanced view: Viticchi et al. [40] reported that “early-risers” experienced lower migraine frequency, while Im et al. [41] found that eveningness was associated with a higher attack frequency and later attack timing. Such discrepancies highlight the need for further research into the genetic and biological links between circadian regulation and migraine susceptibility.

The link between chronotype and migraine may have a profound genetic and biological basis. GWAS studies have identified chronotype-associated loci enriched for core circadian clock genes (e.g., PER, CRY1, ARNTL) and pathways involving glutamate and insulin signaling, predominantly expressed in the hypothalamus and pituitary gland [42]. The clinical relevance of these findings is underscored by a meta-analysis showing that over 50% of migraine patients exhibit a circadian pattern of attacks, and 65.5% of migraine susceptibility genes show circadian expression [21]. Specific mutations, such as those in the CSNK1D gene, have been linked simultaneously to migraine and advanced sleep phase syndrome, with experimental models suggesting these mutations lower the threshold for cortical spreading depression [43].

Melatonin signaling represents another crucial link. Reduced nocturnal melatonin excretion has been observed in women with menstrual migraine [44], and lower levels of its main metabolite have been reported during acute attacks in both episodic and chronic migraine [45]. This suggests that melatonin secretion is particularly impaired during the ictal phase. Clinical support for this hypothesis comes from a recent meta-analysis where melatonin was found superior to placebo in reducing headache frequency [46], although data for TTH remain inconclusive [47].

Modern models suggest that chronotype is not merely a preference for sleep timing but a complex phenotype reflecting the interaction between biological rhythms and environmental factors [48]. From this perspective, the relationship between chronotype and migraine is likely bidirectional: migraine may be modified by a specific chronotype, but the disease itself, and the resulting lifestyle adjustments, may also influence the behaviors used to assess chronotype. This interaction is further complicated by “social jetlag,” which occurs when social demands, such as working hours, are out of phase with an individual’s biological time [49, 50].

Migraine, particularly when characterized by high attack frequency and a substantial interictal burden, can severely limit professional, familial, and social functioning [51–54]. Such impairments often necessitate alterations in physical activity, meal times, and sleep-wake schedules. Consequently, the observed shift toward eveningness in migraine patients may not exclusively represent a primary trait of the circadian system, but rather a secondary behavioral adaptation to the disease, characterized by sleep disturbances, activity avoidance, or social isolation.

This study has notable strengths. First, it was based on a large, nationwide sample of over 5,000 adults, which provided the statistical power necessary to estimate sleep-related differences with high precision while adjusting for key sociodemographic covariates. Second, headache phenotypes were classified using the standardized HARDSHIP protocol, which is a validated instrument derived from ICHD-3 criteria for population-based research. Furthermore, by comparing mutually exclusive groups (migraine, TTH, unclassified headache, and headache-free controls), this study provides a more comprehensive overview of sleep across the headache spectrum, moving beyond the traditional migraine-versus-control paradigm.

Several limitations of this study should be acknowledged. First, the cross-sectional design precludes drawing firm conclusions regarding the causality or directionality of the observed associations between headache and sleep-related measures. Second, data were collected using self-reported questionnaires, which may be subject to recall or reporting bias. Although headache classification followed the standardized HARDSHIP diagnostic module, phenotypes were not confirmed by clinical examination, which introduces a potential risk of misclassification.

A common limitation of CAWI-based recruitment is the potential for selection bias, specifically the limited participation of individuals without internet access or with lower digital literacy. We addressed this by employing quota sampling to approximate the distribution of the Polish adult population in terms of age, sex, and place of residence. Furthermore, the lack of objective sleep measures, such as polysomnography, meant we were unable to evaluate sleep architecture, objectively measured sleep efficiency, fragmentation, or sleep-disordered breathing. We also did not collect granular data on coexisting primary sleep disorders or the use of sedative-hypnotic medications and other drugs that potentially influence sleep patterns. Another limitation is the higher average age of participants in the non-headache group, which may have affected sleep quality and chronotype comparisons, as sleep patterns and sleep quality change with age, which could therefore act as a potential confounder. Finally, while we assessed perceived stress, validated measures of clinical depression and anxiety were not included, despite their known influence on both headache burden and sleep dysfunction. Consequently, our secondary and exploratory analyses should be interpreted as hypothesis-generating rather than indicative of causality.

Additionally, the study did not account for seasonal variations, as data collection occurred between May and June, which might have influenced sleep-wake patterns and MEQ responses due to extended daylight hours. Also, behavioral and occupational factors known to influence sleep were not assessed. These include alcohol intake, caffeine intake, use of blue-light-emitting devices close to bedtime, and shift work. Furthermore, we did not collect information on Body Mass Index (BMI) or hormonal factors, such as the menstrual cycle phase or use of hormonal contraceptives, both of which are known to modulate sleep quality and headache susceptibility.

Conclusions

Poor sleep quality is a hallmark of primary headache disorders, being most pronounced in patients with migraine. This dysfunction is primarily characterized by prolonged sleep latency, frequent sleep disturbances, and significant daytime dysfunction. Furthermore, migraine is associated with a distinct shift toward eveningness and a lower prevalence of morning chronotypes compared to both tension-type headache and headache-free individuals. These findings underscore the profound link between headache disorders, sleep quality, and circadian preferences, although sleep quality and chronotype patterns in the present study were assessed using subjective measures only. Given the complex interplay between these factors, future prospective studies utilizing objective markers are essential to elucidate causality and inform targeted therapeutic interventions.

Abbreviations

CAWI

Computer-Assisted Web Interview

HARDSHIP

Headache-Attributed Restriction, Disability, Social Handicap and Impaired Participation

ICHD-3

International Classification of Headache Disorders, 3rd edition

IQR

Interquartile ranges

MEQ

Morningness-Eveningness Questionnaire

MHD

Monthly headache days

MIDAS

Migraine Disability Assessment scale

NREM

Non-rapid eye movement

PACAP

Pituitary adenylate cyclase-activating polypeptide

PSG

Polysomnography

PSQI

The Pittsburgh Sleep Quality Index

REM

Rapid eye movement

SD

Standard deviations

TTH

Tension-type headache

UH

Unclassified headache

WHOQoL-8

World Health Organization Quality of Life scale

YLDs

Years lived with disability

Author contributions

Karol Marschollek - conceptualization; methodology; formal analysis; statistical analysis; writing - original draftHelena Martynowicz - writing - review and editing. Sławomir Budrewicz - writing - review and editing. Marcin Straburzyński - writing - review and editing. Marta Waliszewska-Prosół - conceptualization; project administration; methodology; data curation, writing - review and editing.

Funding

The study was funded from ‘the Initiative of Excellence – Research University’ provided by Wroclaw Medical University, Wroclaw, Poland (task number: IDUB.C230.24.001).

Data availability

The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study was approved by the Ethics Committee at Wrocław Medical University (No. KB/39/2025). The study was conducted according to the principles set out in the Declaration of Helsinki. All data was kept confidential and in accordance with data protection regulations, while the anonymity of respondents was maintained.

Competing interests

Karol Marschollek has received honoraria from AbbVie for speaker activities.Sławomir Budrewicz and Helena Martynowicz declare no competing interests related to the present study.Marcin Straburzyński - reports personal fees from AbbVie, Pfizer and Teva for speaker activities. Marta Waliszewska-Prosół is a member of the Editorial Board for The Journal of Headache and Pain and reports personal fees from AbbVie, Pfizer, and Teva for speaker activities. The authors declare no other relevant affiliations or financial interests with any organization or entity that has a financial interest in or conflict with the subject matter or materials discussed in this manuscript, apart from those disclosed.