The acute effect of night work‐related circadian misalignment on headache episodes: Results from the 1001 nights‐cohort

Abstract

Objectives

We aimed to investigate the prevalence of headaches on days with night shifts compared with days with day shifts within the same individuals, accounting for work‐related psychosocial stressors, physical job demands, and sleep duration and quality. This approach allowed us to isolate the impact of circadian misalignment due to night work from other potential headache triggers.

Background

Night work has been suggested to increase the risk of headaches, primarily due to circadian misalignment and disturbed sleep. Most previous studies compare night workers with day workers, but differences in job characteristics and tasks between these groups may introduce bias. To minimize this potential bias, we examined headache occurrence under different working conditions (night vs. day shifts) within the same individuals.

Methods

We used data from 14 days of repeated measurements in the 1001 nights‐cohort, which includes female employees from the Danish hospital sector. Data were collected from September 2022 to April 2024. Participants completed diaries for 14 consecutive days, providing daily information on working hours, sleep, work‐related psychosocial stressors, physical job demands, and headache occurrence (yes/no). Participants with data from at least one day shift and at least one night shift were eligible for inclusion in the analyses. In total, 522 participants contributed 3348 measurement days (1926 day shifts and 1422 night shifts). We estimated prevalence ratios (PRs) for headache occurrence while accounting for repeated measures within individuals and with adjustment for possible confounders (adjusted prevalence ratio [aPR]).

Results

Headache was reported on 21.5% of measurement days with day shifts and on 27.9% of measurement days with night shifts. Working a night shift was associated with a significantly higher headache prevalence (aPR, 1.31; 95% confidence interval, 1.13–1.52) compared with day shifts when adjusting for work‐related psychosocial stressors, physical job demands, and sleep duration and quality. For consecutive night shifts, the similarly adjusted headache prevalence was highest on the measurement day with the second night shift (aPR, 1.20; 95% confidence interval, 1.02–1.42), using the first night shift as a reference.

Conclusion

This study is the first to investigate the headache prevalence when working night shifts compared with day shifts while accounting for work‐related psychosocial stressors and physical job demands. Neither these factors, nor shorter sleep duration or lower sleep quality, explained the increased headache prevalence observed when participants worked night shifts. Thus, other (cascading) effects and underlying mechanisms of night work‐related circadian misalignment may be the primary drivers of headache in night shift workers.

Plain Language Summary

Night work has been linked to headaches, but it is unclear whether this is due to night shifts themselves, other job‐related factors, or sleep. We followed over 500 female hospital employees working day and night shifts, and we found that headaches were more common on night shifts than day shifts, even after accounting for work‐related stressors, physical demands, and sleep. These findings suggest that night work may independently increase the risk of headaches, highlighting a need to consider shift schedule adjustments, especially for workers prone to headaches.

Abbreviations

aPR

adjusted prevalence ratio

CI

confidence interval

PR

prevalence ratio

TST

total sleep time

INTRODUCTION

Headache disorders are among the most prevalent neurological disorders and constitute a significant public health concern. ¹ , ² , ³ Tension‐type headache and migraine are the most prevalent primary headache disorders ¹ , ² with severe consequences, as migraine is the leading cause of disability among individuals under 50 years, ⁴ and frequent headache is associated with lower work ability and reduced work productivity. ⁵ , ⁶ , ⁷ , ⁸

Physical activity increases migraine pain, ⁹ and among men with physically demanding jobs, the prevalence of migraine is higher than in those with less physically demanding jobs. ¹⁰ Other occupational and environmental factors, such as night work, psychosocial stressors, and indoor environmental parameters (e.g., noise, temperature, humidity, and carbon dioxide) have also been related to migraine or headache occurrence. ¹¹ , ¹² , ¹³ , ¹⁴ Night work implies circadian misalignment, ¹⁵ which encompasses shifts in the timing of sleep and activity and changes in eating patterns. Night work‐induced misalignment of the sleep–wake cycle with the solar day frequently results in poor sleep and substantial sleep debt, ¹⁶ , ¹⁷ which are well‐established risk factors for both headache and migraine. ¹⁸ , ¹⁹ , ²⁰ Population‐based surveys among occupationally active citizens in the United States, Canada, and Europe report an overall night work prevalence of 7.4%, 12%, and 14%, respectively. ²¹ , ²² , ²³ However, in certain sectors such as healthcare, transportation, industry, and law enforcement (including police and prison services), an even higher proportion of the employees work night shifts, with possible implications for health and well‐being.

Nevertheless, findings from previous longitudinal studies on night work and headache disorders are inconsistent. For example, one study found higher odds of headache disorders among shift workers compared with day workers, ²⁴ another study yielded mixed results, ²⁵ and yet another study found no association between headache and night work. ²⁶ Similarly, a cross‐sectional study also observed no significant association. ²⁷ These null findings may reflect health‐related selection into or out of night work, a phenomenon known as the healthy shift worker effect, which is a common methodological limitation in studies comparing night and day workers. ²⁸

As an alternative methodological approach to group comparisons in longitudinal studies, daily self‐reported measurements of working hours and headache can be used to compare the occurrence of headache episodes within the same individual when working night shifts and day shifts, respectively. Using such data, a Norwegian study found higher odds of headache when working night shifts compared with day shifts. ²⁹ Furthermore, data from the same study also indicated that the odds of headache were higher on the third than the second night shift. ³⁰ However, these studies did not account for potential differences in work‐related psychosocial stressors and physical job demands, even though these occupational factors likely vary between shift types due to differences in job tasks ³¹ , ³² and may either trigger headache episodes or heighten pain perception.

Against this background, the overall aim of this study was to investigate the acute effect of night work on headache episodes using self‐reported diary data with repeated measurements. As a secondary aim, we analyzed whether an association between night shifts and headache could be attributed to work‐related psychosocial stressors and physical job demands as well as sleep duration and quality. Third, we investigated if more consecutive night shifts were associated with more frequent headache. Based on prior research, we hypothesized that night work was associated with a higher prevalence of headache, and that this association could be partly explained by psychosocial stressors, physical job demands, and poorer sleep. We did not have any predefined hypothesis regarding the association between consecutive night shifts and headache due to limited prior research.

MATERIALS AND METHODS

Design, study population, and inclusion criteria

We used data from 14 days of repeated measurements in the 1001 nights‐cohort for secondary analyses of night work and headache. Our data include female employees from a nationwide Danish cohort, consisting mainly of health professional (e.g., nurses, midwifes, and assistant nurses) working in hospitals (see Nabe‐Nielsen et al. ³³ for details). In the hospital sector, women are overrepresented among occupational groups with night work, and therefore the cohort was restricted to women. The cohort was established to investigate health effects of night work‐related circadian disruption, and it was approved by the Danish National Committee on Health Research Ethics (H‐21077744). Participants received verbal and written information about the project, and all participants signed a declaration of consent before formal inclusion and the data collection commenced.

To be eligible for inclusion in the 1001 nights‐cohort, the employee had to be employed for at least 28 h/week and not be pregnant or attempting to conceive during the data collection period. A total of 1075 participants were enrolled, and the data collection took place from September 2022 to April 2024. For the present analyses, we used data from the background questionnaire and diaries, which the participants completed once a day for 14 consecutive days after enrollment. Participants were instructed to complete the diaries immediately after waking up from their primary sleep (Figure 1).

FIGURE 1.

Example of timing of collection of diary data after a day shift and a night shift. Participants were instructed to complete the diary after their primary sleep.

The sample size for the present article was based on available data, and no statistical power calculation was conducted before this study. Of the 1075 participants enrolled in the cohort, we included those who provided data from at least one day shift and one night shift during the 14‐day follow‐up period. Specifically, 93 participants had no measurement days with day shifts, 425 had no measurement days with night shifts, and 35 had neither a day shift nor a night shift recorded. Thus, the final analytical sample comprised 522 participants who contributed one or more day shifts and one or more night shifts. For these participants, a total of 3348 diaries were completed, covering 1926 day shifts and 1422 night shifts. Diaries from evening shifts and nonworking days were excluded.

Working hours

In the diaries, participants indicated their working hours during the preceding 24 h (first diary) or since the last diary entry (diaries 2–14). We categorized working hours into night shifts (at least 3 h between 23:00 and 06:00), evening shifts (at least 3 h between 18:00 and 02:00), and day shifts (at least 3 h between 6:00 and 21:00). The coding of shifts was based on prior research and considerations about the timing of the biological night. ³⁴ , ³⁵ Shifts that fulfilled more than one criterion were coded according to a predefined hierarchy: “night shift” had the highest priority, followed by “evening shift” and then “day shift.” Because 24‐hour shifts inherently involve night work, they were classified a priori as night shifts, even though some employees may have been on call rather than actively working during the night. Additionally, we coded the number of consecutive night shifts for each participant starting from the first night shift after a day without a night shift (i.e., following a day shift, evening shift, or nonworking day). Each participant could experience multiple consecutive night shift episodes during the 14‐day follow‐up period. For the analyses of consecutive night shifts, we excluded night shift number 6–14 due to less than 10 observations for each night shift number, as these groups were too small to be analyzed separately. The number of consecutive night shifts (1‐5) was treated as a categorical variable in the statistical analyses. Double shifts were identified based on participants' reports of two separate shifts in the same sleep diary.

Headache

We used headache data from the diaries as the outcome variable. Participants were asked “During the last 24 h/since the last diary entry, have you experienced any of the symptoms listed below? (You may choose more than one option).” The response categories included: “No symptoms; Headache; Pain in the neck; Pain in the shoulders/arms; Back pain; Pain in the hips/legs/knees/feet; Other symptoms.” If participants reported headache or reported “headache” or “migraine” as free text under “Other symptoms,” this information was included in the analyses. The latter yielded five extra observations of headache. Headache was coded as a binary variable (headache vs. no headache). No further information about timing, intensity, frequency, or accompanying symptoms was available.

Psychosocial work‐related stressors and physical job demands

Psychosocial work‐related stressors were measured by assessing emotional and quantitative job demands, as reported in the diaries. We asked “How emotionally demanding was/were your shift/your shifts?” and “How busy was/were your shift/your shifts?”. Responses were given on a scale from 1 to 10 (1 = not demanding/busy; 10 = extremely demanding/busy). These questions were inspired by The Danish Psychosocial Work Environment Questionnaire. ³⁶ Physical job demands were assessed using the question “How physically demanding was/were your shift/your shifts?” on a scale from 1 to 10 (1 = not demanding; 10 = extremely demanding). This question was developed with inspiration from a survey conducted by the Danish Working Environment Authority. ³⁷

Sleep duration and sleep quality

Each diary included the following sleep‐related questions: “When did you go to bed?,” “How long did it take you to fall asleep? (indicate minutes),” “When did you wake up?,” and “During the last 24 h, have you taken a nap/powernap one or more times?” For primary sleep, participants reported the start time, sleep latency (minutes from going to bed until sleep onset), and end time. For naps, participants reported start and end time (up to four naps per diary). Total sleep time (TST) was calculated as the sum of primary sleep duration (excluding sleep latency) and the duration of naps. Sleep quality was rated by asking “How would you describe your sleep?” from 1 to 5 (1 = very good; 5 = very poor). Sleep variables were treated as continuous in the statistical analyses.

Covariates

Age was obtained from personal identification numbers including birth year. Additional background information was collected via a questionnaire including the following questions: “At what time of day do you usually work in your main occupation?,” “How many hours per week do you work in your main occupation (including additional hours, if any)?,” “What is the highest level of education you have achieved?,” “Overall, how would you rate your health?,” and “All things considered, how strenuous do you find it having night shifts? (Choose 1 for not at all strenuous and 7 for very strenuous).” Response options are shown in Table 1.

TABLE 1.

Description of the study population (n = 522).

	N	%	Mean	SD
Age (years)	522		39.6	11.4
4‐week headache prevalence
Not at all	102	20.9
A little/somewhat troubled	313	64.0
Considerably/very much troubled	74	15.1
Self‐reported migraine diagnosis (yes)	69	14.2
Usual work schedule
Permanent daytime work (mainly between 6:00 and 18:00)	17	3.3
Permanent night work (mainly between midnight and 5:00)	15	2.9
2‐shift work without night work	6	1.2
2‐ or 3‐shift work with night work	484	92.7
Weekly working hours
28–36	154	29.5
37	260	49.8
>37	108	20.6
Highest level of education
Danish upper secondary school	5	1.0
Danish vocational education	26	5.3
Short higher education (<3 years) (e.g., assistant nurse)	31	6.3
Medium length higher education (3–4 years) (e.g., nurse)	370	75.5
Long higher education (>4 years) (e.g., medical doctor)	58	11.8
General health
Excellent/very good	41	8.4
Good	406	83.0
Less good/poor	42	8.6
Finds night work strenuous on a scale from 1 (not at all) to 7 (very much)
Not strenuous (1, 2)	65	13.8
Medium strenuous (3–5)	306	65.0
Very strenuous (6, 7)	100	21.2

Note: Participants with missing values were excluded analysis by analysis.

Abbreviation: SD, standard deviation.

For descriptive purposes, we included migraine and 4‐week headache prevalence data. Migraine was assessed by asking if the participant had ever been diagnosed with migraine by a medical doctor (yes/no/do not know). Headache was assessed by asking “During the last 4 weeks, how much have you been troubled by…” following a list of symptoms including headache with five response categories: “not at all,” “a little,” “somewhat troubled,” “considerably,” and “very much troubled.”

Statistical analyses

We described the study population and the characteristics of the included day and nights shifts. Data are presented as counts and percentages, means with their standard deviation (SD) for symmetrically distributed data, or medians with their interquartile range for skewed data. Participants/measurement days with missing information were excluded analysis by analysis.

We used proc genmod to analyze the association between headache (yes/no) as outcome and night shifts (yes/no) or the number of consecutive night shifts (1–5) as exposures. In the latter analyses of consecutive night shifts, we performed two different analyses: one with the first night shift as reference and a second with all day shifts as reference. To estimate prevalence ratios (PRs) for headache, we used Poisson regression, implemented through generalized estimating equations to account for repeated measurements within participants and with log as the link function. An independent working correlation structure was specified, as it provided the best model fit based on the quasi‐likelihood under the independence model criterion compared with alternative structures. This model also produced robust error variance as described by Linquist ³⁸ and Zou. ³⁹ Our model check showed no sign of overdispersion. Statistical testing was two‐tailed and considered significant at p value <0.05. SAS 9.4 was used for all analyses.

Because each participant is her own control and because individual characteristics were time‐invariant during the 14‐days' follow‐up, we only performed minimal adjustment. Thus, model 1 was age‐adjusted, model 2 was further adjusted for work‐related psychosocial stressors and physical job demands, and model 3 was further adjusted for TST and sleep quality. Results are presented as adjusted prevalence ratios (aPRs) with their 95% confidence intervals (CI), and can be interpreted as the ratio of the probability of reporting headache, when working a night shift compared with a day shift. Age, work‐related psychosocial stressors, physical job demands, and sleep variables were treated as linear variables. To assess the assumption of linearity, we included quadratic terms for each variable in the fully adjusted model; none were statistically significant, supporting the use of linear specifications.

We tested the methodological robustness in sensitivity analyses by running the following sensitivity analyses with (1) adjustment for prior‐day headache to approximate incident headache episodes in a sample restricted to observations with nonmissing information on the previous calendar day (model 1); (2) adjustment for shift duration and excluding all days with double shifts (i.e., working two shifts on the same day, typically a day shift and a night shift) to account for differences in total working time during one measurement day (model 1); (3) exclusion of all shifts longer than 12, 16, and 20 h to assess how much of the association would be due to night shifts being very long (model 3); (4) replacement of TST with primary sleep duration as sleep patterns differ between day and night shifts (model 3); and finally (5) replacement of TST with a measure of time in bed, (i.e., without the subtraction of sleep latency) (model 3). Results of the sensitivity analyses are provided in the Appendix (Tables A1, A2, A3, A4).

RESULTS

Of the 1075 female participants recruited for the 1001 nights‐cohort, 522 participants met the inclusion criteria and were included in the analyses. On average, each individual contributed data from 3.7 day shifts and 2.7 night shifts. Participants had a mean age of 39.6 years at enrollment. Regarding headache prevalence over the past 4 weeks, 20.9% of participants reported no headache, 64.0% were somewhat troubled by headache, and 15.1% were considerably or very troubled by headache. Additionally, 14.2% had a physician‐diagnosed migraine. Most participants (92.7%) usually worked 2‐ or 3‐shift work with night shifts, and 49.8% worked 37 h/week (Table 1). Headache was reported on 21.5% of measurement days with day shifts and on 27.9% of measurement days with night shifts. Participants had a median TST of 7.57 h when working day shifts and 6.23 h when working night shifts (Table 2).

TABLE 2.

Description of the 3348 observed measurement days included in the statistical analyses.

	Day shifts (n = 1926)				Night shifts (n = 1422)
	N	%	Mean (SD)	Median (IQR)	N	%	Mean (SD)	Median (IQR)
Starting time (hh:mm)	1926			7:15 (01:00)	1422			23:00 (03:30)
Ending time (hh:mm)	1926			15:15 (00:50)	1422			07:15 (00:30)
Duration (decimal h)	1926		8.09 (1.38)		1422		9.74 (3.13)
TST (decimal h)	1921		7.57 (1.69)		1422		6.23 (2.56)
Primary ST (decimal h)	1921		7.40 (1.58)		1327		5.08 (1.90)
Quantitative demands a	1923		4.66 (2.22)		1419		4.43 (2.27)
Emotional demands a	1915		4.26 (2.32)		1411		3.75 (2.22)
Physical demands a	1920		3.67 (1.97)		1412		3.77 (2.07)
Headache (yes)	1926	21.5			1422	27.9
Sleep quality b	1916		2.22 (0.88)		1324		2.45 (1.01)

Note: Participants with missing values were excluded analysis by analysis. For day shifts, measurement days with missing data ranged from 0 to 11. For night shifts, measurement days with missing data ranged from 0 to 11 for most variables, with the exception of primary sleep and sleep quality which were missing for 95 and 98 measurement days.

Abbreviation: hhmm, hours and minutes; IQR, interquartile range; SD, standard deviation.

^a Range 1–10.

^b Range 1–5.

We found a significant association between night shifts and headache across all three models, and the estimates were robust to adjustments for work‐related psychosocial stressors, physical job demands, TST, and sleep quality. In the fully adjusted model (model 3 in Table 3), we found a 31% higher prevalence of headache when working night shifts compared with day shifts (aPR, 1.31; 95% CI, 1.13–1.52) (Table 3).

TABLE 3.

The occurrence of headache when working night shifts compared with day shifts.

	Model 1 a				Model 2 b				Model 3 c
	aPR	Lower 95% CI	Upper 95% CI	p	aPR	Lower 95% CI	Upper 95% CI	p	aPR	Lower 95% CI	Upper 95% CI	p
Night shift vs. day shifts	1.29	1.11	1.49	0.0006	1.33	1.15	1.54	0.0001	1.31	1.13	1.52	0.0003
Consecutive night shifts
1	1.00				1.00				1.00
2	1.13	0.97	1.32	0.1262	1.17	0.99	1.36	0.0513	1.20	1.02	1.42	0.0302
3	0.84	0.65	1.08	0.1760	0.87	0.67	1.12	0.2689	0.95	0.73	1.22	0.6784
4	0.68	0.43	1.10	0.1162	0.73	0.46	1.16	0.1851	0.68	0.42	1.10	0.1145
5	0.62	0.25	1.52	0.2936	0.65	0.27	1.16	0.3330	0.66	0.27	1.58	0.3473

Note: Participants with missing values were excluded analysis by analysis. Results are presented as aPR with their L and U 95% CI. Night shifts versus day shifts: model 1, n = 3348; model 2, n = 3316; and model 3, n = 3206. Consecutive night shifts: model 1, n = 1278; model 2, n = 1267; and model 3, n = 1175.

Abbreviations: aPR, adjusted prevalence ratio; CI, confidence interval; L, lower; U, upper.

^a Adjusted for age.

^b Adjusted for age, work‐related psychosocial stressors (emotional and quantitative job demands), and physical job demands.

^c Adjusted for age, work‐related psychosocial stressors (emotional and quantitative job demands), physical job demands, total sleep time, and sleep quality.

When analyzing consecutive night shifts, we found that headache was most frequent on the second night shift. This was observed both when using the first night shift as reference (aPR, 1.20; 95% CI, 1.02–1.42) (Table 3, model 3) and when using day shifts as reference (aPR, 1.54; 95% CI, 1.28–1.84) (Figure 2). For night shift 3–5, the results indicated a decreasing prevalence of headache.

FIGURE 2.

The adjusted prevalence ratio of headache on consecutive night shifts (day shifts as a reference), with adjustment for age, work‐related psychosocial stressors (emotional and quantitative job demands), physical job demands, total sleep time, and sleep quality (n = 3074 measurement days).

All sensitivity analyses (see Appendix) implying (1) adjustment for headache on the previous day (model 1), (2) excluding double shifts and adjusting for shift duration (model 1), (3) excluding shifts longer than 12, 16, and 20 h (model 3), and replacing TST with (4) primary sleep duration or with (5) time in bed (model 3), showed that the association between night shifts and headache was remarkably robust.

DISCUSSION

Main findings

In this diary‐based study, we found a significantly higher prevalence of headache when working night shifts compared with day shifts within the same individual. The association remained robust after adjustment for age, work‐related psychosocial stressors, physical job demands, sleep duration, and quality. The highest occurrence of headache was observed on the day with the second consecutive night shift. After two consecutive night shifts, the prevalence of headache declined.

Comparison with previous research

A diary study of 679 Norwegian nurses using daily measures of shift type, sleep, and pain complaints suggested that night shifts were associated with higher odds of headache compared with day shifts (odds ratio, 1.23; 95% CI, 1.06–1.43). ²⁹ Another study in the same population reported a borderline significant increase in headache after three consecutive night shifts compared with two consecutive night shifts, but also that longer sleep duration could protect against this effect. ³⁰ Longitudinal studies with 2‐ and 3‐year follow‐ups have yielded mixed results: a Danish population‐based study of 2952 public employees (municipality or hospital) found higher odds of nonspecific headache or migraine among shift workers compared with permanent day workers after 2 years when adjusting for headache/migraine at baseline. This association remained after adjusting for perceived stress and sleep problems. ²⁴ Conversely, a Norwegian study of 1560 nurses found that night shifts did not predict headache onset 3 years later. ²⁶ However, a later study in the same population found that reducing night shifts or quitting night work was associated with a lower risk of monthly headaches. Interestingly, the study also found that starting night work was linked to an increased likelihood of recovery from monthly headaches. ²⁵

Additionally, several cross‐sectional studies have reported an association between shift work or night shifts and headache or migraine, ¹⁰ , ⁴⁰ , ⁴¹ findings further supported by a recent meta‐analysis. ⁴² In general, previous studies have reported a higher occurrence of headache among shift and night workers compared with day workers. The only study that did not observe such an association used a 3‐year follow‐up period, ²⁶ which may have introduced exposure misclassification due to potential changes in work schedules—such as transitions between night and day work—during the follow‐up.

Overall, several previous studies support the present finding of an association between night work and headache. However, there is currently no strong support for an increasing risk of headache with an increasing number of consecutive night shifts, mainly due to the paucity of studies. Our findings suggest that the misalignment of the sleep–wake cycle with the solar day that occurs, when initiating a period of night shifts and thereby making a shift from day orientation to night orientation, may contribute more to headaches than the number of consecutive shifts. This nonlinear trend in the association between consecutive night shifts and headache, peaking on the second night shift, may reflect several co‐occurring mechanisms. These include an initial sleep deficit due to an extended period without sleep, cumulative sleep debt during the first two night shifts, and potential improvement in daytime sleep quality or partial adaptation to night work after the third consecutive shift. For example, a gradual improvement in objective and subjective sleep parameters has been observed in oil rig workers. ⁴³ However, studies of permanent night workers in warehouses and manufacturing settings suggests that sleep duration decreases with more consecutive night shifts. ⁴⁴ Although speculative, it is also possible that individuals experiencing headaches during earlier night shifts may be more likely to call in sick before the third or subsequent shift, thereby contributing to the observed decrease in headache prevalence beyond the second consecutive night.

Work‐related psychosocial stressors and physical job demands are possibly related to headache occurrence ⁴⁵ , ⁴⁶ or aggravate migraine pain. ⁹ These factors vary across the 24‐hour work cycle ³¹ , ³² and could have influenced our findings. Likewise, sleep plays a crucial role in headache and migraine risk. ¹⁸ , ²⁰ In our study, headache remained more frequent when working night shifts compared with day shifts even when adjusting for work factors, suggesting that the association is not primarily driven by work‐related psychosocial stressors or physical job demands. Additionally, we adjusted for TST and sleep quality, which did not attenuate the association between night shift and headache, suggesting that these sleep‐related mechanisms do not account for the higher prevalence of headache during night shifts. Thus, other shift‐related or physiological mechanisms are likely to be causing the observed associations.

Strengths and limitations

A key strength of this study is its repeated‐measures design, which allowed us to observe the same individual when working at least one night shift and at least one day shift to examine differences in the occurrence of headache. This method allowed us to track day‐to‐day fluctuations in the occurrence of headache, and it enabled us to use participants as their own controls, thereby reducing the risk of unmeasured confounding due to group differences in individual factors. Furthermore, we adjusted for work‐related psychosocial stressors, physical job demands, and sleep. The study's relatively large sample size, with 522 participants and 3348 measurement days, strengthened the statistical power.

A limitation of our study is that we lacked information about the timing of the initiation of the headache episodes, making it unclear whether they started before, during, or after the shifts. Consequently, we cannot fully determine the temporal relation between the shift type and headache episodes. To address this limitation, we performed sensitivity analyses where we adjusted for prior‐day headache, which did not change the observed association.

Recall bias and the consequent misclassification of exposure and outcome was minimized because participants only had to recall their working hours, work factors, sleep start, sleep end, sleep quality, and headache since they last filled out a diary. However, in some cases, participants delayed diary completion and thereby had to recall multiple days at once. Yet, 90.5% of the diaries in the cohort study were filled out on the planned date, thus we expect that this bias had minor implication for the findings. ³³ Furthermore, we cannot rule out the possibility that particularly strenuous shifts—especially when followed by headache—reduced the likelihood of completing a sleep diary. If “missingness” is related to both the exposure (e.g., night shift strain) and the outcome (e.g., headache), this could lead to an underestimation of the association between night work and headache.

Another limitation is that we did not use validated headache diagnosis measures or clinical classification of the type of headache participants experienced (e.g., migraine or tension‐type headache). We solely asked about the presence of headache in the diaries, without any further details about accompanying (migraine) symptoms or pain intensity. Thus, we cannot distinguish between subtypes of headache (e.g., tension‐type headache and migraine). We did not apply a more thorough assessment of headache due to the brief diary format. Although participants likely accurately reported severe headache presence, the absence of pain intensity or associated symptoms may have resulted in underestimation of headache prevalence. In the background questionnaire, 69 individuals reported being diagnosed with migraine, thus, we had insufficient statistical power for a meaningful subgroup analysis.

To assess sleep, we used self‐reported information about bedtimes and wake times with sleep latency subtracted to approximate total sleep time. By doing so, we approach a proposed standardized method of constructing diaries for research purposes. ⁴⁷ Taking sleep latency into account is particularly important, as shift workers find it easier to fall asleep after night shift than after day shifts. Furthermore, to account for differences between day‐ and nighttime sleep, apart from duration, we adjusted for sleep quality, which did not change the findings. Neither did the use of “time in bed” as an alternative measure of sleep duration.

Although our study design used individuals as their own controls, the potential influence of the healthy shift worker effect ²⁸ cannot be entirely ruled out. It is possible that individuals prone to headache may avoid night shifts, resulting in fewer headache episodes altogether among those working night shifts. On one hand, this could explain that the headache prevalence did not increase with the number of consecutive night shifts as individuals prone to headache may avoid working consecutive night shifts and night shifts altogether. On the other hand, when combining migraine and frequent headache in our study population, we found that 23.0% were diagnosed with migraine and/or responded being considerably/very much troubled by headache. This number is similar to the frequency of headache disorders in Danish occupationally active women, among whom 27.2% have self‐reported migraine or frequent headache. ⁴⁸ Furthermore, in the 1001 nights‐cohort, the prevalence of self‐reported migraine diagnosis and the 4‐week prevalence of headache were similar between permanent day workers and shift workers with night work. Altogether, this suggests that the burden of headache disorders in our sample does not differ substantially from that of the general female working population or permanent day workers, thereby supporting the generalizability of our findings beyond the studied group. However, it is important to note that most existing research in this field has been conducted among nurses or hospital staff. Although we have no reason to believe that the associations observed between night work and headache are unique to our cohort, their applicability to other occupational groups remains to be empirically confirmed. Moreover, our findings pertain solely to female employees working day and night shifts, whereas the association between headache and work schedules involving permanent night shifts remains to be determined.

CONCLUSION

This is the first study using daily measurements for investigating the acute effect of night work on headache, while also accounting for work‐related psychosocial stressors and physical job demands. Our results indicate that the adjusted prevalence of headache is 31% higher when working a night shift compared with a day shift. This difference cannot be explained by the investigated work factors or sleep. We found no evidence to support the notion that an increase in consecutive night shifts systematically leads to higher headache frequency. Headache appears to be most frequent on the second night shift, which may both be a consequence of circadian misalignment, accumulation of sleep debt, and behavioral mechanisms (e.g., sickness absence on later shifts). Based on the current findings, we cannot provide evidence‐based recommendations of certain threshold shifts per month or maximum consecutive nights. Therefore, this issue warrants further investigation to support nonpharmacological prevention of headache through alterations in the number of consecutive night shifts.

Nevertheless, the findings of this study may inform strategies for mitigating headache risk among occupationally active women, particularly in sectors with high night work prevalence, such as healthcare, transportation, manufacturing, and law enforcement. Although the results do not clearly identify modifiable factors that could reduce this risk, interventions might focus on personalized scheduling for headache‐prone workers. Additionally, individualized strategies, developed in consultation with headache specialists, could be considered for workers experiencing frequent and disabling headaches associated with night work; however, the potential effectiveness of such interventions remains uncertain based on the current findings.

AUTHOR CONTRIBUTIONS

Rikke Harmsen: Conceptualization; formal analysis; investigation; methodology; writing – original draft. Jakob Møller Hansen: Conceptualization; methodology; supervision; writing – review and editing. Dagfinn Matre: Conceptualization; methodology; writing – review and editing. Anne Emily Saunte Fiehn Arup: Formal analysis; investigation; writing – review and editing. Anne Helene Garde: Conceptualization; data curation; funding acquisition; investigation; methodology; project administration; supervision; writing – review and editing. Kirsten Nabe‐Nielsen: Conceptualization; data curation; formal analysis; investigation; methodology; project administration; supervision; writing – original draft; writing – review and editing.

FUNDING INFORMATION

The data collection for the 1001 nights‐cohort is supported by funding from The Danish Working Environment Research Fund (26‐2020‐09; 19‐2022‐09), The Health Foundation (22‐B‐0452), the European Union's Horizon 2020 Research and Innovation Program (grant agreement 874703), and The National Research Centre for the Working Environment.

CONFLICT OF INTEREST STATEMENT

Rikke Harmsen, Dagfinn Matre, Anne Emily Saunte Fiehn Arup, and Kirsten Nabe‐Nielsen declare no conflicts of interest. Jakob Møller Hansen reports receiving personal fees from Pfizer and AbbVie. Anne Helene Garde has received funding for the 1001 nights‐cohort from The Danish Working Environment Research Fund and declares no conflict of interest.

APPENDIX E. RESULTS OF SENSITIVITY ANALYSES (NIGHT WORK AND HEADACHE)

TABLE A1.

The occurrence of headache when working night shifts compared with day shifts with adjustment for prior‐day headache in a sample restricted to observations with nonmissing information on the previous calendar day (n = 2947).

	Sensitivity analysis (i)
	aPR	L 95% CI	U 95% CI	p
Night vs. day	1.29	1.14	1.46	<0.0001

Note: Results are presented as aPR with their L and U 95% CI. Adjusted for age.

Abbreviations: aPR, adjusted prevalence ratio; CI, confidence interval; L, lower; U, upper.

TABLE A2.

The occurrence of headache when working night shifts compared with day shifts with adjustment for shift duration and excluding all days with double shifts (i.e., working two shifts on the same day, typically a day shift and a night shift) (n=3181).

	Sensitivity analysis (ii)
	aPR	L 95% CI	U 95% CI	p
Night vs. day	1.29	1.10	1.51	0.0021

Note: Results are presented as aPR with their L and U 95% CI. Adjusted for age.

Abbreviations: aPR, adjusted prevalence ratio; CI, confidence interval; L, lower; U, upper.

TABLE A3.

The occurrence of headache when working night shifts compared with day shifts with exclusion of all shifts longer than 12 h (n = 2948), 16 h (n = 3092), and 20 h (n = 3191).

	Sensitivity analysis (iii)
	12 h				16 h				20 h
	aPR	L 95% CI	U 95% CI	p	aPR	L 95% CI	U 95% CI	p	aPR	L 95% CI	U 95% CI	p
Night vs. day	1.27	1.08	1.49	0.0035	1.32	1.14	1.54	0.0003	1.31	1.13	1.52	0.0003

Note: Results are presented as aPR with their L and U 95% CI. Adjusted for age, work‐related psychosocial stressors (emotional and quantitative job demands), physical job demands, total sleep time, and sleep quality.

Abbreviations: aPR, adjusted prevalence ratio; CI, confidence interval; L, lower; U, upper.

TABLE A4.

The occurrence of headache when working night shifts compared with day shifts with replacement of TST with primary sleep time as sleep patterns differ between day and night shifts (n = 3206) and replacement of TST with a measure of time in bed, i.e., without the subtraction of sleep latency (n = 3206).

	Sensitivity analysis (iv)				Sensitivity analysis (v)
	aPR	L 95% CI	U 95% CI	p	aPR	L 95% CI	U 95% CI	p
Night vs. day	1.38	1.18	1.63	<0.0001	1.32	1.14	1.53	0.0002

Note: Adjusted for age, work‐related psychosocial stressors (emotional and quantitative job demands), physical job demands, primary sleep duration/time in bed, and sleep quality.

Abbreviations: aPR, adjusted prevalence ratio; CI, confidence interval; L, lower; TST, total sleep time; U, upper.

Harmsen R, Hansen JM, Matre D, Arup AESF, Garde AH, Nabe‐Nielsen K. The acute effect of night work‐related circadian misalignment on headache episodes: Results from the 1001 nights‐cohort. Headache. 2025;65:1554‐1564. doi: 10.1111/head.15054