Baseline sleep quality, stress and depressive symptoms and subsequent headache occurrence in a six-week prospective cohort study of patients with episodic migraine

Abstract

Background:

Despite the high prevalence of sleep disturbance, stress, and depressive symptoms among patients with episodic migraine, there has been limited prospective research examining how these comorbid symptoms relate to future headache risk.

Methods:

We conducted an a priori secondary analysis of a prospective cohort study of 98 adults with episodic migraine recruited through Harvard-affiliated medical centers and local college student clinics in Boston, MA. At baseline, participants completed validated questionnaires on sleep quality, stress, and depressive symptoms. Over the next six weeks, they recorded headaches on twice-daily diaries. We conducted time-to-event analyses to evaluate whether these baseline symptoms are associated with headache recurrence.

Results:

At baseline, 45/98 (46%) participants had poor sleep quality, 51/98 (52%) reported moderate/high stress levels, and 18/98 (18%) had high depressive symptom scores. Over 4,406 person-days, we observed 823 discrete headaches. In multivariable models, the hazard ratios for headache recurrence were: 1.22 (95% CI 1.02, 1.46) for people with baseline poor sleep, 1.12 (95% CI 0.93, 1.35) for those with baseline moderate/high stress compared to lower levels, and 1.31 (95% CI 1.05, 1.65) for the combination of poor sleep and moderate/high stress compared to the combination of good sleep and low stress. There was no association between depression scores and headache risk.

Discussion:

Among patients with episodic migraine, poor sleep was associated with a higher rate of headache recurrence over the next six weeks, especially among those with coexisting moderate/high stress.

Introduction:

The clinical evaluation of migraine often includes the assessment of comorbid psychiatric symptomatology, including poor sleep, stress, and depressive symptoms. Over the long-term, these comorbidities may confer a higher risk of migraine chronification ¹ and in those with chronic migraine, they may alter the migraine treatment approach ². However, whether baseline reports of poor sleep quality, stress, and depressive symptoms are associated with headache burden in the short-term in episodic migraine is unknown.

Poor sleep quality is highly prevalent in those with migraine ^3–9 and is cross-sectionally associated with higher headache frequency among those with episodic migraine ^{5, 10}. We previously found that among people with episodic migraine, nightly poor sleep quality is not associated with headache risk the next day ¹¹. However, the time frame (acute vs. chronic) by which sleep may influence migraine may differ. Improving the understanding of the temporal relationship between poor sleep quality and migraine has important implications for elucidating the neurobiologic mechanisms of migraine as well as informing intervention targets. Therefore, we examined whether longer periods of poor sleep quality may be associated with a higher risk of subsequent headache occurrence.

Poor sleep quality shares bidirectional relationships with self-perceived stress and mood symptoms ^12–14 and investigation of all three and their timecourse provides insight into their individual and collective influence on migraine. Perceived stress appears to be a more immediate precedent to headache days in those with episodic migraine ¹⁵, and it influences headache severity in those with chronic migraine ^{16, 17}. The exact timing of this relationship may be more nuanced, however, as high stress may precede migraine by more than one day ^16–18. Lastly, there is a bidirectional risk between depression and migraine, such that patients with migraine over a longer period (8–14 years) are at a higher risk of incident clinical depression and vice-versa ^{19, 20}. Interestingly, these associations are attenuated when controlling for chronic stress ¹⁹. Amongst those with episodic migraine, depressive symptoms may also confer a higher risk of transformation to chronic migraine migraine¹.

We examined clinically-based thresholds on validated scales for sleep quality, stress, and depressive symptoms and the rate of migraine recurrence over a subacute timeframe, the following six weeks, among patients with episodic migraine. Our a priori hypothesis is that those with poorer sleep quality, higher stress or higher depressive symptoms at baseline would have a higher rate of headache recurrence during the prospective study period. This information may be useful to a practicing clinician and add to the understanding of the timing of these relationships.

Methods:

We conducted an a priori secondary analysis of a prospective cohort study whose primary objective was to examine the day-to-day temporal associations of sleep and migraine in participants with episodic migraine, the details of which have been previously published¹¹. Since this is a secondary analysis of existing data, no statistical power calculation was conducted, and the sample size was based on the available data.

Study Setting

Participants were recruited through three academic centers in Boston, MA (Beth Israel Deaconess Medical Center [BIDMC], Massachusetts General Hospital, and Brigham and Women’s Hospital), and local college student health clinics from March 2016 to August 2017. Baseline and follow-up visits were conducted at BIDMC, and prospective electronic diaries using web-based REDCap (Research Electronic Data Capture)²¹ or paper diaries, if preferred (n=5), twice daily for six weeks. The BIDMC Committee on Clinical Investigations approved the study, and all participants provided written informed consent.

Study population

Potential participants were screened by telephone and then completed one week of run-in diaries, with the final determination of eligibility based on physician interview at the baseline visit. Participants with episodic migraine were enrolled if they were ≥18 years of age, reported a history of migraine for at least three years, experienced at least two migraine headaches per month during the three months before enrollment, were able to communicate in English, and give informed consent. Enrolled participants met the criteria for the International Classification of Headache Disorders-3-beta for episodic migraine with or without aura ²² based on an in-person physician interview at the baseline visit. Exclusion criteria included: ≥15 headache days per month for the previous three months, chronic pain condition, current opioid use, high-risk of obstructive sleep apnea ²³ or known untreated obstructive sleep apnea, pregnancy, uncontrolled medical problems that precluded participation, or failure to complete at least 4 out of 7 days of run-in diaries. Among the 126 individuals expressing interest in the study, 101 met inclusion criteria and agreed to participate. Three withdrew with <21 days of data, resulting in a study sample of 98 participants.

Baseline Assessments of Sleep Quality, Depression, Stress, and Covariates

At the baseline visit, participants reported socio-demographics, medical and headache history and sleep quality, stress, and depressive symptoms using standardized questionnaires. We quantified sleep quality and disturbance in the last month used the Pittsburgh Sleep Quality Index (PSQI)²⁴, a 19-item validated questionnaire with sub-scores for sleep efficiency, perceived sleep quality, and daily disturbances (ie, frequency of nighttime sleep disturbance, use of sleep medications and difficulty staying awake during the day) ⁸. We measured depressive symptoms over the past week with the 20-item Center for Epidemiological Studies Depression scale (CES-D) ²⁵. We measured self-perceived stress over the last month using the ten-item Perceived Stress Scale (PSS-10)²⁶. We obtained information for potential confounders from self-reported questionnaires at the baseline visit, including their recall of typical caffeine intake, alcohol intake, exercise frequency, and headache days per month for the past three months.

Six-week prospective daily reporting of headache occurrence

During the six weeks of follow-up, participants completed twice-daily diaries that asked “Did you have a headache since the time you completed your last diary?”. Additional features such as headache duration, maximum pain severity, and associated migrainous features (photophobia, phonophobia, nausea/vomiting, etc.) were also captured. When patients reported two headaches within the same day (n=79), headache duration was defined as the start of the first headache and the end of the second headache. Maximum pain intensity was determined using the highest pain scores. Rates of adherence to daily data capture were excellent (93.5 ± 6.0%). Since data from headaches were captured from diary to diary, this provided information on headache status on days that a participant did not submit a diary entry. Therefore, observation days with missing diary data were not censored.

Statistical Analysis

We summarized baseline characteristics by calculating means, standard deviations (SDs) and proportions of participants by demographic characteristics, headache characteristics and scores on questionnaires for the whole sample. For the PSQI and CES-D total scores, we used published cutoffs to create binary indicators for poor sleep quality (PSQI ≥5) and high depressive symptom scores (CESD ≥16) ^{25, 27}. In analyses using the PSQI 3-factor model ²⁸, we dichotomized the scores for sleep efficiency, perceived sleep quality, and daily disturbances at the 75^th percentile of the sample distribution, with lower scores serving as the referent category. For the PSS-10, we combined the moderate and high stress categories (PSS-10 ≥14) since only five participants reported high stress.

To quantify the associations between psychological symptoms and headache occurrence (defined as the presence of headache on a particular day), we conducted time-to-event analyses using an Andersen-Gill formulation of the Cox proportional hazard models to estimate hazard ratios (HRs) and 95% confidence intervals (CIs). ²⁹ This allowed us to accommodate multiple events (since participants were expected to have more than one headache day during the prospective 6 week period). We constructed separate models for baseline sleep quality, depressive symptoms, and stress, respectively, and an additional model that included all three of these exposures. In all models, we adjusted for sex, age, and baseline report of caffeinated beverage intake, alcohol intake, exercise frequency, and headache days per month. Because the baseline report of headache days per month may be a confounder or a surrogate for the outcome, we conducted the analysis with and without this variable in the model.

Since scores of sleep quality and stress were highly correlated with each other, we examined their combined impact on the primary outcome. Specifically, we created a four-level variable: low stress/good sleep quality (referent group); poor sleep quality only; moderate/high stress only; and both poor sleep quality and moderate/high stress. We included this term in a model adjusting for all potential confounders mentioned above. Two-sided p-values of <0.05 were considered statistically significant. Analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, NC).

Results

Participants had a mean age of 35±12 years, 86/98 (88%) were women, and 81/98 (83%) identified as white/Caucasian. At baseline, participants reported an average of five headaches per month, and 26/98 (27%) were taking a migraine preventive medication. At baseline, 45/98 (46%) of the participants reported poor sleep quality (PSQI ≥5), 18/98 (18%) reported high depressive symptom scores (CES-D ≥16), and 51/98 (52%) reported moderate/high stress levels (PSS-10 ≥14). Most participants reported a modest intake of caffeinated and alcoholic beverages and exercised three or more times per week (Table 1).

Table 1. Baseline characteristics of 98 participants with episodic migraine, 2016–2017.

Pittsburgh Sleep Quality Index (PSQI); Center for Epidemiological Studies depression scale (CES-D); Perceived Stress Scale-10 (PSS-10)

	N (%) or Mean (± SD)
Age (years)	35.1 ±12.1
Women	86 (88%)
Premenopausal (% women)	68 (79%)
Race
White/Caucasian	81 (83%)
Black/African-American	4 (4%)
Asian	4 (4%)
Other	9 (9%)
Smoking Status
Never	81 (84%)
Past	13 (14%)
Current	2 (2%)
Age at migraine onset (years)	16.3 ±8.3
Headache attacks/month	5.0 ±3.6
Migraine with aura	48 (49%)
Any migraine prophylactic	26 (27%)
Caffeine use/day
0	20 (21%)
1–2	65 (67%)
3–4	12 (12%)
Alcoholic beverages/week
0	31 (32%)
1–3	46 (47%)
4–7	16 (16%)
8 or more	4 (4%)
Exercise (times/week)
0	3 (3%)
1–2	26 (27%)
3–4	41 (43%)
5 or more	25 (26%)
High depressive symptoms (CES-D ≥16)	18 (18%)
Moderate/high stress (PSS-10 ≥14)	51(52%)
Poor sleep quality (PSQI≥5)	45 (46%)
Poor sleep efficiency (PSQI 3 factor ≥1)	31 (32%)
Poor perceived sleep quality (PSQI 3 factor ≥ 4)	22 (23%)
Worse daily disturbances (PSQI 3 factor ≥ 3)	18 (18.4)

Over 4,406 person-days, we observed 823 discrete headaches, representing an average of 8.5 headaches per person during follow-up. The average headache duration was 11.1±9.4 hours, and the average maximum pain severity was 46±14(0–100 scale). Of note, rates of adherence to daily data capture were excellent (93.5 ± 6.0% diary).

In fully adjusted models, baseline report of poor sleep quality was associated with a 22% higher rate of headache recurrence (HR 1.22, 95% CI 1.02, 1.46; Table 2). Moderate/high stress was not significantly associated with a higher rate of headache recurrence (HR 1.12, 95% CI 0.93, 1.35) nor were depressive symptoms (HR 0.93, 95% CI 0.69, 1.26). Results did not differ with or without the adjustment for potential confounders. We observed a 31% higher rate among those reporting both poor sleep quality and moderate/high stress compared to those with good sleep quality/low stress (HR 1.31, 95% CI 1.05, 1.65; Figure 1).

Table 2. Associations between baseline poor sleep quality, stress and depressive symptoms and headache reccurrence over 4,406 days among 98 participants with episodic migraine, 2016–2017.

Scores dichotomized using published clinical cutoffs: CES-D ≥16, PSS-10 ≥14 and PSQI ≥5.

	Hazard Ratio (95% Confidence Interval)
	Minimally adjusteda	Adjustedb	Fully adjustedc
High depressive symptoms (CES-D ≥16)	0.96 (0.74,1.25)	0.99(0.75,1.30)	0.93 (0.69,1.26)
Moderate/high stress (PSS-10 ≥12)	1.19(0.94,1.51)	1.16(0.92,1.47)	1.12 (0.93, 1.35)
Poor sleep quality (PSQI≥5)	1.21(0.98,1.48)	1.20 (0.98,1.46)	1.22 (1.02, 1.46)

^aAdjusted for sex and age.

^bFurther adjusted for caffeinated beverages per day, alcoholic beverages per week and number of times of moderate/high intensity exercise per week.

^c.Further adjusted for baseline report of headache days per month.

Figure 1. Hazard ratios for associations between baseline reported sleep quality and stress with headache occurrence over the following six weeks, 2016–2017.

Hazard ratios for combined sleep quality and stress categories in a model adjusted for sex, age, caffeinated beverages per day, alcoholic beverages per week and number of times of moderate/high intensity exercise per week, baseline report of headache days per month were run as continuous variables.

In separate models for the PSQI subscores, poor sleep quality (HR 1.22, 95% CI 0.94, 1.57), poor sleep efficiency (HR 1.06, 95% CI 0.87,1.28), and worse daily disturbances (HR 1.16, 95% CI 0.89, 1.52) were associated with higher headache recurrence, though none reached statistical significance

Discussion

Among adults with episodic migraine, poor sleep quality at baseline was associated with a 22% higher daily rate of headache over the following six weeks. Concurrent report of higher levels of stress and poor sleep quality over the prior month was associated with a 31% higher rate of headache recurrence. These results suggest that in patients with episodic migraine, positive screens using brief validated questionnaires for poor sleep quality and stress at a single time point may identify those at higher risk of headache recurrence over the short-term.

Our study is among the first longitudinal studies to report that baseline poor sleep quality over the past month is associated with a higher rate of subsequent headache recurrence in patients with episodic migraine followed over time. These results are consistent with cross-sectional studies showing that poor sleep quality (i.e., PSQI >5) is associated with higher self-reported headache frequency in those with episodic migraine ^{5, 10}. Although other studies did not have a short-term prospective design, poor sleep quality was more prevalent in patients with chronic migraine compared to episodic migraine and more prevalent in patients with episodic migraine and probable migraine compared to those with no history of headaches ^3–9. Given that most patients with chronic migraine have episodic migraine for several years, factors such as sleep quality may play a role in migraine chronification. Our subscale analysis of the PSQI suggests that daytime dysfunction may have an equally important role as poor sleep quality, and further research is needed to disentangle the potential independent contribution of sleep quality and daytime dysfunction on migraine risk. As previously reported in this cohort of participants with episodic migraine, prospective day-to-day analysis of nightly poor sleep quality was not a temporal precedent of next-day migraine ¹¹. Taken together with our current findings, it appears that among patients with episodic migraine, poor sleep quality on a given day is not associated with immediate headache, but having poor sleep quality over the last month is associated with a higher migraine burden over the next six weeks.

We also observed that the combination of poor sleep and self-perceived moderate/high stress at baseline was associated with a higher risk of subsequent headache recurrence compared to good sleep/low stress over the following six weeks. Two prior cross-sectional studies reported higher PSS-10 scores in patients with chronic migraine only (i.e., not episodic migraine) compared to those without migraine ³⁰, and higher PSS-10 scores in those with high-frequency episodic migraine (9–14 days/month) than those with a low frequency of episodic migraine (1–4 days per month) ³¹. Several prior studies have examined the day to day associations between self-perceived stress and migraine. In those with episodic migraine, perceived daily stress levels may have a more direct and immediate impact on migraine occurrence,^{15, 32}, although the timing of this may differ amongst individuals and vary from attack to attack³². Two landmark studies have investigated the day-to-day associations of stress and migraine occurrence ¹⁷ and pain intensity ¹⁶ in those with chronic migraine. In a longitudinal daily diary study of stress and sleep in those with chronic migraine, Houle et al identified poor sleep, high stress or the combination of both as a risk factor for worse headache severity two days later. ¹⁶ In another longitudinal daily diary study by Lipton et al, a decline in stress (ie “stress letdown”) was associated with headache occurrence the next day ¹⁷. Our results provide additional insights on temporality, suggesting that headache recurrence is the highest for those reporting both moderate/high stress and poor sleep over the last month. Future research may focus on the complementary benefits of treatments targeting subacute stress and poor sleep quality.

We did not find an association between depressive symptoms and subsequent headache recurrence. In contrast, the American Migraine Prevalence and Prevention (AMPP) study reported that depressive symptoms in people with episodic migraine were associated with subsequent chronic migraine diagnosis in the following year ¹. Although the prevalence of depressive symptoms in our study (18%) was similar to the prevalence of those in the AMPP study (20%), depressive symptoms and diagnosis of depression were quantified using different methods. We used CESD-20 cutoffs to determine depressive symptom scores, and the AMPP study defined depression as either clinically diagnosed depression or depression diagnosis using PHQ-9 cutoffs. Therefore, the prior study included participants with diagnosable depression and more severe symptoms of depression ³³. Furthermore, we studied the rate of headache recurrence among participants with episodic migraine over a short period (six weeks). In contrast, the AMPP study assessed depression as a risk factor for chronic migraine at one year among participants with episodic migraine at baseline ¹.

To summarize, our findings suggest that among patients with episodic migraine, poor sleep was associated with a higher rate of headache recurrence over the next six weeks, especially among those with coexisting moderate/high stress. This indicates that poor sleep quality, or its perception, and stress may either influence headache frequency or may be markers of a worsening migraine burden. Relationships between migraine, stress, and sleep disturbances localize to common brain structures, including the hypothalamus, and implicate HPA axis activation and circulating inflammatory markers as biologically plausible links. Prior studies of day to day changes in diurnal cortisol in those with migraine have been inconsistent, although those with chronic migraine may have higher baseline diurnal cortisol³⁴. Prior studies of pain threshold testing in headache-free healthy adults with poor sleep quality and insomnia demonstrated lower pain thresholds ³⁵ and increased cortisol reactivity ³⁶ compared to participants with sleep quality in the normative range. Inclusion of patients with episodic migraine in pain research is needed to provide insight into the relationship between potential HPA axis abnormalities and migraine.

Our study’s strengths included the use of validated baseline assessments of sleep, stress, and depressive symptoms and the daily, prospective assessment of headaches for at least six weeks. Our study was limited in that our sample comprised mostly white/Caucasian premenopausal women with relatively healthy sleep habits and modest sleep disturbances. Thus these results may not be generalizable to other populations.

Including validated assessments of stress and sleep quality that quantify symptoms over the prior month as part of the evaluation of headache may help identify those at higher risk for headaches who may benefit from more aggressive management. Completing these brief assessments is a feasible and informative addition to the initial clinical evaluation. Further research is needed to explore how treatment may affect the independent and combined effects of sleep quality and stress on headache frequency in those with episodic migraine.