Association between sleep disorders, organic diseases, and mental disorders in 1442 subjects with chronic morning headaches. All numbers indicate percentages.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Ohayon MM. Prevalence and Risk Factors of Morning Headaches in the General Population. Arch Intern Med. 2004;164(1):97–102. doi:10.1001/archinte.164.1.97
To determine the prevalence of chronic morning headaches (CMH) in the general population and their relationship to sociodemographic characteristics, psychoactive substance use, and organic, sleep, and mental disorders.
A telephone questionnaire was submitted to 18 980 individuals 15 years or older and representative of the general populations of the United Kingdom, Germany, Italy, Portugal, and Spain. It included a series of questions about morning headaches, organic disorders, use of psychoactive substances, and sleep and mental disorders in accordance with the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV).
Overall, the prevalence of CMH was 7.6% (n = 1442); CMH were reported to occur "daily" by 1.3% of the sample, "often" by 4.4%, and "sometimes" by 1.9%. Rates were higher in women than in men (8.4% vs 6.7%) and in subjects aged between 45 and 64 years (about 9%). The median duration for CMH was 42 months. Various conditions and disorders were found positively associated with CMH. The most significant associated factors were comorbid anxiety and depressive disorders (28.5% vs 5.5%), major depressive disorder alone (21.3% vs 5.5%), dyssomnia not otherwise specified (17.1% vs 6.9%), insomnia disorder (14.4% vs 6.9%), and circadian rhythm disorder (20.0% vs 7.5%). Sleep-related breathing disorder (15.2% vs 7.5%), hypertension (11.0% vs 7.2%), musculoskeletal diseases (14.1% vs 7.1%), use of anxiolytic medication (20.1% vs 7.3%), and heavy alcohol consumption (12.6% vs 7.7%) were also significantly associated with CMH.
Morning headache affects 1 individual in 13 in the general population. Chronic morning headaches are a good indicator of major depressive disorders and insomnia disorders. Contrary to what was previously suggested, however, they are not specific to sleep-related breathing disorder.
Waking up with a headache is traditionally associated with sleep disorders. Clinical studies have reported a high association between morning headaches and obstructive sleep apnea syndrome (OSAS)1,2 and snoring.1,3 Between 18% and 41% of patients with OSAS have experienced headaches upon awakening in the morning.1,2,4 The prevalence of morning headaches in the general population is not well known, although, according to a Swedish study, 5% of the population awakens often or very often with headaches.1
Other sleep disorders have also been associated with headaches upon awakening. Two studies reported morning headaches associated with bruxism.5,6 A study that examined the sleep of women living with heavy snorers found that these women, in addition to suffering disturbed sleep and daytime sleepiness, also had morning headaches.7 Other studies reported associations between morning headaches and periodic limb movement disorders.8,9 Morning headaches were also found in persons with hypertension.10
Surprisingly, although there is considerable literature about the association between the different subtypes of headaches (migraine, cluster headaches, tension headaches) and mental disorders, especially depressive disorders and anxiety disorders,11-15 to our knowledge no study has attempted to determine the status of mental disorders in the report of morning headaches.
This study examines the importance of 5 categories of factors (sociodemographic determinants, use of psychoactive substances, organic diseases, sleep disorders, and mental disorders) in relationship to morning headaches (ie, headaches present upon awakening) using a sample of 18 980 individuals representative of 5 European countries.
The participants were interviewed by telephone between 1994 and 1999 in 5 countries16: the United Kingdom, Germany, Italy, Portugal, and Spain. Ethical and research committees at the Imperial College (London, England), the Regensburg University (Regensburg, Germany), the San Raffaele Hospital (Milan, Italy), the Santa-Maria Hospital (Lisbon, Portugal), and the University Vall d'Hebron Hospital General (Barcelona, Spain) approved the study.
The target population consisted of noninstitutionalized residents 15 years or older, with the exception of Portugal where the minimum age was set at 18 years at the recommendation of the Portuguese ethics committee. This represented about 206 million Europeans. A 2-stage design was used for all countries. The population of each country was first divided according to a geographic distribution that followed official census data, then telephone numbers were randomly drawn in each geographic area. Second, within each household, a member was selected by age and sex using the Kish method17 to maintain the representativity of the sample and avoid bias related to noncoverage error.
Participants had to grant their verbal consent prior to proceeding with the interview. For subjects younger than 18 years, verbal parental consent was also requested. We excluded potential participants who had insufficient fluency in the national language, a hearing or speech impairment, or an illness that precluded an interview.
The participation rate was 79.6% (4972 of 6249 eligible individuals) in the United Kingdom; 68.1% (4115 of 6047) in Germany; 89.4% (3970 of 4442) in Italy; 83% (1858 of 2234) in Portugal; and 87.5% (4065 of 4648) in Spain. A total of 18 980 subjects participated in the study. The overall participation rate was 80.4%.
The Sleep-EVAL expert system,18,19 a software specifically designed by the author to conduct epidemiological studies in the general population and to administer questionnaires, was used to perform the interviews. Lay interviewers read to the subjects the questions displayed on the computer screen and entered the answers in the system.
The knowledge base of the system comprised a standard questionnaire and diagnostic pathways covering the International Classification of Sleep Disorders (ICSD)20 and the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV).21 The questionnaire consisted of sociodemographic information, sleep/wake schedule, physical health queries, and questions related to sleep and mental disease symptoms. Interviews typically began with general questions about demographic characteristics, followed by questions about sleeping habits. The interviews progressed to more private questions regarding mental health.
The system used the answers to select a series of plausible diagnostic hypotheses (causal reasoning process). Further questioning and deducting consequences from each answer allowed the system to confirm or reject these hypotheses (nonmonotonic, level-2 feature). The differential diagnosis process was based on a series of key rules allowing or prohibiting the co-occurrence of 2 diagnoses in accordance with ICSD and DSM-IV guidelines. The interview ended once all diagnostic possibilities were exhausted.
The system has been tested in various contexts. In clinical psychiatry, overall κ values between the diagnoses of 4 psychiatrists and those of the system ranged from 0.44 with 1 psychiatrist to 0.78 with 2 psychiatrists (n = 114 cases).22 Another study involved 91 forensic patients. A majority of the patients (60%) received a diagnosis in the psychosis spectrum. The κ value between diagnoses given by the system and by psychiatrists was 0.44 for specific psychotic disorders (mainly schizophrenia). In a study performed in the general population (n = 150), the diagnoses obtained by 2 interviewers inexperienced in sleep and psychiatric assessments who used the Sleep-EVAL system were compared with the diagnoses obtained by 2 clinician psychologists; κ values of 0.85 and 0.70 were obtained in the recognition of sleep problems and insomnia disorders, respectively. Validation studies performed in sleep disorders clinics (Stanford University, Regensburg University, and Toronto Hospital) testing the diagnoses of the system against those of sleep specialists using polysomnographic data gave κ values of 0.93 and 0.92 with the diagnosis of OSAS and 0.78 and 0.71 with that of insomnia.23,24
The duration of the interviews ranged from 10 to 333 minutes, with an average ± SE of 40 ± 20 minutes. The longest interviews involved subjects with multiple sleep and mental disorders. Interviews were completed over 2 or more sessions if their duration exceeded 60 minutes.
The presence of morning headaches was assessed using the following question:
"Do you have headaches when you wake up in the morning?" The subject had the following answer choices: "always," "often," "sometimes," "rarely," "never," "does not know." Subjects who answered with "always," "often," or "sometimes" were considered to have morning headaches.
Morning headaches were analyzed in relationship to demographic variables and use of alcohol, tobacco, caffeine, and psychotropic medication. Organic disorders as well as DSM-IV sleep and mental disorder diagnoses were also analyzed.
A weighting procedure was applied to correct for disparities in the geographic, age, and sex distribution between the sample and each studied country. This procedure was to compensate for any potential bias from such factors as an uneven response rate across demographic groups. Results were based on weighted n values. Percentages for target variables were given with 95% confidence intervals (CIs) or SEs. Bivariate analyses were performed using the χ2 test. Each class of variables was analyzed using logistic regressions,25 and 95% CIs were calculated for prevalence rates and odds ratios (ORs). Reported differences were significant at .05 or less.
The sample consisted of 18 980 individuals 15 years or older; more than half of them were married, one third were single, 51.3% were women, one quarter were from the United Kingdom, 1 in 5 was Italian, German, or Spanish, and 1 in 10 was Portuguese.
Overall, 7.6% of the sample (n = 1442) reported waking up with headaches: 1.3% said they always woke up with headaches; 4.4% said it occurred often, and 1.9% reported sometimes waking up with headaches. The median duration in which morning headaches occurred was 42 months.
Women reported waking up with headaches more frequently than men (8.4% vs 6.7%; OR, 1.3 [95% CI, 1.1-1.4]). The prevalence of morning headaches also changed with age: it was higher among subjects aged between 45 and 64 years (Table 1). Morning headaches were significantly higher in married than in single individuals (Table 1), and higher in unemployed individuals and homemakers than in workers and students (Table 1).
The prevalence of morning headaches was linked to the body mass index (BMI; calculated as weight in kilograms divided by square of height in meters) of subjects. A higher prevalence of morning headaches was found in subjects with a BMI less than 20 (9.0%) and in obese subjects (BMI >27) (8.1%) than in subjects with a BMI in the normal range (20-25) (7.0%) (P<.005).
Morning headaches were also examined in relationship to the most frequently reported diseases. We found a higher prevalence of morning headaches in subjects who reported musculoskeletal diseases (14.1%) than in other subjects (7.1%; P<.001). Subjects with a heart disease more frequently reported morning headaches than other subjects (10.0% vs 7.5%; P<.05). Subjects with a thyroid disease also had a higher prevalence of morning headaches (13.1% vs 7.6%; P<.05). Subjects with an upper airway disease significantly more often reported waking up with headaches (11.3% vs 7.5%; P<.05). Subjects with high blood pressure, treated or not, had a higher prevalence of morning headaches (11.0%) than the remaining sample (7.2%; P<.001). Other types of physical diseases were not significantly related to morning headaches.
Morning headaches were more prevalent in heavy alcohol drinkers (≥6 glasses/day) (12.6%) than in subjects who did not drink alcohol (7.7%) or who drank less (1 or 2 glasses/day [5.6%] and 3 to 5 glasses/day [6.5%]) (P<.01).
Subjects who did not drink coffee daily were more likely to report morning headaches (8.8%) than those who drank at least 1 cup of coffee daily (7.1%) (P<.005).
Morning headaches occurred more frequently in subjects taking hypnotic (13.1% vs 7.5%; P<.01), anxiolytic (20.1% vs 7.3%; P<.001), antidepressant (17.5% vs 7.5%; P<.001), or neuroleptic (17.7% vs 7.6%; P<.05) medications. Tobacco use was not related to morning headaches.
Morning headaches were more frequent in subjects with DSM-IV insomnia disorders than in those without (18.4% vs 6.9%; P<.001). Subjects with a DSM-IV circadian rhythm disorder also reported having morning headaches more often than other subjects (20.0% vs 7.5%; P<.001). Similarly, morning headaches were more common in subjects with a dyssomnia not otherwise specified (17.1% vs 6.9%; P<.001) or with a breathing-related sleep disorder (15.2% vs 7.5%; P<.001) than in the remaining sample.
The prevalence of morning headaches was higher in subjects with a DSM-IV major depressive disorder alone (ie, without comorbid anxiety) than in nondepressed and nonanxious subjects (21.3% vs 5.5%; P<.001). It was also higher in subjects with an anxiety disorder (10.8% vs 5.5%; P<.001). The prevalence was highest in subjects with comorbid anxiety and depressive disorders (28.5% vs 5.5%; P<.001).
The prevalence also varied according to the experienced stress. Subjects who said that their life was very stressful more often reported having morning headaches (9.7%) than those who reported an average level of stress (7.7%) or little or no stress (6.9%; P<.001).
Significant variables were introduced into a multivariate model to determine which variables made an independent contribution to morning headaches. Nonsignificant variables were intake of caffeine and antidepressant or hypnotic medications, heart disease, upper airway disease, and other nonpainful diseases.
Subjects younger than 25 years and those aged between 45 and 54 years, women, the unemployed, and homemakers were significantly more likely to report morning headaches (Table 2). Subjects with a low BMI (<20), a musculoskeletal disease, or hypertension were also more likely to report morning headaches. Similarly, heavy alcohol drinkers (≥6 glasses/day) and subjects taking an anxiolytic medication were more likely to report morning headaches. Among subjects with sleep disorders, those with insomnia diagnoses or dyssomnia not otherwise specified were at higher risk of having morning headaches. Nightmares, circadian rhythm disorders, sleep-related breathing disorders, and loud snoring were also positively related to morning headaches. Among subjects with mental disorders, those with comorbid anxiety and major depressive disorders had the highest risk of having morning headaches, followed by subjects with a major depressive disorder alone, an anxiety disorder alone, and high stress (Table 2).
Another multivariate model was calculated using daily morning headache (vs no headache) as the dependent variable to verify whether the associated factors remained the same. In this model, age and occupation became nonsignificant. Level of stress, anxiety disorders, and loud snoring also became nonsignificant. Hypertension (OR, 1.42) and musculoskeletal diseases (OR, 1.54) remained significantly associated with daily morning headaches. The OR related to daily consumption of 6 or more alcoholic beverages increased to 4.06 and the OR related to the use of anxiolytic medications increased to 2.12. The OR related to insomnia disorder decreased to 1.6 while the ORs of the other sleep disorders increased: to 2.8 for circadian rhythm disorder, to 2.50 for sleep-related breathing disorder, and to 2.61 for dysomnia not otherwise specified. Major depressive disorder alone (OR, 2.06) and comorbid anxiety and depressive disorders (OR, 2.89) remained positively associated with daily morning headaches, but with lower ORs. The OR related to nightmares occurring at least 1 night per week increased to 2.98.
Subjects with morning headaches more frequently reported feeling anxious (12.0% vs 3.8%; P<.001), depressed (16.4% vs 6.0%; P<.001), and inefficient (21.4% vs 8.6%; P<.001) during daytime than the other subjects. They also more frequently reported being irritable (21.7% vs 8.9%; P<.001) than the other subjects. Daytime fatigue was also more frequent in subjects with morning headaches (24.2% vs 9.4%; P<.001), as was being oversensitive to light, touch, and sound (13.9% vs 4.2%; P<.001).
This study is the first to explore the associated factors of morning headaches in the general population using a large sample (N = 18 980). We found a prevalence of 7.6% of subjects who said that they woke up at least sometimes with headaches, and the median duration in which these morning headaches occurred was 42 months.
Our study is not without shortcomings. The primary purpose of these community-based surveys was to investigate sleep and mental disorders. Therefore, full description of the headaches and the localization and intensity of the pain were not assessed. A study that attempted to classify morning headaches into the different categories of headaches (migraine, tension-type, cluster, and cervicogenic headaches) was unable to fit almost half of the patients with morning headaches into one of these categories.2
Risk factors related to morning headaches were examined according to 5 main categories: sociodemographic determinants, use of psychoactive substances, organic diseases, sleep disorders, and mental disorders.
Among sociodemographic determinants, we found that being a woman, being middle-aged, and being unemployed or a homemaker were positively related to morning headaches. A study that examined the frequency of morning headaches in a community-based sample did not find significant differences between men and women, but limited the analysis to subjects with heavy snoring and OSAS.1
We also found a positive association between heavy drinking—at least 6 alcoholic drinks per day—and morning headaches, and between the use of an anxiolytic medication and morning headaches. Subjects using these psychoactive substances were twice as likely to report morning headaches. In the studies that analyzed factors related to morning headaches, these 2 factors were rarely taken into account, although they have well-known depressive effects on the respiratory system. Lack of sufficient oxygenation during sleep may favor the occurrence of headaches upon awakening.
Among organic disorders, we found that subjects with hypertension or musculoskeletal diseases had a higher risk of reporting morning headaches. Changes in blood pressure are likely to cause headaches. Hypertensive headaches do not have specific diagnostic features, but are known to be the most pronounced on awakening in the morning.26 However, as the results showed, the presence of hypertension alone is not sufficient to explain the presence of morning headaches: not all subjects with morning headaches have hypertension, and, conversely, not all subjects with hypertension reported morning headaches.
Morning headaches have been mostly studied in relationship to heavy snoring or OSAS, which limits the airflow during sleep, causing repeated episodes of hypoxia and alterations in blood pressure control mechanisms. These 2 mechanisms can provoke headaches during sleep that may still be present upon awakening. Many believe that morning headaches are specific to sleep breathing disorders. This assertion is based on clinical trials that report reduction of the severity of morning headaches in patients whose OSAS is treated with continuous positive airway pressure.2 However, most of these studies lack adequate control groups and/or assessment of possible confounders to provide strong support to whether morning headaches are specific to OSAS. Our results do not support the uniqueness of this association. We found that OSAS and heavy snoring made an independent contribution to morning headaches, but that they were neither the only nor the strongest predictors of morning headaches. Other studies also reported the nonspecificity of morning headaches to sleep breathing disorders.27,28
In this study, we extended the associations to other diseases and disorders that may influence the occurrence of morning headaches. Among sleep disorders, dyssomnia not otherwise specified, which included restless legs syndrome, periodic limb movement disorder, and sleep disorders with multiple possible causes, had the highest association with morning headaches. The association between morning headaches and insomnia disorders or other sleep disorders involving a lack of sleep is not surprising because sleep deprivation is a well-known cause of headaches.29
We also found that subjects with a major depressive disorder were at high risk of waking up with headaches. This disorder had one of the highest ORs in the multivariate model; however, because our data are cross-sectional, we cannot make causal reference between the morning headaches and depression. Some longitudinal studies also had explored the association between depression and migraine or chronic headaches. For example, Breslau et al14 found that subjects with migraine and those with severe headaches were 3 times more likely to have a major depression in their lifetime and, conversely, that subjects with migraine or severe headache were 2 to 3 times more likely to develop a major depression. Similarly, in a longitudinal study, Pine et al15 reported that adolescents with major depression at the first evaluation had nearly a 10-fold increased risk of developing headaches during the next 7 years. It appears that the association of migraine and depression is a bidirectional relationship, the presence of one increasing the risk of appearance of the other. However, the extent to which waking up with headaches is related to depressive disorders is little documented. It is likely that the relationship is similar to that reported for migraine or severe headache and depression. Morning headaches can be a somatic manifestation of depression, or morning headaches can be a cause of depression.
As we show in Figure 1, almost all subjects with headaches have an organic disease, a sleep disorder, or a mental disorder, but most often several factors are involved. Our results clearly show that it is misleading to relate morning headaches only to sleep-related breathing disorder. An effort should be made to better describe morning headaches in subjects with OSAS before concluding to the specificity of this association.
Recurrent morning headaches in about 80% of cases are related to an identifiable organic, mental, or sleep disorder. Physicians should be aware of the multiplicity of factors that can be involved in the complaint of morning headaches and the necessity of conducting a thorough interview with the patient to identify all possible factors.
Corresponding author: Maurice M. Ohayon, MD, DSc, PhD, Stanford Sleep Epidemiology Research Center, Stanford University School of Medicine, 3430 W Bayshore Rd, Suite 102, Palo Alto, CA 94303 (e-mail: email@example.com).
Accepted for publication February 21, 2003.
This study was supported by an unrestricted educational grant from the Sanofi-Synthelabo Group and by grant 971067 from the Fond de la Recherche en Santé du Québec.