Increased risk of chronic fatigue syndrome in patients with
migraine:
A retrospective cohort study
Chi-Ieong Lau
a,b,c, Che-Chen Lin
d,e, Wei-Hung Chen
a,f, Han-Cheng Wang
a,f,g,
Chia-Hung Kao
h,i,⁎
Introduction
Migraine is a common disorderwith relapsing episodes of moderate to severe headache, nausea, vomiting, and sensory hypersensitivities
that last for hours to days [1]. Chronic fatigue syndrome (CFS), however,
is a less common and less recognized illness characterized by profound, unexplained fatigue formore than 6 months that signifcantly interferes
with daily activities [2].
By defnition, an addition of 4 or more symptoms from a list of 8, including postexertion malaise, unrefreshing sleep, impaired memory, muscle pain, joint pain, sore throat, tender cervical nodes and headaches,
must also be met to fulfll the criteria of CFS [2]. Among these ancillary
symptoms, headache is commonly reported [3]. Nevertheless, the types
of headache in CFS are poorly designated. Thus far, only one study explored the types of headache in CFS and revealed a higher prevalence
of migraine in CFS patients than in controls [4]. Conversely, fatigue is
also common among migraineurs, with one study reporting an increased
prevalence of CFS in chronic migraineurs [5]. The common concurrence of
the 2 disorders leads to the speculation that their pathophysiology may
overlap [4].
Moreover, in addition to headaches contributing to the ancillary criteria of CFS, migraine and various disorders, such as irritable bowel syndrome, fbromyalgia, and temporomandibular joint disorder, are
common comorbidities of CFS [6]. Consequently, these disorders may
be variable manifestations of the same biological processes [6]. Investigating
whether migraineurs are at a higher risk of developing subsequent CFS may therefore provide additional evidence and shed light
on the potential overlapped pathophysiology of this spectrum of disorders.
The current study conducted a nationwide, population-based cohort study by using data fromthe National Health Insurance Research
Databases (NHIRD) of Taiwan to address the question. Methods
Data source
The Taiwan government reorganized 13 government-managed insurance programs into the nationwide, single-payer Taiwan National Health Insurance (NHI) in 1995. The NHI cover rate has reached over 99% of the 23 million residents of Taiwan since 1998. The Taiwan government entrusted the National Health Research Institutes (NHRI) to manage all reimbursement claims data from the NHI and to establish the National Health Insurance Research Databases (NHIRD). All personal identifcation information is encrypted to safeguard patient privacy before being released for research. This study was approved to fulfll the condition for exemption by the Institutional Review Board (IRB) of China Medical University (CMUH-104-REC2-115). The IRB also specifcally waived the consent requirement.
The study population was constructed from a subset of the NHIRD called the Longitudinal Health Insurance Database (LHID). The LHID was built from one million randomly sampled insured individuals
from 1996 to 2000. According to an NHRI report [7], there is no difference
in age or sex distribution between the populations of the LHID and NHIRD. The LHID consists of annual historical claims data of one million people. The claims data includes benefciary registries, outpatient and inpatient records, prescriptions, and records of other medical services. The NHRI created an anonymous identifcation number to link the fles of each insurant because the identifcation information is encrypted.
In the NHIRD, disease records are based on the International Classifcation of Diseases, Ninth Revision, Clinical Modifcation (ICD-9-CM). To survey the disease history of a patient,we collected outpatient and inpatient data from the disease records.
Study population
This study employed a population-based, retrospective cohort design. The migraine cohort consisted of migraine patients newly diagnosed (ICD-9-CM code: 346) from 2006 to 2010 and aged older than 20 years. The index date of the migraine cohort was fxed as the initial migraine diagnosis date. The comparison subjects were individuals without migraine diagnoses selected from the LHID and randomly frequency
matched by age (per 5 years) and sex with 1:4 ratios. Those in the comparison cohort were randomly assigned the same index date as the matched cases. Both cohorts excluded individuals with CFS diagnosed
before the index date. Study follow-up continued until December 31, 2011 or until a subject withdrew from the health insurance program or experienced CFS occurrence.
Outcome and comorbidity assessment
An instructive study outcome was the occurrence of CFS (ICD-9-CM code: 780.71). The study also collected comorbidity histories as a confounding factor. A comorbidity was defned as a participant having a comorbidity diagnosis before the index date. The comorbidities included hypertension (9-CM codes: 401–405), diabetes mellitus (DM; ICD-9-CM code: 250), hyperlipidemia ICD-9-CM code: 272), anxiety (ICD-9-CM codes: 300.0, 300.2, 300.3, 308.3, and 309.81), and depression (ICD-9-CM codes: 296.2–296.3, 300.4, and 311).
Statistical analysis
A chi-squared test was performed to demonstrate the number and proportions of the sex and comorbidity distributions between the migraine and comparison cohorts and to test their differences. A t test was conducted to show the mean and standard deviation (SD) of the age distributions of the 2 cohorts and to test their differences. The cumulative incidence of CFS was calculated by dividing the total number of CFS occurrences by the total sum of follow-up for each cohort by 10,000 per person-years. The cumulative incidence curves were measured using the Kaplan–Meier method. A log-rank test was performed
to assess the difference in incidence curves between the participants with and without migraine. To determine the risk of migraineurs developing CFS, single-variable and multivariable Cox proportional hazards
regression models were used to estimate crude and adjusted hazard ratios (HRs) and 95% confdence intervals (CIs). The sensitivity test of the Cox proportional hazards regression model was used to analyze the effect ofmigraine under the various demographic and comorbidity conditions. In this study, data management and analysiswere performed using
SAS Version 9.3 software (SAS Institute, Cary, NC, USA). The level of signifcance for 2-sided testing was P b .05.
Results
The study population comprised a cohort of 6902 migraine patients and a comparison cohort of 27,608 subjects with similar mean ages
(45.5 years) and the same sex ratio (male: 26.7%) (Table 1). Except for
DM proportion, the comorbidity proportions in the migraine cohort were much higher than those in the comparison cohort.
CFS incidence was 52.72 per 10,000 person-years in the migraine cohort and 28.85 per 10,000 person-years in the comparison cohort (Table 2). The CFS cumulative-incidence curve for the migraine cohort
was also greater than the curve for the comparison cohort (Fig. 1; P b
.0001). After adjustment for age, sex, hypertension, DM, hyperlipidemia, anxiety, and depression, patients with migraine had a 1.50-fold increased risk of CFS compared with subjects without migraine (HR = 1.50, 95% CI= 1.20–1.88).
Table 3 shows the association between the frequency of migraine and CFS risk. There was no difference of CFS risk between the comparison cohort and patients with a lower frequency of medical visits due to migraine (HR=0.92, 95% CI=0.68–1.23). A higher frequency of medical visits due to migraine (average frequency 3–5 and ≥6) was signifcantly associated with an increased risk of CFS. The results also revealed that CFS risk was increased with an increase in frequency of medical visits due to migraine (for trend, P b .0001).
Table 4 shows the results of the sensitivity analysis, which estimated the effect of migraine according to the statuses of various demographic factors and comorbidities. Relative to the subject without migraine, the patient with migraine was signifcantly associated with an increased risk of CFS according to the statuses of various demographic factors and comorbidities, except for the status of young age (b45 years), patients without hypertension, patients with DM, and patients with depression.
However, the results showed that the trend of CFS incidence for patients with migraine was higher than that of the incidence for subjects without
migraine for each demographic factor and comorbidity status. Discussion
Our nationwide, population-based cohort study revealed that
migraine was associatedwith a 1.5-fold increased risk of CFS after adjustment for age, sex, and medical comorbidities. The only study in the literature conducted by Silberstein and colleagues showed a 66.7% prevalence
of CFS in sixty-three chronicmigraine patients [5]. Based on our research,
the present population-based study is the largest to demonstrate an increased risk of CFS in migraine patients.
The major strength of the present study is that we analyzed a large, national dataset containing a representative cohort of one million residents covered by the NHI of Taiwan. The large sample size and the long
Moreover, this study is the frst to examine the association between the 2 disorders by searching for subsequent incidences of CFS in a migraine cohort. In addition, we found a trend showing higher incidence
rates of CFS in migraineurs with migraine-related, health-care-seeking behaviors that were more frequent after adjustment for age, hypertension, diabetes, hyperlipidemia, anxiety, and depression. The increased
risk of CFS in the migraine cohort appeared to be highest in the oldest group (i.e., over 65 years old), which had an adjusted HR of 2.11. Fatigue is common among many disorders, including migraine, but is a vague and subjective symptom that is diffcult to study. CFS is a more specifcally defned disorder, but little is known about its cause and pathophysiology. Despite many proposed hypotheses (from the earlier emphasis on viral and psychiatricmechanisms to the subsequent investigations on neuroendocrine, immunology, brain function and
sleep), the pathophysiology underlying CFS remains elusive [8].
Potential mechanisms
Mitochondria dysfunction in migraine and CFS
An increasing amount of evidence shows thatmitochondrial dysfunction may play a vital role in CFS. Vermeulen et al. showed that low oxygen uptake by muscle cells causes exercise intolerance in CFS, suggesting that an insuffcientmetabolic adaptation to exercise may underlie the mechanism
of CFS [9]. Compared with controls, CFS patients examined using
magnetic resonance spectroscopy (MRS) displayed higher intracellular
lactate levels with exercise [10], abnormal increases in lactate with
minor exercise [11], lower ATP resynthesis rates during recovery
[11],and reduced intracellular ATP inmuscles [12]. These fndings suggest
that mitochondrial dysfunction may underlie the pathophysiology of reduced physical endurance in CFS patients. Similarly, results from MRS studies have generally indicated abnormal mitochondrial function in migraine. Early studies have shown reduced PCr and increased Pi concentrations in migraine during ictal or interictal phases, suggesting a low
availability of free energy in the cell [13,14]. Migraineurswithout aura
(MwoA) also exhibited increased ADP in the occipital lobe, suggesting a
higher metabolic rate in the brain region [15]. In addition, phosphorylation
potential, an index of mitochondrial functionality, was found to be consistently lower in migraineurs than in the control, indicating an
impairment of energy oxidative metabolism in the brain [14].
states in both CFS and migraine. Postexertional malaise in CFS and the development of migrainous attacks may be related to mitochondrial dysfunction. Impaired oxidative phosphorylation resulting inmitochondrial dysfunctionmay therefore underlie themechanistic pathophysiology of both disorders.
Central sensitization of CFS and migraine
A compelling amount of evidence suggests that a migraine brain has
dysregulated cortical excitation–inhibition balance [16]. Neurophysiological
studies on migraine between attacks demonstrated an increased cortical excitability and a lack of habituation of evoked responses, suggesting
an altered interictal corticalmodulation of sensory inputs [17]. Thesefndings
have been supported by animal studies and neuroimaging studies by
using visual [18,19] and other inputs as stimuli [20,21]. Intriguingly, an
altered cortical excitability may also be a vital component in the pathophysiology of CFS. For example, the postexercise cortical excitability of
CFS elicited by transcranial magnetic stimulation (TMS)was signifcantly
reducedwhen comparedwith responses in healthy subjects [22]. Various
functional neuroimaging studies including single photon emission computed
tomography [23], positron emission tomography, near-infrared
spectroscopy [24] and fMRI [25] showed reduced cerebral perfusion as
well as altered cerebral activations in CFS, suggesting that patients with CFS may require increased brain activation to complete complex mental
tasks [26]. In summary, fatigue and postexertional malaise in CFS may
be explained by malfunctioned cerebral excitability.
One plausible mechanism in both migraine and CFS is a shared
underlying pathophysiology of central sensitization [27], which is a
process of amplifed cerebral responsiveness that produces symptoms including all odynia and systemic hyperalgesia. Chronic cortical-spreadingdepression– like depolarization of the cerebral cortex in migraine may
facilitate central sensitization and thus develop the functional and morphological
changes of a migraine brain [28]. Intriguingly, similar patterns
of cerebral gray and white matter abnormalities in CFS appear to support
a shared pathophysiology [29]. For example, MRI with voxel-based morphometry
revealed gray matter volume reduction in patients with CFS
[29]. White matter changes in CFS was found to be correlated with the
severity of fatigue [30]. An increasing amount of evidence suggests a
broader spectrumof overlapping disorders including migraine, CFS, irritable bowel syndrome, fbromyalgia, temporomandibular dysfunction,
posttraumatic stress syndrome, and primary dysmenorrhea through
which central sensitization syndrome may play a role [31,32].
Limitations of the current study
The current study was subject to several limitations. First, the accuracy of the migraine and CFS diagnoses might be a concern. However, general use in Taiwan of the ICHD-2 diagnostic criteria for migraine and the diagnostic criteria published by the Centers for Disease Control and Prevention of the United States for CFS might have improved the accuracy of the coding. The validity of the diagnoses was further improved by the routinely reviewing of NHI medical records to ensure
the eligibility of health insurance claims. Second, it was impossible to elucidate the effects of potentially confounding factors such as lifestyle, habits, body mass index, physical activity, socioeconomic status, and family history of the subjects because such information is not detailed in the NHIRD. Similarly, the records of the NHI claims primarily serve the purpose of administrative billing and do not undergo verifcation for scientifc purposes. We were unable to contact the patients directly to obtain additional information on their use of medications because of the anonymity assured by the identifcation numbers. Consequently, additional large population-based, prospective studies are necessary to replicate our fndings before frm conclusions can be drawn.
Conclusion
Our nationwide, population-based cohort study showed that
migraineurs exhibit an increased risk of developing CFS. Compelling evidence suggests that both disorders may share common pathophysiological
mechanisms such as mitochondrial dysfunction and central
sensitization. The vast spectrumof disorders in the family of central sensitization syndrome is increasingly recognized to represent seemingly
disparate but overlapping disorders. Future research is needed to explore
the impact of CFS in migraine, the prevalence of CFS in various primary headaches as well as their shared genetic and environmental
causes. Understanding the relationship among migraine, CFS and other overlapping disorderswould potentially determine their obscure primary pathophysiological etiologies as well as provide better diagnostic criteria. Identifying unrecognized comorbid conditions in these disorders would offer additional treatment options to achieve more satisfactory overall outcomes.
