Modifiable Risk Factors for Migraine Progression


  • Marcelo E. Bigal MD, PhD,

  • Richard B. Lipton MD

  • From the Department of Neurology, Albert Einstein College of Medicine, Bronx, NY, (Drs. Bigal and Lipton); Epidemiology and Population Health, Albert Einstein College of Medicine, Bronx, NY (Dr. Lipton); The Montefiore Headache Center, Bronx, NY (Drs. Bigal and Lipton); and The New England Center for Headache, Stamford, CT (Dr. Bigal).

Address all correspondence to Dr. Marcelo E. Bigal, Department of Neurology, Albert Einstein College of Medicine, 1165 Morris Park Avenue, Bronx, NY 10461.


Migraine is a chronic-recurrent disorder that progresses in some individuals. Transformed migraine is the result of this progression. Since migraine does not progress in most patients, identifying the risk factors for progression has emerged as a very important public health priority. If risk factors can be identified, that might provide a foundation for more aggressive preventive intervention. Risk factors for progression may be divided into non-remediable (gender, age, race) and remediable categories. In this paper, we focus on several already identified remediable risk factors, including frequency of migraine attacks, obesity, acute medication overuse, caffeine overuse, stressful life events, depression, and sleep disorders. We present the evidence for each risk factor and discuss possible interventions to address them.

Recent evidence suggests that a subgroup of migraineurs may have a clinically progressive disorder characterized by an ongoing increase in migraine attack frequency, sometimes leading to nearly constant pain.1 This state is often referred to as transformed or chronic migraine (referenced herein simply as TM), a subtype of the chronic daily headaches (CDHs).2

TM has been the subject of numerous clinical and population-based studies. Though not included in the International Classification of Headache Disorders (ICHD-2), we focus on it here because of the nature of the available data. In a longitudinal epidemiologic study, the Frequent Headache Epidemiology (FrHE) study, Scher et al showed that over the course of 1 year, 3% of individuals with episodic headache in the general population progressed to CDH.3 The authors concluded that the incidence of CDH in subjects with episodic headache is 3% per year.3–5 In a second study, Katsarava et al followed-up 532 consecutive patients with episodic migraine (<15 days/month) recruited from subspecialty care. Sixty-four patients (14%) developed CDH over 1 year of follow-up.6 Differences in the incidence of CDH are explained by patient source and baseline frequency.

Although migraine progresses in some individuals, it does not progress in most. Migraine is best understood as a chronic disorder with episodic manifestations, progressive in some individuals. Consistently, identifying risk factors for progression has emerged as a very important public health priority. If risk factors can be identified, that might provide a foundation for more aggressive preventive intervention.

Though available evidence is limited, risk factors for migraine chronification can be divided into non-modifiable and modifiable categories, based on the prospects for addressing them with behavioral or medical intervention (Table 1). Examples of non-modifiable risk factors include age, gender, and socioeconomic status (SES). In the FrHE study, the prevalence of CDH decreased slightly with age and increased in women and in those of low SES.4

Table 1.—. Risk Factors for Migraine Progression
Not Readily ModifiableModifiableOther Candidates
AgeAttack frequencyAllodynia
Low education/ObesityPro-inflammatory
 socioeconomicMedication overuse states
 statusStressful life eventsOther pain syndromes
Head injuryCaffeine overusePro-thrombotic states

In this review, we focus on the modifiable risk factors, which are potential targets for interventions aiming to modify the natural history of migraine, and therefore preventing chronification. We first briefly discuss potential mechanisms of migraine progression. We then present the limited evidence supporting some of the modifiable risk factors, as well as potential interventions to address them. It must be emphasized that the topic of this article is largely speculative, and no solid evidence exists to support the concept that addressing risk factors actually avoids disease progression. Ultimately, the notion of disease modification in migraine must be proven in well designed clinical trials.


Migraine probably results from dysfunction of brainstem involved in the modulation of craniovascular afferents.7,8 Brainstem activation may also lead to activation of ascending and descending pathways, with initiation of a perimeningeal vasodilatation and neurogenic inflammation. The pain is understood as a combination of altered perception (due to peripheral or central sensitization) of stimuli that are usually not painful, as well as the activation of a feed-forward neurovascular dilator mechanism in the first (ophthalmic) division of the trigeminal nerve. Cortical spreading depression is the presumed substrate of migraine aura; spreading depression may also occur in migraine without aura.

Burstein and Jakubowski showed that around 75% of migraineurs develop central sensitization (sensitization of the second order trigeminal neuron, which is clinically manifested by the development of cutaneous allodynia) during the course of a migraine attack.9,10 Central sensitization appears to be associated with triptan refractoriness. However, central sensitization may also play a role in the progression of the disease itself, although this is speculative at this point. Welch and colleagues, in an imaging study, showed that iron deposition occurs in the periaqueductal gray (PAG) area in subjects with chronic headaches.11 The PAG area is related to descending analgesic network and is important in controlling pain and providing endogenous analgesia. It is closely related to the trigeminal nucleus caudalis and with areas that are eventually important in migraine generation or modulation. In this study, the iron levels were increased in migraineurs, compared with controls, and in CDH headache patients, compared with migraineurs. These findings may be directly attributable to iron-catalyzed, free-radical cell damage. The authors suggested that iron deposition may reflect progressive neuronal damage related to recurrent migraine attacks. Consistently, it may be hypothesized that repetitive central sensitization of the trigeminal neurons correlate with iron deposition in the PAG area and, therefore, migraine attacks predispose to disease progression.


Frequency of Headache Attacks.— In the FrHE study, individuals with episodic headache were followed-up over the course of 1 year. The baseline characteristics of those who developed CDH were compared with the characteristics of those who did not. One of the most important risk factors for progression was the frequency of headache attacks at baseline. The risk of new-onset CDH increased in a nonlinear manner with baseline headache frequency (Fig. 1); elevated risk for developing CDH occurred in subjects who experienced 3 or more headaches per month. The risk exponentially increased with attack frequency.

Figure 1.—.

Risk factor for chronic daily headaches according to the baseline headache frequency.

Although there is some circularity in reasoning when interpreting these results, since that, compared to individuals with low frequency of attacks, migraineurs with high frequency are obviously at a higher chance of developing CDH, this observation also agrees with the hypothesis that repetitive episodes of pain may lead to central sensitization and generation of free radicals and anatomical changes to the PAG area.10,11

Intervention.— If headache frequency is a strong risk factor for migraine progression, interventions should focus on decreasing attack frequency. This can be achieved by using pharmacological and nonpharmacological strategies (for a review, see reference 12). It also suggests that early effective acute treatment, prior to the development of sensitization, might reduce the risk of CDH.

Recently, as a part of a large epidemiological study focusing on patterns of prevention in the United States (the AMPP study), headache specialists operationalized criteria for preventive treatment.13 Recommendations were based on the number of headache days per month experienced by migraineurs, as well as the level of attack-related impairment caused by the headaches. Impairment during a headache attack was divided into severe impairment (unable to function or requiring bed rest), some impairment (able to function, but with reduced performance), and no impairment (able to function normally). According to expert consensus, 3 groups were identified, based on their need of preventive treatment.13

  • 1Prevention should be offered: Migraineurs reporting either 6 or more headache days per month; or 4 or more headache days with at least some impairment; or 3 or more headache days with severe impairment or requiring bed rest.
  • 2Prevention should be considered: Migraineurs with 4 or 5 migraine days per month with normal functioning; or 3 migraine days with some impairment; or 2 migraine days with severe impairment.
  • 3Prevention is not indicated: Migraineurs with less than 4 headache days per month and no impairment; subjects with no more than 1 headache day per month regardless of the impairment.

Evidence regarding the preventive drugs is most common to the β-blockers, antidepressants, calcium channel antagonists, and anti-epileptic agents.14 Most of these effective agents were discovered serendipitously after use for other purposes, and still represent the majority of prescriptions written for migraine prevention. Most of these medications described herein are associated with >50% efficacy in >50% of the subjects. Selected preventive medications and their target doses are described in Table 2.

Table 2.—. Choices of Preventive Treatment in Migraine
DrugEfficacyAdverse EventsComorbid Condition
Relative ContraindicationRelative Indication
  1. Ratings are on a scale from 1+ (lowest) to 4+ (highest) based on strength of evidence. From the Headache Consortium Guidelines.

β-Blockers4+2+Asthma, depression, congestive heart Raynaud's disease, diabetesHypertension, angina
 Methysergide4+4+Angina, vascular diseaseOrthostatic hypotension
Ca channel blockers
 Verapamil2+1+Constipation, hypotensionAura, hypertension, angina, asthma
 Flunarizine4+2+Parkinson's, depressionDizziness, vertigo
 TCAs4+2+Mania, urinal retention, heart blockDepression, anxiety, insomnia, pain
 SSRIs or SNRIs2+1+ManiaDepression, anxiety
 MAOIs4+4+Unreliable patientRefractory depression, refractory panic disorder
 Divalproex/Valproate4+2+Liver disease, bleeding disordersMania, epilepsy
 Gabapentin2+2+Liver disease, bleeding disordersEpilepsy, neuropathic pain
 Topiramate4+2+Kidney stonesObesity, epilepsy
 NSAIDs2+2+Ulcer disease, gastritisArthritis, other pain disorders

Obesity.— Obesity is comorbid with a number of chronic pain syndromes.15 In fibromyalgia, obesity is associated with increased pain severity and refractoriness to treatment; weight loss is followed by improved physical functioning.16 Obesity is also associated with arthritis and with back and neck pain in the population.16

In addition, obesity is a risk factor for CDH. In the FrHE study, the relative odds of CDH were 5 times higher in obese individuals (BMI ≥ 30) than in the normal weighted.4 Overweight individuals (BMI [25 to 29]) had a 3-fold increased risk of developing CDH.

We recently conducted 2 additional studies to assess the relationship between obesity and migraine progression. In the first study, we focused on the relationship of BMI to episodic migraine.17 We interviewed 30,215 participants and identified migraineurs. Subjects with 15 or more headache days per month were excluded. BMI was not associated with the prevalence of migraine, but was associated with the frequency of headache attacks. In the normal weighted group, just 4.4% of migraineurs had 10 to 15 headache days per month. This increased to 5.8% of the over weight group (odds ratio [OR]= 1.3), 13.6% of the obese (BMI [30 to 34], OR = 2.9), and 20.7% of the morbidly obese (BMI ≥ 35, OR = 5.7) (Fig. 2). The proportion of subjects with severe headache pain also significantly increased with BMI, doubling in the morbidly obese relative to the normal weighted (OR = 1.9).17

Figure 2.—.

Proportion of migraineurs with 10 or more headache days per month according to the body mass index.

In the second study, we assessed the relationship of BMI to CDH, confirming the association with obesity.18 We also showed that obesity is a much stronger risk factor for TM than for chronic tension-type headache (CTTH). For TM, the prevalence ranged from 0.9% of the normal weighted, to 1.2% of the overweight (OR = 1.4 [1.1 to 1.8]), 1.6% of the obese (OR = 1.7 [1.2 to 2.43]), and 2.5% of the morbidly obese (OR = 2.2 [1.5 to 3.2]). The effects of the BMI on the prevalence of CTTH were not significant, except in the morbidly obese group. Adjusted analyses (accounting for the use of daily medications, depression, sleep disorders, gender, age, and other demographic features) showed that obesity was associated with CDH and TM but not CTTH (Fig. 3).18

Figure 3.—.

Prevalence of transformed migraine according to the body mass index.

The mechanisms of the inter-relationships of headache frequency and obesity are complex and not yet understood. Attack frequency, per se, is a risk factor for chronification.4 Furthermore, as discussed above, the frequency and severity of attacks appear to be associated with the risk of sensitization at the level of the trigeminal caudalis.10,11 In addition, obesity is recognized as a pro-inflammatory and pro-thrombotic state. Adipocytes secrete a variety of cytokines, including interleukin 6 (IL-6) and tumor necrosis factor-α (TNF-α). Markers of inflammation, including leukocyte count, TNF-α, IL-6, and C-reactive protein, are also increased in obesity.19 Moreover, obesity is associated with an increase in adipose tissue macrophages, which also participate in the inflammatory process through the elaboration of cytokines.20 Plasma calcitonin gene-related peptide (CGRP) levels are also elevated in obese individuals, particularly in women, and fat intake may be associated with increased CGRP secretion.21 Finally, recent data suggest that hypothalamic neuropeptides orexin A and orexin B play a role in nociception, and stimulate the prejunctional release of CGRP from trigeminal neurons.22,23 It is also demonstrated that orexin A is important in the regulation of energy metabolism in humans, and that in obesity the activity of these peptides is disturbed.24 It may be speculated that the dysmodulation in the orexin pathways may be associated with increased susceptibility to neurogenic inflammation and consequent migraine attacks.

Intervention.— Although evidence suggests that obesity is a modifiable risk factor for progression, it is unknown if weight loss is related to decrease in headache frequency. Based on the data, migraineurs should be motivated to maintain the weight (if normal weight) or decrease it (if overweight or obese). This goal requires diets and behavioral support. Orlistat and sibutramine are approved by the FDA for the long-term treatment of obesity.25 Both low carbohydrate and low fat diets promote weight loss.

Since obesity and headache frequency are related, the use of migraine preventive medications that are weight neutral or that are associated with weight loss in those who are overweight or obese may be recommended. Of the FDA-approved migraine medications, only topiramate has been shown to promote weight loss (Fig. 4).26 Beta-blockers and calcium channel blockers are generally weight neutral. Divalproex sodium and tricyclic antidepressants promote weight gain.

Figure 4.—.

Weight loss in participants of 2 clinical trials of topiramate in the prevention of migraine.

Medication Overuse.— Medication overuse occurs in 80% of TM patients in specialty care,27,28 but in almost one-third of TM patients in the population.

It is not entirely clear if medication overuse is a cause or consequence of TM. Medication rebound headaches (worsening of the pain pattern after discontinuation of the offending analgesic) is widely observed in clinical practice but not demonstrated in placebo-controlled trials. Withdrawal headache has been shown in a controlled trial of caffeine discontinuation.29 Clinical observation suggests that both preventive and acute treatment for the primary headache usually fail unless the offending medication or medications are terminated.28

Two studies significantly helped in our current understanding of medication. First, Wilkinson et al looked for CDH in 28 patients who underwent total colectomy for ulcerative colitis.30 All patients with a previous history of migraine who overused opiates developed CDH (19%), whereas the patients without migraine did not. In a second study, Bahra et al showed that when nonsteroidal anti-inflammatory drugs (NSAIDs) are used daily for a variety of medical conditions, they did not induce CDH in subjects without pre-existing primary headache disorders. However, in those with pre-existing migraine, NSAIDs were a strong risk factor for CDH.31

Both studies established 2 principles of medication overuse as a risk factor for CDH: 1—Even when the overused medication is used for reasons other than headache, it may still be associated with the development of CDH; 2—Acute medication overuse induces CDH just in those biologically predisposed to it (ie, those with pre-existing episodic migraine).

Favoring the hypothesis that medication overuse is a consequence instead of a cause of CDH, it is clear that some patients with CDH do not improve after medication withdrawal.32 Furthermore, some subjects clearly report that the increase in headache frequency preceded the overuse of medication. Third, population-based and clinical studies show that high frequency of attacks is, per se, risk factor for CDH and consequent overuse of medication.3 Finally, in the population, medication overuse is not the most important risk factor for CDH.5 However, in the clinic-based longitudinal study of Katsarava et al, attack frequency and medication overuse were independent risk factors for the development of CDH.6

The controversy about analgesic overuse as a cause or consequence of migraine chronification is far from over and these viewpoints may not be mutually exclusive. It is possible and very likely that while in some individuals medication overuse was the risk factor for migraine chronification, in others it was merely a response to an already refractory headache syndrome. In either regard, the clinical take-home message is that in order for improvement to occur, medication must be withdrawn.

Intervention.— Medication overuse must be prevented. From a theoretical perspective, this goal may be achieved by:

1—Effectively treating migraine attacks before central sensitization is developed. As previously discussed, repetitive episodes of central sensitization may cause permanent neuronal damage in areas that modulate pain.10,11 Therefore, it can be argued that if central sensitization acute medication is taken early, central sensitization is less likely to occur, reducing the chance of escalation in migraine frequency.

2—Decreasing the frequency of headache attacks, which is the major goal of nonpharmacologic therapies, as well as preventive medications. Subjects being effectively treated with preventive medications will have less frequent migraine attacks and consequently will use acute medications less frequently, decreasing the chance of medication overuse.

In subjects already overusing medication, most studies suggest the benefit and necessity of detoxifying, followed by an intensive, long-term preventive treatment plan. If patients discontinue their overused medications, they frequently improve considerably, and if they do not, they are usually difficult to treat effectively.28,33,34

Basically, there are 3 outpatient approaches to detoxification. One approach is to taper the overused medication gradually while an effective preventive therapy is established. The second strategy is to abruptly discontinue the overused drug, institute a transitional medication to break the cycle of headaches, and subsequently taper the transitional medication. The third approach is to combine the 2 strategies by eliminating the rebound medication rapidly, adding a preventive medication rapidly, as also supplying a temporary bridge, to give the patient the maximum chance to improve without drastically worsening first.35

No matter what medication is being tapered, a very useful technique is to use a 3 to 7 day taper of oral steroids, either prednisone starting at 60 mg/d, dexamethasone starting at 4 to 12 mg/d, or methylprednisolone. The mechanism of action is unknown, but presumed to be related to decreased neurogenic inflammation in the meninges. Alternatively, triptans, DHE, tizanidine, naproxen sodium, and botulinum toxin may be used as bridge therapies.36–40 Some medications (especially those that include barbiturates and opioids) should not be discontinued abruptly in the outpatient setting. For barbiturates, a useful approach is to switch butalbital for long-acting agents, such as Phenobarbital.

Caffeine.— Caffeine is the only substance shown to cause withdrawal headache under placebo-controlled double-blind conditions.29 Dietary caffeine exposure to coffee, tea, chocolates, and soft drinks is remarkably common. Caffeine-withdrawal headache has been reported in individuals who consume the equivalent of about 1 cup of coffee/day.41

A population-based study investigated current and past dietary caffeine consumption and medication use for headache.42 High caffeine exposure was defined as being in the upper quartile of dietary consumption or using a caffeine-containing over-the-counter analgesic as the preferred headache treatment. Compared with episodic headache controls, CDH cases were more likely overall to have been high caffeine consumers before the onset of CDH (OR = 1.50, P= .05). No association was found for current caffeine consumption (ie, post CDH) (OR = 1.36, P= .12). In secondary analyses, associations were confined to younger (age < 40) women (OR = 2.0, P= .02) and those with chronic episodic (as opposed to chronic continuous) headaches (OR = 1.69, P= .01), without physician consultation (OR = 1.67, P= .04) and of recent (<2 years) onset (OR = 1.67, P= .03).

Intervention.— Some patients are very sensitive to caffeine and can develop withdrawal symptoms when using as little as 1 to 2 cups per day. It is important to taper caffeine slowly, over the course of several weeks. For example, a patient should decrease the caffeine intake no faster than 5 ounces of coffee every 3 to 5 days. Generally, patients may state that they are having 2 cups per day when they actually mean mugs containing 8 to 16 ounces of coffee. It is important to identify all the caffeine a patient is consuming, whether it be in coffee, tea, or soft drink or medication.

Snoring/Sleep Apnea.— Case studies of specialty patients and 2 population-based studies suggest that headache is associated with sleep disorders in general, including sleep-disordered breathing.43,44 Sleep-disordered breathing has also been strongly associated with cluster headache.45

Scher et al compared the prevalence of snoring in a group of CDH subjects (n = 206) with a control group of episodic headache subjects (n = 507). Overall, CDH subjects were more likely to be habitual (daily) snorers than control subjects (OR = 2.02 [1.2 to 3.3, P < .005]). The OR for habitual snoring remained elevated after adjustment for factors associated with sleep-disordered breathing (age, gender, marital status, diagnosed hypertension, alcohol intake, and BMI; OR = 2.86 [1.7 to 5.0, P < .005]). Further adjustment for current depression or caffeine intake attenuated the OR for habitual snoring. Adjustment for opioid analgesic use and educational level did not significantly modify the OR.46

The main limitation of this study was that snoring was self-reported. However, the accuracy of self-reported snoring has been validated in few population-based studies. In 2 such studies, the positive predictive value of self-reported snoring was 50% to 80%.47

Interventions.— Snoring emerges as an independent risk factor for CDH, though many individuals who snore are also obese. Interventions require encouraging patients to lose weight and to stop smoking (smoking worsens snoring). Subjects should omit sedative drugs and reduce alcohol intake. Pillows or specific products (modern variants of the “tennis ball in a sock sewn to the back of a pyjama shirt”) are effective for some people.48

From a pharmacological perspective, nasal congestion should be treated with decongestant and steroid nasal sprays. Surgical procedures include uvulopalatopharyngoplasty and radiofrequency stiffening of the soft palate. Finally, if a sleep study shows obstructive sleep apnea syndrome, treatment may involve continuous positive airway pressure. Mandibular advancement devices can help snoring and mild obstructive sleep apnea; this is best assessed by a dentist or an oral surgeon.48

Depression and Stressful Life Events.— Cross-sectional associations and bi-directional associations between migraine and a variety of psychiatric and somatic conditions have been reported.49–51 In a recent study, overall prevalence of major depression was 28.1% for migraine, 19.5% for probable migraine, 23.9% for migraine and probable migraine pooled together, and 10.3% for the control group.52 The prevalence of major depression was elevated in all migraine groups as compared to controls, for both crude and adjusted (by age, sex, education) analyses. These findings confirm previous studies by Breslau et al and other authors.49 Finally, Scher et al found that recent history of stressful life events is an independent risk factor for CDH.3

Interventions.— Since depression is at the same time comorbid with TM and also a risk factor for migraine progression, subjects with migraine should be actively screened for depression, which should be managed with nonpharmacological or pharmacological techniques (or both).

Among the nonpharmacological techniques, cognitive-behavioral therapy focuses on the cognitive and affective components of headache and is typically administered in conjunction with relaxation training.53 Biofeedback refers to any procedure that feeds back biologic information about physiological processes (usually through the use of electronic instrumentation).54,55 Interventions alert patients to the role their thoughts play in generating stress responses and to relationships between stress-coping efforts and headaches. Meta-analytic literature reviews of these behavioral interventions have consistently identified clinically significant reductions in recurrent headaches. Across studies, behavioral interventions have yielded approximately 35% to 50% reduction in migraine and tension-type headache activity.56 A recent study aiming to determine whether combined treatment using medication and biofeedback would be more effective than drug treatment alone for treating chronic migraine with analgesic overuse concluded that the combination is more effective than drug therapy.57

The same rationale that we used to propose that obese migraineurs should be treated with drugs that do not increase weight can be used herein. Subjects with migraine and depression should be treated preferentially with antidepressants, which may also help anxiety and sleep problems. As a cautionary note, it is important to emphasize that the antidepressant doses are usually higher than the antimigraineur doses. The target dose should be therapeutic to both conditions.


At present, TM has been shown to develop as a consequence of migraine progression. A number of remediable risk factors have been identified and a number of other potential risk factors have been suggested. While awaiting additional data, it is worthwhile to consider risk factor modification as part of migraine management. Clinicians should aspire to not just relieve current pain and disability, but to avoid migraine progression. Reducing attack frequency, avoiding medication overuse, appropriately using preventive drugs and behavioral therapies, and encouraging weight loss should be part of migraine therapy, to improve our patients' lives today and prevent progression.

Conflict of Interest:  Dr. Bigal and Dr. Lipton are on the advisory board and/or received research support and/or are on the speaker bureau of OMP, AZ, GSK, Pfizer, Merck, Pozen, Endo, among other pharmaceutical companies.