Clinical Aspects of Medication Overuse Headaches


  • Arnaldo Neves Da Silva MD,

    Corresponding author
    1. MHNI, 3120 Professional Drive, Ann Arbor, MI, USA
    • Address all correspondence to A.N. Da Silva, Cleveland Clinic, Neurology, 26500 Amhearst Circle, Ap. 203, Beachwood, OH 44122, USA.

    Search for more papers by this author
  • Alvin E. Lake III PhD

    1. MHNI, 3120 Professional Drive, Ann Arbor, MI, USA
    Search for more papers by this author

  • Conflict of Interest: None.


Medication overuse headache (MOH) is a subset of chronic daily headache, occurring from overuse of 1 or more classes of migraine abortive medication. Acetaminophen, combination analgesics (caffeine combinations), opioids, barbiturates (butalbital), non-steroidal anti-inflammatory drugs, and triptans are the main classes of drugs implicated in the genesis of MOH. Migraine seems to be the most common diagnosis leading to MOH. The development of MOH is associated with both frequency of use of medication and behavioral predispositions. MOH is not a unitary concept. The distinction between simple (type 1) vs complex (type 2) forms is based on both the class of overused medication and behavioral factors, including psychopathology and psychological drug dependence.

MOH is a challenging disorder causing decline in the quality of life and causing physical symptoms, such as daily and incapacitating headaches, insomnia, and non-restorative sleep, as well as psychological distress and reduced functioning. MOH is associated with biochemical, structural, and functional brain changes. Relapse after detoxification is a challenge, but can be addressed if the patient is followed over a prolonged period of time with a combination of prophylactic pharmacotherapy, use of abortive medication with minimal risk of MOH, withholding previously overused medication, and providing psychological (cognitive-behavioral) therapy.

Medication overuse headache (MOH), previously called rebound headache, drug-induced headache, and medication-misuse headache, is a subset of chronic daily headache (CDH), occurring on 15 days or more per month, 4 or more hours per day, for 3 or more months (Table 1), with the added feature of overuse of at least 1 class of abortive drug.[1] It is important to note that in the International Classification of Headache Disorders, 2nd Edition revised,[1] MOH diagnosis no longer requires the improvement of headaches after withdrawal of the overused medication. In clinical practice, the common scenario is a patient with episodic migraine (EM) that transforms to chronic migraine (CM) in the setting of overusing 1 or more classes of abortive drugs. There is an increase in frequency and intensity of headache attacks and enhanced sensitivity to stimuli that would trigger these attacks.[2]

Table 1. International Classification of Headache Disorders-2nd Edition (ICHD-2) revised[1] Criteria for Medication Overuse Headache
8.2 Medication Overuse Headache 
Diagnostic Criteria 
AHeadache present on ≥15 days/month in a patient with a preexisting headache disorder
BRegular overuse for >3 months of 1 or more acute/symptomatic treatment drugs
CNot better accounted for by another ICHD-2 diagnosis

Since 2010, the US Food and Drug Administration in the prescribing information for onabotulinumtoxinA approved a definition of chronic migraine as a headache occurring 15 or more days per month, for 4 or more hours per day. This definition covers both primary and secondary headaches, and includes CM, CDH, as well as MOH.[3]

Prevalence and Incidence

CDH has a prevalence of 4-5%, and the annual average incidence of new-onset CM in patients with EM is 2.5%.3-6 Pediatric CM prevalence in the United States is 1.7%.[7]

MOH prevalence is estimated in about 1-2% of the general population[8] and is overwhelmingly more prevalent in women than in men.9-11 An epidemiological study in Taiwan of adolescents between the ages of 12 and 14 (N = 7900) found a prevalence of 1.5% for CDH. MOH was present in 20% of the CDH group, representing 0.3% of the study population.[12]

In specialized headache centers, the prevalence of MOH can be as high as 70% among referred patients,[13] and if its high socioeconomic impact is taken into account (work absenteeism, recurrent emergency room visits, hospital admissions, and unnecessary diagnostic tests), MOH is likely to be one of the most if not the most costly neurological disorder known.[14]

Piazza et al, studying a group of children and adolescents coming for treatment at a tertiary care center in Italy (n = 118), found MOH in 9.3% of patients. When only the subset of patients with CDH was analyzed, the incidence of MOH increased to 20.8%, showing that MOH is prevalent in children and adolescents.[15]

The MOH population, if compared with patients with EM, are more likely to be women, have lower education level, married, unemployed, have migraine remission during pregnancy, be menopausal, constipated, do not use oral contraceptives, have a higher utilization of health care resources, and be on polypharmacy, especially sedative-hypnotics and antihypertensive medications.[16]

The general quality of life of MOH patients is worse than patients with episodic headaches, as measured by the General Health Questionaire-28.17 Also, the results from the quality of life short form-36 (SF-36) health survey revealed a decreased score in all health-related domains for patient with MOH compared with healthy individuals, with highest differences for generalized body pain and physical activity.[10] The Migraine Disability Assessment Score (MIDAS) is about threefold higher in MOH patients than in patients with EM and is comparable to those with other forms of CDH.[18] Current opioid users with probable dependence have MIDAS scores over twice as high as occasional users, with significantly higher rates of emergent care than non-users.[19]

A meta-analysis of MOH series revealed that the most frequent headache diagnoses at onset are the following: migraine in 65%, tension-type headaches in 27%, and mixed or other headaches in 8%.[20] Other publications also endorse migraine as the most common diagnosis leading to MOH.[10, 11, 18, 21] Migraine diagnosis may be associated with a better prognosis among other episodic primary headache diagnoses once they chronify to CDH.[22] It also appears that migraine starts earlier in the life of patients with MOH than in those with EM.[16] Cluster headache with coexisting migraine disorder can also be susceptible to developing MOH, and triptans along with opioids are the most common offending drugs.[23, 24] This is a situation in which the migraine is presumed to transform to CDH, while the cluster headaches themselves do not.

Mechanisms of MOH

Understanding of the mechanisms of MOH continues to unfold, and these mechanisms may include a combination of pronociceptive pain facilitation with weakened descending pain inhibition. Mechanisms may also differ from 1 class of overused medication to another. Opioid-related MOH may be related to opioid-induced hyperalgesia, resulting from recurrent headaches activating nociceptive pathways, interacting with pain facilitation due to glial activation, and creating a pro-inflammatory state.[25] Prolonged sumatriptan exposure causes elevations in calcitonin gene related peptide (CGRP) levels in blood.[26] Overuse of triptans may induce a state of latent sensitization, characterized by persistent pronociceptive neural adaptations in dural afferents and enhanced susceptibility to migraine triggers.[27] Overuse of acetaminophen (paracetamol) may alter cortical excitability, leading to an increased susceptibility of cortical spreading depression and facilitation of trigeminal nociception.[28]

Compared with healthy controls, patients with MOH have a significant increase of gray matter volume in the periaqueductal gray of the midbrain, thalamus, and ventral striatum, as well as a significant decrease in gray matter volume in the frontal regions, including the orbitofrontal cortex, anterior cingulate, the insula, and precuneus. These findings are consistent with dysfunction of antinociceptive systems in MOH, influenced by anxiety.[29] Some patients may be more genetically susceptible to the development of MOH.[30] Comorbidity between MOH and substance use disorders also suggests possible genetic linkages.[31]

Clinical Features of MOH

Clinically, over time, many patients experience increased frequency of migraine attacks, and this may be due to life stressors and/or unintentional use of triggers. The overuse of medications becomes consistent, and is anticipated with increased refractoriness and subsequent escalation of the number of headache days.

Bigal et al, in a seminal paper reporting longitudinal population-based research (the American Migraine Prevalence and Prevention study, or AMPP), found that butalbital combinations predict the 1-year transformation of EM to MOH if used only 5 or more days per month, opioids if taken 8 days or more days per month, and triptans if taken 10 or more days per month. Non-steroidal anti-inflammatory drugs (NSAIDs) if used 5 or fewer days per month seem to be protective against MOH, but can cause MOH if taken 10 or more days per month.[5] A population-based sample (the Head-HUNT study) found that weekly reliance on analgesics at baseline predicted CM 11 years later, with a relative risk (RR) of 13.3 (confidence interval [CI] = 9.3, 19.1). Analgesic overuse at baseline predicted continued to CDH at follow-up (RR = 19.6, CI = 14.8, 25.9).[32]

Barbiturates for a short period of time were the leading drugs causing MOH in the world but were removed from the market in European countries, although they are still available in the United States.[11]

Triptans currently are the most common MOH producers in high-income countries.[5] Triptan overuse appears to cause MOH faster and at lower doses than other drugs.[2] Limmroth et al demonstrated that triptan use can cause MOH after only 1.7 years of use, while analgesics average 4.8 years of use to precipitate MOH.[33] Triptans also showed the lowest monthly intake frequency necessary to cause MOH (18 single doses per month) and shorter duration of withdrawal headaches during detoxification, while analgesics had the highest monthly intake frequency (114 doses per month) and longer duration of withdrawal headaches.[34, 35]

A prospective 11-year longitudinal study reported by Hagen et al identified several behavioral risk factors for MOH that were not risk factors for CDH without medication overuse, as shown in Table 2.[36] Of a sample of 25,596 individuals who did not have CDH at baseline and were recontacted 11 years later, 0.8% had developed MOH, and 1.0% had CDH without medication overuse.

Table 2. Behavioral Risk Factors for Medication Overuse Headache
Risk FactorOdds Ratio95% Confidence Interval
  1. From Hagen et al.
Regular use of tranquilizers5.23.0-9.0
Anxiety/depression (Hospital Anxiety and Depression Scale >10) with musculoskeletal and gastrointestinal complaints4.72.4-9.0
Analgesic use for any condition3.02.0-4.5
Inactivity vs high activity2.71.2-6.3
Smoking daily1.81.2-2.5

It is worth emphasizing that for reasons not fully understood, a patient without migraine genotype who uses daily opioids for other types of chronic pain may develop drug dependency and can develop MOH, but not always. This was found by Bahra et al, who observed that in patients using daily analgesics (62.5% used opiates) for rheumatologic conditions, those with EM developed MOH at a significantly higher rate than those without headache.[37] Wilkinson et al, studying patients who underwent total colectomy for ulcerative colitis and were using daily opioids to control bowel movements, found that only patients with EM developed MOH.[38]

Pain location in MOH varies and frequently differs from the original pain location when migraines were episodic. Patients tend to describe different types of headaches, such as on the forehead, temples, back of the head, unilateral, and bilateral, and they think that they have different types of headaches and seek explanations for this variability.[3] Neck pain is found in the vast majority of MOH patients, leading to an incorrect diagnosis of cervicogenic headache. Therefore, patients may be submitted to unnecessary and costly neck interventions that are frequently ineffective.

Expansion of the headache area and cutaneous allodynia may imply sensitization of central nociceptive neurons in the trigeminal pathway in addition to cells of the periaqueductal gray.[39] Repetitive activation of the trigeminal nerve can lead to functional changes in neurons at the trigeminal nucleus caudalis, characterized by a decrease in nociceptive threshold and expansion of the receptive field.[39, 40]

Headaches may be more frequent in the morning secondary to nocturnal withdrawal or to a non-restorative sleep also related to drug withdrawal, but perhaps more due to increased caffeine consumption (combination analgesics usually contain caffeine).[41]

As well emphasized by the Teppers, it is not the quality of headaches but rather the quantity that makes an MOH diagnosis.[3, 41] Some patients erroneously assume they can distinguish between features of a “rebound” headache and their typical migraine, failing to recognize that an increasing frequency of headaches correlated with increasing analgesic/abortive use is a red flag for MOH.

Refractoriness to preventive and abortive medications in the setting of MOH is frequently seen.[42] A post hoc analysis of the regulatory trials of onabotulinumtoxinA suggested that its use in patients with MOH is beneficial even before the discontinuation of the overused drug, although the trial excluded patients with continuous headache and discouraged inclusion of opioid users.[43] Topiramate was also shown to reduce the number of headache days in patients not undergoing detoxification in 1 of 2 randomized controlled trials for CM.[44] In our opinion, these trials offer insufficient data against preemptive detoxification from the offending drugs.

MOH patients frequently have a long list of medications that were tried without success, and many of those drugs were used for insufficient time and in doses not effective for migraine prevention or treatment. In addition, the preventive trials were almost always done without concomitant and complete detoxification for overused medications. After weaning the offending drugs, prophylactic treatment may be more effective even before an episodic headache pattern is reestablished.[1, 3]

Around 90% of MOH patients use more than 1 drug for acute attack treatment; therefore, it is difficult to differentiate characteristics of MOH subtypes according to the overused drug.[11] Patients who overuse ergotamine and analgesics may be more likely to have a daily headache with tension-type features, while triptan-induced MOH may induce a daily migraine-type headache or have an increase in migraine frequency.[33] MOH caused by the combined overuse of triptans and analgesics typically manifests higher headache frequency and intensity, and more additional symptoms, than the overuse of triptans alone.[45]

Autonomic and vasomotor symptoms, such as nasal drainage and congestion, lacrimation, vasomotor instability, and gastrointestinal hypermotility, can represent medication withdrawal, in particular opioid withdrawal. Patients complaining of sinus symptoms are frequently treated for sinus infection, or self-medicated with decongestants or cold medications, which can worsen MOH.[3]

Most acute drugs when overused may decrease the efficacy of preventive medications. An example is NSAIDs (such as ibuprofen), interfering with serotonergic antidepressant activity.[46]

Psychopathology and MOH

Psychopathology manifested as depression and anxiety is comorbid with MOH. It has to be treated in addition to the weaning of overused medication. Opioids and barbiturates have strong reinforcing and anxiolytic properties, in addition to their addictive potential, and intuitively might be expected to pose particular problems in treating MOH. One of the many challenges of dealing with MOH patients is to determine the presence or absence of confounding factors related to the medication overuse.

Certain behaviors and psychological states, such as fear of headache (cephalgiaphobia), anticipatory anxiety, catastrophizing, low headache-related self-efficacy, obsessional drug-taking, and psychological drug dependence, seem to be of particular importance in provoking and sustaining medication overuse.[31, 47] A history of obsessive–compulsive behavior is more common in patients with MOH, and may predispose patients to obsessive drug-taking.[48, 49] Psychopathology may play a role in convincing physicians to prescribe opioids. A secondary analysis of the Healthcare for Communities Survey (N = 9279) found that the presence of major depression, dysthymia, panic, or generalized anxiety predicted the regular use of opioids, with an odds ratio (OR) of 6.15 (95% CI = 4.1, 9.1). Moreover, the presence of psychiatric disorders increased the odds for prescription opioids in patients who reported low levels of pain interference (OR = 3.12; CI = 1.7, 5.9), suggesting that patients with psychopathology may have lower pain tolerance, or may be using opioids to medicate both pain and psychological distress.[50]

A large proportion of patients with CDH and patients with potential to develop MOH fit criteria for substance dependence in the Diagnostic and Statistical Manual of Mental Disorders Fourth Edition (DSM-IV).[30, 51] DSM-5 replaced the distinction between “dependence” and “abuse” (terms with pejorative connotations and defining criteria that deviated at times from intuitive lay definitions) with the simpler concept of “substance use disorder” – for example, “opioid use disorder” – with additional diagnoses of intoxication, withdrawal, other substance-induced disorders, and unspecified substance-related disorders.[52] The 2009 AMPP study revealed that 15.9% of migraineurs reported opioid use in the past 3 months for control. Of the opioid users, 16.6% met criteria for probable opioid dependence, representing 2.6% of the total sample of migraineurs. The prevalence of anxiety and depression was highest among current opioid users with probable dependence.[19]

Recent studies in neuroimaging and pharmacogenetics suggest an overlap between the pathophysiological mechanisms of MOH and substance-related disorders.[53] Functional magnetic resonance imaging data during the execution of a decision-making under risk paradigm found that MOH patients showed dysfunction in the mesocorticolimbic dopamine circuit – particularly in the ventromedial prefrontal cortex and in the substantia nigra/ventral tegmental complex – when compared with several control groups. The ventromedial prefrontal cortex dysfunctions appear to be reversible based on comparison to a control group of MOH patients 6 months after drug withdrawal. In contrast, the substantia nigra/central tegmental area complete dysfunctions persist even 6 months after withdrawal, based on a comparison to CM and non-migraine controls.[54]

Moreover, headache sufferers with personality disorders (PDs), in particular cluster B entities, may act out particular behaviors, such as defiance of limits regarding medication usage, battles of control over treatment, and attitudes of entitlement regarding pain control, resulting in medication overuse.[55] DSM-5 defines a personality disorder as “an enduring pattern of inner experience and behavior that deviates markedly from the expectations of the individual's culture, is pervasive and inflexible, has an onset in adolescence or early adulthood, is stable over time, and leads to distress or impairment.”[52] PDs continue to be grouped in 3 clusters: A (paranoid, schizoid, and schizotypal), B (antisocial, borderline, histrionic, and narcissistic), and C (avoidant, dependent, and obsessive–compulsive). The cluster B disorders are marked by self-serving traits that often emerge in the doctor–patient relationship, such as deceit (antisocial), dramatic mood swings from idealizing to intense anger at the physician (borderline), seductiveness and drama (histrionic), and entitlement (narcissistic). Of potential interest is a new PD diagnosis in DSM-5: “personality change due to another medical condition,” which refers to medically related persistent personality changes that deviate significantly from the individual's previous characteristic personality pattern.

In a sample of 267 patients who completed a multidisciplinary inpatient treatment program, MOH was more common in those with a PD diagnosis (62%) than those with no PD (38%).[55] The presence of borderline PD, moderate-severe depression, and MOH may be an unfortunate trifecta associated with poor outcomes from outpatient treatment programs.[56]

Two Types of MOH

Saper and Lake after several years of experience classified MOH patients as type I (simple) and type II (complex) in an attempt to develop better treatment strategies and outcome measuring tools. Type I would refer to relatively uncomplicated cases, such as patients overusing non-opioid- and non-barbiturate-containing medication, and the absence of significant psychopathology. Type II would refer to patients overusing opioid- or barbiturate-containing medications, and/or the presence of significant psychopathology.57-59 Several other researchers have supported the importance of drawing a distinction between MOH subtypes.[60] Rossi et al proposed a distinction between simple vs complicated MOH based on the presence of at least 1 of the following: (1) diagnosis of coexistent, significant, and complicating medical illnesses; (2) current diagnosis of mood, anxiety, eating, or substance addiction disorder; (3) relapse after previous detoxification; (4) significant psychosocial and environmental problems; and (5) daily use of multiple doses of symptomatic medication.[61] Radat and Lanteri-Minet emphasize the distinction between MOH with minimal psychological contribution and MOH where addictive behavior plays a central role.[31] Caution should be exercised in generalizing from successful outcomes with simple MOH or primarily triptan-related MOH to the more complicated form.

Remission and Relapse

CM patients can transition back to EM in a rate of 57% over 1 year and 66% over 2 years.[62, 63] However, relapse to medication overuse can be high: 28-31% within 6 months of withdrawal, to 41% at 1 year, and 45% at 4 years. Relapse rates for analgesics can be as high as 71% 4 years after treatment. Adding a behavioral component to treatment can significantly reduce relapse, to as low as 12.5% at 3-year follow-up.[64] Risk factors for relapse include high baseline intake of overused drugs, return to use of previously overused drugs, failure to improve at 2 months post-withdrawal, smoking, and alcohol use.[65] Opioid- and barbiturate-related MOH increase the risk for relapse, particularly if the patient is given even limited access to these drugs as rescue medications, a very slippery slope. Alternative rescue medications with low risk for MOH should remain the first choice for breakthrough pain. Although psychiatric comorbidity was unrelated to relapse at 1 year, patients with less depression and anxiety had the most favorable outcome at 4 years post-withdrawal.[65, 66] Appropriate therapy can be rewarding, but patients should be seen frequently and over prolonged periods of time to assess their progress. Although relapse is common, it can be treated effectively if the patient remains in treatment.[58] Even those who have argued that detoxification may not always be necessary for effective treatment – based on research samples participating in studies of topiramate and onabotulinumtoxinA – agree that the trifecta of withdrawal therapy, pharmacological prophylaxis, and behavioral treatment should be a significant component of care in those who fail to respond and in complex cases.[67]