Published Ahead of Print on August 18, 2017 as 10.1212/WNL.

0000000000004371
VIEWS & REVIEWS

Medication overuse headache

An entrenched idea in need of scrutiny

Ann I. Scher, PhD ABSTRACT

Paul B. Rizzoli, MD It is a widely accepted idea that medications taken to relieve acute headache pain can paradox-
Elizabeth W. Loder, MD, ically worsen headache if used too often. This type of secondary headache is referred to as
MPH medication overuse headache (MOH); previously used terms include rebound headache and
drug-induced headache. In the absence of consensus about the duration of use, amount, and type
of medication needed to cause MOH, the default position is conservative. A common recommen-
Correspondence to
Dr. Scher: dation is to limit treatment to no more than 10 or 15 days per month (depending on medication
ann.scher@usuhs.edu type) to prevent headache frequency progression. Medication withdrawal is often recommended
as a first step in treatment of patients with very frequent headaches. Existing evidence, however,
does not provide a strong basis for such causal claims about the relationship between medication
use and frequent headache. Observational studies linking treatment patterns with headache fre-
quency are by their nature confounded by indication. Medication withdrawal studies have mostly
been uncontrolled and often have high dropout rates. Evaluation of this evidence suggests that
only a minority of patients required to limit the use of symptomatic medication may benefit from
treatment limitation. Similarly, only a minority of patients deemed to be overusing medications
may benefit from withdrawal. These findings raise serious questions about the value of withhold-
ing or withdrawing symptom-relieving medications from people with frequent headaches solely to
prevent or treat MOH. The benefits of doing so are smaller, and the harms larger, than currently
recognized. The concept of MOH should be viewed with more skepticism. Until the evidence is
better, we should avoid dogmatism about the use of symptomatic medication. Frequent use of
symptom-relieving headache medications should be viewed more neutrally, as an indicator of
poorly controlled headaches, and not invariably a cause. Neurology® 2017;89:1–9

GLOSSARY
CVD 5 cardiovascular disease; ICHD 5 International Classification of Headache Disorders; MOH 5 medication overuse
headache; PCORI 5 Patient-Centered Outcomes Research Institute.

It is a widely accepted idea that medications taken to relieve headache pain can paradoxically
worsen headache when used too often. This concept of medication overuse headache
(MOH) is embedded in diagnostic criteria and is extensively reflected in guidelines, treatment
recommendations, and proposed quality standards for headache care,1–4 and ultimately dissem-
inated to the general public at patient-oriented websites.5–8
In the absence of consensus about the duration of use, amount, and type of medication
needed to cause MOH, the default position is conservative. Most pain medications, including
acetaminophen and aspirin, are believed to cause it.1,2 Many authorities recommend that
symptom-relieving medications should be used no more than 10 or 15 d/mo depending on
medication type.1,4
Sometimes called rebound or drug-induced headache, the current version of the International
Classification of Headache Disorders (ICHD) defines MOH as headache occurring $15 d/mo
in a patient with a preexisting headache disorder who has regularly exceeded specific thresholds
Editorial, page XXX of symptomatic medication use (table 1). Previous versions required that headache had “devel-
oped or markedly worsened” during medication overuse, and resolved or reverted “to its
Supplemental data
at Neurology.org
From the Department of Preventive Medicine and Biostatistics (A.I.S.), Uniformed Services University, Bethesda, MD; and Department of
Neurology (P.B.R., E.W.L.), Brigham and Women’s Hospital, Boston, MA.
Go to Neurology.org for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article.

© 2017 American Academy of Neurology 1

ª 2017 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.


Table 1 Subtypes of medication overuse headache (MOH) (International
Classification of Headache Disorders-3-beta)1
Intake threshold required
Subtype to diagnose MOH, d/mo
Ergotamine overuse headache 10
Triptan overuse headache 10
Simple analgesic overuse headache 15
Paracetamol (acetaminophen) overuse headache
Acetylsalicylic acid overuse headache
Other nonsteroidal anti-inflammatory drug overuse
headache
Opioid overuse headache 10
Combination analgesic overuse headache 10
MOH attributed to multiple drug classes not individually 10
overused
MOH attributed to unverified overuse of multiple drug 10
classes
MOH attributed to other medication 10
previous pattern within 2 months” of discon-
tinuation.9 These requirements have been
eliminated in favor of the presumption that
MOH exists if arbitrary thresholds for head-
ache frequency and medication intake are met,
while still allowing for physician judgment.1
The authors of the ICHD identify medication
overuse as “the most common cause of symp-
toms suggestive of chronic migraine.”1
THE SHORTCOMINGS OF EXISTING EVIDENCE
In this Viewpoint, we argue that the well-established
dominance of the MOH narrative stands in consid-
erable contrast to the strength of the underlying sci-
entific evidence—evidence that does not meet the
standards that are usually applied to other treatments
for headache. We argue that the evidence does not
provide a strong basis for causal claims about the role
of medication overuse in worsening headaches. We
discuss harms that result from the incorporation of
poorly supported views about MOH into clinical
thinking and practice, including stigmatization and
undertreatment of pain. Finally, we propose reeval-
uation of this belief system and make recom-
mendations for future research.
Evidence of cause and effect is weak. High-quality evi-
dence supporting the existence of MOH is inherently
difficult to obtain. A gold standard clinical trial would
randomly assign individuals with episodic headaches
to overuse or not overuse medication and compare
rates of headache progression in the 2 groups. Such
a study has not been done nor will it ever be done.
Observational studies provide data related to the nat-
ural history of headache frequency as well as describe risk
2 Neurology 89 September 19, 2017
ª 2017 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
factors that predict headache improvement or worsen-
ing. Some of these studies have calculated prevalence
or incidence of chronic headache or migraine in individ-
uals who use different types of medication or who take
medication above a critical threshold.10–13
Only a few such studies are longitudinal, such as
that done by Bigal et al.10 In this well-designed pop-
ulation-based study, individuals with episodic
migraine were followed for a year. The outcome of
chronic migraine was compared among those who
used particular types of medication at several levels
of frequency. People with episodic migraine who used
medication containing opioids or barbiturates were
more likely to progress to chronic migraine than the
reference group of acetaminophen users after control-
ling for sex, headache frequency and severity, and
preventive medication use. Frequency of medication
use by itself was not associated with chronic migraine
incidence after controlling for headache frequency,
although there was a dose-response relationship for
frequency of use of barbiturates. In a separate remis-
sion analysis based on the same study population by
Manack et al.,14 use of preventives predicted a worse
prognosis although this association disappeared after
controlling for baseline headache frequency. Frequent
medication use, defined as $10 days per month of
over-the-counter or prescription medication, did not
predict remission in crude or adjusted analysis.
Observational studies that show an association
between frequency or type of medication used and
worsening headache can provide useful prognostic
information. They cannot, however, answer the ques-
tion of whether treatment patterns are a cause or, alter-
natively, a consequence of frequent headaches—or
whether the association is bidirectional. We argue that
causal inferences will be threatened by confounding by
indication, even when there is control for factors that
drive treatment preference or headache occurrence (e.
g., baseline headache frequency or important comor-
bidities such as depression).
Confounding by indication (and the related con-
cepts of confounding by severity or confounding by
contraindication) is a concept that is relevant to
observational studies that measure or compare the
efficacy of different disease therapies.15 Since treat-
ments are not assigned randomly in practice, choice
of treatment will be based in part on factors that may
be related to how well the patient would have done
even without treatment. Factors could include disease
severity, previous failed treatments, contraindica-
tions, or comorbidities, among others. Thus, if pa-
tients with refractory disease are more likely to be
prescribed third-line medication A, then treatment
A may look worse than other treatments in observa-
tional studies even though it may be efficacious.
While it is possible to control for known prognostic
 factors or indications using various adjustment strat- prognostic variables. Would this level of evidence be
egies such as propensity score matching,16–18 the effec- sufficient to recommend that patients limit their use
tiveness of statistical control always depends on how of these medications? Is not the more plausible inter-
well-understood (and how well-measured) these other pretation simply that medication overuse is an indi-
factors are.15 In the present example, the indication cator of poorly controlled asthma, not a cause of it?21
(e.g., refractory disease) may have a much larger influ- A number of studies have identified comorbidities
ence on treatment success than the treatment itself, so and biomarkers that are associated with medication
it is critically important that the chosen analytic strat- overuse, including genetic polymorphisms as well as
egy effectively separates the effect of treatment from electrophysiologic and neuroimaging abnormalities.
the effect of the indication for treatment. Relevant As with medication overuse itself, however, all of
examples of observational study results that may be these things may simply be markers of severe or
due in part to confounding by indication include the poorly controlled headache conditions.22
finding that migraineurs who use migraine preventive We note that many people with frequent head-
medications have a worse prognosis than those who aches are not overusing medication (table 2), and
do not12,14 and lack of increased risk of incident car- overuse of medication does not invariably produce
diovascular disease (CVD) in migraineurs who were MOH.10 The authors of a recent systematic review
prescribed triptans compared to migraineurs who identified 18 population-based studies that reported
were not prescribed triptans.19,20 The confounding the prevalence of medication overuse among people
would be positive in the first example since the indi- with very frequent headache.34 Prevalence estimates
cation for preventives (frequent headaches) is itself based on raw, unadjusted data ranged from 11% to
a strong risk factor for headache progression. The 68%. This illustrates substantial uncertainty about
confounding would be negative in the second exam- the magnitude of the problem. It also suggests that
ple since triptan use is contraindicated in patients even if medication overuse produces worsening head-
with CVD risk factors and therefore patients for ache in some people, it is not the only and possibly
whom doctors prescribe triptans would be expected not even the most frequent cause of chronic head-
to have a lower risk of CVD than other migraineurs. ache. Table 2 lists a subset of these studies.
The concept of confounding by indication in Finally, evidence is inconsistent about the associa-
MOH studies can be illustrated by a simple thought tion between specific types of symptomatic medica-
experiment. Suppose the Bigal et al.10/Manack et al.14 tions and increased frequency of headaches. The
studies or a similar observational study had instead evidence that simple analgesics might produce
been conducted in patients with asthma, and it had MOH is especially weak. ICHD criteria allow a diag-
been observed that the patients who frequently used nosis of MOH to be made in people with frequent
symptom-controlling medications had a worse prog- headache who are using simple analgesics such as aspi-
nosis than others, even after adjusting for all known rin or ibuprofen 15 or more days per month. Yet in
one observational study, headache patients who re-
ported regular use of aspirin or ibuprofen had
Table 2 Selected population-based studies examining the proportion of
a decreased risk of headache progression.29 This
individuals with chronic daily headache (‡15 headache d/mo) who are may be an example of confounding by indication or
overusing medication reverse causality, since patients with very frequent
headaches may avoid these medications due to side
Proportion overusing
Study Age range, y medicationa effects. However, evidence from 2 randomized
Castillo et al.23 14 and older 28 placebo-controlled trials of daily aspirin showed
improvement rather than worsening in migraineurs
Dyb et al.24 13–18 36
who were assigned to the aspirin group.35,36 A guide-
Katsarava et al.25 16 and older 11
line issued in 2012 by the American Academy of
26
Lu et al. 15 and older 34
Neurology and the American Headache Society con-
Lundqvist et al.27 30–44 46 cluded that several nonsteroidal anti-inflammatory
Prencipe et al.28 65 and older 38 drugs, including ibuprofen, naproxen, and ketopro-
Scher et al. 29
18–65 23 fen, are “probably effective and should be considered
Wang et al.30 65 and older 25
for migraine prevention.”37 It is difficult to reconcile
31
this evidence with the widespread belief that regular
Wang et al. 12–14 20
use of these medications worsens headache.
32
Wiendels et al. 25–55 63
Four interesting studies have considered whether
Zwart et al.33 20 and older 45 MOH occurs when pain medications are used for
a
Based on various definitions. Some figures have been estimated from the indicated other conditions. The populations evaluated were
publication. postcolectomy patients,38 rheumatology patients,39,40

Neurology 89 September 19, 2017 3

ª 2017 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

 and patients with degenerative musculoskeletal dis-
ease.41 Two studies40,41 concluded that there was no
association between regular or frequent use of medi-
cation for nonheadache pain and the development of
chronic headache. The other 2 studies38,39 concluded
that frequent analgesic use for nonheadache pain was
associated with the development of chronic migraine
only in those with a preexisting history of migraine.
All 4 studies are small and have important limitations
that do not support strong conclusions at this time.
Medication withdrawal does not help most patients with
frequent headaches. Another line of evidence support-
ing the existence of MOH relates to treatment by
withdrawing medications (sometimes referred to as
detoxification). If MOH is reversible, then with-
drawal of overused medications (alone) should
improve headache frequency, yet the evidence it does
so is weak. Despite this, the authors of ICHD-3-beta
say “the majority” of patients with frequent headaches
who meet criteria for MOH will improve after
withdrawal.
A systematic review of treatment strategies for
MOH evaluated withdrawal alone or with other pre-
ventive strategies. The authors of the review con-
cluded, “The level of evidence to support early
discontinuation of overused medications alone is
low due to the absence of controlled studies.”42 They
recommended instead the addition of preventive ther-
apy to the regimens of patients with very frequent
headache and presumed MOH.
The 4 withdrawal-only studies identified in this
review are listed in table 3 along with 2 relevant studies
published after the date range encompassed by the
review.43–48 The 6 studies used a variety of withdrawal
strategies ranging from simple advice to reduce
medication intake to inpatient withdrawal. Only one
study had a nontreatment control group and none was
blinded. Success, usually defined as .50% improve-
ment in headache frequency, was #33% in 5 of the 6
studies at the follow-up time period closest to 2–3
months, using intention-to-treat analysis.
Intention-to-treat analyses are particularly important
in the evaluation of these studies. Some studies show
success rates for those patients who complete drug with-
drawal treatment, but this can be misleading because
many patients do not tolerate drug withdrawal. Patients
who cannot or will not tolerate withdrawal are not a ran-
dom subset of participants. Excluding them when re-
sults are calculated inflates estimates of treatment
success, because those who remain in treatment are pre-
sumably more likely to be placebo responders or patients
who would have improved anyway. Improvement in the
patients who can tolerate withdrawal is then interpreted
incorrectly as evidence that medication withdrawal was
effective. Returning to the asthma analogy, this is equiv-
alent to conducting an uncontrolled interventional trial
on asthma patients who can tolerate going without res-
cue inhalers for 2–3 months, and then extrapolating the
results to all people with asthma. Reasons for nonpar-
ticipation in the 6 studies are described in appendix e-1
at Neurology.org and the reader can evaluate the extent
to which exclusion of the dropouts/noncompleters
would have led to an overestimate of treatment effec-
tiveness. Our intention-to-treat analysis in table 3 uses
the most conservative assumption, which is that those
who did not successfully withdraw or were otherwise
lost to follow-up were treatment failures.
Given that most studies described in table 3 lacked
a control group, we are left to wonder whether
a ;33% response rate is greater than the percentage
of patients who might have improved due to natural

Table 3 Proportion of patients with presumed medication overuse headache who improved with medication withdrawal as the sole treatment

Follow-up Proportion Proportion of

duration, completing completers who Proportion of all
Study, year/no. of patients moa withdrawala improved patients who improveda Study definition of improvement

Studies identified in Chiang

et al.42 review

Grande et al.,43 2011/n 5 140 18 Unknown 42 33 ,15 Headache d/mo

44
Hagen et al., 2009/n 5 20 3 91 15 (Intention-to-treat) $50% Reduction in headache days
and no medication overuse

Rossi et al.,45 2011/n 5 100 2 79 87 69 $50% Reduction in headache days

Zeeberg et al.,46 2006/n 5 337 2 64 45 29 $1% Reduction in headache days

Relevant studies published after

Chiang et al.42 review

Sarchielli et al.,47 2014/n 5 44 3 100 24 24 (Withdrawal 1 $50% Reduction in headache days

placebo group)

Pijpers et al.,48 2016/n 5 416 2–3 68 42 27 $50% Reduction in headache days

a
Calculated or estimated from original publication in some cases. See appendix e-1 notes for details of data extraction or calculation.

4 Neurology 89 September 19, 2017

ª 2017 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

 Table 4 Medication overuse headache remission in placebo arm of randomized controlled trials of preventive treatments

Study (treatment) No. (placebo) Population Follow-up, wk Response rate, % Outcome

51
Silberstein et al. (botulinum toxin) 459 CM 1 MO 24 32 .50% Reduction in headache days
52 a
Sandrini et al. (botulinum toxin) 35 CM 1 MO 12 30 .50% Reduction in headache days

Diener et al.53 (topiramate) 23 CM 1 MO 4 0 .50% Reduction in headache days

54
Silvestrini et al. (topiramate) 14 CM 1 MO 8 0 .50% Reduction in headache days

Abbreviation: CM 1 MO 5 chronic migraine with medication overuse.

a
Estimated.

history, regression to the mean, or nonspecific pla-
cebo effects. Some clues can be gleaned from natural
history studies and placebo-controlled trials. Head-
ache frequency is highly variable when assessed over
multiple time points and a substantial proportion of
people with very frequent headaches, probably the
majority in the general population at least, will drop
below the arbitrary 15 headache days per month
threshold at follow-up interviews.14,30,31,49,50
One can also observe response rates in the placebo
arms of 4 randomized controlled trials for chronic
migraine that reported results for the subgroup with
medication overuse (table 4).51–54 In these trials, the
proportion of MOH patients in the placebo arm who
experienced a greater than 50% reduction in head-
ache days over the course of the trial ranged from 0%
in 2 small (based on the size of the MOH subgroup)
topiramate studies to 30%–32% in the onabotuli-
numtoxinA trials. In the aggregate, these population
and clinical studies illustrate the considerable uncer-
tainty about the course of high-frequency headache
and underscore the critical need for rigorous con-
trolled trials before medication withdrawal can be
recommended as an effective treatment.
THE HARMS AND IMPLICATIONS OF AN MOH
DIAGNOSIS OR TREATMENT The harms of an
MOH diagnosis or treatment for it have not been
well-studied, but deserve consideration. These
include (1) unnecessary suffering, (2) blame and
stigmatization, and (3) diversion of research and
clinical attention.
Suffering. In the case of a patient with daily headaches
—not an uncommon scenario in specialty headache
care—ICHD-3-beta criteria suggest that, depending
on medication type, MOH should be suspected in
patients who treat one-third or one-half of attacks.
Limiting treatment for acute attacks to below this
level is thought to protect patients from headache
frequency progression, although there is no evidence
to support this view. In fact, it is plausible that, for
most patients with frequent headache, strict medica-
tion limitation involves harm from undertreatment of
pain and does not result in benefit in terms of reduced
frequency. Table 5 shows the hypothetical number of
migraine patients who must forgo or limit symp-
tomatic treatment of headaches in order to avoid one
case of incident chronic migraine, based on the risk
estimates from the observational Bigal et al.10 study.
Assuming causality (that is, assuming headache
prognosis was solely related to treatment patterns
rather than unmeasured confounders or natural his-
tory), the hypothetical number needed to undertreat
ranges from 3.4 to 4.7 for the high-frequency
headache group.
Medication withdrawal also involves a poor bal-
ance of benefit and harm, especially since evidence
suggests that preventive treatment to reduce headache
frequency may in fact be effective even if symptom-
relieving medications are not withdrawn.44,55 Further-
more, many patients who undergo withdrawal do not
complete the treatment but still must endure long
periods without effective relief of pain. In the largest
study of this approach (table 3), patients were advised

Table 5 Number needed to undertreat to avoid one case of incident chronic (transformed) migraine (after
Bigal et al.10)

One-year incidence of One-year incidence of No. needed to

chronic migraine in chronic migraine undertreat to avoid
Medication type Headache, d/mo unexposed, % in exposed,a % one case

Barbiturate-containing 10–14 2.5 32 3.4

medication

Opioids 10–14 2.5 24 4.7

a
One-year incidence of chronic migraine in group using indicated medication type 10 days per month, estimated from
figure 3 of the observational study by Bigal et al.10 of episodic migraine transition to chronic migraine as a function of
treatment frequency, treatment type, and headache frequency. Assumes causality and that the incidence of chronic
migraine in the unexposed is 2.5%, i.e., the total chronic migraine incidence rate for the population.

Neurology 89 September 19, 2017 5

ª 2017 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

 that they would not be offered specific headache
treatment unless they were able to do without all
symptomatic medication including caffeine for 2–3
months.48 Intention-to-treat analysis (including the
approximate one-third of patients who could not or
would not complete withdrawal) shows a treatment
response of just 27%, defined as .50% reduction in
headache frequency.
Very similar results were seen in the second larg-
est study, with about one-third of patients not com-
pleting withdrawal and an intention-to-treat
response of 29%, defined in this study as . 0%
reduction in headache frequency.46 This corre-
sponds to a number needed to undertreat of roughly
3. Put another way, at best 3 patients had to suffer
months of unrelieved headache pain in order for one
patient to be helped—this assuming that all
improvement was due to withdrawal per se and that
none improved due to natural history, regression to
the mean, or placebo response.
Blame and stigmatization. Patients may suffer if they
are viewed as the architects of their own headache
problem. Caregivers may believe that help is not
deserved or, as in the study described above, even
withhold other treatments until patients complete
a medication withdrawal program. Patients may expe-
rience a sense of failure if they are unable to limit
medication use or may actually conceal medication
use from their doctor.
The stigma associated with perceptions of medica-
tion overuse only adds to the sense of shame experi-
enced by those with chronic headaches, who already
“worry about what their doctors will think of them if
they complain or fail their treatments.”56 Moreover,
this burden of blame is not equally distributed: pa-
tients with the most severe, unusual, or intractable
headache disorders would be affected by an inflexible
view of medication overuse as a cause of chronic or
refractory headache.
Diversion of attention. A less apparent but important
harm is the diversion of time, attention, and resources
away from activities that would benefit patients more
than medication limitation or withdrawal. Clinicians
may be more interested in limiting or reducing symp-
tomatic medication than investigating alternative
evidence-based explanations or treatments for
chronic headaches. Pressed to recommend treatment
of some sort, they may suggest treatments that are
expensive, are invasive, or lack good quality evidence
of benefit. These options may actually increase costs
and produce worse outcomes. Researchers may be
dissuaded from studying other causes of headache
progression, or find it difficult to secure money from
funders who believe strongly in the explanation of
MOH. These opportunity costs are an important
6 Neurology 89 September 19, 2017
ª 2017 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
argument against overly broad generalizations about
the connection between medication use and frequent
headaches.
A NEED FOR REASSESSMENT MOH is an en-
trenched belief that has gained plausibility and accep-
tance through repetition.57 Why has the concept of
MOH proved so popular and enduring despite the
weakness of the scientific evidence supporting its exis-
tence? Why is the evidence supporting MOH treat-
ment not evaluated as critically as the evidence for
other interventions, such as, for example, patent fora-
men ovale closure and other surgical treatments?58–60
There are several possible explanations for its
appeal. First, the idea of MOH aligns with the clinical
experience of medical practitioners: patients taking
a lot of medication tend to be those who do poorly
and whose headaches do not respond well to treat-
ment. Second, MOH provides a convenient explana-
tion for treatment failure—although it does so by
placing blame on the patient for overusing medica-
tions rather than on the inadequacies of existing
treatments.
Finally, there are many good reasons other than
fear of MOH to avoid excessive use of symptomatic
medication. MOH provides an explanation for physi-
cians to avoid prescribing medications such as opioids
or barbiturates that (for many good reasons) they do
not want to use. Opioid or barbiturate-containing
medications impair alertness and are associated with
a risk of addiction or dependence syndromes; nonste-
roidal anti-inflammatory medications may produce
gastrointestinal bleeding or kidney damage, for exam-
ple. We are not suggesting that frequent use of symp-
tomatic medication is desirable. Our point is
a narrower one, which is that the specific harm of
MOH has been overstated.
The MOH narrative also ignores what we know
about the potency of other risk factors and biological
phenomena. How can it be possible that exposure
beyond an arbitrary threshold of medication intake
could produce chronic headache in most patients?
The complexity of the interaction among individual
susceptibility, the pharmacology of different medi-
cations, and diverse underlying causes of headache
suggests this is an oversimplification. If MOH
is real, it is more plausible that a spectrum of suscep-
tibility exists so that not everyone exposed develops
the problem, and that other factors must
be involved.
FUTURE DIRECTIONS Better evidence is needed.
One encouraging development is the recent
announcement by the Patient-Centered Out-
comes Research Institute (PCORI) that it will fund
a large pragmatic trial that compares 2 treatment
 strategies for patients with chronic migraine who
are deemed to be overusing medications to abort
headache attacks. The study will compare early
discontinuation of the overused medication plus
migraine prophylaxis with a strategy of migraine
prophylaxis without early discontinuation of over-
used medication.61 Consistent with PCORI
research priorities (e.g., comparative effectiveness
of accepted therapies), the study does not include
a control group, so it will not be able to demon-
strate the extent to which either strategy is prefer-
able to maintaining the status quo.
While evidence in this area is inherently difficult
to obtain, there are some methodologic considera-
tions that could improve the quality of research.
As previously mentioned, it is essential that studies
of medication withdrawal include dropouts in anal-
yses in accordance with standard methodology.
Another hindrance to research in this area has been
the use of arbitrary cutpoints for defining chronic
headache and medication overuse.62 As previously
described, use of such cutpoints results in spurious
“remission” and “incidence” estimates, which can be
substantial depending on the extent of natural vari-
ability or measurement error.63 While this source of
error may be unbiased, it is important to consider
when interpreting uncontrolled interventional stud-
ies (which may have an artificially high response
rate) or observational studies (in which the true
effect of risk factors may be obscured). A number
of headache researchers have considered these and
other methodologic issues in detail and have made
suggestions to improve the rigor and usefulness of
observational and interventional studies in chronic
headache research including MOH.62–67
The concept of MOH has been embraced too
enthusiastically and should be viewed with more
skepticism. Until the evidence is better, we should
avoid dogmatism. It is not intellectually honest, or
fair to patients, to pretend that the rigor of the evi-
dence is strong. Instead, frequent use of symptom-
relieving headache medications should be viewed
more neutrally, as a marker of poor headache control.
This in turn may reflect illness severity, the shortcom-
ings of current therapies, social or economic effects,
or the complex interplay of these factors. Not all of
these are amenable to intervention. Meanwhile, pa-
tients and physicians should recognize that we cannot
yet answer with certainty the question “How much
medication is too much?”
DISCLAIMER The views expressed in this article are
those of the authors and do not necessarily reflect
the official policy of the Uniformed Services Univer-
sity of the Health Sciences, the Department of
Defense, or the US Government.
ª 2017 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
AUTHOR CONTRIBUTIONS
Ann Scher: study concept and design, analysis and interpretation,
approval of final version. Elizabeth Loder: study concept and design, crit-
ical revision of the manuscript for important intellectual content,
approval of final version. Paul Rizzoli: critical revision of the manuscript
for important intellectual content, approval of final version.
STUDY FUNDING
No targeted funding reported.
DISCLOSURE
A. Scher: consultant/advisory board member for Allergan. P. Rizzoli and
E. Loder report no disclosures relevant to the manuscript. Go to
Neurology.org for full disclosures.
Received January 24, 2017. Accepted in final form May 10, 2017.
REFERENCES
1. Headache Classification Committee of the International
Headache Society (IHS). The International Classification
of Headache Disorders, 3rd edition (beta version). Ceph-
alalgia 2013;33:629–808.
2. Evers S, Jensen R. Treatment of medication overuse head-
ache: guideline of the EFNS headache panel. Eur J Neurol
2011;18:1115–1121.
3. Ross S, Wall E, Schierman B, et al. Quality improvement
in neurology: primary headache quality measures.
Neurology 2015;84:200–203.
4. Loder E, Weizenbaum E, Frishberg B, Silberstein S.
Choosing wisely in headache medicine: the American
Headache Society’s list of five things physicians and pa-
tients should question. Headache 2013;53:1651–1659.
5. Medication-overuse headache [online]. Available at:
migrainetrust.org/about-migraine/types-of-migraine/other-
headache-disorders/medication-overuse-headache. Ac-
cessed August 1, 2017.
6. Medication overuse headache [online]. Available at:
americanmigrainefoundation.org/medication-overuse-headache.
Accessed August 1, 2017.
7. Rebound headaches [online]. Available at: patientslikeme.
com/symptoms/show/2023. Accessed August 1, 2017.
8. Away with rebound headache! [online]. Available at:
relieve-migraine-headache.com/rebound-headache.html.
Accessed August 1, 2017.
9. Headache Classification Committee of the International
Headache Society (IHS). The International Classification
of Headache Disorders: 2nd edition. Cephalalgia 2004;24
(suppl 1):9–160.
10. Bigal ME, Serrano D, Buse D, Scher A, Stewart WF,
Lipton RB. Acute migraine medications and evolution
from episodic to chronic migraine: a longitudinal
population-based study. Headache 2008;48:
1157–1168.
11. Lipton RB, Serrano D, Nicholson RA, Buse DC, Runken
MC, Reed ML. Impact of NSAID and triptan use on
developing chronic migraine: results from the American
Migraine Prevalence and Prevention (AMPP) study. Head-
ache 2013;53:1548–1563.
12. Katsarava Z, Schneeweiss S, Kurth T, et al. Incidence and
predictors for chronicity of headache in patients with epi-
sodic migraine. Neurology 2004;62:788–790.
13. Zwart JA, Dyb G, Hagen K, Svebak S, Holmen J. Anal-
gesic use: a predictor of chronic pain and medication over-
use headache: the Head-HUNT Study. Neurology 2003;
61:160.
Neurology 89 September 19, 2017 7
 14. Manack A, Buse DC, Serrano D, Turkel CC, Lipton RB.
Rates, predictors, and consequences of remission from
chronic migraine to episodic migraine. Neurology 2011;
76:711–718.
15. Joseph KS, Mehrabadi A, Lisonkova S. Confounding by
indication and related concepts. Curr Epidemiol Rep
2014;1:1–8.
16. Andrade C. Propensity score matching in nonrandomized
studies: a concept simply explained using antidepressant
treatment during pregnancy as an example. J Clin Psychi-
atry 2017;78:e162–e165.
17. Kyriacou DN, Lewis RJ. Confounding by indication in
clinical research. JAMA 2016;316:1818–1819.
18. Sturmer T, Joshi M, Glynn RJ, Avorn J, Rothman KJ,
Schneeweiss S. A review of the application of propensity
score methods yielded increasing use, advantages in
specific settings, but not substantially different estimates
compared with conventional multivariable methods. J Clin
Epidemiol 2006;59:437–447.
19. Hall GC, Brown MM, Mo J, MacRae KD. Triptans in
migraine: the risks of stroke, cardiovascular disease, and
death in practice. Neurology 2004;62:563–568.
20. Velentgas P, Cole JA, Mo J, Sikes CR, Walker AM. Severe
vascular events in migraine patients. Headache 2004;44:
642–651.
21. Slejko JF, Ghushchyan VH, Sucher B, et al. Asthma
control in the United States, 2008–2010: indicators of
poor asthma control. J Allergy Clin Immunol 2014;133:
1579–1587.
22. Diener HC, Holle D, Solbach K, Gaul C. Medication-
overuse headache: risk factors, pathophysiology and man-
agement. Nat Rev Neurol 2016;12:575–583.
23. Castillo J, Munoz P, Guitera V, Pascual J. Kaplan Award
1998: epidemiology of chronic daily headache in the gen-
eral population. Headache 1999;39:190–196.
24. Dyb G, Holmen TL, Zwart JA. Analgesic overuse among
adolescents with headache: the Head-HUNT-Youth
Study. Neurology 2006;66:198–201.
25. Katsarava Z, Dzagnidze A, Kukava M, et al. Primary head-
ache disorders in the Republic of Georgia: prevalence and
risk factors. Neurology 2009;73:1796–1803.
26. Lu SR, Fuh JL, Juang KD, Wang SJ. Migraine prevalence
in adolescents aged 13-15: a student population-based
study in Taiwan [in process citation]. Cephalalgia 2000;
20:479–485.
27. Lundqvist C, Aaseth K, Grande RB, Benth JS, Russell MB.
The severity of dependence score correlates with medica-
tion overuse in persons with secondary chronic headaches:
The Akershus Study of Chronic Headache. Pain 2010;
148:487–491.
28. Prencipe M, Casini AR, Ferretti C, et al. Prevalence of
headache in an elderly population: attack frequency,
disability, and use of medication. J Neurol Neurosurg
Psychiatry 2001;70:377–381.
29. Scher AI, Lipton RB, Stewart WF, Bigal M. Patterns of
medication use by chronic and episodic headache sufferers
in the general population: results from the Frequent Head-
ache Epidemiology Study. Cephalalgia 2010;30:321–328.
30. Wang SJ, Fuh JL, Lu SR, et al. Chronic daily headache in
Chinese elderly: prevalence, risk factors, and biannual
follow-up. Neurology 2000;54:314–319.
31. Wang SJ, Fuh JL, Lu SR, Juang KD. Outcomes and
predictors of chronic daily headache in adolescents: a 2-year
longitudinal study. Neurology 2007;68:591–596.
8 Neurology 89 September 19, 2017
ª 2017 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
32. Wiendels NJ, Neven AK, Rosendaal FR, et al. Chronic
frequent headache in the general population: prevalence
and associated factors. Cephalalgia 2006;26:1434–1442.
33. Zwart JA, Dyb G, Hagen K, Svebak S, Stovner LJ,
Holmen J. Analgesic overuse among subjects with headache,
neck, and low-back pain. Neurology 2004;62:1540–1544.
34. Westergaard ML, Hansen EH, Glumer C, Olesen J,
Jensen RH. Definitions of medication-overuse headache in
population-based studies and their implications on prevalence
estimates: a systematic review. Cephalalgia 2014;34:409–425.
35. Bensenor IM, Cook NR, Lee IM, Chown MJ,
Hennekens CH, Buring JE. Low-dose aspirin for migraine
prophylaxis in women. Cephalalgia 2001;21:175–183.
36. Buring JE, Peto R, Hennekens CH. Low-dose aspirin for
migraine prophylaxis. JAMA 1990;264:1711–1713.
37. Holland S, Silberstein SD, Freitag F, Dodick DW,
Argoff C, Ashman E. Evidence-based guideline update:
NSAIDs and other complementary treatments for episodic
migraine prevention in adults: report of the Quality Stand-
ards Subcommittee of the American Academy of Neurology
and the American Headache Society. Neurology 2012;78:
1346–1353.
38. Wilkinson SM, Becker WJ, Heine JA. Opiate use to con-
trol bowel motility may induce chronic daily headache in
patients with migraine. Headache 2001;41:303–309.
39. Bahra A, Walsh M, Menon S, Goadsby PJ. Does chronic
daily headache arise de novo in association with regular use
of analgesics? Headache 2003;43:179–190.
40. Lance F, Parkes C, Wilkinson M. Does analgesic abuse
cause headaches de novo?Headache 1988;28:61–62.Letter.
41. Bowdler I, Kilian J. The association between analgesic
abuse and headache: coincidental or causal? Headache
1988;28:494.Letter.
42. Chiang CC, Schwedt TJ, Wang SJ, Dodick DW. Treat-
ment of medication-overuse headache: a systematic review.
Cephalalgia 2016;36:371–386.
43. Grande RB, Aaseth K, Benth JS, Lundqvist C, Russell
MB. Reduction in medication-overuse headache after
short information: the Akershus Study of Chronic Head-
ache. Eur J Neurol 2011;18:129–137.
44. Hagen K, Albretsen C, Vilming ST, et al. Management of
medication overuse headache: 1-year randomized multi-
centre open-label trial. Cephalalgia 2009;29:221–232.
45. Rossi P, Faroni JV, Nappi G. Short-term effectiveness of
simple advice as a withdrawal strategy in simple and com-
plicated medication overuse headache. Eur J Neurol 2011;
18:396–401.
46. Zeeberg P, Olesen J, Jensen R. Probable medication-
overuse headache: the effect of a 2-month drug-free
period. Neurology 2006;66:1894–1898.
47. Sarchielli P, Messina P, Cupini LM, et al. Sodium
valproate in migraine without aura and medication overuse
headache: a randomized controlled trial. Eur Neuropsy-
chopharmacol 2014;24:1289–1297.
48. Pijpers JA, Louter MA, de Bruin ME, et al. Detoxification
in medication-overuse headache, a retrospective controlled
follow-up study: does care by a headache nurse lead to
cure? Cephalalgia 2016;36:122–130.
49. Scher AI, Stewart WF, Ricci JA, Lipton RB. Factors asso-
ciated with the onset and remission of chronic daily head-
ache in a population-based study. Pain 2003;106:81–89.
50. Lu SR, Fuh JL, Chen WT, Juang KD, Wang SJ. Chronic
daily headache in Taipei, Taiwan: prevalence, follow-up
and outcome predictors. Cephalalgia 2001;21:980–986.
 51. Silberstein SD, Blumenfeld AM, Cady RK, et al. Onabo-
tulinumtoxinA for treatment of chronic migraine: PRE-
EMPT 24-week pooled subgroup analysis of patients
who had acute headache medication overuse at baseline.
J Neurol Sci 2013;331:48–56.
52. Sandrini G, Perrotta A, Tassorelli C, et al. Botulinum
toxin type-A in the prophylactic treatment of
medication-overuse headache: a multicenter, double-
blind, randomized, placebo-controlled, parallel group
study. J Headache Pain 2011;12:427–433.
53. Diener HC, Agosti R, Allais G, et al. Cessation versus
continuation of 6-month migraine preventive therapy
with topiramate (PROMPT): a randomised, double-
blind, placebo-controlled trial. Lancet Neurol 2007;6:
1054–1062.
54. Silvestrini M, Bartolini M, Coccia M, Baruffaldi R,
Taffi R, Provinciali L. Topiramate in the treatment of
chronic migraine. Cephalalgia 2003;23:820–824.
55. Diener HC. Detoxification for medication overuse head-
ache is not necessary. Cephalalgia 2012;32:423–427.
56. Kempner J. What biology can’t do. Headache 2016;56:
1047–1052.
57. Greenberg SA. Understanding belief using citation net-
works. J Eval Clin Pract 2011;17:389–393.
58. Diener HC, Bingel U. Surgical treatment for migraine: time
to fight against the knife. Cephalalgia 2015;35:465–468.
59. Schurks M, Diener HC. Closure of patent foramen ovale
in the prevention of migraine: not enough evidence in
favor. Nat Clin Pract Neurol 2009;5:22–23.
ª 2017 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
60. Tariq N, Tepper SJ, Kriegler JS. Patent foramen ovale and
migraine: closing the debate: a review. Headache 2016;56:
462–478.
61. Determining the optimal treatment strategy for patients
who have chronic migraine with medication overuse
[online]. Available at: pcori.org/research-results/2016/
determining-optimal-treatment-strategy-patients-who-
have-chronic-migraine. Accessed August 1, 2017.
62. Lipton RB, Penzien DB, Turner DP, Smitherman TA,
Houle TT. Methodological issues in studying rates and
predictors of migraine progression and remission. Head-
ache 2013;53:930–934.
63. Houle TT, Turner DP, Smitherman TA, Penzien DB,
Lipton RB. Influence of random measurement error on
estimated rates of headache chronification and remission.
Headache 2013;53:920–929.
64. Turner DP, Smitherman TA, Penzien DB, Lipton RB,
Houle TT. Rethinking headache chronification. Headache
2013;53:901–907.
65. Houle TT, Turner DP, Houle TA, et al. Rounding behav-
ior in the reporting of headache frequency complicates head-
ache chronification research. Headache 2013;53:908–919.
66. Hagen K, Jensen R, Boe MG, Stovner LJ. Medication
overuse headache: a critical review of end points in recent
follow-up studies. J Headache Pain 2010;11:373–377.
67. Hagen K, Stovner LJ, Zwart JA. Potentials and pitfalls in
analytical headache epidemiological studies: lessons to be
learned from the Head-HUNT study. Cephalalgia 2007;
27:403–413.
Neurology 89 September 19, 2017 9
 Medication overuse headache: An entrenched idea in need of scrutiny
Ann I. Scher, Paul B. Rizzoli and Elizabeth W. Loder
Neurology published online August 18, 2017
DOI 10.1212/WNL.0000000000004371

This information is current as of August 18, 2017

Updated Information &
Services
Supplementary Material
Subspecialty Collections
Permissions & Licensing
Reprints
including high resolution figures, can be found at:
http://www.neurology.org/content/early/2017/08/18/WNL.0000000000
004371.full.html
Supplementary material can be found at:
http://www.neurology.org/content/suppl/2017/08/18/WNL.000000000
0004371.DC1
http://www.neurology.org/content/suppl/2017/08/18/WNL.000000000
0004371.DC2
This article, along with others on similar topics, appears in the
following collection(s):
All epidemiology
http://www.neurology.org//cgi/collection/all_epidemiology
Natural history studies (prognosis)
http://www.neurology.org//cgi/collection/natural_history_studies_prog
nosis
Secondary headache disorders
http://www.neurology.org//cgi/collection/secondary_headache_disorde
rs
Information about reproducing this article in parts (figures,tables) or in
its entirety can be found online at:
http://www.neurology.org/misc/about.xhtml#permissions
Information about ordering reprints can be found online:
http://www.neurology.org/misc/addir.xhtml#reprintsus

Neurology ® is the official journal of the American Academy of Neurology. Published continuously since
1951, it is now a weekly with 48 issues per year. Copyright © 2017 American Academy of Neurology. All
rights reserved. Print ISSN: 0028-3878. Online ISSN: 1526-632X.