Treatment of Medication Overuse Headache: Effect and Predictors After 1 Year

A Randomized Controlled Trial

Louise N. Carlsen MD, PhD; Carolien Rouw MD; Maria L. Westergaard MD, PhD; Mia Nielsen MD; Signe B. Munksgaard MD, PhD; Lars Bendtsen MD, PhD, Dr.Med.Sci.; Rigmor H. Jensen MD, Dr.Med.Sci.


Headache. 2021;61(7):1112-1122. 

In This Article


This study presents 1-year follow-up results for the treatment of 96 patients with MOH. In line with the 6-month follow-up results, the present follow-up underlines that MOH can be treated effectively by a combination of withdrawal and preventive medication. The question is the timing of the interventions: whether withdrawal and preventive treatment should be done together at the start or whether one form of treatment should be started first, then supplemented by the other treatment later.

On Combination of Treatment Strategies

The management of MOH has been discussed for decades, and several studies have investigated the effect of withdrawal and preventive treatment.[18–21] In this study, we compared three treatment strategies that all consisted of options for withdrawal therapy and preventive treatment, but in different combinations. Although our hypothesis was that the treatment strategy including withdrawal with immediate preventive treatment would have the largest treatment effect, all three strategies proved equally effective after 1 year. However, as previously reported, the combination of withdrawal and preventive medication from the start led to the fastest improvement within the first 6 months, based on the number of patients with resolved MOH and number of patients reverting to episodic headache.[6,7] The most important message from this 1-year follow-up study is that this rapidly obtained effect in the W+P group is a long-lasting and stable treatment effect. Therefore, in line with the newest European treatment guidelines for MOH, we conclude that the strategy with withdrawal and preventive medication from the start should still be the recommended treatment of MOH.[8]

Moreover, an early intervention with a quick onset of effect is crucial for patients and is also more cost-effective, as reported in the ComoEstas study, where a combination of withdrawal and preventive medication considerably reduced direct and indirect costs.[22]

A strategy with withdrawal and delayed preventive medication can be necessary for patients with an unclear preexisting headache diagnosis. The withdrawal period may help clarify the diagnosis, and a relevant pharmacological preventive can be prescribed after an observation period. However, a strategy with preventive medication before starting withdrawal may be helpful for some patients to ease the withdrawal process.

On Measurable Outcomes

After 1 year, several parameters showed a statistically significant improvement for the entire study population: monthly headache days, monthly migraine days, days with acute medication, headache pain intensity, SDS score, overall QoL, self-rated overall health, and HURT score. Furthermore, 89% had resolved MOH, which corresponds to a relapse rate of only 11%. Prior studies found relapse rates between 13% and 41% after 1 year,[8,23,24] and most relapses occurred within the first year after successful withdrawal.[25,26]

The low relapse rate might be explained by frequent follow-up during the study period. All participants received patient education, 71% went to a physiotherapist, and one-third were seen by a psychologist at the Danish Headache Center. Because prevention of relapse is an important part of the treatment of MOH, we suggest a combination of frequent follow-up, patient education, and nonpharmacological preventive treatment. This multimodal and multidisciplinary approach has also been advocated in other studies.[27–29]

In contrast to the results of the ComoEstas study, no significant reduction was seen in the HADS depression or anxiety score from baseline to 1-year follow-up in our patients.[18] Interestingly, the HADS depression and anxiety scores at 6-month follow-up in the ComoEstas study (4.1 [SD 3.7] and 7.1 [SD 4.4], respectively) are close to the scores obtained in our study at baseline (4.8 [95% CI: 4.0–5.6] and 7.60 [95% CI: 6.7–8.5], respectively) and again at 1-year follow-up (4.35 [95% CI: 3.7–5.1] and 7.14 [95% CI: 6.1–8.1], respectively). This indicates that our study population had low HADS scores compared with the population in the ComoEstas study.

On Predictors

The third aim of this study was to identify predictors of nonresponse to treatment, that is, still having chronic headache after 1 year. This may help to identify patients requiring more intensive care and frequent follow-up. We found that high headache frequency, more frequent use of acute medication, high pain intensity, and depression at baseline were predictors of nonresponse.

In the ComoEstas study, tension-type headache and tension-type headache plus migraine were found to be predictors of having chronic headache after 6 months.[30] We did not observe this in our study. Unexpectedly, we did not find the SDS score to have a prognostic relevance in our study. Lundqvist et al. reported that a specific cutoff SDS score (≥4 for men and ≥5 for women) could predict a successful prognosis related to withdrawal as treatment for MOH.[31,32] The SDS score might be more useful as an indicator of behavior during withdrawal, rather than a predictor of who will benefit from withdrawal and revert to episodic headache.

This study suggests that higher self-rated health at the start of the treatment reduces the risk of still having chronic headache after 1 year. A high self-rated health may have a beneficial psychological effect, or this might indicate a proportion of the study population who are least disabled by MOH and more responsive to treatment.

Interestingly, daily caffeine consumption more than 200 mg may have a protective effect against chronic headache. This may be a spurious finding but could also be explained by the effect of caffeine on nociception. Caffeine is known to have an effect on nociception through its antagonism of the adenosine receptors,[33,34] and there is some evidence that caffeine monotherapy is associated with more pain relief than placebo in treating patients with migraine or with migraine and tension-type headache.[33,35] Another hypothesis could be that daily caffeine consumption may be a proxy for other protective factors, for example that these people share good coping strategies. Nonetheless, high caffeine consumption has been found to be a risk factor for MOH when comparing a daily caffeine intake of more than 540 mg with an intake of less than 240 mg.[36] It could be hypothesized that a moderate dose of caffeine has a protective effect on patients with severe headache conditions. Further investigation about the role of caffeine as a predictor of remission to episodic headache is needed.

Strengths and Limitations

An important strength of this study is its high clinical relevance. All three treatment strategies were outpatient programs, and the results of this study are, therefore, easily applied in both primary and secondary care. The study was carried out at the Danish Headache Center, which is a tertiary center specialized in treating MOH, which has ensured a systematic and unified approach for all patients. In addition, the study was a randomized, controlled trial, and we followed the participants for a whole year. That said, there are some limitations that must be addressed.

This study was not blinded because withdrawal is impossible to blind. Another limitation is the relatively small sample size, which yields some risk of a type 2 error, possibly resulting in nonsignificant findings. Sample size calculation to achieve a power of 80% was done to minimize this risk. Although the study overall had a high adherence rate of 80%, an unequal dropout rate between the three treatment groups was seen, with the highest dropout rate of 28% in the W+P group. This may have led to attrition bias, but no specific patterns in baseline characteristics were seen between the patients, who completed the 1-year follow-up, and the dropouts. Due to ethical reasons, we did not set up a control group. For the same reason, patients without improvement and continuous MOH in the P+pW group were offered a 2-month withdrawal period after 6 months. As part of our multidisciplinary treatment of MOH, a large percentage of all three treatment groups went to see a physiotherapist and/or a psychologist. This may have influenced and have added to the effect of the treatment strategies and the low relapse rate after 1 year.

All three treatment strategies in this study were conducted as outpatient programs. Patients unsuitable for outpatient treatment were excluded, for example, because of severe physical or psychiatric comorbidities and most often because of significant opioid overuse (Figure 2). This means that our results cannot necessarily be applied to the subgroup of complex patients with MOH.

Preventive medication was prescribed for every patient based on the existing guideline at the Danish Headache Center. Even though the preventive medication prescribed for the patients varied, selection of treatment reflects clinical practice in most countries. Calcitonin gene-related peptide-targeting monoclonal antibodies were not available at the time of the study.


This randomized, controlled study with 1-year follow-up of patients with MOH demonstrates that MOH is a treatable condition with a high resolution rate, and all treatment strategies proved to be effective in treating MOH. Treatment should include both withdrawal, including education, and preventives, and an initiation of both elements from the start leads to the fastest effect and to a long-lasting treatment effect.