Probable medications overuse headaches: validation of a brief easy-to-use screening tool in a headache centre

The prevalence of chronic daily headaches with medication overuse is around 1.5% of the general population, with prevalence figures being very similar across different countries. They account for 10% to 78% of patients seen in headache centres [1]. Medication overuse headache (MOH) decreases the quality of life and accounts for reduced efficiency at work. Population-based and clinical data reflects the importance of this public health problem [2].

The 2nd edition of the International classification of headache disorders (ICHD-II) defined MOH as a migraine-type headache:1 present on ≥15 days per month with at least 1 of the following characteristics: bilateral, pressing/tightening quality, mild or moderate intensity; [3] associated with use of acute treatment for headache on a regular basis for more than 3 months [1] that developed or markedly worsened during medication overuse [4] that resolved or reverted to its previous pattern within 2 months after discontinuation of the offending medication [5]. Subsequent revision proposed to eliminate the headache characteristics [3] and to delete the last criterion in order to allow the establishment of MOH diagnosis when the patient suffers from the disorder and not only after the medication withdrawal [4]. This change suggested by Olesen has not been officially implemented and correct formulation for MOH before withdrawal should be probable MOH (pMOH). The IHS classification is an explicit clinical criterion so that the diagnosis of pMOH does not require any specific examination that is needed only to exclude a symptomatic origin. Application of the operational diagnostic revised criteria of the ICHD-II allows making the diagnosis on the basis of consumption of medication (number of days monthly) and a worsening of the headache during the time of medication intake. However, this clinical diagnosis with a face-to-face interview is the gold standard and requires clinical experience for the physician, and enough time to question patients precisely, to specify ICHD-II criteria, and to specify the diagnosis of primary headache, more frequently migraine, which has been the basis for the development of the MOH. This is time-consuming.

In this clinical context, a less time consuming tool would be of interest. We have developed a 4 item questionnaire evaluating the frequency of headaches and the amount of medication consumption used to treat the attacks and compared sensitivity and specificity of this in migraine patients and MOH patients. Indeed, such a tool would be useful in clinical practice and for research, to pre-include patients for epidemiological studies.

Patients were enrolled between September 2009 and December 2009. Eighty-nine patients have been seen for headache in the headache centre during this period.

Seventy-seven patients were pre-included, with twenty-one men (27.3%) and fifty-six women (72.7%). Ages varied from 16 to 71 years, with mean age of 42.5.

After being interviewed by the neurologists, 42 patients (55%) have been considered as suffering from probable MOH, 35 patient having migraine without medication overuse. There were more women with probable MOH than men (60.7% vs. 39.3% p = 0.08). The mean age in pMOH group was higher than in migraine group (46.7 (12.2) years vs. 37.4 (13.6) years, p = 0.002). Patients were heterogeneous concerning the duration of the disease and concerning the kind of medication they overused (Table 2).

Next we determined sensitivity, specificity, positive predictive value, negative predictive value for each item, for each pair of item and for the items 1 + 2 + 4.

The third question about the onset and duration of the drug intake was clearly not discriminating enough: 90.5% of patients have answered yes to the third question, that’s why item 3 has been marked out. Results are showed in Table 3.

Concerning patients with pMOH, 19 patients had a duration of medication overuse > 10 years, 13 patients up to 10 years. Duration of pMOH didn’t have any influence on answers to questions. Twenty-four patients had pMOH type 1 as mentioned Saper and Lake [6], 18 type 2. This had no impact on answers. Eight patients overused acetaminophen or NSAID, 10 overused an association of medications, 14 overused combinations, 9 overused triptans and one ergotamine. The size of the sample did not allow to study the impact of the type of medication on the response profile. Duration of pMOH and classification of pMOH type 1 or 2 had no impact on answers to questionnaire (p = 0.07 and p = 0.01 respectively).

The main results of our study validation was the following: the association of question 2 (do you take a treatment for attacks more than 10 days per month?) and question 4 (is this intake on a regular basis?) had a sensitivity of 95.2% and a specificity of 80%. The advantages of such questions are their simplicity, rapidity, and low cost. Our sample of migraineurs, and of pMOH sufferers were heterogeneous in terms of age, sex, and duration. Moreover, we kept equivocal cases.

Several limitations of the results have to be discussed: At first, possible selection bias could exist: in a tertiary centre, patients could have a better knowledge of their disease, and could answer easier yes to the different questions, with better understanding, and therefore would be more able to quantify their medication intake. That could increase positive answers and then explain an increase of their sensibility in a tertiary centre. Moreover, patients seeking care in tertiary centres could have more severe disease, more severe medication overuse leading again to an increase in the questionnaire sensibility. For these reasons, generalisability to a population-based setting need to be further tested. Secondly problems with the measures have to be considered: concerning the construction of the questionnaire, the criteria specifying the necessity of having developed a worsening of the headaches during overuse was not kept. Indeed, patients usually answered that if they take medication, it is because they have headaches and not the opposite [7]. Patients usually think that medication overuse is the result of an unrecognized or inadequate treatment of a primary headache disorder. Moreover it is a retrospective question, about a period sometimes many years before the diagnosis of pMOH. However, to evaluate the reliability of transforming the criteria of having developed a worsening of the headaches during overuse , we did a pretest in a sample of 10 patients, asking them if they have developed worsening of the headaches during overuse asking them if they understood the question. The results of this pretest are shown in Table 4. Five patients answered that they didn’t develop a worsening of headaches during overuse, 2 patients answered that they did not know. That’s why this criterion was not kept. In the same way, we evaluated the ability to understand the other questions by the patients. Are expressions such as treatment for attacks intake on a regular basis easily understandable for patients without any medical knowledge? We did a second pilot study on a sample of 24 patients testing the comprehensibility of both these expressions: namely the expression treatment for attack was understood by 22 patients. The expression “intake on a regular basis” was understood by all, except one patient. At the end, if we do the hypothesis on a decrease of sensitivity in the general population, what could be the outcome of false positives, i.e. patients with an intake below 10 days per month, on the border of medication overuse: could it have negative implications for the patients wrongly classified as having pMOH by the questionnaire? In fact, it could have positive repercussions for the patients, with a primary or secondary preventive effect on the risk of developing pMOH. Concerning false negative patients, they will continue their overuse, until they see a neurologist or another physician. In our study, sensitivity and specificity could be high because of the sole inclusion of migraine sufferers, without including other primitive headaches, such as cluster headache and tension-type headaches. But, for a pilot study, it is acceptable to have included only migraine patients and CH frequency in tertiary centre only. What improvements does this questionnaire have in comparison to other published questionnaire: Table 5 presents data about the methods used in main epidemiological studies concerning MOH [5, 8‐13]. For each study, we specified the questions or criteria used for the MOH diagnosis, and specified if these questions were validated or not. One can notice that the only author using a validated questionnaire was Hagen [14]. In the general population he determined the diagnostic performances of the association of 2 questions: headaches which are present for more than 14 days per month and the use of analgesics 4 times per week or more. With such an association, a lower sensitivity was found: 75%, specificity was 100%. Other questionnaires do exist [14‐17], but they concern research on dependence [15, 16]. Radat et al. [15] constructed and validated a questionnaire measuring dependence on analgesics and on migraine treatments in headache patients. It measures with 21 questions a global dimension of dependence in MOH sufferers, but does not allow for giving a cut-off on which patients should be considered to have MOH. Such a dependence questionnaire which focus on the dimension of dependence are more time-consuming, and maybe more difficult to administrate in current care, neither in tertiary nor in primary care settings. Maizels in 2003 [18] proposed a paradigm for the clinical application of a brief headache screening tool. Questions were open and medication overuse was correctly identified in 86% of patients with a specificity of 79%. In this context, the advantages of this screening tool are its shortness, it’s easy administration without much paper work and low cost. We must consider its diagnostic performances outside tertiary centre, i.e. neurologists outside hospitals, general practitioners, and maybe for patients themselves outside of any health place. In particular, predictive values have to be determined in the general population. To conclude, we validated a screening tool in the French version for the diagnosis of pMOH. In the future, an English version should be validated.