Sex differences in the clinical manifestations related to dependence behaviors in medication-overuse headache

Medication-overuse headache (MOH) is one of the leading causes of disease-related disability among all neurological disorders [1]. The estimated prevalence of MOH is about 1% in the general population [2‐5], and it is associated with tremendous socioeconomic impacts [1, 5]. Timely diagnosis and management are indispensable in reducing the disease burden of MOH. In particular, it was found in the Chronic Migraine (CM) Epidemiology and Outcomes (CaMEO) study that medication overuse (MO) remained an important issue in about three fourths of CM patients, despite having had medical consultation, accurate diagnosis, and minimally appropriate medical treatment [6]. Therefore, MOH deserves more attention from the general public and medical professionals.

There is a growing interest in sex-related differences in medicine, especially in epidemiology, pathophysiology, clinical manifestations, and treatment response [7]. In clinical neurology and psychiatry, such a topic is gaining attention in headache disorders and substance use disorders (SUDs) [8‐10]. Similar to CM, MOH is more common in women than in men [11‐13]. However, there is little information about sex differences in the clinical manifestations of MOH. In particular, behaviors of substance dependence were reported to be present in up to two thirds of MOH patients [14‐17]. In addition to acute medications, use of tobacco and other psychoactive substances is more common in patients with MOH than in those without [18, 19]. These findings are suggestive of shared pathophysiology with SUDs. Recently, there were reports pointing out the roles of sex-specific mechanisms in reward and addiction [20]. Whether the associations between MOH and clinical manifestations related to dependence behaviors could also be different in women and men is yet to be clarified. Such sex differences could have important clinical implications, and need to be elucidated.

The Leeds Dependence Questionnaire (LDQ) and the Severity of Dependence Scale (SDS) are two neuropsychological instruments commonly used in the clinical studies of dependence behaviors in SUDs [21, 22] and in MOH [23‐26]. It was demonstrated that the LDQ and the SDS were useful to detect the presence of medication overuse (MO) or MOH [23‐26], and the scores were correlated with the outcomes [17, 27, 28]. However, in many of the studies, comparisons were made between chronic daily headache (CDH) or CM patients with MOH and those with episodic migraine (EM), episodic cluster headache, or episodic tension-type headache (TTH) or even healthy controls. Besides, some of data were from population-based studies, namely the Akershus study [25, 26, 28]. The comparative performance of these two instruments in the detection of MOH among CM patients in the clinical settings is uncertain. More importantly, whether there could be sex differences in the diagnostic utilities of these two instruments for the detection of MOH among CM patients remains to be determined.

The aims of the present study were (1) to determine sex differences in the association between MOH and the current use of tobacco, alcohol, caffeine, or hypnotics, (2) to compare the diagnostic utilities of the LDQ and the SDS in the detection of MOH between women and men with CM, and (3) to evaluate between-sex differences in the clinical presentations of MOH.

In the current study, it was found that smoking was independently associated with MOH in men, but not in women, and frequent hypnotic use was associated with MOH, regardless of sex. Besides, the LDQ and the SDS were of excellent diagnostic performances for the diagnosis of MOH in patients with CM of both sexes, although the cutoff scores for both of the instruments were one point lower in men than in women. In addition, MOH in the current cohort was characterized by a more severe phenotype in women, and a shorter latency between migraine onset and CM onset in men. Sex should be considered as an important factor in the initial evaluation of patients with suspected MOH.

Important strengths of the present study included sample size, quality of the data, and study design. The current study recruited more than 1400 consecutive CM patients. Therefore, selection bias could be reduced, and more accurate estimates could be derived. In particular, the number of male patients in the present study was larger than those in the majority of prior studies. Besides, the data were of high quality and reliability. The diagnoses of CM and MOH were made by headache specialists according to the ICHD-3 criteria. The data were collected systematically by using a specifically designed questionnaire. In addition, the clinical utilities of the LDQ and the SDS in CM patients with and without MOH were compared directly. In addition, dependence behaviors and related manifestations were compared between the sexes. Therefore, the present study could provide more comprehensive and complete information that is more relevant to CM patients evaluated in the clinical settings.

It was found that smoking was associated with MOH, and the association was stronger in men. An association between smoking and MOH, as well as SUDs, was reported in the literature [19, 35‐37], although whether there could be sex differences has rarely been discussed upon. Since smoking shares some clinical features with SUDs, it is possible that tobacco use could reflect the presence of disturbances in the reward circuit such that these patients were at risk of developing MOH. On the other hand, frequent hypnotic use was also associated with MOH, which was in keeping with the association between MOH and tranquilizers reported in the literature [19, 38]. However, there was no sex difference. Interpretations for the associations with alcohol and caffeine use could be more complicated since these agents could have direct impacts on migraine attacks [39]. In addition to the mechanisms related to reward circuit, it is also possible that the identified association between MOH and smoking and hypnotic use could also be attributed to certain shared factors, such as socioeconomic status, stress levels, lifestyles, etc. Further studies are needed to clarify the roles of these potential confounders. Interestingly, smoking cessation could have a positive effect on the treatment outcome of SUDs [40]. Whether smoking cessation would reduce the risk or improve the prognosis of MOH, especially in men, deserves further exploration.

The current MOH cohort was characterized by a low rate of triptan use, and there was an association between overuse of triptans or ergots and more severe dependence behaviors. The rate of triptan use or overuse in our MOH patients was lower than estimates from some studies [41‐43], although the trend was consistent with prior reports from Taiwan, which could be attributed to the costs of tritpans and the restrictions on their use, e.g. ≤ 400 mg/month for sumatriptan tablets, by the reimbursement regulations of the Taiwan NHI [29, 44]. In the present study, triptan or ergot users scored higher on the LDQ and the SDS. In fact, it takes shorter latencies and lower doses for patients overusing triptans or ergots to develop MOH than those with simple analgesic overuse [41]. These findings could be suggestive of an association between overuse of these two classes of acute medications and a greater tendency to develop MOH. Alternatively, it is also possible that patients with different headache severities or personality traits would preferably overuse certain categories of acute medications [45, 46]. However, interpretations could be cautious since the categories of triptan or ergot overuse included patients using these agents alone or in combination with other acute medications. More studies are needed to provide more insight into the underlying pathophysiology.

In the current study, the LDQ and the SDS were of comparable accuracies in the detection of MOH in CM patients of both sexes. It was reported that the severity of dependence behaviors, as measured by the LDQ, in CDH patients with MO was greater than those with episodic primary headache disorders, and was similar to that in SUDs involving alcohol or illicit drugs [23]. Besides, it was demonstrated that an SDS score of ≥ 5 was correlated with the presence of MO in the general population [25], as well as behaviors of substance dependence in the clinical settings [17], and the cutoff score identified in our study was the same. In MOH patients who received detoxification, responders had greater decrease in the LDQ scores [27]. Besides, the SDS scores at baseline were predictive of the outcome in MOH associated with primary headache disorders [28], and the SDS could decrease following withdrawal of acute medications [47]. In addition to preventive medications, behavioral therapy targeting at dependence behaviors could also be an important treatment option to improve the treatment outcome [14]. It was demonstrated that brief intervention, which consisted of feedback based on the initial scores of SDS and consultations to reduce the use of acute medication, was an effective for MOH patients [48]. It is possible that both the LDQ and the SDS could have important roles not only as screening instruments, but also in the treatment and prognostication.

In the present study, it was found that cutoff scores for the diagnosis of MOH were lower in men than in women for both of the instruments, and there were differences in clinical manifestations of MOH and medication use between the sexes. Actually, it was demonstrated in a population-based study that SDS ≥ 5 for women and ≥ 4 for men could help detect MO in a group of patients with primary chronic headache disorders, of whom 95% had chronic TTH [25], and the cutoffs identified in our study were exactly the same. In fact, there have been some reports on sex-specific cutoffs in the diagnosis a variety of diseases, such as right bundle branch block, myocardial infarction, venous thromboembolism, etc. [49‐52], which indicate that an uniform diagnostic cutoff score may not always be the best strategy. On the other hand, it was found that MOH in women was associated with more severe clinical manifestations, which is consistent with the trend in migraine and CM [9, 12, 53]. Besides, MOH in men was characterized by a shorter interval between migraine onset and CM onset. In fact, among patients with EM at baseline, men were more likely to have CM onset at 6, 9, and 12 months in the CaMEO study [54]. This might indicate that migraine chronification could take a shorter period of time to develop in men. Since there is a female preponderance in both CM and MOH [11‐13], data in male patients could be under-presented in the literature. Clinical decisions for men with MOH based on available evidence could be biased. Further studies are needed to provide more insights on sex-specific differences on such an issue.

There are some limitations. First of all, the patients were recruited from the headache clinic of a tertiary medical center, and only patients who were 20–65 years old were included in the study. There could be concerns about generalizability. However, there is no strict regulation on the process of referral in the healthcare system of our country, and the majority of patients came in directly without referral. Therefore, patients involved in the present study could reflect those in the general population to a certain extent. Second, both the LDQ and the SDS were administered by the patients themselves, and there could be concerns about reliability. However, the responses in these instruments were verified by headache specialists at face-to-face interviews. Besides, the self-administered version of SDS has been validated against headache specialist interviews by telephone [55]. Third, dependence behaviors in the current study were defined based on the scores on neuropsychological instruments, rather than formal psychiatric evaluation. In fact, although MOH shares many clinical and neuroimaging features with SUDs, certain important elements of SUD are not seen in patients with MOH, such as craving, drug-associated cues, impulsivity, and compulsive behaviors [56, 57]. Besides, controversies remain as to whether overuse of acute medications could just reflect the severity of the underlying headache disorders [45], since most of the commonly used acute medications are generally considered to be of low abuse potential [58, 59]. Therefore, the findings should be interpreted with caution. Fourth, the use of tobacco, alcohol, and caffeinated beverages was defined by dichotomous variables, and whether the amount or frequency of their use to have an impact is yet to be determined. Finally, as a cross-sectional study, only associations, rather than causal relationships, could be identified, and the findings should be interpreted with caution.

In conclusion, the present study demonstrated differential risks of MOH associated with tobacco use between men and women and identified sex-specific cutoff scores of the LDQ and the SDS for MOH among patients with CM. Besides, compared with men, women with MOH seemed to have longer latencies of migraine chronification, except for those who smoke. Sex should be taken into account in the evaluation of MOH.