Abstract
We carried out a population-based case-control study to evaluate the association between multiple sclerosis (MS) and headache. We had previously determined the incidence of MS during 1990-1999 in Catania, Sicily, identifying 155 incident MS patients; these subjects underwent a telephone interview using a standardized questionnaire for headache. Diagnosis and classification of headaches were made according to International Headache Society criteria (1988). A control group was selected from the general population through random digit dialling. One hundred and one (65.2±) MS patients, of the 155 identified, and 101 controls were screened for headaches. Fifty-eight (57.4±) MS patients and 38 (37.2±) controls fulfilled the diagnostic criteria of headache. A significant association between MS and headache was found with an adjusted odds ratio, estimated by logistic regression, of 2.18 (95± confidence interval 1.27, 3.93). Frequency of headaches in our MS population is higher than in the general population, supporting the hypothesis of a possible association between these two conditions.
Introduction
Headache is one of the most prevalent symptoms in the adult population (1). Migraine and other headaches are frequently chronic disorders involving repeated attacks, sometimes leading to poor performance at school, impairment of work and reduction of quality of life. An epidemiological survey to determine the prevalence of chronic headache among the affiliates of general practitioners has been recently carried out in two Italian areas estimating, according to the diagnostic criteria proposed by the International Headache Society (IHS) in 1988 (2), a prevalence rate of 40% in Varese and 38% in S. Giovanni Rotondo (3).
Multiple sclerosis (MS) is a chronic neurological disease with a wide spectrum of symptoms and signs, and the relationship between MS and headache is poorly understood. Headache is not generally considered a symptom of MS, and studies investigating the relationship between the two conditions have produced conflicting results. In particular, the frequency of headache among MS patients ranges from 4 to 37.5% in studies carried out before 1990 (4–9), and up to 50% considering only the most recent studies (10–12), suggesting a higher prevalence of headache in MS patients than in the general population. This fact has generated the fascinating hypothesis of the existence of aetiopathogenic and physiopathological common pathways between MS and headache. It has been suggested that the reduced numbers of T8 lymphocytes, which characterize both conditions, may arise from a shared immune mechanism (13–15). It has also been hypothesized that repeated demyelinization of brainstem structures could cause migraine-like headache in MS patients (16, 17, 19). The evidence that several brainstem areas are involved in the pathophysiology of migraines supports the latter hypothesis. Animal studies, in fact, suggest that the trigeminocervical complex, the rostral brainstem and periaqueductal grey matter have primary roles in the complex mechanisms underlying migraine attacks (18–21), even if the role of the brainstem in the pathogenesis of migraine is still debated.
In particular, the hypothesis of an involvement of the periaqueductal grey matter has been recently supported by evidence of an increased risk of migraine-like headache among MS patients with evidence of midbrain lesions [odds ratio (OR) 3.91, 95% confidence interval (CI) 2.01, 7.32] (22).
Nevertheless, published data concerning the relationship between headache and MS have to date come only from clinical series, and are, consequently, prone to selection bias. We evaluated the prevalence of headaches in a population-based incidence cohort of MS patients identified in the city of Catania from 1990 to 1999 (23, 24), in an attempt to increase our knowledge regarding the relationship between MS and headache. To allow comparison, we adopted the criteria proposed by the IHS in 1988 (3). To our knowledge, this represents the first population-based case–control study carried out to estimate the possible association between MS and HA.
Materials and methods
At the end of the 1990s we carried out an epidemiological survey to determine the prevalence and incidence of MS in the city of Catania, Sicily, from 1975 to 1995 (23). A further follow-up survey was carried out at the end of 2004 to estimate the incidence during 1990–1999 (24).
The official population Catania, in 2001, date of the last census, was 313 110 (165 065 women, 148 045 men) and is slightly lower than the official population in 1991 (333 075 inhabitants; 174 267 women, 158 808 men) (25). Since immigrants from other countries represent only 1% of the entire province of Catania (official census 2001), its population can be considered ethnically stable.
We considered as prevalent and incident cases all patients who satisfied Poser's criteria (26). Incidence risk of MS during 1990–1999 was calculated considering all patients who had the onset of disease during this study period (onset adjusted incidence risk). All patients were followed up at least to 5 years after the onset of disease. All MS patients resident in the study area who had experienced the onset of MS during the last decade (1990–1999) represented the population-based incidence cohort. Methods for the MS case ascertainment have been extensively reported elsewhere (24).
The prevalence of headache was determined in the cohort of MS patients. To ascertain whether there was a history of headache, all MS patients underwent a semistructured telephone interview performed by a team of trained neurologists, guided by a standardized questionnaire. The questionnaire was similar to that adopted in the epidemiological survey on headache carried out in two Italian communities in 2003 (Varese and S. Giovanni Rotondo) (3). The questionnaire consisted of 17 items to sought information on the presence and clinical characteristics of headache, current treatments, frequency of attacks, aggravation by movement and routine physical activity, followed by three items regarding interferon (IFN) treatment. Headache suffers were identified among those answering affirmatively the question: ‘Do you usually suffer from headache?’ (3), and, to allow international comparison with other epidemiological surveys, diagnosis was made according to the diagnostic criteria proposed by the IHS in 1988 (2). When diagnosis or classification of headache were doubtful, patients were invited to undergo a complete neurological examination performed in our department by trained neurologists on headache.
The prevalence of headache was determined as point prevalence defined as the proportion of patients with headache in the MS population at a specified time [prevalence day (PD), 1 January 2005].
The prevalence of headache among the MS population was compared with that in the Italian population, i.e. the prevalence of headache reported in Varese and S. Giovanni Rotondo in 2003 (3).
Nevertheless, due to the different age structure of our MS cohort with respect to the general population, a direct comparison between these two rates appeared inadequate. For this reason one control, not affected by neurological disorders, per case was selected from the general population through random digit dialling (RDD). Controls were group-matched by age and sex.
As for the MS patients, control subjects underwent a semistructured telephone interview performed by the same team of trained neurologists, guided by a standardized questionnaire. The headache's questionnaire was the same (17 items) as that administered to the MS subjects, but to avoid a greater participation rate among subjects suffering from headache the question ‘Do you usually suffer from headache?’ was preceded by three other questions concerning other common diseases. Furthermore, to exclude the presence of neurological disorders, at the end of the questionnaire, soon after the headache-specific items, nine items of the screening instrument for neurological disorders adopted and validated in the Sicilian Neuroepidemiological survey (27) were included.
Statistical analysis
Data were analysed using Epi-Info 6.04 and STATA 6.0 software (StataCorp, College Station, TX, USA) (28, 29).
Quantitative variables were described using mean and standard deviation (
Results
One hundred and fifty-five MS patients living in the study area had had clinical onset of the disease within the studied decade (1990–1999). According to Poser's diagnostic criteria, 130 of these were classified as clinically definite multiple sclerosis, three as laboratory-supported definite multiple sclerosis and 22 as clinically probable multiple sclerosis. The mean age at onset was 33.6 ± 10.8 years (range 9–61 years). During the incidence study period the average population was 331 509 (157 987 men, 173 787 women). The average annual onset-adjusted incidence risk was 4.7/100 000 (95% CI 4.0, 5.5), 5.7/100 000 (95% CI 4.6, 6.9) for women and 3.5/100 000 (95% CI 2.7, 4.5) for men. The age-specific incidence risk increased in both sexes in the first groups of age with a peak in the group aged 25–34 years, 11.7/100 000 (9.3/100 000 for men and 14.1/100 000 for women) and showed a steep decline in the other age groups. These patients represented our incidence cohort of MS patients. Detailed data on prevalence and incidence of MS in Catania have been reported extensively elsewhere (24).
Of the 155 MS patients who had experienced onset of the disease during 1990–1999, our incidence cohort, 104 (67.1%) participated in the telephone survey. Three MS patients were excluded due to cognitive impairment, because they were not able to give accurate information; 101 (65.2%) completed the screening questionnaire for headache. Fifty-four MS patients were not interviewed because four refused and 47 were not traced. No significant differences were found among the baseline characteristics of the incidence cohort of MS patients (155 subjects) and the 101 screened subjects enrolled; furthermore, no significant differences were found between those MS subjects not screened (n = 54) and the incidence cohort (155 MS subjects), suggesting that our screened population was representative of the population-based incidence cohort previously identified. Baseline characteristics of the screened and not screened population are reported in Table 1.
Baseline characteristics of eligible and screened population
P-value = screened (101 subjects) vs. MS population (155 subjects).
P-value = not screened (54 subjects) vs. MS population (155 subjects).
CDMS, clinically definite multiple sclerosis; LSDMS, laboratory-supported definite multiple sclerosis; CPMS, clinically probable multiple sclerosis; RR, relapsing remitting; SP, secondary progressive; PP, primary progressive.
Of the 101 patients who participated, 58 (57.4%), 18 men (31%) and 40 women (69%), fulfilled the diagnostic criteria of headache proposed by the IHS in 1988.
According to the IHS classification proposed in 1988, of the 58 patients 48 (82.8%) were affected by primary headache [28 (48.3%) by tension-type headache and 20 (34.5%) by migraine] and four (6.9%) by secondary headache, whereas six were unclassifiable. Headache classification is reported in Table 2.
Frequency of headache subtypes according to the classification of the International Headache Society proposed in 1988
The crude prevalence rate of headache at the PD (1 January 2005) was 57.4% (95% CI 47.5, 66.8), higher for women (61.5%, 95% CI 49.3, 72.7) than for men (50%, 95% CI 34.0, 66.0). Considering the primary form of headache, 20 patients (three men and 17 women) were affected by migraine, giving a prevalence rate for migraine of 19.8% (95% CI 12.8, 28.4), whereas 28 MS patients (10 men, 18 women) were affected by tension-type headache, giving a prevalence of 27.7% (95% CI 19.7, 37.0). The mean age at onset for headache was 24.9 ± 12.6 years and 24.2 ± 11.4 years considering only the primary forms (19.5 ± 7.4 for migraine and 27.5 ± 12.7 for tension type). Characteristics of the attacks are presented in Table 3. As regards the specific treatment for MS, of the 101 subjects 49 (48.5%) were receiving IFN (33 out of the 58 affected by headache), five (4.9%) copolymer, eight (7.9%) azathioprine and one (9.9%) methotrexate. Seven (14.2%) MS patients of the 49 under IFN treatment developed headache after the initiation of therapy, whereas four cases affected by headache referred to a worsening of their attacks in terms of frequency and intensity after the initiation of IFN treatment. One of these seven patients was classified as having migraine, whereas six were classified as having tension-type headache.
Baseline characteristics of MS patients and controls
A total of 101 control subjects (40 men and 61 women; mean age 35.4 ± 11.5 years) group-matched for age and sex were randomly selected from the general population. Baseline characteristics (sex and age) of the MS population and control group are reported in Table 3.
In particular, we selected a RRD telephone sample of 210; 85 (40.5%) refused to participate and 24 where excluded due to lack of an available age- and sex-matched control. None of the screened subjects was positive at the screening questionnaires for neurological disorders, or reported to be affected by neurological examination.
Of the 101 controls, 38 (37.6%) fulfilled the diagnostic criteria for headache. According to the IHS classification 33 control subjects presented primary headache [16 migraine (39.5%) and 17 tension-type headache (47.2%)], whereas five were unclassifiable.
Crude prevalence of migraine (15.8%; 95% CI 9.6, 23.9) as well as of tension-type headache (16.8%; 95% CI 10.4, 25.1) in the control group was lower, even if not significantly, than in the MS population (19.8 and 27.7%, respectively; P = 0.5 for migraine and 0.06 for tension-type headache). Frequency of headache subtype in the control group is reported in Table 2. Characteristic of attacks are presented in Table 4.
Characteristics of attacks among MS and control subjects affected by headache
Prevalence of headache was significantly higher in the MS group, giving a crude OR of 2.24 (95% CI 1.27, 3.93; P = 0.005). The strength of association was the same also after adjustments for age and sex, the latter being considered as an a priori confounder, according to a multivariable model using logistic regression (OR 2.18, 95% CI 1.19, 3.97; P = 0.01). When analysis was restricted to the primary headache the OR, adjusted for age and sex, was 2.04 (95% CI 1.10, 3.79; P = 0.02). Analysis was also performed excluding the seven patients who developed headache only after IFN treatment; also in this case a positive borderline significant association was found (OR 1.69, 95% CI 0.97, 2.96; P = 0.06).
The prevalence of headache among the 52 MS patients who were never treated with IFN was higher than in the control group (48% vs. 37.6%), even if the association was not significant (OR 1.53, 95% CI 0.78, 3.02; P = 0.2); however, the reduced sample size probably led to a lack of power.
Discussion
Primary headaches are the most common forms of headache in the adult population.
Some studies have investigated the possible relationship between headache and MS, and the few published data suggest an increased frequency of primary headaches among MS patients, supporting the hypothesis of the existence of a common aetiopathogenic pathway between MS and headache. Unfortunately, most of these data come from clinical series and are consequently prone to selection bias (4–12). For these reasons, interpretation and comparison of results are often difficult. Furthermore, headache is a common condition and coincidental association can also occur, making a causative link difficult to establish.
To our knowledge, this study represents the first population-based study to evaluate the frequency of headache among a population-based incidence cohort of MS patients. Prevalence of headache found in our MS cohort (57.4%) was significantly higher than that found in the general Italian population (P < 0.0001), even if this comparison could be limited by the different age structure of the MS cohort (3). For this reason, we also selected a control group from the general population; prevalence of headache in the control group was close to that observed in the Italian population (37.6% vs. 39%) (3), and significantly lower with respect to our MS population. A significant positive association was found between MS and headache with an adjusted OR of 2.19 (P = 0.01) and a positive borderline significant association (OR 1.69, 95% CI 0.97, 2.96; P = 0.06) when the analysis was performed excluding the seven patients that developed headache only after IFN treatment. Our results are in agreement with those obtained in a large hospital-based case–control study recently carried out in Italy, where a prevalence of headache of 51% was reported among MS patients vs. a prevalence of 23% in the controls group, resulting in a close OR of 2.14 (30).
However, it should be emphasized that the trained interviewers were unmasked with respect to the disease status and possible interviewer bias cannot be entirely excluded, even if several strategies have been adopted during data collection to minimize the risk (standardized questionnaires, use of closed questions, training of the interviewers).
Concerning the possible association between headache and IFN-beta treatment (12), in our cohort only seven MS patients (14.3%) of the 49 under IFN treatment developed headache after the initiation of IFN therapy. This percentage is close to that reported in a study of 167 patients during immunomodulatory treatment (17%) (12), but lower if compared with the frequency found in a study recently carried out in Italy in a series of 137 MS patients (41%) (11). Nevertheless, due to the different study design, comparison with other hospital-based studies is difficult.
This represents the first population-based case–control study carried out to assess the relationship between MS and headache and, in agreement with more recent data coming from hospital series (9–11, 22), our data confirm that the prevalence of headache among MS patients is close to 60% and significantly higher than that reported in the general population.
The increased risk of primary headache in our MS cohort supports the hypothesis of a common pathway between these conditions; as suggested by other studies, the higher frequency of headache in MS subjects could be related to brainstem lesions (16, 17, 19). However, it should be noted that the role of brainstem in migraine pathogenesis is still controversial, and other types of study are needed to confirm this hypothesis.