Abstract
We present the results of a community survey based on the diagnostic criteria of the International Headache Society (IHS) describing headache prevalence and symptomatology in the Singapore population. A questionnaire administered by trained personnel was completed by 2096 individuals from a randomized sample of 1400 households. The overall lifetime prevalence of headache was 82.7%. The migraine prevalence was 2.4% in males and 3.6% in females; for episodic tension-type headache and chronic tension-type headache the corresponding figures were 11.1%/11.8% and 0.9%/1.8%, respectively. Inclusion of borderline cases (IHS codes 1.7 and 2.3) resulted in prevalences of 9.3% for migraine, 39.9% for episodic tension headache and 2.4% for chronic tension headache. Headaches described by 31.2% of the respondents were unclassifiable. The different premonitory symptoms, precipitants and aggravating factors in migraine and tension-type headache in our study population suggest that they represent two distinct syndromes rather than opposite ends of a clinical spectrum.
Introduction
Several community-based studies (1–11) in various countries have explored the prevalence and associated features of the primary headache syndromes according to the operational diagnostic criteria of the International Headache Society (IHS) (12). We have previously used a self-assessed questionnaire to investigate headache features in a local undergraduate population fluent in the English language (13), and now present the findings of the first study of headache diagnosis and prevalence in the highly urbanized, multiethnic (79.0% Chinese, 14.4% Malay, 6.2% Indian) Singapore community. The relative contribution of individual IHS criteria in diagnosis of headache in the study population as well as the relationship of headache type to age, gender, premonitory symptoms and precipitants are given special attention.
Materials and methods
A questionnaire incorporating demographic data, headache symptomatology and management as well as the operational diagnostic criteria of the IHS was administered during door-to-door visits by undergraduates of a local university to 2096 individuals ≥12 years of age. The sampling frame was based on data from the 1990 Singapore Population Census and involved 1400 randomized households nationwide equivalent to a estimated study population of 3000. Interviewers attended a training session to ensure an adequate understanding of the information required from each question, and to standardize questions in Mandarin, Mandarin dialect and Malay where the respondents could not understand English. In order to encourage an adequate response rate, the local newspaper and television media provided publicity regarding the survey as well as on headaches in general. Interviewers were instructed to visit households after office hours to permit inclusion of working and schooling family members in the interview process. A household was considered non-respondent after failure of three separate attempts at contact on 3 different days. A research co-ordinator subsequently confirmed each visit to a household and rectified obvious errors and omissions in the interview form with a telephone call.
The characteristics of the ‘usual’ headache experienced by the interviewee were taken into account in the questionnaire. Blood pressure, urinalysis and visual acuity assessments were obtained in consenting individuals, but no formal physical examination was performed.
Data were entered into a database program (Access 2.0; Microsoft) and subsequently analysed with the Statistical Package for the Social Sciences (SPSS for Windows, Chicago, IL, USA) and Epistat programs. The χ2 test or Fisher’s exact test was used to determine the significance of differences in proportions, while Student’s t-test and analysis of variance with Duncan post hoc testing were used to determine the significance of differences of means. The χ2 test for trend was used to determine the significance of age-related changes in proportions. P-values < 0.05 were regarded as statistically significant.
Headaches were classified as migraines without aura (IHS code 1.1), episodic and chronic tension-type headaches (IHS code 2.1 and 2.2, respectively), as well as migrainous headache and tension-type headache not fulfilling all the diagnostic criteria of the previous categories (IHS codes 1.7 and 2.3, respectively). The identification of migraines with aura according to the IHS classification in a questionnaire-based study was not attempted because of the complex temporal criteria involved. For this reason, only approximations of the prevalence of the latter condition were possible. Specific inquiries were made regarding visual scotomata, ‘basilar’ symptoms (vertigo/diplopia) and focal sensory/motor deficits. No attempt was made to identify further diagnostic categories because a physician-directed interview and examination was not performed.
The age of headache onset for each individual was taken to be the midpoint of the range provided if an exact figure could not be provided. ‘Environmental factors’ that precipitated headaches were defined as changes in ambient temperature, lighting and air quality.
Results
Population profile
An estimated 69.9% (2096/3000) of the sample population underwent an interview. Their mean age was 35.6 ± 11.9 years (range 14–74 years). Males comprised 47.2% of the respondents, females 52.8%. The racial distribution of our study population (79.1% Chinese, 14.4% Malay, 6.2% Indian, 0.7% other races) did not differ significantly (P < 0.05) from that of the population at large (76.9% Chinese, 14.0% Malays, 7.7% Indians and 1.1% other races) as found in the 1990 National Census. The overall lifetime prevalence of headache was 82.7%: 80.0% in males and 85.1% in females, a statistically significant gender difference (P = 0.002). The corresponding figures for point prevalence, defined as the proportion of respondents with headache at the time of interview, were 5.1% and 5.2%, respectively. No significant gender predisposition was therefore present (P = 0.95). Table 1 and Table 2 summarize the age-related changes in gender, headache diagnosis and prevalence as well as mean age of headache onset. Lifetime prevalence diminished with increasing age (P = 0.001), but point prevalence did not significantly correlate with age (P = 0.11), gender (P = 0.47), or mean age of headache onset (P < 0.05). The mean age of headache onset was earlier in men (17.0 ± 8.9 years) than women (18.4 ± 8.9 years) (P = 0.001). The age of onset of headache did not differ between individuals experiencing migrainous (IHS codes 1.1 and 1.7) and tension-type headaches (IHS 2.1–2.3) (P = 0.53).
Gender, headache prevalence and age of onset by age group within the study population
Age- and gender-specific prevalence rates for migraine without aura and episodic tension-type headache in the Singapore population (n = 2096)
Headache classification
A female preponderance was present amongst headache sufferers as a whole (80.1%:85.1%, P = 0.003), as well as in those with migrainous (IHS codes 1.1 and 1.7) headaches (7.1%:11.1%, P = 0.002). Males and females were equally likely (41.9%:42.5%, P = 0.78) to experience tension-type (IHS codes 2.1–2.3) headaches (Table 3).
The relationship of the IHS headache classification with gender
∗Significant (χ2 test) gender difference within each diagnostic group.
Symptoms possibly due to a migrainous aura were reported by 275 individuals (13.1%). These included 25 respondents otherwise classifiable as having migraines without aura, 12 otherwise classifiable as having chronic tension-type headaches and 24 otherwise classifiable as having episodic tension-type headaches. No aura types other than those included in Table 3 were reported in association with a headache. Without taking into account the temporal features of the aura, migraine with aura had a potential prevalence ranging from 1.2% (if all the features of migraine without aura were taken as a prerequisite) to 13.1% (if all other features were disregarded). The prevalences of other headache entities were calculated without taking into account the diagnostic implications of the possible aura. Three respondents fulfilled the IHS criteria for both chronic tension-type headache and migraine without aura; they had chronic (>6 months), frequent (>180 days/year) headaches that were bilateral, throbbing, worse with exertion, mild or moderate in intensity and associated with nausea but not with vomiting, phonophobia nor photophobia.
Table 4 shows that no single criterion used in the diagnosis of either migraine or episodic tension-type headache was associated with both high sensitivity and specificity. The great majority (87.2%) of borderline (IHS codes 1.7 and 2.3) cases of migraine and tension-type headache failed to meet IHS criteria for previous number of similar episodes (43.9%) or for duration of the episode (43.3%) (Table 5).
Specificity and sensitivity of the IHS criteria in migraine and episodic tension headache
There were a total of 1734 headache sufferers in the study population.
The IHS diagnostic criteria in subjects with headache not completely fulfilling the criteria for migraine and tension-type headache (IHS codes 1.7 and 2.3, respectively)
Headache frequency, associated symptoms and precipitants
Over the previous year, 34.4% of the migraineurs experienced headache episodes ‘every few months’, 53.1% ‘a few times a month’ and 10.9% ‘a few times a week’. Migraines occurred significantly more frequently than episodic tension-type headaches (P < 0.001) or headaches in general (P < 0.001).
Insufficient sleep was the commonest precipitant of tension-type headache and headaches in general, and was second only to ‘mental stress’ in importance as a headache precipitant in migraineurs (Table 6). Alcohol was the commonest single precipitant mentioned by those whose headaches were brought on by dietary factors (20.4%, 21/103 respondents), followed by spicy food in 9.7% (n = 10), ‘heaty’ food in 7.7% (n = 8), coffee in 5.8% (n = 6) and ‘oily’ food in 4.9% (n = 5). No instance of food craving was noted preceding headaches. Table 6 also shows several significant differences in the other precipitants and associated symptomatology of migraines and tension headaches.
Headache precipitants and associated symptoms in migraine and tension-type headache
∗Significant difference (χ2 test) in the prevalence of a particular symptom between migraines and tension headaches.
Discussion
Headache classification may be difficult in questionnaire-based prevalence studies, varying from that obtained from a clinical interview even when the two are performed on the same occasion (14). The IHS classification requires exclusion of a secondary headache syndrome with a physical examination, but the rarity of isolated symptomatic headache syndromes (15) ensures that, while a clinical assessment is important in individuals, it assumes less significance in epidemiological studies.
It is especially difficult to diagnose migraines with aura in a non-physician directed context (14, 16). The relatively high rate of aura-like symptoms in our population may be due to a degree of suggestibility in the interview process, since the visual, basilar and hemisensory/hemiparetic symptoms specifically inquired for were common, but no other symptoms (such as speech difficulty) were mentioned. Visual and focal sensorimotor, but not ‘basilar’, symptoms do occur much more frequently in those with migrainous pain than in those with tension-type headache, and it is likely that some of these are true auras. In view of the known relative prevalence of migraine with and without aura, and because pain features of the latter are usually present in the former entity (17), its inclusion as an additional diagnostic category is unlikely to affect greatly the proportion of unclassifiable headaches in our population.
The diagnosis of migraine without aura is a less subjective process, and the subset of respondents classified as migraineurs do share symptoms unrelated to the IHS criteria that set them apart from those with episodic or chronic tension headache. Together with other epidemiological evidence (18), these differences between migraine and tension headache suggest that they are less likely to represent opposite ends of a spectrum of a single disorder, but are conditions with distinct clinical features and origins.
Several reasons may account for the relatively high proportion of atypical and unclassifiable headaches in our study population. The absence of a physician to guide the process of history taking and diagnosis must rank as the most important. The sole published study of headache prevalence in a general population based on a clinical interview (4) found that the IHS criteria successfully classified all but two of 740 cases. The request for respondents to provide the ‘usual’ details of only their ‘usual’ headaches would also have resulted in loss of information and the possibility of confusion in recall where more than one headache type, or overlap headache entities, was present in an individual. Because tension-type headache is more common than migraine, it is reasonable to assume that migraine prevalence would be underestimated because of cases with overlap headaches or multiple headache types. It is also possible that the duration of many episodes of ‘true’ migraine and tension headache in the Singaporean population is shorter than that demanded by the IHS classification. Finally, the diagnostic requirement for at least 10 similar episodes of episodic tension-type headache is probably unnecessarily restrictive in a community-based study.
The increasing number of published community-based studies using the diagnostic criteria of the IHS has made it evident that the wide geographical variation in migraine prevalence cannot be adequately accounted for by differences in case definition. The 3.1% prevalence of migraine in our study population is consistent with the finding of a generally lower prevalence of migraine in predominantly Chinese (0.63–1.5%) (8, 19, 20) compared with Caucasian (8.1–16%) (4–10) or African (3.0–7.2%) (11, 21–23) communities. The relative importance of genetic and environmental/cultural factors remains unclear. There is a small, but definite (M:F = 1:1.5) female preponderance in our migraineurs, rising to 1:1.75 if all migrainous headaches are included. This is less marked than the 1:2.2–3.1 usually described elsewhere. Migraineurs in Singapore have a higher frequency of attacks than the ‘less than once a month’ described in a recent review (24), but similar to that found in a German (9) and in an Ethiopian (11) study. In view of the relatively greater disability from migraines compared with most other headache types, this finding may have important socio-economic implications.
The epidemiology of tension-type headache is not as well studied as that of migraine, and less comparative data based on the IHS criteria are available. Other investigators have found a prevalence of 27–78% for the episodic form (3, 4, 8, 9) and 3% for the chronic form (24), but both types of tension-type headache in Singapore are relatively less common. Some (1, 3, 4, 8, 11, 24) have found a female predominance in tension headache, but others (9, 21, 25), like ourselves, have found no significant gender differences. Unlike the case in migraine, the frequency of tension headache episodes in Singapore is probably lower than that found elsewhere (4, 8, 9).
Our findings concur with those of other investigators (1, 3, 11, 20, 26–28) in identifying the chief headache precipitants as mental/emotional stress, physical ever-exertion, weather and insufficient sleep. These appear to trigger both migraines and tension-type headaches. Conversely, menstruation, alcohol and food appear more likely to predispose to migraine rather than tension-type headache. The foods rich in bioactive amines (cheese, red wine, chocolate) which are usually thought to be important (29) in triggering migraines are not prominent in the diet of the majority Chinese population, which may account for the relatively low importance of food and alcohol as headache precipitants in Singapore. It is interesting that a number of headaches are attributed to ‘heaty’ food in which the ‘yang’ (or ‘male’) element predominates over the ‘yin’ (or ‘female’) element. This is a concept found in traditional Chinese medicine but not in Western dietetics.
The local prevalence of migraine peaks in the fifth decade, similar to the pattern found in several other population studies (24). The relationship of tension headache with age is less clear. There is a fall in lifetime headache prevalence with increasing age, and the markedly different mean age of headache onset reported by the oldest population subgroup suggests that the most likely cause for this is poor recall. An increasing tendency for younger cohorts to have more headaches or for headaches to remit with advancing age cannot be excluded, and the third alternative – that headache sufferers die at a younger age – is not supported by available evidence (30).
Headache is a common symptom in the Singaporean population, with a prevalence intermediate between that found in most international community studies and in studies of other predominantly Chinese populations. The IHS criteria successfully classified more than two-thirds of the headaches reported by the respondents in a questionnaire-based study and were associated with minimal diagnostic overlap. Failure to meet the criteria with respect to number of previous similar headache episodes and/or duration of headache episodes was a common occurrence, suggesting that a more liberal approach to diagnostic classification could be taken in community-based prevalence studies. Our data suggest that tension headache and migraine are distinct entities with different precipitants and associated symptoms.
Footnotes
Acknowledgements
This study was supported by a grant from Glaxo-Wellcome (Singapore).
Excerpts from the questionnaire relevant to the IHS classification
