Prevalence, burden, and risk factors of migraine: A community-based study from Eastern India
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.217979
Source of Support: None, Conflict of Interest: None
Background: Headache is common in communities; however, epidemiological research regarding its prevalence is infrequent in India.
Keywords: Disease burden, India, migraine, prevalence, risk factor
Headache is a common neurological complaint in communities. It may present as a primary headache syndrome, such as migraine, tension type headache (TTH) or cluster headache, or, the headache may be secondary to various illnesses. Community-based epidemiological studies are scarce in India and other developing countries. Based on new initiatives to make headache management universally accessible, studies on epidemiology of headaches have been reported from several developing countries such as Georgia, Pakistan, and China.,,, The prevalence of headache peaks in midlife, and is low among adolescents and after the age of 60 years. Migraine ranks among the top 20 causes of the disease burden as per the years lived with disability (YLD) criteria.
Cross-national studies report a low prevalence of headache among Africans [Table 1]. The regional variations in epidemiology as well as social, financial, and cultural factors influence the prevalence of headache experienced amongst individuals, particularly in a resource-poor setting. This makes it pertinent to study the phenomenon of headache separately in different populations.
In India, there is dearth of focused epidemiological studies on primary headache and its subtypes. The available data emerges from studies assessing headache as part of a generalized assessment of neurological disorders, resulting in methodological limitations.,, Headache is common in urban areas, afflicting 4.13% of the population; however, a recent focused survey found the 1-year prevalence rate of primary headache as 62.0% and that of migrainous headache as 25.2% from a southern state. As India is multiethnic, it is essential to determine the prevalence of headache and its subtypes in other regions also. Information about headache disorders is insufficient and variable, although migraine was identified to be more common in the rural population, with its prevalence ranging between 1.37% and 72%. Migraine was more frequent in female patients aged more than 21 years.
Hence, we planned a study to determine the 1-year prevalence of migraine in an urban setting. We also conducted a case-control analysis to determine the putative risk factors and measured their burden utilizing the disability adjusted life year (DALY) criteria.
The study was a cross-sectional, urban, community-based survey in Kolkata, the largest metropolis in eastern India. Approval was obtained from the Institutional Ethics Committee, and a written informed consent was collected from all respondents before the interview was conducted.
Kolkata spans 186 km 2 in area, with a population of 4.58 million (source: Census of India, 2001). It includes 141 municipal wards and is demarcated into 5200 blocks with 75–150 households per block on an average, according to the National Sample Survey Organization (NSSO). The block is the smallest socioeconomic unit as indicated by the documentation released by the Government of India Considering that Kolkata has a slum population of approximately 35–40%, we included NSSO blocks with predominant slum-type dwellings, called stratum I (total blocks: 727). Rest of Kolkata is divided into five strata depending on the north-south location and the presence of housing complexes [stratum-2 and 3 from northern part (total blocks: 2032); stratum-4 and 5 from southern part (total blocks: 1403), and stratum-6 from central part of the city (total blocks: 1038)]. To ensure sample heterogeneity, we adopted the stratified random sampling strategy. Proportionate number of blocks were selected randomly from each stratum using a random number table. A total of 44 blocks were selected. Atleast 50% of the families in each block were visited and screened. The study period was from 01.03.2011 to 31.07.2011.
Sample size calculation
The prevalence of headache reported in studies from different parts of the world varies from less than 10% to nearly 90%. Assuming a crude prevalence of 50% (without there being any definitive data from large scale, community-based studies from India), it was calculated that 2400 participants needed to be screened to determine the prevalence of headache with ±2% margin of error and 95% confidence level.
Participants aged between 20 and 50 years were chosen because of the maximal prevalence of headache in this age range. Patients of this age group are also in a productive period of their life. Their productivity has a considerable impact on both the individual as well as his/her family. India has a demographically young population (with 86.77% of the population below the age of 50 years, and 44.62% between 20 and 50 years). Hence, we planned to estimate the disease burden in this productive range.
Two control subjects (from the screened participants with a negative history of headache) without any history of headache were chosen randomly from the same family or from the neighborhood and were sex and age (±4 years) matched with the patient population. The family history of the control subjects was not taken into consideration while selecting them. The same questionnaire was administered to them to determine the putative risk factors for headache. The two controls were selected with an objective of narrowing the confidence interval.
The stress-arousal checklist (SACL) was used to assess the level of subjective reactions to various stressful situations in cases as well as in controls. The checklist has 19 items, with a Likert-type subjective scoring scale ranging from the response of “definitely feel” to “definitely do not feel;” in this scale, a low total score implies a greater tendency to be tense and physically stressed.
Questionnaire and data administration
We prepared our protocol based on the International Classification of Headache Disorders 2nd edition (ICHD-II), which was translated in the vernacular language, and test–retest reliability was determined for language validity. Subsequently, it was tested amongst the participants in a pilot study to measure the sensitivity and specificity of the screening questionnaire.
Based on the protocol, data were collected from selected families in two stages. First, trained field workers administered validated questionnaire in the vernacular language. All field workers had 15 years of formal education and were trained for 1 month at the institute regarding the assessment of headache, its subtypes, the associated symptoms, and the communication skills of their subjects. They approached the head of the family, or alternatively, an informed person within the family if the head of the family was absent, and made a list of the family members. From among those individuals suffering from headache, consenting participants in the appropriate age group determined for this study were listed if they were residing at the same address for at least 12 months prior to the initiation of the survey. If more than one person were having headache in the family, only one sufferer from each household who was articulate and agreeable was included in the study and individually interviewed. The key screening question was: “Have you experienced headache in the last one year?” In case the participant with headache was not available, visits were made in the evening following prior intimation to the family.
Respondents who responded with an affirmative response to the presence of headache were asked to describe the severity of headache using the visual analogue scale and to provide details of the demographic and clinical variables, including the duration, frequency, location, character of the headache, the associated symptoms accompanying it, the related psychiatric complaints, and the use of regular, pain-relieving medications. Questions were asked about the headache “triggers,” that is, the specific conditions that increased the chances of headache. Each individual with headache was subsequently scrutinized by neurologists and psychiatrists who performed both the physical and the mental status examinations that also included a fundoscopic examination. The available medical reports were reviewed to exclude secondary causes of headache. Neuroimaging was not a part of the protocol. The method of data collection and assessment is illustrated in [Figure 1]. The diagnosis of the types of headache was established utilizing the criteria of ICHD-II. Based on the pilot study and considering the neurologist's assessment as the gold standard, the sensitivity and specificity of the screening questionnaire in diagnosing the primary headache disorder was 80.7% and 100%, respectively.
The definition of the subtype of migraine and tension headache was based in accordance with the ICHD-II diagnostic criteria.
The prevalence was considered as an 1-year prevalence and was calculated as the ratio of number of cases with headache in the last year prior to the initiation of the survey to the total number of individuals seen, expressed in percentage. Poisson distribution was assumed for calculating 95% confidence intervals (CI) for the key data.
Numerical and categorical variables were compared between cases and controls using Student's independent samples t-test, Pearson's chi-square test, or Fisher's exact test, as appropriate. Analysis was two-tailed and statistically significant at P < 0.05. To identify the risk factors responsible for the migrainous headache, the statistically significant factors (P < 0.1) based upon the univariate analysis were subjected to the binary logistic regression model. Statistical version 6 (Statsoft Inc., Tulsa, Oklahoma, 2001) and Statistical Package for the Social Sciences (SPSS) version 17.0 (Illinois, Chicago: SPSS Inc., 2008) software were used.
Disability adjusted life year calculation
The DALY metric was used for calculating the burden of migraine within the community. DALY is the sum of years of life lost due to premature mortality (YLL) and the years of life lived with disability (YLD). There was no YLL component in the subjects suffering form primary headache as no deaths were attributable to headache per se. For calculation of the YLD value, disability weights (mild 0.05, moderate 0.16, severe 0.735) for migraine were selected based on the severity of the headache, according to the World Health Organization Global Burden of Disease (GOB) 2000 estimates. No weighting based on age or time discounting were incorporated in the calculation of the DALY values attributable to headache.
The step-wise case detection of patients suffering from primary headache as well as specifically the migrainous headache are illustrated in [Figure 1]. The figure also includes information regarding non-participating patients and those who refused to undertake the study.
Demographic characteristics of migraineurs
Among the subjects suffering from migraine (n = 342), the mean age was 32.2 years (standard deviation [SD] ±7.85). Two hundred and eighty participants (81.87%) were female patients; 161 (47.08%) of the participants resided in slum areas. Majority (204; 59.65%) of them were housewives, 277 (80.99%) were married, and 260 (76.02%) belonged to nuclear families. Most of the participants had (81.38%) an average monthly family income between Indian rupee (INR) 2501 and 10000 (US $38.47 to US $154); 3.22% were unemployed. Approximately half (52.92%) of the migraine sufferers had studied until the secondary level whereas 13.29% of them were illiterate. Complaints of headache based on the proband subject's history were present in the first-degree relatives in 64.89% of the cases. This family history was corroborated from other family members also to ensure the veracity of information obtained. [Table 1] shows the various epidemiological studies reported from India and abroad. [Table 2] shows the frequency of the overall proportion suffering from headache and also the frequency of those specifically suffering from migraine in the study. [Table 3] reveals the various factors triggering, aggravating, or relieving migraine. Spontaneous pain relief without recourse to any interventions occurred in 41.22% of the individuals. The most important triggering factors were 'prolonged exposure to sun', followed by 'sleep deprivation' and 'long distance travel'. The most common relieving factors were 'sleep' followed by 'drinking hot caffeinated drinks'.
The subanalysis [Table 4] revealed that individuals having migraine associated with an aura were significantly older (P = 0.004) than those without an aura. Significant differences existed between groups in relation to the site of headache and its perceived intensity.
[Table 5a] shows the comparison of different variables between the cases and the matched controls. Among migraineurs, a significant difference was observed in the prevalence of migraine based on factors such as whether or not the subject was a slum dweller; the family income; the family structure; the literacy status; the marital status; the presence or absence of diabetes mellitus, anxiety and depression stress or environmental exposure; an erratic sleep timing; prolonged and regular television viewing; long distance travel; and, the use of oral contraceptives. [Table 5b] lists the univariate odds ratios and adjusted odds ratios based on the logistic regression analysis A lower educational status, adverse environmental exposure (to sun, pollutants or strong odors), long-distance travel, and the use of oral contraception emerged as significant risk factors responsible for the development of migraine in our cohort. Logistic regression analysis returned a Nagelkerke's R2 value of 0.744, suggesting that nearly 75% of migraine risk was explained by the predictors selected for the current model and that 92.6% of the cases were correctly predicted by the present methodology.
[Table 6] provides the age and sex distribution of primary headache burden for the study population based upon the DALY criteria. The DALY figures projected per 100,000 population reveal that the maximum burden of headache is experienced by individuals aged between 25 and 30 years among male subjects and between 30 and 34 years among the female ones.
This was a two-stage, cross-sectional, urban community-based study from eastern India where a randomly selected sample of individuals, aged between 20 and 50 years, were assessed for migraine using the globally-accepted criteria. About 14.12% of the respondents had migraine. The maximum burden was among women aged between 30 and 34 years. The triggering, aggravating, and relieving factors of the migrainous episodes were identified, and the case-control analysis suggested that approximtaely 75% of putative risk factors were responsible for migraine with a high predictability. To the best of our knowledge, this is first population-based survey to estimate the prevalence of migraine from eastern India that provides a comprehensive estimation of the disease burden in India based upon the global burden of disease 2012 study.
A house-to-house survey is considered as the gold standard for estimating prevalence of a disease. The stratified random sampling, training of interviewers, two-stage assessment, standardized case ascertainment, and expert diagnostic evaluation also helped in overcoming methodological challenges. In the absence of biological markers or routine neuroimaging for primary headaches, the ICHD-II criteria, as applied worldwide, helped in achieving a relatively high precision in the diagnosis of headache and its subtype.
The prevalence of migraine varies from country to country. In Europe, the 1-year prevalence of migraine was estimated to be 14% with the peak incidence occurring between the age range of 20 and 50 years. In USA, the 3-month overall prevalence of migraine was 14.2%, with the maximum incidence occurring in subjects aged between 18 and 44 years. In Latin America, the 1-year prevalence of migraine varied from 6.1% to 17.4% among women and 2.9 to 7.8% among men across different countries. In Africa, the pooled migraine prevalence was 5.6% among the general population. In Asia, the prevalence has been estimated to be between 8.4% and 12.7%. Thus, our data is similar to that obtained in most other countries except Africa. The mildly varying rates may be related to the differing methodology adopted, differences in defining the criteria of headache prevalence (1 year vs. 3 months), coexisting environmental factors, urban/rural differences, or ethnicity of the studied population.
One study from south India has also documented a higher prevalence of migraine among the population studied. A plausible reason for this high incidence is that both in that study as well as ours, the sample group belonged to an age range where a peak prevalence of migraine has been reported. Possibly, migraine is more prevalent in communities but remains unnoticed as the patients suffering from it do not seek treatment, or alternatively, practice self-medication. The spontaneous remission of the attack discourages them from visiting doctors. Thus, headache becomes an integral part of the migraineurs' lives. Furthermore, the socioeconomic conditions in India dictate a higher priority for other problems of daily living. One of the possibilities related to the high prevalence rate for migraine detected in the community is the misdiagnosis of episodic tension headache as migraine, that may be difficult to differentiate in a prevalence study of this type.
The nature of the triggers of migraine in our study was comparable to those found in other studies, although the relative frequencies of the triggers varied. A condition may not be a consistent trigger for each individual. Among dietary triggers, irregular mealtimes and intake of food, such as chocolate and cheese, were more commonly associated in patients with migraine than in controls. Interestingly, fasting was less frequent among those patients who were suffering from migraine and was possibly avoided due to its triggering effect in precipitating an episode of migraine. Intake of spicy food was higher in the control group, raising the possibility that ingredients in such foods may have protective effect. Most of these dietary factors, though significant on univariate analysis, did not remain so on the multivariate logistic regression analysis.
Migraine was more prevalent in women, with a female:male ratio exceeding twice or more. A higher prevalence among women may have been related to the hormonal differences between the genders or due to specific psychosocial stressors affecting women., Regular medications, primarily the intake of oral contraceptive pills, was significant in the migraine group.
Slum dwelling, a poor income, lack of education, absence of family support as commonly noticed in a nuclear family, and the marital status were more associated with migraine on univariate analysis. Effective stress management skills has the potential to reduce the impact of these factors on migraineurs. The factors evaluated here could explain the risk of migraine in nearly 75% of the patients.
Previous studies have described clinical differences between migraine with and without aura, apart from obvious differences in the headache phase itself. A sub-analysis among the two migraine cohorts revealed the existing clinical differences although psychosocial factors were comparable [Table 3].
The burden of migraine in the present study represented only the episodic migrainous headache. However, it represented a significant part of the disease burden because the most common types of headache and the more active age groups suffering from headache were surveyed. The age and sex distribution of disability adjusted life years in patients suffering from primary migrainous headache were different based on the severity of the migraine they were suffering from [Table 6]. Studies on the headache burden are limited. The variable reports suggest that migraine represents a major portion of the burden of the disease and is more severe in mid-life amongst women.,
The prevalence of headache varies in different continents and is possibly related to the methodology of assessment, the environmental factors, or ethnicity. In India, headache, being common, is often not reported if it is trivial. It may be due to cultural and social influence of concealing health problems, leading to under-reporting of cases. Mild headache is mostly due to tension headache and is often ignored. This can explain the low reporting of overall patients suffering from headache in this study. A recall bias may also influence the outcome. Hence, in previous studies, the actual prevalence of headache was documented to be low. Due to exclusion of cases, both the first and second phases of the survey in this study might have underestimated the prevalence. No case of medication overuse headache was detected, which may be related to the self-procurement of drugs by the patients. Similarly, no case of trigeminocephalic headache was diagnosed, indicating the need for a larger sample size to actually determine its prevalence.
The prevalence of migraine found in the present study was similar to that reported in studies from both developing and developed countries, except Africa. The presence of significant risk factors indicates the need for a better understanding of the problems of the headache sufferers and their proper counseling to reduce the burden of headache. The significant burden of headache, especially in the most productive age-groups of the population, emphasizes the urgent need for institution of public health measures to mitigate the effects of the disease.
We are thankful to the Indian Council of Medical Research (ICMR), New Delhi, India, for partially sponsoring this study (Project No. SWG/Neuro/20/2005/NCD-I). Dr. Sujata Das drew remuneration as a Senior Research Fellow from this project. No other authors have financial disclosures to make. We are also grateful to the community participants for their active involvement in this study.
Financial support and sponsorship
partial funding from ICMR, New Delhi.
Conflicts of interest
There are no conflicts of interest.
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5a], [Table 5b], [Table 6]