Prevalence of Migraine Headache in Canada: A Population-Based Survey

Transcription

1 International Journal of Epidemiology O International Epidemlologlcal Association 994 Vol. 23, 5 Printed In Great Britain Prevalence of Migraine Headache in Canada: A PopulationBased Survey BERNIE O'BRIEN,*" RON GOEREE* AND DAVID STREINER** O'Brien B (Department of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, Ontario, Canada), Goeree R and Streiner D. Prevalence of migraine headache in Canada: A populationbased survey. International Journal of Epidemiology 994; 23: Background. The aim of the present study was to estimate the prevalence of migraine headache among Canadian adults (aged * years) and anatyse vanation by age, gender, household income and province of residence. Methods. A populationbased survey was undertaken using telephone interviews with 2922 adults who were randomly selected from households across Canada by stratified regional sampling. The questionnaire asked about frequency and characteristics of headaches experienced and other symptoms. The diagnostic critena of the International Headache Society were used to classify people as migraineurs (with or without aura), headachers or nonheadachers. Results. Of 92 random calls to households, 4235 were eligible and 2922 interviews were successfully completed (response rate 66%) The prevalence of migraineurs, headachers and nonheadachers among males was 7.%, 76.%, 6.% and among females was 24.9%, 65.6%, 9.4%. For females prevalence appears to increase with age, peaking at 4044 years and declining thereafter. Sexspecific prevalence for males and females, controlling for age, was significantly lower in the province of Quebec compared to other provinces. We found no association between migraine prevalence and household income. Of 500 people classified by IHS criteria as migraineurs only 232 (46%) reported any migraine diagnosis by a physician Conclusion. We estimate that 2.6 million adult females and 0. million adult males in Canada are migraineurs, but only half are likely to have been diagnosed by a physician. Contrary to a recent US survey, people from lower income households in Canada are not at greater risk of migraine. The lower prevalence of migraine in Quebec was unexpected and remains unexplained, but it may be influenced by language/translation problems. There is wide variation in the reported prevalence of migraine in both adult men (9%) and women (329%).' Variation in prevalence reported in earlier studies is likely due to four main factors: first, the lack of standardized diagnostic criteria which would permit consistent differentiation between migraineurs and people with other types of severe headaches; second, variation in populations surveyed, some estimates being based on those seeking treatment for migraine while others are based on more representative population samples; 2 third, variation in the time period used for prevalence estimates, some studies being based on lifetime prevalence 3 ' 4 and others on the respondents' experience in the previous year; 3 ' 6 and fourth, true variation in the underlying prevalence of the disease between countries or regions studied. Department of Clinical Epidemiology and Biostaustics, McMaster University, Hamilton, Ontario, Canada. ** Center for Evaluation of Medicines, St Joseph's Hospital, Hamilton, Ontario, Canada. * Department of Psychiatry, McMaster University, Hamilton, Ontario, Canada. Reprint requests to" Dr Bernie O'Brien, Centre for Evaluation of Medicines, St Joseph's Hospital, Martha Wing, Room H329, 50 Chartton Avenue E, Hamilton, Ontario, Canada LN 4A An important advance in the standardization of migraine diagnosis was the publication of criteria by the Headache Classification Committee of the International Headache Society (IHS) in 9. 7 The aim of the present study is to use the IHS diagnostic criteria to estimate the prevalence of migraine in the Canadian adult population by telephone interview of people randomly selected from households across Canada. We also wished to test some prior hypotheses concerning demographic features associated with migraine: in particular, previous studies have found (i) prevalence to be higher in females than males, 3 ' 6 ' ' 9 (ii) prevalence to peak (particularly for females) between the ages of 35 and 55 ' 9 and (iii) prevalence to be higher in those with lower household income. In addition, we were interested in examining regional variation in migraine prevalence within Canada and to compare our Canadian data with published estimates from other countries. METHODS A populationbased telephone survey was conducted among adult (aged ^ years) Canadians between May and July 992 to estimate the prevalence of migraine using IHS criteria.

2 MIGRAINE IN CANADA 02 IwBfBile* n>4274 Rrfoab n 05* D tni DIIIAWI cidi ESfibb HourhcJdi o» 4235 Completed IutcTTiewi n 2922 ESfEMfitj t Determined n 42 ftnimn^flil CallBaefci B57 Household eligibility rate = 52% (4235/ ) Overall response rate = 66% (2922/ (0.52 x 42)) Includes nonresidential and not in service numbers, unable to speak English or French etc. FIGURE Household contacts, completed interviews and response rate Survey Methods The survey fieldwork was conducted by the Institute for Social Research, York University, Toronto. Using IHS criteria, a questionnaire was developed for computerassisted telephone interviewing (CATI) to classify people as migraineurs, headachers or nonheadachers. The method of random digit dialling was used to randomly select households within regional blocks. Within each randomly selected household we randomly selected one adult the person with the most recent birthday. 0 Respondents were interviewed in either English or French. A total of 92 phone numbers were called, of which 4235 were eligible households (i.e. residential phone and able to speak English or French) and agreed to be interviewed (Figure ). A total of 2922 interviews were completed for an overall response rate of 66%. The distribution of sampling between provinces was in proportion to population. Following the convention in survey research, 0 " 2 adjustments were necessary to make the final sample representative of the Canadian population: cases were weighted to adjust for household size (otherwise individuals in smaller households would be oversampled) and regional under or oversampling. Further details of case weighting, call backs, survey design and the questionnaire are available from the authors. International Headache Society (IHS) Criteria In a large population survey of migraine it is not feasible to base diagnosis upon clinical examination. The IHS criteria are valuable for population research because they offer a standardized and structured A. Headache Characteristics B. Concurrency of Headache Characteristics C. Concomitant Symptoms D. Duration E. Frequency Al: Unilateral location (one side of head) A2: Pulsating quality A3: Moderate or severe intensity (limits daily activities) A4: Aggravated by physical activity Headaches must be characterized by two or more of A to A4 at the same time (e.g. unilateral and pulsating headache) During headache at least one of the following must be present: C Nausea and/or vomiting C2: Photophobia and phonophobia Headaches fulfilling AC must last 472 hours if untreated or unsuccessfully treated At least 5 attacks in a person's lifetime fulfilling AD Source: Adapted from International Headache Society. 7 FIGURE 2 International Headache Society Criteria for defining migraine with or without aura method for taking a medical history as the basis for diagnosis. The IHS categories for migraine are the general class of migraine without aura (IHS.) and the less common form of migraine with aura (IHS.2). This latter is a subset of the former with the additional requirement of aura; since we did not seek to differentiate on the basis of aura our estimates are for migraine with or without aura (IHS. and IHS.2). The IHS criteria for migraine are presented in Figure 2. In summary, to be classified as a migraineur respondents must have experienced five or more headaches in their lifetime fulfilling the various pain characteristics, duration and concomitant symptoms indicated in Figure 2. Our phone questionnaire comprised a series of questions about headaches and other symptoms with the intent of determining whether the respondent met the cumulative criteria (AE) to be classified as an IHS migraineur. Where appropriate, the survey questions were rephrased in everyday language; for example, unilateral headache was phrased as 'headache on one side of your head'. Although the IHS criteria can be used to subclassify people experiencing nonmigraine headaches into those with tensiontype and cluster headaches, this differentiation was not an objective of our study and would have added substantially to interview time and cost.

3 022 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY Therefore people reporting nonmigraine headaches were simply classified as headachers. The third class of respondents were those who reported no headache in the past year (nonheadachers). Definition of Prevalence Point prevalence is the percentage of a defined population who have a disease at a single point in time; period prevalence is the same concept relating to a period of time. Strict application of the IHS criteria leads to an estimate of lifetime (period) prevalence because the headaches for people to qualify as a migraineur can have occurred at any point in their lives (criterion E, Figure 2). Limitations of lifetime prevalence are that it is retrospective, prone to recall bias, and gives no clear indication of those who currently have active disease. An alternative reported in some migraine studies 5 ' 6 is period prevalence based on the year previous to the survey. We estimated both lifetime and previousyear prevalence people who are lifetime migraineurs and who experienced one or more migraines in the previous year. Respondents were kept blind to their classification by IHS criteria and all questions were framed in terms of the respondent's headaches and symptoms, avoiding the use of the label 'migraine'. The exceptions were two direct questions on selfreport of migraine: (i) Had the respondent ever experienced migraines? (ii) Had migraine ever been diagnosed by a physician? Statistical Methods Prevalence is expressed as a proportion and 95% confidence intervals (CI) were calculated based on the binomial approximation to the normal distribution. Sexspecific migraineur status was modelled as a binary dependent variable (yes/no) using multiple logistic regression to examine associations with covariates of age, region and household income. For sexspecific prevalence we report the adjusted relative odds of migraine between levels of each covariate. Statistical significance is reported at the 5% level of probability (twotailed). RESULTS The demographic characteristics of our 2922 survey respondents were very similar to those of the Canadian adult population, with 54% being female and 4% in the age range 2544 years. The demographic characteristics of the caseweighted sample were also very similar to the unweighted sample. The majority of respondents (62%) were in either fulltime or parttime employment. TABLE I Classification of migraineurs, headachers, and nonheadachers based on two prevalence criteria for migraine: lifetime and previous year Migrajneurs b Headachers nheadachers Total Males. (% prevalence) Previous year* 05 (7.) 025(76.) 27(6.) 347(00) 00(7.4) 030 (76.5) 27(6.) 347(00) Females. (% prevalence) Previous year 392 (24.9) 032(65.6) 4 (9.4) 572(00) 344(2.9) 00 (6.7) 4(9.4) 572(00) People who fulfil IHS lifetime criteria for migraine and who experienced one or more migraines in the previous year. b Sex of respondent was missing for 2 cases. c Sex of respondent was missing for case. ialfraport«d migraine (n5) migraine by IHS criteria (n500) raportad doctor.diagnosed mlgratna in335) FiouRE 3 Relationship between selfreported lifetime migraineur, selfreport of doctor diagnosed migraineur and IHS criteria migraineur Migraine Prevalence by IHS Criteria Classification of respondents by IHS migraine criteria is presented in Table. The estimated lifetime prevalence of migraine is 7.% in males (95% CI: ) and just over three times higher in females at 24.9% (95% CI: ). In addition, a total of 025 males (76%) and 032 females (65.6%) reported headaches but did not fulfil criteria A to E (Figure 2) to be classified as migraineurs. Our more conservative definition of previousyear prevalence yielded slightly lower rates in both males (7.4%) and females (2.9%). SelfReported Migraine Selfreported migraine is illustrated as a Venn diagram in Figure 3. Of 500 people who were classified as IHS migraineurs, only 232 (46%) reported having migraine diagnosed by a physician. Of the 5 who selfreported

4 MIGRAINE IN CANADA 023 that they experienced migraines only 293 (57%) were classified as IHS migtaineurs. Of the 335 people reporting migraine diagnosis by a physician, 232 (69%) were classified as IHS migraineurs. 00 o fwntlit InB72I mil«i (n347) ±95* a Frequency and Duration of Headaches Migraineurs reported a median of seven migraine headaches in the previous 2 months compared to a median of five headaches in headachers; but the distributions were highly skewed with 9% of migraineurs reporting >52 in the previous year. The reported typical duration of headaches was also skewed with a median of 24 hours for migraineurs and 2 hours for headachers. When migraineurs were asked when they had last experienced one of 'their headaches' (i.e. migraine) the median interval was 4 days; 3 respondents (%) reported that they had a migraine at the time of interview. Demographic Covariates Sexspecific lifetime and previousyear prevalence of migraine by 5year age groupings is shown in Figures 4a and 4b. Although 95% CI overlap for males and females at the two age extremes, for the majority of the life cycle the lifetime prevalence clearly differs by gender; male prevalence peaks in the early twenties and declines thereafter, while female prevalence follows an inverted 'U' shape, peaking at around age 3550 years. For previousyear prevalence the pattern is similar but many CI overlap, except for ages 3539, 4554 and 6064 where prevalence is clearly higher among females. Analysis of sexspecific lifetime prevalence by covariates of age, region and household income is presented in Table 2. The most striking and consistent regional result is that prevalence appears to be significantly lower in Quebec (5.2% males, % females) than in other provinces. For example, the logistic regression suggests that the relative odds of being a migraineur in Quebec (relative to Ontario) is 0.4 in males (95% CI: ) and 0.59 in females (95% CI: ). There appears to be no consistent linear relationship with household income; the data are not consistent with a prior hypothesis (from the US) that migraine prevalence is greater in lower income households. DISCUSSION Our survey data indicate that, by the lifetime prevalence criteria of the International Headache Society, 2.6 million adult females and 0. million adult males are migraineurs in Canada. Furthermore, the majority of these had experienced one or more migraines in the FIGURE 4a 5 a. jc if O n. FIGURE 4b by age C M M c n ^ ^ m i a e ^ Q i i i i i A I ii A " age In year* Sexspecific lifetime prevalence of migraine by age CM N en en I I I I a m o m CM CM en en age in years * o + T V S Sexspecific previousyear prevalence of migraine previous year, so that previousyear and lifetime prevalence definitions yield very similar estimates. Our analysis of variance suggests that sexspecific prevalence appears to vary systematically with age and geography; both of these observations are worthy of comment. The distinctive age distribution of migraine prevalence for females (Figures 4a and 4b), which was particularly elevated at ages 3554 years and then declined, has been found in other surveys and is usually explained in terms of hormonal changes associated with the reproductive cycle. 34 But observing such an age distribution for lifetime prevalence is puzzling: by definition, lifetime prevalence should be cumulative and increasing with age (i.e. people who have experienced five migraines in their life cannot 'go back' to experiencing four as they age). There are (at least) three explanations for this phenomenon. First, these are crosssectional not

5 024 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY TABLE 2 Sexspecific lifetime migraine prevalence by age, region and annual household income Males Migraineur prevalence Females Migraineur prevalence. % Adjusted relative 95% odds confidence interval. % Adjusted relative 95% odds confidence interval Total Age: Region: Ontario Mantimes Quebec Prairies British Columbia Household income ($CDN): < ^ > /347 2/25 3/340 25/2 3/220 7/30 2/05 47/49 9/05 /349 /29 4/3 3/93 7/96 7/94 /59 6/34 27/ /572 44/24 97/42 5/ /57 9/63 37/2 76/42 49/57 7/27 59/0 62/279 6/2 46/97 49/76 3/24 77/ te: Adjusted relative odds are odds ratio for migraineur status relative to the reference group (= ) for each covanate, after adjusting for all covariates by logistic regression. longitudinal data and this may be a cohort effect with successive generations being more or less predisposed to migtaine. Second, the apparent decline in lifetime prevalence with older age may be due to excess mortality among migraineurs. Third, because lifetime prevalence relies heavily upon the respondents' memory of events, it is prone to recall bias particularly among the elderly who may not remember migraines which occurred decades ago. The similarity between the previousyear and lifetime prevalence age distributions suggests to us that a person's recall may be strongly influenced by their most recent experience. This threat of recall bias raises questions about the reliability of the IHS lifetime prevalence definition. We were intrigued by the unexpected observation that, for males and females, migraine prevalence is significantly lower in the province of Quebec than for other provinces in Canada, even after controlling for age. The first concern was that this may be a spurious inference due to interview language problems; Quebec being a predominantly Frenchspeaking province. Of 726 French interviews in Quebec, 7 (2%) people were classified as migraineurs, compared to 44 English interviews where seven (6%) people were classified as migraineurs. Although this is suggestive of a language difference the numbers are too small to reach statistical significance (x 2 = 0.6, P = 0.4). The response rate for Quebec (5.6%) was also the lowest of all the regions; overall mean response in Canada being 66% (including Quebec). Two observations offer some corroboration for lower prevalence in Quebec. First, Quebec has the lowest per capita expenditure on prescription medicines for migraine (e.g. Fiorinal) of all Canadian provinces. 5 Second, a recent survey in France by Henry et al} 6 suggests an IHS migraine prevalence of only 4% in males and 2% in females rates lower than those previously published for most other countries (Table 3). We draw no conclusion from these observations but leave it to others to speculate or study whether these are facets of the Francophone diet or lifestyle that predispose people to reduced risk of migraine. A recent US survey of migraine prevalence by IHS criteria found a clear association between increased risk

6 MIGRAINE IN CANADA 025 TABLE 3 Summary of migraine prevalence studies Study Year Country Populationbased survey Sample size Age Time period for prevalence Migraine prevalence % Male Female Total Waters" Waters DalsgaardNielson " Waters 20 Markrush 2 Philips 22 Ziegler 4 Cnsp 3 Hollnagel 23 Schnarch Nikiforow 6 D'Alessandro 5 Duckro 24 Rasmussen 23 Stewart PrysePhillips 9 Henry 6 O'Brien" The present study Denmark Denmark Finland Italy Denmark Canada France Canada of migraine and lower household income. The authors speculated that this may be due to either reduced access to effective therapy for those with lower incomes, or reversecausality such that people experiencing migraines may lose income because of absence from work and/or unemployment due to sickness. We found no evidence of an incomemigraine association in our data for Canada, although the lowest household income category we used was CDN$ whereas the US study went down to US$ However, this result should be viewed with some caution because we had an 3% response rate to the income question; it is not clear whether this nonresponse was systematic with respect to level of income, migraine status or both. The observation that 54% of IHS migraineurs in our survey reported no doctor diagnosis of migraine raises two points for comment. The first is that a large part of the management of this disease is likely to be through selfmedication with overthecounter pharmacy purchase of analgesics for headache pain, rather than use of family doctors and prescription medicines. The second point is that use of health care utilization data (e.g. doctor visits or diagnoses) as a basis for estimating the prevalence of migraine will result in underestimates compared to IHS criteria. In conclusion, our Canadian survey data are broadly consistent with other international prevalence studies > >5 Adults >5 >7 > »5 *5 > Previous 6months t stated published since the IHS criteria appeared in 9 (Table 3), suggesting lifetime migraine prevalence is age dependent and ranges between 2% in males and 233% in females. ACKNOWLEDGEMENTS This study was funded by a grant from Glaxo Canada Inc. We are grateful to David rthrup and Richard Myles of the Institute for Social Research at York University who helped design and execute the survey fieldwork. Thanks also to Lori Scapinello and Lori Houghton for research assistance and secretarial support. REFERENCES Linet M S, Stewart W F. The epidemiology of migraine headache. In: Blau J N (ed.) Migraine: Clinical and Research Aspects. Baltimore, MD: Johns Hopkins University Press, Schnarch D M, Hunter J E. Migraine incidence in clinical vs nonclinical populations. Psychosomatic! 90; 2: Crisp A H, Kalucy R S, McGuinness B et al. Some clinical, social and psychological characteristics of migraine subjects in the general population. Postgrad Med J 977; 53: Ziegler D K, Hassanein R S, Couch J R. Characteristics of life headache histories in a nonclinic population. Neurology 977; 27:

7 026 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY 5 D'Alessandro R, Bcnassi G, Lenzi P L et al. Epidemiology of headache in the Republic of San Marino. J Neurol Neurosurg Psychiatry 9; 5: Nikiforow R. Headache in a random sample of 200 persons: A clinical study of a population in rthern Finland. Ccphalalgia 9; : Headache Classification Committee of the International Headache Society. Classification and diagnostic criteria for headache disorders, cranial neuralgias and facial pain. Ccphalalgia 9; (Suppl. 7). Stewart W F, Lipton R B, Celantano D D, Read M L. Prevalence of migraine headache in the United States. JAMA 992; 267: PrysePhilhps W, Findlay H, Tugwell P, Edmeads J, Murray T J, Nelson R F. A Canadian population survey on the clinical, epidemiologic and societal impact of migraine and tensiontype headache. Can J Neurol Sci 992; 9: O'Rourke D, Blair J. Improving random respondent selection in telephone surveys. J Marketing Res 93; 20: " Kalton G Introduction to Survey Sampling. Newbury Park, California Sage, Kish L. Survey Sampling. New York: John Wiley and Sons, Kappius REK, Goolkasian P. Group and menstrual phase effect in reported headaches among college students. Headache 97; 27: Silberstein S D, Meman G R. Estrogens, progestins and headache. Neurology 99; 4: Intercontinental Medical Statistics. Canadian Compuscnpt Henry P, Michel P, Brochet B, Dartigues J F, Tison S, Salamon R, and the GRIM. A nationwide survey of migraine in France: prevalence and features in adults. Cephalalgia 992; 2: Waters W E, O'Connor P J. Epidemiology of headache and migraine in women. J Neurol Neurosurg Psychiatry 97; 34: 453. Waters W E. Headache and migraine in general practitioners. The Migraine Headache and Dixaril. Bracknell, England: Boehnnger Ingelheim, 972, pp " DalsgaardNielsen T, Ulrich J. Prevalence and heredity of migraine and migrainoid headaches among 46 Danish doctors. Headache 973; 2: Waters W E. The Pontypridd Headache Survey. Headache 974; 4: Markrush R E, Herbert R K, Hyeman A, O'Fallon W M. Epidemiologic study of migraine symptoms in young women. Neurology 975; 25: Philips C. Headache in general practice. Headache 977; 6: Hollnagel H, rrelund N. Headache among 40 yearolds in Glostrup. UgeskrLaeger 90; 42: Duckro P N, Tait R C, Margolis R B. Prevalence of very severe headache in a large US metropolitan area. Cephalalgia 99; 9: Rasmussen B K, Jensen R, Schroll M. Epidemiology of headache in a general population a prevalence study. J Clin Epidemiol 99; 44: {Revised version received March 994)