Abstract
The characteristics of disturbing primary headache and the occurrence of headache types were studied by sending a questionnaire to 1132 Finnish families of 6-year-old children. Children with headache in the preceding 6 months and their controls were clinically examined at the ages of 6 and 13. During the follow-up, half of the headaches, classified as migraine at age 6 years, were unchanged and 32% turned into tension-type headache. In children with tension-type headache, the situation was unchanged in 35%, and in 38% of children the headache type had changed to migraine. At preschool age the most common location of headache was bilateral and supraorbital, and at puberty bilateral and temporal. During the follow-up, symptoms concurrent with headache, such as odour phobia, dizziness and balance disturbances became more typical, whereas restlessness, flushing and abdominal symptoms became less marked. The early manifestation of both migraine and tension-type headache predict equally often migraine in puberty with marked changes in concurrent symptoms and pain localization.
Introduction
The first virtual description of the different types and clinical characteristics of headache in children was given by Vahlquist in 1949 (1). His set of criteria (1) was then applied to several prospective studies of the natural history of childhood headache. These studies showed that different types of headache were related to age and developmental factors (2–7). Childhood headache and migraine often manifest before the age of 10 years (5) and the prevalence rates gradually increase from the age of 11 years (6, 8). The prevalence of headache is greater in girls than in boys after the age of menarche (9, 10).
The classification of the different primary headache types is based on their typical characteristics and associated symptoms. Therefore, data on the changes of the characteristics with age and on differences between children and adults are vitally important for a correct and prognostic diagnosis also in children. The diagnosis is always based on information from the patients and/or their relatives. Some features, such as unilaterality, vomiting, nausea, aura, pulsatility, photophobia and phonophobia are considered more typical of migraine than of tension-type headache. On the other hand, several studies have found none of these characteristics specific for migraine alone (11, 12).
A changing pattern of headache seems to be typical of childhood headache (13). However, the nature and role of age-related factors in the differentiation of headaches and identification of prognostic factors are still largely unknown. The aim of the present study was to clarify the picture of age-related characteristics of primary disturbing headache among children from the age of 6 to 13 years. The main questions addressed were: do the quality, triggers and intensity factors show differences in preschool and pubertal headache? Do primary headache types change during the first few school years? How do the characteristics of migraine headache and tension-type headache change during follow-up?
Methods
The present study is part of the Finnish Competence study. The recruitment of subjects began in 1986 in the province of Turku and Pori, south-western Finland, and was based on stratified randomized cluster sampling. Each cluster consisted of the population of a health authority area. Randomization was based on selecting by lot 11 of the 35 health authority areas of the province. All maternity health clinics (n = 67) and well-baby clinics (n = 72) of the 11 health authority areas participated in the study. The study population originally consisted of families expecting their first child and paying their first visit to a maternity clinic at the first signs of pregnancy in 1986. Invitations to participate in the original study were sent to 1582 families, of which 1443 (91%) gave informed consent. There were 1294 deliveries: three of the children died during the neonatal period, eight in infancy. Five children moved abroad and 146 families could not be traced. Thus, 1132 children were eligible for the study at the age of 6 years. When the child was 6 years old, the parents answered, among other things, the following question about their child's headache: ‘Has your child had headache in the preceding 6 months disturbing his/her daily activities?’.
When the children were 6 years old, the question about headache was answered by 968 (86%)/1132 parents. Of the 968 children, 204 (22%) had had previous headache disturbing their daily activities. Of these 204 children, 144 (71%) had suffered from disturbing headache during the preceding 6 months (= present headache) (14). A control child, matched for age, sex and domicile, without prior or present headache, was randomly chosen from the sample for each index child. The children with present headache and control children were invited to participate in the clinical paediatric and neurological examinations at the Departments of Child Neurology and Public Health, Turku University Central Hospital. The invitation was by telephone or, if the family was not reached by telephone, by mail. At least one parent was requested to accompany the child. Of the 144 children, 106 (74%) with present headache and 106 matched control children participated in the study. After the clinical examination, the child's headache was classified using the Classification of the International Headache Society (IHS) (Classification and diagnostic criteria for headache disorders, cranial neuralgias and facial pain, 1988) (15). Children with secondary headaches were excluded from the clinical examinations. Finally, 96 headache children and 96 controls participated in the clinical examination (14).
When the 96 children with headache and 96 control children were 13 years old, they were again invited to participate in a clinical examination. The first author invited them by telephone or, if the family was untraceable by telephone, by mail. Again, it was suggested that the child should be accompanied by at least one parent. Six (6%) of the index children and five (5%) of the control children could not be traced. Another six (6%) families of the index children and three (3%) families of the control children refused to participate in the clinical examination. The reasons for refusal were living far away from the university hospital (n = 3), lack of time (n = 4), the child's refusal (n = 2) or recent death of a family member (n = 1). Questions on the quality of the child's headache could in all these cases be asked by telephone. At 13 years old, the final number of index children was 90 and that of control children 91 (Fig. 1). After the clinical examination, the child's headache was again classified using the IHS criteria (15). The mean age of the children was 13.0 years (range 12.6–13.6 years). Girls accounted for 49% and boys for 51% of the subjects, of whom 90% came to the clinical examination with their mother and 10% with their father or grandparents. Of the families, 57% lived in urban, 29% in suburban and 14% in rural areas.
Flow chart of the study cohort.
The differences between participants (n = 90) and non-participants (n = 144 − 90 = 54) were analysed to determine the generalizability of the results. The variables used for this dropout analysis were the same as those in the initial study of 6-year-old children: mothers's age (P = 0.538), educational level (<9 years, 9 years, >9 years; P = 0.356), degree of urbanization of the domicile (P = 0.131), having siblings (P = 1.000), family composition (parents married, P = 0.078; parents divorced, P = 0.605; one-parent family, P = 0.660) and the presence of child's long-term disease (excluding headache; P = 0.027). There were more children with long-term disease among the non-participants than among participants.
Because the children had been clinically examined at age 6 years by one of the authors (M.A.) and at age 13 by another author (R.V.), the interexaminer reliability test was done in seven children, who were separately examined by the above two authors. No more than four of the 173 questionnaire items showed differences in the recordings of the examiners. All these four items, where the interpretation of the recorded answers showed differences between the examiners, were categorical questions with the options ‘yes’ or ‘no’. Interexaminer reliability was also studied by calculating the κ coefficient of the four variables. The κ values showed a range of 0.70–0.76, which was considered reasonably good, considering the small number of subjects.
Changes in the concurrent symptoms of present headache
Both at the age of 6 and 13 years, the following characteristics were studied: headache frequency, intensity of pain, triggers of headache, predominant time of headache onset, aura symptoms, location of pain, duration of untreated attacks (sleep period included), characterization of pain, concurrent symptoms, alleviating and worsening factors and use of medication.
Changes in the concurrent symptoms of migraine and tension-type headache
After classification of the child's headache, the changes in concurrent symptoms of migraine and in non-migrainous headache were studied using all above-mentioned headache characteristics.
Changes in factors associated with the occurrence of headache
Both the children and their parents were asked open questions about the child's general health status, including long-term disease (active illness that had lasted >6 months, not headache), e.g. diabetes mellitus, bronchial asthma or allergy and question about travel sickness, syncope and head trauma.
Statistical methods
Changes in the characteristics of headache and changes in variables associated with headache between the ages of 6 and 13 years were studied using cross-tabulation. For dichotomic variables, McNemar's test and for polytochomous variables, Bowker's test (generalization of McNemar's test) were used in the statistical analyses. For frequency tables with small (<5) frequencies, exact P-values were calculated. Odds ratios (OR) and exact 95% confidence intervals (95% CI) for changes were calculated using exact logistic conditional regression analysis (16). Statistical computation was based on the SAS System for Windows, release 8.1/2001 (SAS Inc., Cary, NC, USA). A P-value <0.05 was used as the cut-off point of significance.
Results
Changes in headache types
During the 7 years of follow-up, the headache type remained unchanged in 28 children (53%) of the 53, whose headache was classified as migraine at age 6 years, in 17/53 children (32%) migraine had turned into tension-type, and 8/53 children (15%) had become totally headache free. Of children with tension-type headache (n = 37) at age 6 years, 13 (35%) had tension-type headache, in 14 (38%) the headache type had changed into migraine and 10 (27%) were headache free at age 13 years. In the control group, five (5%) children had begun to suffer from migraine and seven (8%) from tension-type headache (Table 1).
Changes in the headache types of 90 index children (headache at age of 6) and 91 control children (headache free) from age of 6 to 13 years
At the age of 13, 47 children (56%) had migraine and 37 (44%) tension-type headache. Migraine occurred in 26 children (31%) without aura and in six children(7%) with aura, 16 (19%) had migrainous disorder not completely fulfilling the IHS criteria. Episodes of tension-type headache occurred in 21 children (25%) and the other 15 children (18%) had tension-type headache not completely fulfilling the IHS criteria.
Changes in concurrent symptoms of present headache
Present headache was reported by 72 index children at the ages of both of 6 and 13 years. Figure 2 shows the changes in the location of the child's headache pain. Among 6-year-old children the most common location of headache was the forehead bilaterally above the eyes (14), whereas among 13-year-olds it was bilaterally on the temples. Figure 3 shows the courses of headache episodes with increasing and decreasing intensity of pain at the ages of 6 and 13 years.
Locations of pain in children with present headache at ages of both 6 and 13 years (n = 72).
The course of headache episode in children with present headache at ages of both 6 and 13 years (n = 72).
Table 2 shows the changes in headache characteristics. Headache characterized as ‘occasionally aggravating’ at the age of 6 was reported more often as ‘gnawing’ headache at age 13 years. Symptoms concurrent with headache such as odour phobia, dizziness and balance disturbances were more common, whereas restlessness, flush and abdominal symptoms were less marked at the age of 13 years. Feelings of hunger and sensation of cold air as headache triggers were more typical, whereas worries, fears and anxiety were significantly less important among 13-year-old children than younger children. Coughing increased the headache pain intensity significantly more often and physical exercise less often at the age of 13 than at the age of 6 years (Table 2). Headache frequency, intensity or duration did not change significantly during follow-up. No statistically significant differences were seen between the ages of 6 and 13 years in aura symptoms, the predominant time of headache onset.
Changes in the characteristics of headache between 6 and 13 years among 72 children with present headache at age of 6 and 13 years
P-values from the McNemar test and odds ratios and confidence intervals from exact conditional logistic analysis; inf., infinite: one-sided 97.5% CI.
During the follow-up the number of children using headache medication did not change statistically significantly (P = 0.109). The most commonly used medication at the age of 13 years was ibuprofen (61%). Other medications used were paratacetamol (30%), salicylic acid (2%), naproxen (2%), ketoprofen (2%) and deksibuprofen (2%). Of the 6-year-old children, 1% did not use any headache medication at the age of 13 years.
Changes in concurrent symptoms of migraine and tension-type headache
After classification into different headache types, 28 children had migraine and 13 children had tension-type headache at the ages of both 6 and 13 years. Both in the migraine and tension-type headache groups, headache pain in the temples was reported more often at the age of 13 than at the age of 6 years (Table 3). Odour phobia was the only concurrent symptom more typical at the age of 13 than at the age of 6 in both migraine and tension-type headache sufferers. The feeling of hunger and external cold stimuli became more marked as triggers, whereas the role of fears and anxiety became less important in children with migraine during follow-up. Other concurrent pain, flushing and abdominal symptoms were more atypical for children with migraine at puberty than at preschool age. In the tension-type headache group, no such changes were found. Coughing and nose blowing were typical pain-worsening factors in migraineurs at age 13 years, whereas no changes were found in the tension-type headache group.
Changes in the characteristics of headache between 6 and 13 years among children with migraine (n = 28) or tension-type headache (n = 13) at ages of both 6 and 13 years
P-values from the McNemar test and odds ratios and confidence intervals from exact conditional logistic analysis; inf., infinite: one-sided 97.5% CI.
Changes in factors associated with the occurrence of headache
Children suffering from headache at age 13 years had statistically significantly more often long-term disease, especially allergy (P = 0.028, OR 2.6, 95% CI 1.0–7.2) and head trauma (P = 0.018, OR 3.5, 95% CI 1.1–14.6) than children with headache at age 6. Control children showed no such change (P = 0.317, OR 0.7, 95% CI 0.3–1.6 and P = 0.818, OR 0.9, 95% CI 0.3–2.5, respectively).
Discussion
The likelihood of migraine at puberty was practically equal among children presenting with tension-type headache or migraine at 6 years of age. Increasing age also seems to be linked to considerable changes in pain localization, concurrent symptoms and triggers.
Childhood headache seems to have a high tendency to remission periods and improvement (8). In their prospective study among adolescents aged 12–26 years, Guidetti et al. found that in 45% of the children (45/100) headache was relieved and 35% (35/100) were headache free, whereas the situation was unchanged in 15% (15/100) after 8 years of follow-up. The type of headache tend to change mostly from migraine to tension-type headache (17). One possible explanation for this changing pattern of childhood headache is the ‘continuum severity theory’, which considers primary headache a continuum between tension-type headache and migraine. In this model, headache is labelled as tension-type when the pain is mild, as common migraine when the pain is more severe and as classic migraine when the pain is associated with neurological symptoms (18). Another explanation may be that many children may suffer from two or three different types of headache, and different types may be mixed. Although many studies have been carried out, it is not known whether the criteria used for the diagnosis of the different headache types in childhood are valid or not. Our results showed that the early manifestation of migraine seems to signify a 50% remission, whereas one-third of the tension-type headaches turn into migraine at puberty.
In a population-based study of 4000 children aged from 4 to 15 years, Mavromichalis et al. found that all headache attacks in children with migraine lasted from 2 h to 3 days (19). Several studies have shown that migraine attacks are less prolonged in children (<2 h) than in adults (11, 20). IHS criteria (15) do not include attacks lasting <2 h. This has been one reason for the development of new IHS criteria for migraine in children (21). We found no statistically significant increase in the duration of headache attacks from the age of 6 to 13 years, but many adolescents also reported headache attacks of short duration (<2 h).
Bilateral migraine pain seems to be more common in young children, becoming unilateral with increasing age (10, 22). In adolescence, most children with migraine report unilateral pain (23), whereas bilateral location is associated with tension-type headache (11). Maytal et al. found that bilateral pain is common in childhood, whereas unilaterality has a high specificity of 86% and a positive predictive value for migraine of 85% (20). Studies on the sensitivity and specificity of the diagnostic criteria of the IHS for childhood headache (15) only address criteria such as uni- or bilaterality, without analysing the specific localization of headache pain. The present study showed that headache pain, which most commonly occurred in the forehead above the eyes, changed location to the temples, and this change was typical of both migraine and tension-type headache. No age-related changes from bilaterality to unilaterality were found.
Pulsating headache has not been found to be any specific feature of paediatric migraine (11, 12), even though this pain feature may help to exclude other headache types (23). The pulsation reported by migraineurs was significantly related to older children (>10 years of age) (22). In the present study pulsating headache was more typical of puberty.
The role of the autonomic nervous system in the pathophysiology of migraine has been investigated in several previous studies (24–26), because neurological symptoms are common in patients with migraine. Migraine may be attributed to neurotransmitter imbalance in the brainstem nuclei, in contrast with patients with chronic tension-type headache, who may have dysfunction of the peripheral autonomic nervous system presenting as a decrease of sympathetic activity (27). Other neurological symptoms such as vertigo and dizziness are frequently reported by patients with migraine. The prevalence of migraine according to the classification of the IHS was 1.6 times higher in adult dizziness clinic patients than in controls from an orthopaedic clinic (28). The benign paroxysmal vertigo of childhood is an early manifestation of migrainous vertigo. In their population-based study, Abu-Arafeh and Russell found that the prevalence of recurrent vertigo associated with migraine was 2.8% among children aged between 6 and 12 years (29). We found that neurological symptoms were more typical of adolescence, whereas flushing and abdominal symptoms were more common at preschool age. Aggravation by physical activity and abdominal symptoms have been found more typical of childhood migraine than adult migraine (11), confirming our results.
In our study odour phobia as a concurrent symptom at puberty was typical of both migraine and tension-type headache, whereas the role of strong emotions diminished as a precipitating factor only in migraine. In previous studies psychological factors such as stress, pressures and fear have been found to be prime triggering factors for headache in children (30) as well as in adults (31).
The present study was based on a prospective follow-up of a sample from an unselected population. The basic data of the parents and their children were collected during the follow-up years based on the case or control status of the study subjects (14, 32, 33). When the children were 6 years old, they were classified to cases and controls on the basis of a question about the occurrence of headache previously reported (14). At the ages of both 6 and 13 years the classification of different headache types and characteristics of headache was based on an interview and clinical examination. Therefore, the authors were not blinded to the case/control status, which may have limited the generalizability of the results. To maintain comparability, the same version (1998) of the IHS classification was used at both 6 and 13 years. This may have caused some underestimation of migraine diagnoses. The variables used in the present study were derived from the specific characteristics of headache and the clinical examination. They were modified as categorical questions with the options ‘yes’ or ‘no’ in an attempt to minimize differences in the interpretations of the two examiners. Because standardized questions were used in the questionnaire, interexaminer reliability can be considered very good.
The limitation of the study is that only the most prominent headache type and its characteristics were studied at the interview and clinical examination, although many of the study children may have suffered from two or more different headache types. This ascertainment factor may have influenced the study results. Data collection was repeated as similarly as possible at both phases of the study. With a headache diary, we would have obtained even more accurate information on headache characteristic than during an interview (34) and would have reduced recall bias. Another limitation of the study is that the accuracy of information on the headache symptoms depends on the age and developmental stage of a child. Six-year-old children may have difficulty in describing certain symptoms, e.g. dizziness or balance disturbances, and in using terms for headache characteristics. The fact that only first-born children of the family were included in the study may also have influenced the results. Parents may be more sensitive to their first child's symptoms.
Conclusions
Many and marked changes in the incidence, prevalence and phenotype of headache occur from prepuberty to puberty. More children with headache become symptom free than children without headache start to have headache. Migraine at puberty appears to be predicted by prepubertal headache equally often, irrespective of whether it is migraine or tension-type headache. Any prepubertal headache should therefore be considered in the prevention of later migraine.
Acknowledgements
Appreciation is expressed to Simo Merne MA for revising the English language of this report and Mrs Inger Vaihinen for secretarial assistance.