Objectives. The aims of the study were to determine whether the prevalence of children's somatic symptoms, such as headache, abdominal pain, other pain, and nausea and vomiting, changed from 1989 to 1999 and to study the similarity of parents' and children's reports of the child's symptoms. Furthermore, the aims were to explore possible comorbidity in somatic symptoms and to investigate the associations between somatic and psychiatric symptoms.
Methods. Two cross-sectional, representative samples were compared. All children born in 1981 (1989 sample, n = 985) and 1991 (1999 sample, n = 962) and living in selected school districts in southwest Finland served as study samples. The response rate for the 1989 sample was 95% and that for the 1999 sample was 86%. Both children and parents were asked about the children's somatic symptoms, whereas parents, children, and teachers were asked about psychiatric symptoms. To study psychiatric symptoms, the Children's Depression Inventory and Rutter's parent and teacher scales were used.
Results. The prevalence of frequent headaches and abdominal pain increased somewhat from 1989 to 1999. Parents often failed to recognize their children's psychosomatic problems. Child-reported somatic symptoms were associated with conduct and hyperactivity symptoms, in addition to a previously well-documented association with depression. In associations between somatic symptoms and psychiatric symptoms, there were some differences between the 1989 and 1999 samples.
Conclusions. In clinical work, questions about somatic and psychiatric symptoms should also be addressed to children themselves, because parents and teachers do not always recognize children's symptoms. When somatic problems are being evaluated, psychiatric symptoms should be asked about, and vice versa. More research is needed to explore the reasons for the increased prevalence of somatic symptoms and their associations with psychiatric symptoms.
Children's functional psychosomatic symptoms, ie, pains without any specific organic reason, such as headache and recurrent abdominal pain, are found to be common. These problems interfere with children's quality of life during a sensitive developmental period1,2 and are costly for individual families and health services.3,4 Headache is the most studied somatic symptom among children, and its prevalence is found to vary from 6%5 to 57%,6 depending on the definition and classification of headache. Prevalence figures for recurrent abdominal pain among children vary from 9% to 25%,7,8 and those for recurrent limb pain vary from 2.6% to 33.6%.9–12 The prevalence of functional somatic symptoms is found to increase with age and to be gender dependent, especially in older age groups.13,14 Many children are found to suffer from multiple pains,10,15 most commonly from the combination of headache and abdominal pain.15 Only a few studies have examined changes in the prevalence of children's somatic symptoms with time. Sillanpää and Anttila16 found an increased prevalence of headache among 7-year-old children from 1974 to 1992, and Hakala et al17 found an increase in adolescents' back and neck pains from the middle 1980s.
Several studies found associations between children's somatic symptoms, such as headache and recurrent abdominal pain, and internalizing psychiatric symptoms or disorders (anxiety and depression), in both clinical18–27 and population-based7,28–35 studies. Fewer studies found associations between somatic symptoms and externalizing symptoms or disorders (oppositional or conduct symptoms) or hyperactivity symptoms.31,33,34 Despite the observed increase in somatic symptoms, no such increase in children's and adolescents' psychiatric symptoms was found in the few studies on the subject.36–38 Studies of adults suggested a temporal increase in the rates of major depression, a decrease in the age of onset, and a narrowing of the differential risks of men and women.39
We were interested in the changing picture of children's symptoms. Can the increase in the prevalence of somatic symptoms found in some studies be confirmed? Are there changes in the strengths of associations between somatic and psychiatric symptoms over time? The aims of this study were (1) to determine whether the prevalence of 8-year-old Finnish children's somatic symptoms in 1999 was different from that 10 years earlier (1989), (2) to study the similarity of parents' and children's reports of the child's somatic symptoms, (3) to explore possible comorbidity in the various somatic symptoms, and (4) to examine associations between somatic and psychiatric symptoms and to determine whether there are differences in the strengths of associations in the years 1989 and 1999.
Finland is a country of 5 million inhabitants, and it is divided into 5 university hospital catchment areas, each covering a population of ∼1 million. The target population of this study was all Finnish-speaking children born in 1981 (for the 1989 sample) and in 1991 (for the 1999 sample) and living in 1 of the 5 university hospital catchment areas, that of Turku University Hospital, in southwest Finland. The sampling methods were identical for the 1989 and 1999 samples. A representative sample of 13% of the age cohort was drawn by selecting a representative sample of school districts in the area; all children living in the chosen school districts belonged to the sample. A child registered in the district belonged to the sample even if she or he attended school outside the district because of a need or desire for special education, for example, special language classes (a desire to start English, French, or German studies earlier than others) or classes for behaviorally disturbed or disabled children. The sample collected in 1989 was a subsample of the Epidemiologic Multicenter Child Psychiatric Study in Finland, the main results of which were published previously.40
There were altogether 1038 children in the 1989 sample and 1035 children in the 1999 sample. In the 1999 sample, 3 teachers refused to participate in the study; therefore, the children in the corresponding 3 classes in the 1989 sample were dropped from the analysis of this study. Three informants were used, ie, a parent and a child were asked about somatic symptoms, and the child, a parent, and a teacher were asked about psychiatric symptoms.
In the 1989 sample, the questionnaires were distributed to 985 children. The children (n = 936, 95%) for whom parent, teacher, and child questionnaires were all returned were included in the analysis. In the 1999 sample, the questionnaires were distributed to 962 children; for 831 (86%), parents returned the parent questionnaire and gave their consent for participation. Altogether 820 child questionnaires (85%) and 818 teacher questionnaires (85%) were returned. Forty-seven of the 1038 children in the 1989 sample and 107 of the 1035 children in the 1999 sample attended school outside their own district. The response rates among these children were 96% and 70%, respectively. In the 1999 sample, the biggest losses came from the 2 schools offering foreign languages and Swedish (the second official language of the country) in the first school years and from the Steiner pedagogic school. Only 23 of the 39 children in these schools participated in the study in 1999. Because of a technical mistake, identification codes and information about gender were lost for 180 child questionnaires (Children's Depression Inventory [CDI]) in the 1999 sample. Therefore, the data from the 180 child questionnaires could not be linked to parent and teacher questionnaire data and could not be included in the analyses in which boys and girls were analyzed separately.
The procedures for data collection were similar for both times and were organized through schools and teachers. The principal and teachers were contacted, the study procedure was explained, and study questionnaires were delivered to the teachers by the researchers. The teacher sent a study questionnaire to be completed by a parent and an information sheet, via the child, to the parent, who returned the completed questionnaire to the teacher in a sealed envelope. In 1999 (but not in the 1989 sample), the parents' written consent was required for a child's participation in the study, and the teacher completed a teacher questionnaire about the child only after consent was given. In 1989, it was understood that, when parents returned the parent questionnaire (in a sealed envelope) to the teacher, they gave their consent for participation, and only after that did the teacher complete the teacher questionnaire and the child complete the CDI questionnaire in the classroom. A teacher forwarded the parent questionnaires (in sealed envelopes), consent sheets, child questionnaires, and teacher questionnaires to the researchers. Some parents sent their parent questionnaires directly to the researchers, at the address they found on the information/consent sheet. The researchers gave each teacher a follow-up sheet, on which the teacher marked how many children had returned a parent questionnaire and consent sheet and the birth year of those who did not return the material. These follow-up sheets were also forwarded to the researchers. Information on children belonging to a selected school district but attending school outside it was obtained from the school authorities. Principals and teachers of those children were contacted and asked to carry out the study procedure for the individual child as described above. The delivery of questionnaires was started in November 1989 and in November 1999. The study was approved by the school authorities and by the ethics committee of Turku University Hospital. The study design and procedure were described in more detail in the study by Sourander et al.38
Each child was asked about headache, abdominal pain, other pain, and nausea or vomiting on the basis of 3 alternatives. Table 1 presents the exact wording of the questions and the alternatives. Parent-reported headache and abdominal pain were derived from Rutter's Parent Questionnaire.41
The children completed the CDI, which measures depression. The CDI was developed by Kovacs42 on the basis of Beck's depression studies and consists of 27 items rated 0, 1, or 2. The items include cognitive, affective, and behavioral aspects of depression among children, during the previous 2 weeks.42,43 In both the 1989 and 1999 surveys, the question about suicidal ideation was omitted because it was thought to be possibly problematic to answer in the classroom setting. In the statistical analysis of somatic symptoms and the CDI, all of the items connected closely to somatic symptoms (being tired, worrying about aches and pains, and not feeling like eating) were omitted from the CDI.
To assess psychiatric symptoms, the parent questionnaire included the Rutter A2 scale (31 items) and the teacher questionnaire included the Rutter B2 scale (26 items). The parent questionnaire Rutter A2 scale and the teacher questionnaire Rutter B2 scale assess global child psychiatric symptoms, divided into emotional (5 items on the parent questionnaire and 4 items on the teacher questionnaire), conduct (5 items on the parent questionnaire and 6 items on the teacher questionnaire), and hyperactivity (3 items on both questionnaires) symptoms within the past 12 months, rated 0, 1, or 2.41,44 The Rutter scales also include some somatic items. The emotional subscales include the following items: often worried, worries about many things; tends to be fearful or afraid of new things or new situations; is in tears on arrival at school or has refused to come into the school building in the past 12 months; has stomachache or vomiting (only on the parent scale); has sleeping difficulties (only on the parent scale); and often appears miserable, unhappy, tearful, or distressed (only on the teacher scale). In the analysis in which the associations between somatic symptoms and Rutter parent and teacher questionnaires and their subscales were examined, all of the items dealing with somatic symptoms were omitted from the Rutter scales. The items describing the child having headache, abdominal pain, or asthma were omitted from the Rutter A scale. The item describing the child having abdominal pain was excluded from the subscale of emotional symptoms. The item stating that the child often complains of aches and pains was omitted from the Rutter teacher scale. The validity and reliability of the CDI and the parent Rutter A2 and teacher Rutter B2 scales are well documented, both internationally and in Finland, and have been found to be good or moderate.42,43,45
In the 1989 and 1999 samples, the genders were equally represented (48% boys and 52% girls). In the 1999 sample, more children had mothers who had completed upper secondary school (50% vs 32%) than in the 1989 sample, and they also belonged to a higher socioeconomic class (52% vs 42%).38 In the 1989 sample, 89% of children lived in 2-parent families; in 1999, the proportion was 87%. In our earlier study with the same sample, children's psychiatric symptoms were found to be associated positively with family structure (living in a family other than with 2 biological parents) and mother's lower education and socioeconomic class.38
The somatic symptoms were the outcome variables in the analyses. All 4 outcome variables were measured on an ordinal scale with 3 categories. The analyses of associations between different variables and somatic symptoms were conducted with cumulative logistic regression analysis. Cumulative logistic regression analysis is a comprehensive method suitable for univariate and multivariate analyses when the response variable is measured on an ordinal scale.46 The associations were quantified with their cumulative odds ratios (ORs) and 95% confidence intervals (CIs). The gender differences in associations were investigated by testing interactions in the logistic models. The log-linear models were used to compare the association structure of the 4 outcome variables with each other. To avoid 0 frequencies in the log-linear analysis, the categories of each somatic symptom variable were combined so that the categories indicating occurrence of the symptom every day or often (and often or sometimes for nausea or vomiting) were combined. For all tests, P values of <.05 were considered statistically significant. Statistical computations were performed with SAS for Windows, release 8.2 (SAS Institute, Cary, NC).
Associations Between Somatic Symptoms and Background Variables
When the associations between child-reported somatic symptoms (headache, abdominal pain, other pain, and nausea or vomiting) and background variables (family structure, mother's education, and mother's socioeconomic class) were studied with the χ2 test, it was found that children of mothers from a lower socioeconomic class more often had frequent headaches (P = .0317) and occasional nausea or vomiting (P = .0301) than did children of mothers from a higher socioeconomic class. No other associations between somatic symptoms and background variables were found.
Differences in Prevalence of Somatic Symptoms Between 1989 and 1999 and Between Parent and Child Reports
Significantly more children in 1999 than in 1989 reported frequent headaches and abdominal pain. Also, according to parent reports, more children had occasional headaches in 1999 than in 1989. The differences between the 1989 and 1999 samples were similar for boys and girls (Table 1). There was no statistically significant difference in the prevalence of nausea or vomiting, but the difference in other pains between the 2 samples was almost statistically significant (P = .058). When the differences between the 1989 and 1999 samples were adjusted for the background variables (family structure, mother's education, and mother's socioeconomic class), the results remained practically identical. The frequencies of somatic symptoms were the same for boys and girls, except that parents reported slightly more abdominal pain among girls than among boys (P = .002).
Table 2 presents the differences in parent and child reports in the analysis in which the 2 samples were combined. Of the children who reported headaches almost every day, 14% were considered to be totally headache-free by their parents. Correspondingly, 30% of the children who reported having abdominal pain almost every day were completely symptom-free according to their parents (Table 2).
Comorbidity of Somatic Symptoms
The associations among the 4 psychosomatic symptoms reported by the children were analyzed with log-linear models (both samples combined). To avoid 0 frequencies in the log-linear analyses, the categories of each somatic symptom variable were combined so that the categories indicating occurrence of the symptom every day or often (and often or sometimes for nausea and vomiting) were combined. All of the 3- and 4-variable interactions in the models were nonsignificant, whereas all 2-variable interactions were significant. This means that there were associations between each pair of variables but the association between each pair did not depend on the other 2 variables. For instance, headache was associated with abdominal pain but the association did not depend on other pains or on nausea or vomiting. Both headache and abdominal pain were also associated with nausea or vomiting and other pain. Nausea and vomiting were associated with each other. There were no more than 2 variables clustering among different psychosomatic symptoms (Fig 1).
Association of Child-Reported Somatic Symptoms With Psychiatric Symptoms
In univariate analyses, both girls' and boys' somatic symptoms were most strongly associated with child-reported depressive symptoms (CDI) (Table 3). Among girls, all child-reported somatic symptoms were also found to be associated with parent-reported emotional, conduct, and hyperactivity symptoms (headache only when the analysis was adjusted for mother's education), with the exception of nausea or vomiting, which was not associated with parent-reported emotional problems. Other pain and nausea or vomiting among girls were also associated with teacher-reported conduct and hyperactivity symptoms (Table 3). When the associations were adjusted for the background variables, the results remained practically the same as in the unadjusted analysis, with 2 exceptions (associations of headache and other pain with parent-reported emotional problems) (Table 3).
In the multivariate analysis (cumulative logistic regression analysis) in which overlapping correlations for parent- and teacher-reported psychiatric symptoms were studied separately, the association between headache and parent-reported hyperactivity symptoms remained statistically significant. Abdominal pain remained associated significantly with parent-reported conduct symptoms. Other pain and nausea or vomiting remained associated significantly with parent-reported conduct symptoms and teacher-reported hyperactivity symptoms (other pain also with parent-reported hyperactivity symptoms) (Table 4).
Among boys, both abdominal pain and headache were found to be associated with parent- and teacher-reported conduct symptoms and headache with teacher-reported hyperactivity and emotional symptoms (Table 3). Among boys, other pain and nausea or vomiting were not associated with parent- or teacher-reported psychiatric symptoms. When the associations were adjusted for the background variables, the results remained practically the same.
In the multivariate analysis (cumulative logistic regression analysis) in which overlapping correlations for parent- and teacher-reported psychiatric symptoms were studied separately, headache among boys remained associated significantly with parent- and teacher-reported conduct symptoms and also with teacher-reported emotional problems. Abdominal pain remained associated significantly with parent- and teacher-reported conduct symptoms (Table 5).
There were a few differences in the associations of somatic and psychiatric symptoms between the 1989 and 1999 samples. In the total sample, the associations between headache and CDI scores and between other pain and CDI scores became weaker from 1989 (OR: 3.3; 95% CI: 2.4–4.4; P < .001; and OR: 3.2; 95% CI: 2.4–4.5; P < .001; respectively) to 1999 (OR: 1.1; 95% CI: 0.6–2.0; P = .835; and OR: 0.7; 95% CI: 0.4–1.5; P = .399; respectively). The positive association between abdominal pain and CDI scores in the 1989 sample (OR: 2.7; 95% CI: 1.9–3.9; P < .0001) turned negative (OR: 0.4; 95% CI: 0.2–0.8; P = .0126). The associations between other pains and nausea or vomiting and teacher-reported emotional symptoms became stronger from 1989 (OR: 1.0; 95% CI: 0.8–1.2; P = .954; and OR: 0.9; 95% CI: 0.8–1.1; P = .576; respectively) to 1999 (OR: 1.3; 95% CI: 1.1–1.7; P = .005; and OR: 1.4; 95% CI: 1.1–1.7; P = .002; respectively).
The main results of this study were that the prevalence rates of frequent headaches and abdominal pain increased somewhat from 1989 to 1999 and that parents often fail to recognize their children's psychosomatic problems. Child-reported somatic symptoms were associated with conduct and hyperactivity symptoms, in addition to the previously well-documented association with depression and anxiety. Some clear differences between boys and girls were found. There were some differences in the associations between somatic symptoms and psychiatric symptoms between the 1989 and 1999 samples. In the study of comorbidity of somatic symptoms (headache, abdominal pain, nausea or vomiting, and other pains), it was found that 1 somatic symptom predicted the presence of other somatic symptoms but with no special clustering.
The samples were representative of the study populations because of the high response rates in both study phases. The minor difference in response rates, although rates were very high for both samples, might have biased the findings slightly. Unfortunately, we have no way of analyzing the characteristics of the nonresponders. One reason for the lower response rate in the 1999 sample was probably the explicit request for signed informed consent from the parents.
Our finding of increased prevalence of headache is in line with the study by Sillanpää and Anttila,16 who found an increase in the prevalence of headache among 7-year-old children from 14.4% in 1974 to 51.5% in 1992. As reasons for this increase, Sillanpää and Anttila16 suggested changes in the social environment. Negative life events have been found to be associated with headaches,47 and it has also been found that children's headaches are associated with family functioning.47,48 Another reason for the increase may be the increase in the prevalence of neck/shoulder symptoms,17 which have been found to be associated with both migraine and tension-type headaches.35,49 Children are spending increasing amounts of time with information technology equipment and electronic entertainment, which can cause static strain in the neck/shoulder area easily. Bener et al50 found that, among school children, the most common environmental factor provoking migraine attacks was playing with a computer.
To the best of our knowledge, there are no earlier studies on changes in the prevalence of recurrent abdominal pain. Recurrent abdominal pain was found to be associated with major negative life events,47 daily stressors,51 anxiety,52 and sleep disturbances.53 This study is unable to clarify the reasons for the increase in prevalence, but it seems evident that children's living environments and lifestyles were less favorable in this respect in 1999 than in 1989.
An alarming finding was that, even in cases in which a child reported having headaches or abdominal pain almost daily, many parents (14% for headaches and 30% for abdominal pain) reported that their child never had these symptoms. Our finding emphasizes the importance of interviewing children themselves in both research and clinical settings. In earlier studies, it was shown that children and parents often disagree, especially about issues that are abstract or ambiguous, such as internalizing symptoms.54,55 Parents should be aware of their children's problems because, for instance, Frare et al56 found that headache frequency and duration have a significant impact on a child's quality of life and the family's daily routine significantly influences the child's coping with headaches and quality of life.
Numerous studies have found associations between children's headaches and recurrent abdominal pain and depression and anxiety.7,18–35 In line with the earlier studies, we found that all psychosomatic symptoms seemed to be associated with depression. The associations between somatic symptoms and externalizing symptoms have been studied less extensively. Egger et al33 found associations between headaches and conduct disorder and between abdominal pain and oppositional defiant disorder and attention-deficit/hyperactivity disorder among boys.34 Taylor et al31 found associations between multiple somatic symptoms and conduct disorder. We found that both abdominal pain and headache were associated with conduct and hyperactivity symptoms for both genders. Nausea or vomiting and other pain were associated with both parent- and teacher-reported hyperactivity and conduct disorder symptoms among girls but not among boys.
The comorbidity of headache and other pains was documented previously in some studies. Aromaa et al57 found that children with headaches were more sensitive to pain, avoided games or play more often because they were afraid of hurting themselves, and had recurring abdominal and growing pains more often, compared with control children. In the large study by Perquin et al15 on pain among children, one half of the children who had experienced pain reported having multiple pains. The pain locations asked about were head, abdomen, limb, ear, throat, back, unknown, and elsewhere. In the present study, the log-linear analyses of different somatic symptoms revealed that separate somatic symptoms (for instance, headache and abdominal symptoms) were associated independently with each other and that the association did not depend on the presence of other symptoms, ie, 1 somatic symptom predicted the presence of another somatic symptom, but no special clustering was supported by the data.
On the basis of their large epidemiologic study of associations between somatic complaints and specific psychiatric diagnoses, Egger et al33,34 suggested that there are gender-, illness-, and complaint-specific associations between somatic complaints and psychopathologic conditions, and they presumed that there are differences in psychobiological processes underlying these associations among boys and girls. In our study, a clear gender difference was found in associations of nausea or vomiting and other pain with hyperactivity and conduct disorders, which were significant for girls but not boys. The clear gender differences before puberty cannot be explained as clearly on the basis of biological factors as can differences during and after puberty. The reasons for the observed gender differences may be at least partly related to different role expectations in society for boys and girls. Girls with hyperactivity and conduct disorder symptoms may be under even greater pressure than their male counterparts when they do not meet the traditional, passive, tame role expectations and may therefore also suffer from numerous somatic complaints.
In the strengths of associations between somatic and psychiatric symptoms, some differences were found between the 1989 and 1999 samples. The time interval in this study was rather short, and additional studies are needed to explore possible temporal changes in the associations between somatic and psychiatric symptoms.
Besides the difference in the response rates for the 1989 and 1999 samples, a weakness of the study was the use of questionnaires, which indicate not exact diagnoses but only psychiatric symptom entities and for which sensitivity and specificity are not as high as with interviews. In comparisons of the child and parent reports of somatic symptoms, it should be noted that the wording of the questions was not identical in the 2 questionnaires. A strength of the study was the use of 3 different informants, ie, a child, a parent, and a teacher. Because parents and teachers often do not notice children's internalizing symptoms and the children themselves are not good reporters of externalizing symptoms, all 3 informants were used. A weakness was that questions about somatic symptoms did not use international diagnostic criteria for headache or recurrent abdominal pain.7,58
Children's headaches and abdominal pain have increased. In clinical work, questions concerning both somatic and psychiatric symptoms should be addressed to children themselves, not only to parents, because children's symptoms are not always recognized by either parents or teachers. When somatic problems are being evaluated, psychiatric symptoms should be asked about, and vice versa. A child with 1 psychosomatic symptom should also be asked about other symptoms, because psychosomatic symptoms often cluster with each other. More research is needed to explore reasons for the increased prevalence of somatic symptoms and their comorbidity with psychiatric symptoms.
This study was supported by the Sigrid Juselius Foundation.
- Accepted November 8, 2004.
- Address correspondence to Paivi Santalahti, MD, Department of Child Psychiatry, Turku University Hospital, 20520 Turku, Finland. E-mail:
No conflict of interest declared.
Dr Sourander is a visiting researcher at the Department of Child and Adolescent Psychiatry, Columbia University, New York, NY.
- ↵Frare M, Axia G, Battistella PA. Quality of life, coping strategies, and family routines in children with headache. Headache.2002;10 :953– 962
- ↵Campo JV, Jansen-McWilliams L, Comer DM, Kelleher KJ. Somatization in pediatric primary care: association with psychopathology, functional impairment, and use of services. Am J Acad Child Adolesc Psychiatry.1999;38 :1093– 1101
- ↵Apley J, Naish N. Children with recurrent abdominal pains: a field survey of 1000 school children. Arch Dis Child.1958;3 :165– 170
- ↵Faull C, Nicol A. Abdominal pain in six-year-olds: an epidemiological study in a new town. J Child Psychol Psychiatry.1985;27 :251– 260
- ↵Naish JM, Apley J. ‘Growing pains’: a clinical study of non-arthritis limb pains in children. Arch Dis Child.1951;26 :134– 140
- ↵Oster J. Recurrent abdominal pain, headache and limb pains in children and adolescents. Pediatrics.1972;50 :429– 435
- ↵Abu-Arafeh I, Russel G. Recurrent limb pain in schoolchildren. Arch Dis Child.1996;74 :336– 339
- ↵Hakala P, Rimpelä A, Salminen JJ, Virtanen SM, Rimpelä M. Back, neck, and shoulder pain in Finnish adolescents: national cross sectional surveys. BMJ.2002;325 :743– 745
- Walker LS, Greene JW. Children with recurrent abdominal pain and their parents: more somatic complaints, anxiety, and depression than other patient families? J Pediatr Psychol.1989;14 :231– 243
- ↵Campo JV, Bridge J, Ehmann M, et al. Recurrent abdominal pain, anxiety, and depression in primary care. Pediatrics.2004;113 :817– 824
- ↵Anttila P, Metsähonkala L, Aromaa M, et al. Determinants of tension-type headache in children. Cephalalgia.2002;22 :401– 408
- ↵Almqvist F, Ikäheimo K, Kumpulainen K, et al. Design and subjects of a Finnish epidemiological study on psychiatric disorders in childhood. Eur Child Adolesc Psychiatry.1999;8(suppl 4) :3– 6
- ↵Rutter M, Tizard J, Whitmore K. Education, Health and Behavior. London, United Kingdom: Longman; 1970
- ↵Hosmer DW, Lemeshow S. Applied Logistic Regression. 2nd ed. New York, NY: John Wiley & Sons; 2000
- ↵Anttila P, Metsähonkala L, Mikkelsson M, et al. Muscle tenderness in pericranial and neck-shoulder region in children with headache: a controlled study. Cephalalgia.2002;22 :340– 344
- ↵Aromaa M, Sillanpää M, Rautava P, Helenius H. Pain experience of children with headache and their families: a control study. Pediatrics.2000;106 :270– 275
- Copyright © 2005 by the American Academy of Pediatrics