Introduction: The authors review the literature on anxiety and depression in children and adolescents who experience migraine headaches. For over a century, clinicians and researchers have noticed potential links between migraine and certain psychopathological traits. More recently, rigorous methodological studies have confirmed some of those links. For example, several reviews have shown a strong comorbidity in adults between migraine, major depression and certain anxiety disorders. As for children and adolescents, no recent work has thoroughly reviewed the literature specifically on the topic of migraine, anxiety, and depression.

Methods: For the purposes of this study, research published between January 1980 and January 2007 was examined. In order to be included in the review, studies had to specify the diagnostic criteria used to indicate migraine headaches and also use validated measures for anxiety and depression. Of the eleven remaining articles, ten used a control group matched for age and sex. Only three of the studies used a representative sample of the general population. Carrying out a meta-analysis was not possible due to the dearth of articles and the wide variety of methodologies applied. The studies included in this review do not provide conclusive findings for the comorbidity of migraine, anxiety and depression in children.

Results: The majority of the studies with clinical populations show slightly higher scores on at least one of the anxiety or depression scales in the migraine group as compared to the control group. However, in all eleven studies, the average score on the anxiety and depression scales obtained by children with migraine did not reach a pathological level, according to the norms established by the validated scales. Findings point to above average levels of anxiety or depression, rather than diagnosed psychopathologies. Therefore, certain authors use the term "sub-clinical." One study of a clinical population, paired dimensional assessment with tests and categorical assessment, using diagnostic interviews. In this particular study, children with highly predictive anxiety or depression scores were interviewed by a psychiatrist or psychologist, in order to confirm or deny a diagnosis. No categorical difference was found. Moreover, none of the three studies carried out in the general population revealed differences between the anxiety and depression scores in children with migraine as opposed to children in the control group. The difference in results from studies in the general population and clinical populations can most likely be explained by a recruitment bias. Studies conducted with clinical populations recruit subjects from specialised medical consultations for children and adolescents with migraine, who are probably not representative of the general population. These results contradict those found in the adult population.

Discussion: More studies are needed to better clarify the links between anxiety, depression, and migraine in children, adolescents and adults. To ensure the validity of future studies, the following remarks should be taken into account. The distinction between headache and migraine is not always clear, even when ICHD criteria are used. The children considered to have migraines often have a variety of diagnoses. Future studies should only use the ICHD 2nd edition criteria. Children suffering from migraine are almost always recruited from specialized headache centres in hospitals. This is a very specific population and probably not representative of children with migraine in the general population. In the future, researchers should do their best to avoid this recruitment bias. The questionnaires used in these studies often contain questions related to migraine symptoms such as headache, nausea, vomiting, etc. Several authors have therefore questioned the validity of results from these questionnaires with migraine patients.

Conclusion: Questionnaires created specifically for a research project, or containing vague terminology, such as "psychosocial disorders", should never be used. Future studies should rely on assessment tools validated for the specific population.