Treatment of recurrent tension headache: a meta-analytic review

The primary aim of this study was to investigate which treatment, patient, and study characteristics are related to treatment outcome in tension headache. Literature on the subject was obtained by searching for articles published in English through CD-rom Compact Cambridge and PSYCHLIT (1970/1993) on the subjects of headache and tension or muscle contraction in combination with therapy or treatment. Inclusion criteria were prospective design, tension headache, > or = five subjects per condition; pre- and posttreatment scores available, subjects older than 16 years. Each article was independently reviewed and its data encoded by the two authors. Finally, 78 articles with 175 treated and nontreated conditions (total of 2,866 patients) were selected for this study. Meta-analysis revealed that cognitive therapy, relaxation, or electromyelographic (EMG) biofeedback alone or in combination with relaxation were superior to no treatment and to pseudo/placebo therapy. Pharmacological and other therapies were better than no treatment. However, restricting the analyses to studies using a headache diary, the results of pharmacological therapy were comparable to those of placebo therapy. Moreover, a better treatment outcome was found in studies with a relatively short duration of headache complaints (r = -0.31) and with a relatively younger patient sample (r = -0.22). The year of publication was negatively related with improvement (r = -0.36). No relationship between treatment outcome and other treatment characteristics (duration, transfer of treatment), patient characteristics (gender, method of patient recruitment), and study characteristics (internal validity of the study, criteria for diagnosing tension headache, number of patients, drop-outs) was found. These findings suggest that treatment outcome may be affected more by patient characteristics than by treatment characteristics. Finally, some practical and research implications of these findings are discussed.