Bio-behavioral treatment
of children with chronic headache: A decision analytic model
The rise of evidence-based medicine in the past few decades has paralleled the incorporation of systematic review methodologies such as decision analysis into clinical research. This abstract outlines a research study in progress that attempts to describe a decision analytic approach to the non-pharmacological management of chronic headache in children. We conducted a pilot decision analysis on this population and found that cognitive-behavior therapy had the best probability (0.6) of causing a reduction in pain intensity (mean payoff = 1.7), and this study elaborates and completes that analysis. An iterative literature search strategy using explicit inclusion and exclusion criteria was developed. Two readers (the second and third authors) conducted keyword, MeSH, and subject searches in MEDLINE, PsycInfo, and HealthStar databases, supplemented by hand-searches of selected journals, consultation with an expert in the field, and follow-up of article bibliographies and unpublished data. Over a thousand articles were identified at the title stage, from which abstract and article level exclusions were then performed. Each reader conducted all searches independently, and assessment of inter-reader agreement and final determination of incorporation into the analysis were conducted by the other authors. Each reader kept a log of searches and this was incorporated into a flow-chart of the search strategy. Disagreements regarding article selection between the first two readers were resolved by the first author, and if unsuccessful, by the other authors. Actual articles are being procured and relevant data will be extracted using a data extraction form developed specifically for this purpose. Articles will be excluded at this stage either if they do not have data required by the extraction form or if their score on a validated measure of quality assessment is less than 1. Proportion of participants demonstrating improvement (probability of improvement) and the actual magnitude of improvement (payoffs) will be determined from each study and incorporated into a decision tree using a decision support software program. Individual trees will be folded back and optimal management paths determined. This study attempts to influence clinical healthcare policy of both this technique and its findings.