Methodological Issues in Systematic Reviews of Headache Trials: Adapting Historical Diagnostic Classifications and Outcome Measures to Present-Day Standards


  • Douglas C. McCrory MD,

  • Rebecca N. Gray DPhil,

  • Peer Tfelt-Hansen MD, PhD,

  • Timothy J. Steiner MB, PhD,

  • Frederick R. Taylor MD

  • From the Duke University Medical Center, Center for Clinical Health Policy Research Durham, (Drs. McCrory, Gray); Durham Veterans Affairs Medical Center, Center for Health Services Research in Primary Care, Durham, NC (Dr. McCrocy); Glostrup Hospital, Department of Neurology, Glostrup, Denmark (Dr. Tfelt-Hansen); Imperial College, Division of Neuroscience, London, United Kingdom (Dr. Steiner); and the Park Nicollet Headache Clinic, Minneapolis, MN (Dr. Taylor).

Address all correspondence to Douglas C. McCrory, Duke University Medical Center, Center for Clinical Health Policy Research, Durham, NC.


Recent efforts to make headache diagnostic classification and clinical trial methodology more consistent provide valuable advice to trialists generating new evidence on effectiveness of treatments for headache; however, interpreting older trials that do not conform to new standards remains problematic. Systematic reviewers seeking to utilize historical data can adapt currently recommended diagnostic classification and clinical trial methodological approaches to interpret all available data relative to current standards. In evaluating study populations, systematic reviewers can:

  • (i) use available data to attempt to map study populations to diagnoses in the new International Classification of Headache Disorders; and
  • (ii) stratify analyses based on the extent to which study populations are precisely specified.

In evaluating outcome measures, systematic reviewers can:

  • (i) summarize prevention studies using headache frequency, incorporating headache index in a stratified analysis if headache frequency is not available;
  • (ii) summarize acute treatment studies using pain-free response as reported in directly measured headache improvement or headache severity outcomes; and
  • (iii) avoid analysis of recurrence or relapse data not conforming to the sustained pain-free response definition.