The Diagnosis Of Headache in Primary Care: Factors in the Agreement of Clinical and Standardized Diagnoses


  • Paul E. Stang PhD, PA-C,

    Associate Director, Adjunct Assistant Professor, Corresponding author
    1. Department of Applied Healthcare Research, Glaxo Inc. Research Institute, Research Triangle Park, North Carolina and
    2. Department of Epidemiology, School of Public Health, University of North Carolina Chapel Hill, Chapel Hill, North Carolina
    Search for more papers by this author
  • Michael Von Korff ScD

    Associate Director
    1. Center for Health Studies, Group Health Cooperative of Puget Sound, Seattle, Washington
    Search for more papers by this author

Paul E. Stang, PhD, Associate Director: Applied Healthcare Research, Glaxo Inc. Research Institute, Five Moore Drive, CS 1130, Research Triangle Park, NC 27709



Objective: Comparison of clinical headache diagnoses in primary care to algorithm diagnoses from a standardized headache interview.

Design: A sample of consecutive primary care headache patients (n=779) at the Group Health Cooperative of Puget Sound were interviewed and medical records abstracted.

Measurements: The medical record provided ‘chart’ diagnoses while a structured interview and diagnostic algorithm yielded the ‘algorithm diagnosis’. The patients were also administered the depression sub-scale from the SCL 90-R, in addition to answering questions about pain intensity and disability used to rade the severity of chronic pain.

Results: Among persons diagnosed by the primary care physician as having migraine, 78% received an algorithm diagnosis of migraine. However, study patients were more likely to receive a diagnosis of migraine by algorithm (57%) than by clinical diagnosis (33%). While clinicians infrequently recorded a diagnosis of migraine and tension-type headache in the same patients, tension-type headache was identified among half of the algorithm diagnosed migraineurs. Among patients with algorithm-diagnosed migraine, female gender, increased isability, and the presence of aura increased the likelihood that the clinician would also diagnose migraine, Conversely, the presence of tension-type headache features, high levels of depression, high number of headache days and age over 65 years reduced the likelihood of the clinician diagnosing migraine among algorithm-diagnosed cases.

Conclusions: The co-occurrence of migraine and tension-type headache symptoms are commonly reported by patients, however, primary care physicians usually recorded a single diagnosis of either migraine or tension headache. Among patients reporting migrainous features, the presence of chronic headache, emotional distress and the absence of disability reduced the likelihood that the clinician would assign a diagnosis of migraine. These results suggest the need for diagnostic and management strategies appropriate for patients whose headaches combine tension-type and migrainous features,