From the Division of General Internal Medicine, University of Cincinnati College of Medicine, Cincinnati, OH (Dr. Martin); Department of Psychiatry and Human Behavior, University of Mississippi Medical Center, Jackson, MS (Dr. Penzien); Center for Pain Studies, Rehabilitation Institute of Chicago, Chicago, IL (Dr. Houle); Department of Biostatistics, West Virginia University, Morgantown, WV (Dr. Andrew); and Pain and Rehabilitation Clinic of Chicago, Chicago, IL (Dr. Lofland).
The Predictive Value of Abbreviated Migraine Diagnostic Criteria
Article first published online: 15 SEP 2005
Headache: The Journal of Head and Face Pain
Volume 45, Issue 9, pages 1102–1112, October 2005
How to Cite
Martin, V. T., Penzien, D. B., Houle, T. T., Andrew, M. E. and Lofland, K. R. (2005), The Predictive Value of Abbreviated Migraine Diagnostic Criteria. Headache: The Journal of Head and Face Pain, 45: 1102–1112. doi: 10.1111/j.1526-4610.2005.00234.x
- Issue published online: 15 SEP 2005
- Article first published online: 15 SEP 2005
- Accepted for publication December 1, 2004.
- primary care;
- abbreviated criteria
Objective.—To determine the operating characteristics and predictive value of abbreviated criteria for the diagnosis of migraine headache.
Background.—The International Headache Society (IHS) diagnostic criteria for migraine have been adopted in limited fashion in clinical practice. Primary care physicians in particular deal with innumerable conditions and diagnostic algorithms. Unless the IHS criteria are simplified the recognition of migraine headache in primary care settings will not be apt to improve.
Methods.—This study was a retrospective analysis of four discrete research databases: headache clinic patients (N = 390), private practice neurology patients (N = 290), college students (N = 99), and community-based patients (N = 784). Physicians and psychologists expert in the diagnostic criteria for migraine headache syndromes conducted a standardized diagnostic interview in all patients (N = 1524). Each was later assigned an IHS headache diagnosis by a previously validated computer-based algorithm. The sensitivity, specificity, positive and negative predictive values, and accuracy were calculated for single- and multiple-variable models of migraine predictors. Optimal models were defined as those with positive likelihood ratios (+LRs) of >4.5 and negative likelihood ratios (−LRs) of <0.25 for the combined population.
Results.—The only optimal single-variable model was nausea, which had an overall +LR of 4.8 and −LR of 0.23. None of the two-variable models met criteria for an optimal model. The best of the optimal three-variable models were nausea/photophobia/pulsating (+LR 6.7, −LR 0.23) and nausea/photophobia/worsening with physical activity (+LR 5.9, −LR 0.21). These three models maintained positive predictive values >0.80 in all 4 patient populations and negative predictive values >0.70 in the majority of populations.
Conclusion.—The single-variable model of nausea and the three-variable models of nausea/photophobia/worse with exertion and nausea/phonophobia/pulsating can effectively predict migraine in diverse clinical settings. These models however, should only be applied after a careful exclusion of secondary headache disorders.