Quantifying headache symptoms and other headache features from chart notes

To describe a method for quantifying headache symptoms/features in family practice charts for patients diagnosed with headache NOS (not otherwise specified, ICD-9: 784) and to determine the share of NOS headache diagnoses with clinical data strongly suggestive of migraine or probable migraine headache.

Headache is one of the most common pain symptoms that brings patients to a family physician. However, the majority of headache sufferers do not receive a specific headache diagnosis when they visit physicians.

We examined the chart notes of 454 patients exclusively diagnosed with one or more ICD-9 coded headache NOS diagnoses from July 1, 1995 through December 31, 1999 at a large suburban, university-affiliated practice. We developed a template containing 20 headache items combining International Headache Society diagnostic criteria and additional headache symptoms/features, and decision rules for coding symptoms/features and collected data from patient charts. We then developed decision rules and reclassified NOS headaches into categories strongly suggestive of migraine, probable migraine headache, or other diagnosis. Our main outcome measure is the consistency in the application of decision rules and diagnostic criteria.

With this method we estimate 3 in 10 (29%) headache NOS patients may have had migraine (8%) or probable migraine headache (21%). Reclassified migraine visits averaged 6.5 migraine symptoms and reclassified probable migraine headache visits 4.7 migraine symptoms. Logistic regression analysis supports the consistency of diagnostic criteria for classifying headache based on coded symptoms/features--our model correctly predicted 96% of visits. Evidence of physical examination was recorded at 75% of visits suggesting that physician attention is focused on elimination of secondary headache.

We think the use of our rigorous procedures reveals that a substantial amount of migraine and probable migraine headache may be missed in everyday practice. We hope our findings will provide a basis for the development of diagnostic methods more closely suited to the needs of nonspecialists, and contribute to a better standard of care for headache patients seen in primary care practice. Finally, we are hopeful that other researchers will consider using our template and guideline procedures in their efforts to identify diagnostic patterns and study headache and other health problems.