LOS ANGELES—About 12% of the US population has migraines, and their diagnosis and treatment are primary care functions. However, primary care physicians are often uncomfortable diagnosing migraine, according to research presented at the 56th Annual Meeting of the American Headache Society. Some physicians are too quick to prescribe opiates for migraine symptoms, and others are too quick to provide specialty referrals.

When researchers at the University of Utah in Salt Lake City began a quality improvement project at their headache clinic, the sheer number of referrals prompted a closer look. The clinic had more than 600 patients on a waiting list, but a capacity to handle about 34 new headache patients per month, said K. C. Brennan, MD, Assistant Professor of Neurology and Director of the Headache Physiology Laboratory at the University of Utah. “Our question was, ‘What is generating this tidal flow of patients?’”

Opiate Use Is “Huge”
Dr. Brennan and his colleagues used the university’s electronic medical record (EMR) database to seek information on all patients with a headache–related code treated at any of the university’s 10 primary care clinics between 2008 and 2010. “This kind of data can sometimes be rather low quality, but we had very large numbers,” said Dr. Brennan, with nearly 17,000 headache patients and more than 50,000 encounters during that two-year period.

Analysis of the data, reported at the 2014 Annual Meeting of the American Academy of Neurology, revealed “huge opiate use,” he added. In comparison, triptan use was “disconcertingly low,” and ergot use was also very low. Although usage varied widely from one clinic to another, “the most commonly used medication in our academic clinic network in primary care was opiates,” which were prescribed in 30% of all headache encounters (per-clinic range, 15% to 62%).

How can this be, the researchers wondered, given that migraine is common and physicians know what migraine looks like?

Symptom Coding Used Most Frequently
One problem involved coding. Providers at the primary care clinics used headache symptom codes for 55% of the visits, suggesting a reluctance to diagnose migraine. Migraine diagnostic codes were used for 42% of the encounters, and both triptans and opiates were more likely to be prescribed in such cases. Providers gave triptans to 19% of those with a migraine diagnosis, however, while 36% received opiates. Prescribing an opiate for a patient diagnosed with migraine is especially troublesome, Dr. Brennan noted, as it indicates that “even when you know something is migraine, you’re treating it incorrectly.”

Further analysis revealed that 10% of providers were responsible for two-thirds of opiate prescriptions. Similarly, 9% of patients—designated heavy users because they visited the clinic seven or more times in the two-year study period—generated almost half (45.5%) of the 50,000 headache encounters. Such patients were much more likely than those seen less frequently to receive a migraine diagnosis (OR, 2.46) and a prescription for an opiate (OR, 5.19).

Root Causes and a Promising Intervention
The researchers concluded that the frequent use of symptom codes and the overuse of opiates may be related. In collaboration with primary care physicians, the investigators considered root causes—a dearth of clinical training in headache care, time constraints, patient preference for referral, and the perception that headache patients are difficult—and how best to respond. They decided on an EMR-based intervention.

Diagnostic, treatment, and referral aids were added to the system, along with ID Migraine, a three-question diagnostic aid developed by Richard B. Lipton, MD. A medical assistant can ask the questions, which cover severity of symptoms, nausea, and reaction to light, before the physician even sees the patient, Dr. Brennan said. A positive response to two of three questions suggests a migraine diagnosis and activates decision support, including possible medication choices.

The researchers conducted a pilot study at two of the university’s primary care clinics, both of which are used to train residents and are involved in quality improvement. “To our surprise, we got statistically significant results,” Dr. Brennan reported. “The ratio of bad codes to good codes went down,” the proportion of encounters in which patients were given triptans increased, and prescriptions for opiates declined.

Can the Results Be Duplicated?
The EMR-based intervention is ongoing. The goal is to improve population health, Dr. Brennan said, adding that effecting change at the other primary care clinics in the system will likely take a greater effort. “Going forward, we want to customize, to focus on the needs of special sites and special providers,” and to build “cross-cultural trust” between primary care providers and neurologists. “We are trying to extend this to other pain pathways and possibly to other institutions,” he concluded.

—Helen Lippman

Prescribing Opioids for Migraine—A Problem in Primary Care