"Migraines can really, really hurt but they don't increase your risk for suicide as far as we know," said Grisolia. "We know that people with bipolar have a higher risk of suicide -- I think that that's affecting this data quite a lot."
But Patorno said she believes her study took such complications into account. Indeed, the researchers were careful to select patients without a recent history of suicidal behavior, and they used mathematical modeling to account for differences between the conditions.
"We are pretty confident that we took care of the major confounding factors according to [drug] indication bias," said Patorno.
But even if the study authors accounted for every conceivable difference between epilepsy and migraines, specialists pointed out that they have to account for the mental state of individuals too.
"They [the study authors] control to a large number of factors," Dr. Timothy Lineberry, board chair of the American Association of Suicidology.
"But how do you apply that to an individual patient?" he asked. "Not treating a patient clearly has implications."
For example, a doctor will have to decide whether using a drug that works best for the patient but has a slightly higher suicide risk is better than choosing a less effective drug with lower suicide risk, or no drug at all.
Devinsky echoed that point.
"If the bipolar disorder is poorly controlled there might be other ways patients might have a bad outcome; driving under the influence, or using drugs and alcohol," said Devinsky.
Dr. Taft Parsons III, of the Henry Ford Health System in Detroit, believes the study provided at least one clear implication.
"Pretty much all the diagnoses psychiatrists treat have a risk for suicide," said Parsons, the medical director of Kingswood Hospital in the Henry Ford Health System.
But Parsons argued that this study should inspire doctors who treat pain problems or headaches to also think of the individual patient's psychological profile when prescribing drugs.
"Historically psychiatric problems have been thought of something that doesn't affect your overall [physical] health," said Parsons. "I think this study shows they do."