A low MIDAS score is often a tracking artifact, not mild migraine.

Sixty seconds on a clipboard

The clipboard comes back with a half-page form on it. Five questions, each one starting "On how many days in the last 3 months," and you fill it in somewhere between the blood pressure cuff and the door opening.

That number travels further than anything you say out loud for the rest of the appointment. It goes in the chart. It gets copied onto the prior-authorization packet. Ninety days later a utilization reviewer who has never met you decides whether your migraine is disabling enough to justify a preventive, and that single integer is most of what they see of you.

Most people fill it in by counting the days they had a headache. Not one of the five questions asks how many headaches you had.

The pain-resolved cutoff

Every migraine log has a moment where the attack gets declared over. Pain gone, entry closed, back to baseline in the data even though nothing about the following day is baseline. Call it the pain-resolved cutoff: the point where your record stops accruing and your memory stops rehearsing.

It's a design decision, and nearly every tracker makes the same one, because pain has a clean off switch and the residual photophobia, the word-finding trouble, the neck that won't loosen, and the flattened mood do not.

Ninety days later that cutoff shows up on the form. You're being asked to recall lost function across a whole quarter, in a condition that degrades exactly the kind of memory the question requires. So you reach for the structured record you have. It contains headaches, so you report headaches, and you report each one as if the attack and the headache were the same object.

A low MIDAS score in that situation is an artifact of where the tracking stopped. Your neurologist has no way to see that from the number alone.

Why a low score gets read as a mild patient

MIDAS was built by Stewart and Lipton as a triage instrument: translate burden into a grade, and let the grade argue for treatment intensity. Grade I is 0 to 5 points, Grade II is 6 to 10, Grade III is 11 to 20, and Grade IV is 21 or more. It works well for the thing it was designed to do, which is stop a genuinely disabled patient from being sent home with over-the-counter advice.

The grade only reflects what gets reported into it. And the postdrome literature says a large share of the burden never gets there. Giffin and colleagues' 2016 electronic diary study in Neurology found that around 81% of attacks were followed by at least one non-headache symptom once the pain had resolved. Kelman's earlier postdrome series put the proportion of patients reporting a postdrome at roughly two thirds.

For a typical chronic or vestibular migraine patient, then, the majority of attacks in the recall window carry a functional tail that the score has no visibility into. The candidacy question gets decided on the visible part. CGRP monoclonal antibody eligibility for episodic migraine leans on exactly this kind of disability score, a threshold pulled from MIDAS or HIT-6 that treats the reported grade as the whole picture. A postdrome that never made it onto the questionnaire does not move that threshold, no matter how much function it costs the patient between attacks.

What the five questions actually ask

Read the form closely and the mechanics get interesting. Q2 and Q4 don't ask whether you had a headache that day. They ask whether your output fell by half or more, and they explicitly instruct you not to double-count days you already claimed in Q1 or Q3. A postdrome day is by definition a day you did not already claim as a pain day.

That's the whole opening. Those days are eligible, they are not being counted, and the instrument was never the thing blocking them.

One clinical note that makes this defensible rather than creative: the form attributes days to "your headaches," and ICHD-3 describes the postdrome as a phase of the migraine attack itself, not a separate event that happens to follow one. A day lost to postdrome is a day lost to the attack.

One quarter, scored twice

Take a patient with four attacks a month, so twelve attacks across the three-month recall window. Reported from pain days alone, the form might look like this: two full days of work missed (Q1), three days at half output (Q2), one household day lost (Q3), one household day at half (Q4), three cancelled social plans (Q5). Total: 10. Grade II, mild disability.

Now score the same twelve attacks including the day after. Say eight of them left a next day where getting through a shift meant producing well under half of normal. Those eight days weren't claimed in Q1, so they belong in Q2, which goes from 3 to 11. On five of them nothing got done at home, which takes Q4 from 1 to 6. Four evenings got cancelled while the light sensitivity was still there, taking Q5 from 3 to 7.

Same quarter. Same twelve attacks. Same patient. Total: 27. Grade IV, severe disability.

Nothing in that second pass is inflation. It's the same instrument, filled in against the full attack instead of the headache portion of it.

The denial starts before the appeal does

Most of the conversation about CGRP access happens downstream, at the appeal: the letter of medical necessity, the peer-to-peer, the second submission with more documentation attached. That's a real fight and people win it. But by then the disability grade is already fixed in the record, and the appeal is arguing against a number the patient supplied.

So the upstream fix is worth more than the downstream one. If the grade that goes into the first submission reflects the full attack, a large share of these denials never get filed, and the ones that do get filed are contesting a number that holds up.

There's a second effect that's harder to see. A patient scoring Grade II every quarter for two years reads, in the chart, as stable on current therapy. That's an escalation decision made on incomplete data, and nobody in the room knows it. Payers lean on these disability scores when they decide whether a patient qualifies for a CGRP monoclonal antibody, so a grade that understates the attack doesn't just weaken one appeal. It can keep a patient sitting below the threshold for treatment they actually need.

A record that stays open past pain-resolved

What a MIDAS form actually needs from a tracker is narrow. Keep the attack record open after pain resolution, so the following day belongs to the attack rather than to nothing. Capture function per day rather than symptom severity per day, because Q2 and Q4 are asking about output, not about how bad it felt. And produce a date-stamped ninety-day view you can sit with the night before an appointment, since the whole design flaw of a three-month recall instrument is that it asks for recall from people whose recall is one of the affected functions.

That's the shape Postdrome was built around: a continuous timeline that runs through the after-phase, logging that works one-handed with the screen dimmed, an export you own, and a price paid once. It doesn't argue with your insurer. It gives you a record that can.

Before your next appointment, do one thing: pull whatever log you have, find the last three months, and count the days after each attack where you got less than half of your normal done. Then compare that to the number you were about to write on the form.