Short answer: episodic migraine is fewer than 15 headache days per month. Chronic migraine is 15+ headache days per month for at least 3 months, with at least 8 of those days meeting full migraine criteria. The 15-day threshold isn’t arbitrary — it marks a real clinical and prognostic divide defined by ICHD-3.

This page covers what the threshold means, how the transition happens, why it matters for treatment decisions, and one specific implication for how Hermly’s prediction model behaves.

The ICHD-3 threshold

The formal definition (ICHD-3 1.3 — chronic migraine):

  • A. Headache (tension-type-like or migraine-like) on ≥ 15 days per month for > 3 months.
  • B. Occurring in a person with at least 5 prior attacks meeting full migraine criteria (1.1 or 1.2).
  • C. On ≥ 8 days per month for > 3 months, headache meets full migraine criteria OR is treated and relieved by triptan or ergot.
  • D. Not better accounted for by another ICHD-3 diagnosis.

A few important details:

  • It counts headache days, not attacks. A 36-hour attack spans 2 calendar days.
  • The 3-month minimum prevents over-diagnosing a brief flare as chronic.
  • Headaches don’t all have to be full migraine — tension-type days count toward the 15 if there are also 8+ migraine days.

How transitions happen

The transition from episodic to chronic is called transformation (sometimes “transformed migraine” — a term that pre-dates the chronic migraine ICHD-3 entity but still appears in clinical conversation).

Annual transition rates from population cohorts:

  • Episodic → chronic: roughly 2–3 percent per year for the general migraine population.
  • Chronic → episodic: roughly 25 percent per year with appropriate management.

The major risk factors for transformation:

  • High baseline frequency — episodic patients with 10–14 monthly headache days are at highest progression risk.
  • Medication overuse — see MOH. Crossing the overuse threshold is the most common single trigger of transformation in clinical practice.
  • Obesity — independent risk factor in cohort studies.
  • Depression, anxiety — psychiatric comorbidity is a real prognostic factor, not just a coincidence.
  • Snoring / sleep apnea — addressing sleep apnea reduces chronic migraine frequency in cohorts where it was present.
  • Stressful life events — major life stressors precipitate transformation in many patients.

The reversibility is real. Chronic migraine that’s been chronic for years can revert with the right intervention — usually addressing whichever of the above factors is dominant.

Why the threshold matters for treatment

Several treatment decisions depend on which side of the threshold you’re on:

  • OnabotulinumtoxinA (Botox) — FDA-approved specifically for chronic migraine, not episodic. The cycle is every 12 weeks.
  • CGRP monoclonal antibodies — approved for both, but most insurance prior authorisations require chronic for first-line coverage of some agents.
  • OnabotulinumtoxinA reimbursement typically requires documentation of the 15+ days for 3 months.
  • Behavioural intervention intensity — CBT for chronic migraine studies show larger effect sizes than for episodic.
  • Preventive medication thresholds — episodic guidelines often recommend prevention at ≥ 4–6 headache days/month; chronic always warrants prevention.

If you’re hovering near the threshold (12–15 days/month), documenting accurately matters for treatment access.

How Hermly’s model handles chronic users

Specific design detail worth surfacing: the Empatica 2025 study found that none of 5 chronic migraine participants achieved above-random forecasting performance, while all 5 episodic participants did. The base rate near 50% (every other day) makes binary classification essentially impossible with current methods.

Rather than pretending to forecast for chronic users, Hermly includes a chronic frequency gate:

  • When 30-day attack count reaches the chronic threshold, RiskPrediction.confidence returns .insufficient.
  • The Today screen flips to the “no prediction” layout — same as when permissions are denied.
  • A clear explanation appears: “Hermly hasn’t learned your pattern yet — over [N] attacks in 30 days is outside what this model was trained on.”
  • Attack tracking, history, doctor report, and all other features still work fully.

This is an honest fail-loud choice per our methodology. The alternative — showing a fabricated forecast number that the literature says is no better than chance — would burn user trust.

The gate uses event count (≥ 15 distinct attack records in 30 days) as a proxy for the strict ICHD-3 headache-day count, which needs day-overlap arithmetic Hermly doesn’t yet do. See the methodology page §12.1 for the proxy details.

What “chronic” feels like, and what to do

The clinical experience of chronic migraine is qualitatively different from frequent episodic — most patients describe it as “never quite feeling normal”, with shorter and shorter pain-free intervals between attacks. The day-to-day variability that lets episodic patients plan around attacks largely disappears.

If you suspect you’ve transitioned, or are near the threshold:

  • See a headache specialist if you haven’t. Chronic migraine is the indication for the more advanced preventive treatments, and primary care often hands off here.
  • Audit acute medication use. MOH is the most common reversible driver of transformation.
  • Bring data. Hermly’s monthly report shows attack frequency and medication frequency in the format clinicians want.
  • Address comorbidities. Sleep apnea screening, depression evaluation, and obesity management aren’t tangential — they’re central.

The trajectory isn’t fixed. Many chronic patients revert with the right plan.