2. The TAC Family & Differentials

Treatment response — particularly to indomethacin — functions as a diagnostic tool within the TAC family, not merely as therapy. This is a critical, non-obvious point: if a “cluster-like” daily headache resolves absolutely with indomethacin, the correct diagnosis is very likely paroxysmal hemicrania or hemicrania continua, not CH — and CH itself does not reliably respond to indomethacin peer-reviewed11.

Condition	Attack duration	Frequency	Autonomic features	Indomethacin response	Other diagnostic notes
Cluster headache	15–180 min	Every other day to 8/day	Prominent, ipsilateral	Generally poor/absent — triptans/oxygen work instead peer-reviewed11	Circadian/circannual rhythmicity; alcohol/nitroglycerin/histamine can provoke attacks during a bout peer-reviewed1
Paroxysmal hemicrania	2–30 min	>5/day for at least half the time (≥20 lifetime attacks)	Present	Absolute — sine qua non for diagnosis; effective dose range 25–300 mg/day peer-reviewed1213	Chronic form more common than episodic in most series; some cases take up to a week to show response peer-reviewed12
Hemicrania continua	Continuous, >3 months, with exacerbations	Daily, continuous (remitting or unremitting subtype)	Present during exacerbations	Absolute — required for diagnosis; oral dose often 25 mg tid titrated to 100–225 mg/day	Critical differential for any daily unilateral head pain; easily mistaken for chronic migraine or CCH peer-reviewed1415
SUNCT	1–600 seconds, single/grouped/saw-tooth stabs	≥1/day (often far more; up to 200/day)	Must have BOTH conjunctival injection AND lacrimation	Not indomethacin-responsive; lamotrigine most effective preventive; IV lidocaine for severe bouts	Triggerable without refractory period, unlike trigeminal neuralgia peer-reviewed1617
SUNA	1–600 seconds	≥1/day	Only one, or neither, of conjunctival injection/lacrimation, but ≥1 other autonomic sign	Not indomethacin-responsive; same treatment profile as SUNCT	Considered by many experts to be the same underlying entity as SUNCT (a spectrum) peer-reviewed1819
Migraine with autonomic features	4–72 hours	Variable	Can occur but usually milder, bilateral-leaning	Not indomethacin-responsive	Photophobia/phonophobia/nausea more typical of migraine; overlap with CH is a major misdiagnosis driver peer-reviewed20
Trigeminal neuralgia	Seconds	Variable, often clustered stabs	Absent or minimal	Not indomethacin-responsive; carbamazepine first-line	Has a post-attack refractory period, unlike SUNCT/SUNA; occasional diagnostic overlap reported with SUNCT peer-reviewed21
Secondary causes (e.g. pituitary lesions, vascular lesions, sinusitis mimics)	Variable	Variable	Can mimic CH exactly	N/A	Pituitary/suprasellar masses account for a disproportionate share (~28.6% of secondary CH cases; 77.3% of mass lesions) of structural mimics — this is why baseline MRI (with dedicated pituitary/sella views) is recommended for every CH diagnosis, and pituitary hormone testing considered in refractory cases even with negative imaging peer-reviewed2223

Two structural points worth internalising: first, SUNCT is now widely regarded as possibly a subform of SUNA rather than a fully distinct entity, though ICHD-3 still classifies them separately pending further study peer-reviewed1719. Second, “cluster headache” itself carries a strong positive likelihood ratio (~11) for detecting significant intracranial abnormality on imaging, and red flags such as late age of onset, abnormal neurological exam (especially cranial nerve findings, likelihood ratio 5.3), a changing headache pattern, or pain worsened by Valsalva substantially raise suspicion for a secondary cause peer-reviewed22.

A further, less-discussed complication: a large German cohort study (825 patients, Kiel Pain Clinic) found considerable clinical variability within CH itself — some patients report persistent, lower-grade pain between attacks, a feature technically outside strict ICHD-3 criteria but clinically real and potentially confused with hemicrania continua. The authors argue that overly strict application of ICHD-3 criteria can itself delay diagnosis in atypical presentations peer-reviewed71.