4. Triggers & Patterns

Established triggers

Alcohol — the in-bout vs out-of-bout distinction matters enormously. During an active cluster bout, alcohol reliably triggers attacks in the majority of drinkers — one Japanese cohort found alcohol triggered a new attack in 95% of habitual drinkers who were in-bout peer-reviewed26. Outside of a bout (in remission), the same patients typically tolerate alcohol without triggering attacks — this in-bout-only sensitivity is a hallmark feature distinguishing CH from most other headache-alcohol relationships peer-reviewed2627. More than 50% of CH patients report alcohol as a trigger during active bouts, though this proportion is notably lower in some Asian cohorts and older Italian data, suggesting either a reporting effect or a genuine population difference tied to drinking habits peer-reviewed27. Curiously, some patients report the opposite effect — large amounts of alcohol producing transient remission or delaying the next attack — an unexplained and minority finding peer-reviewed27. Overall population-level epidemiological data on alcohol and CH risk is genuinely contested: one large cohort found lower CH risk in non-drinkers (RR 0.65), while another found higher risk in non-drinkers (RR 1.54) — directly contradictory results that a meta-analysis could not resolve peer-reviewed28.

Histamine and nitroglycerin are both established provocation-test triggers used experimentally and clinically to confirm CH activity during a bout, per ICHD-3 commentary, though they are not typically encountered as “natural” everyday triggers peer-reviewed1.

Altitude. Unlike the previous pass through this material, a specific CH case report was located: a 40-year-old woman with episodic CH had an attack specifically triggered by high-altitude exposure; her attack was refractory to sumatriptan (which normally worked for her at sea level) but responded to oxygen peer-reviewed8384. This confirms altitude can trigger CH in at least some patients, and that the acute-treatment response profile may shift at altitude — but this is case-report-level evidence (n=1), not a cohort or population study, so the frequency of altitude as a CH trigger across the general CH population remains unquantified. Separately, “high-altitude headache” (HAH) is itself a distinct ICHD-3 secondary headache diagnosis, generally bilateral and exertion-aggravated, and should not be confused with altitude-triggered CH peer-reviewed8586. Population-level data quantifying how common altitude-triggering is specifically among CH patients was not found in this pass and remains a gap.

Barometric pressure and weather — a genuine, unresolved conflict between two different kinds of study. A large, nationwide, peer-reviewed Taiwanese population study (758 episodic CH patients, 2,452 recorded cluster-period episodes, National Health Insurance Research Database, case-crossover design) found that higher mean temperature was significantly associated with the onset of a new cluster period (OR 1.014, 95% CI 1.005–1.023, p = 0.003 on the event day, with similar associations at 7–56 days prior), and that temperature changes following either warm or cold baseline periods could precipitate bout onset, with no such association in tropical climates peer-reviewed8788. By contrast, a 2026 Spanish 14-year primary-care time-series study (99 CH consultations across three centres, ETSX modelling against 14 meteorological variables including barometric pressure, temperature, wind, rainfall, and sunshine) found no statistically significant association between any climatic variable and CH consultation frequency — the authors explicitly note their result diverges from the Taiwanese finding and suggest individual-level triggers (sleep irregularity, stress, behavioural change) may swamp population-level weather signals in a small consultation dataset peer-reviewed89. These two studies are not strictly contradictory — one measures new bout onset against temperature specifically, the other measures consultation frequency against a wider basket of weather variables in a much smaller sample — but the tension between “temperature/weather affects CH periodicity” and “no weather variable predicts CH” should be treated as a live, unresolved question rather than settled science. No peer-reviewed source specifically confirmed barometric pressure (as opposed to temperature) as a validated trigger for CH bout onset or individual attacks; the widely repeated patient and clinical claim that dropping barometric pressure triggers CH attacks remains supported mainly by migraine-literature extrapolation and community report, not CH-specific primary data.

Sleep timing and naps: CH shows strong chronobiological and circadian patterning, with nocturnal attacks common and a documented tendency for attacks to cluster around specific hours and around REM sleep onset (see below) peer-reviewed2930. A South Korean multicentre study found circadian rhythmicity in 55.0% and seasonal rhythmicity in 39.7% of patients, with spring the most commonly cited season for bout onset in a separate Korean cohort (37.5% of those with seasonal propensity) peer-reviewed9091.

Sleep architecture: REM association and obstructive sleep apnoea

Cluster headache has one of the most sleep-entangled profiles of any primary headache disorder:

• Early polysomnography studies found a strong association between nocturnal CH attacks and REM sleep — in one series, almost 60% of recorded CH attacks followed REM sleep despite REM comprising only ~20% of total sleep time peer-reviewed30. Patients with CH also report vivid dream recall when woken by an attack, consistent with REM waking peer-reviewed30.
• However, more recent studies have complicated this picture: several have found CH attacks are not significantly associated with REM sleep after all, and a 2012 review concluded the REM association appears present in episodic CH but not in chronic CH — a genuine, unresolved split in the literature peer-reviewed3130.
• Obstructive sleep apnoea (OSA) is markedly overrepresented in CH populations. A Brazilian study found 58.3% of CH patients had OSA vs 14.3% of controls and 2–4% of the general population, translating to an 8.4-fold increased odds of OSA in CH (rising to 24.4-fold in patients with BMI >25, and 13.5-fold in patients over 40) peer-reviewed32. A separate US study found an even higher rate: 80.6% of episodic CH patients had sleep apnoea on polysomnography peer-reviewed33.
• A 2024 polysomnography study found that greater OSA severity was associated with later CH onset age and longer maximum cluster bout duration, and that even patients with only mild OSA showed elevated susceptibility to hypoxia specifically during REM sleep peer-reviewed31.
• Case reports describe CPAP/BiPAP treatment of comorbid OSA abolishing or greatly reducing nocturnal cluster attacks in some patients, though this remains anecdotal/case-level evidence rather than a controlled trial finding, and the causal direction is explicitly unresolved — current thinking treats CH and OSA as possibly parallel outputs of hypothalamic dysfunction rather than one causing the other peer-reviewed31343536.

Community-reported triggers and protective factors

Smoking: the overwhelming majority (up to ~80%, and 48.3% current/past smoking specifically per the Danish national cohort vs 9.0% in controls) of CH patients are smokers or ex-smokers, one of the strongest lifestyle associations in the condition peer-reviewed3738. However, this looks like association rather than causation: in a pilot survey of 200 patients, active smokers had a more severe CH phenotype (longer active periods, more attacks/day) than never-smokers, but the majority of former smokers reported no change in their CH after quitting — and only around 3% of an internet-surveyed CH population reported improvement after smoking cessation peer-reviewed3938. During an active bout, most patients report either decreasing (45.7%) or keeping stable (45.7%) their smoking, with most of those who cut back reporting a genuinely reduced desire to smoke during the bout peer-reviewed39 [CITIZEN-SCIENCE/COMMUNITY-REPORT for the cessation-outcome framing, PEER-REVIEWED for underlying survey data]. (Editor’s note: the fullest treatment of smoking prevalence and the causality paradox is in Part III, §4.4.)

Psychedelics — citizen science and clinical convergence. This is one of the most striking and well-triangulated findings in the CH literature, originating almost entirely from patient-organised research before formal science caught up. (Editor’s note: the full account — the Clusterbusters origin story, the 2006 Harvard-affiliated survey of 53 patients, the 2015 Clusterbusters Medication Use Survey (496 respondents), the Yale randomised trials and the 2025 case series — is given in Part VI, §§1 and 3, with the current trial pipeline in Part VII, §5.1.3. The duplicate detail that originally sat here was consolidated into Part VI during assembly; this part’s original citations for it remain in the reference list below as 40–46.)

Other citizen-science-flagged patterns: the same Clusterbusters-driven International Cluster Headache Questionnaire (>3,000 respondents, published in Headache, November 2021) also generated formal peer-reviewed papers on oxygen vs sumatriptan effectiveness and patient-reported tolerability citizen science44. Germany’s CSG (Bundesverband der Clusterkopfschmerz-Selbsthilfe-Gruppen) has run its own patient surveys on sex differences in CH presentation and psychological burden, though results from these specific surveys were not retrieved in publicly accessible peer-reviewed form in this pass — flagged as a community data source worth following up directly citizen science9293.