PDF: High Flow Oxygen for Cluster
(2009)Objective
To evaluate whether high-flow 100% inhaled oxygen is effective in aborting cluster headache attacks compared to placebo (air).
Study Summary
Intervention
Randomized, double-blind, placebo-controlled crossover trial. Patients with episodic or chronic cluster headaches received either high-flow 100% oxygen at 12 L/min or air for 15 minutes via face mask during attacks, in alternating order for four attacks.
Inclusion Criteria
Adults aged 18–70 with episodic or chronic cluster headache who had not previously used high-flow oxygen. Required at least 1 attack every other day and up to 5 per day.
Study Design
Arms: High-flow oxygen (12 L/min for 15 min) • Air placebo (12 L/min for 15 min)
Patients per Arm: Approximately 150 attacks per treatment group; 76 patients completed the study
Outcome
Bottom Line
High-flow oxygen therapy (100% oxygen at 12 L/min for 15 minutes) was significantly more effective than placebo air in aborting acute cluster headache attacks, with 78% of oxygen-treated attacks resulting in pain freedom at 15 minutes compared to 20% with air (P<0.001), with no serious adverse events.
Major Points
- First adequately powered, double-blind, placebo-controlled crossover trial of high-flow oxygen in cluster headache
- Oxygen was superior to air for the primary endpoint: 78% pain-free at 15 minutes vs 20% with air (P<0.001)
- Oxygen was numerically superior to air for all secondary endpoints including pain freedom at 30 minutes, pain reduction at all time points, functional disability, and reduced need for rescue medication
- No serious adverse events related to treatment
- Study used crossover design with patients treating 4 attacks alternately with oxygen and air
- Included both episodic (75%) and chronic (25%) cluster headache patients
- 109 patients randomized, 76 completed the study (57 episodic, 19 chronic)
- 298 total attacks treated (150 with oxygen, 148 with air)
- Treatment delivered at home using standard CD-sized 2-liter cylinders with nonrebreathing face masks
- Patients with episodic cluster headache withdrew preventive medications; those with chronic cluster headache maintained stable doses
Study Design
- Study Type
- Randomized, double-blind, placebo-controlled crossover trial
- Randomization
- Yes
- Blinding
- Double-blind; patients and investigators were blinded to treatment assignment. Randomization code remained locked until end of trial
- Sample Size
- 109
- Follow-up
- 60 minutes per attack; patients remained in study until 4 attacks were treated
- Centers
- 1
- Countries
- United Kingdom
Primary Outcome
Definition: Pain free at 15 minutes or, in the absence of a diary, patient-adjudged adequate relief at 15 minutes
| Control | Intervention | HR/OR | P-value |
|---|---|---|---|
| 29/148 attacks (20%, 95% CI 14%-26%) | 116/150 attacks (78%, 95% CI 71%-85%) | - | <0.001 (Wald test χ²₅=66.7) |
Limitations & Criticisms
- Study conducted in patients' homes rather than controlled clinical setting, which may have contributed to withdrawals and missing data
- Only 33% of screened patients were randomized (109/334), and only 23% completed the study (76/334)
- Patients who had previously tried oxygen at 4 L/min or higher were excluded, which became increasingly problematic as the study progressed and may limit generalizability
- Some patients did not complete full diaries, with 9 patients (8% of randomized, 12% of completed) providing only primary endpoint data via telephone/email/follow-up
- Attack duration requirement of minimum 45 minutes does not fully align with ICHD criteria (which allows attacks as short as 30 minutes)
- Some patients took treatment when attack was already underway rather than at commencement as specified in protocol
- Relatively small number of chronic cluster headache patients (21% of completers) limits subgroup analysis for this population
- Crossover design introduces complexity as observations from same patient are not strictly independent (addressed through multilevel analysis)
- Data for periods after rescue medication taken were excluded from secondary analyses, contributing to missing secondary endpoint data
- Many episodic cluster headache patients withdrew because their bout ended before completing 4 attacks
- 17 patients came out of bout, 9 lost to follow-up, 6 withdrew, 1 died before receiving treatment
- Protocol violations included: patients continuing preventive medications, treatments taken out of order, cylinders running out early
- No comparison with active comparator (e.g., sumatriptan); only compared to placebo
- Study does not address optimal flow rate or whether patients with prior oxygen exposure would respond similarly
Citation
JAMA. 2009;302(22):2451-2457