Inhaled Oxygen and Cluster Headache Sufferers in the United States: Use, Efficacy and Economics

Results From the United States Cluster Headache Survey

Todd D. Rozen, MD, FAAN; Royce S. Fishman, BA


Headache. 2011;51(2):191-200. 

In This Article


Inhaled normobaric oxygen has been recognized as an effective acute treatment for CH for about 50 years but its actual use in CH patients in the general population outside of dedicated headache centers has never been truly determined.[2] CH is such a severe condition that if patients do not have adequate acute treatment they can develop suicidal ideations and some individuals will act on these thoughts. Inhaled oxygen is an essential acute therapy for CH patients not only because it is effective, but in many instances CH patients who are known heavy cigarette smokers will eventually develop medical contraindications to the other primary CH abortives triptans and DHE, thus leaving oxygen as a sole abortive choice. In addition, because prescribing information limits the use of injectable sumatriptan to 2 doses per day due to potential adverse reactions and 53% of survey respondents report more than 2 CH attacks per day,[6] the need exists for an abortive such as oxygen that can be safely used for as many attacks per day as a sufferer may have. Several small clinic and community-based investigations have indicated that more than 50% of CH patients have never used oxygen for the treatment of their headaches.[7,8] This statistic is alarming and the reasons why they have not tried oxygen have never been determined. To date, we have no large population-based studies looking at CH and inhaled oxygen usage. The United States Cluster Headache Survey is the largest study ever completed looking at CH sufferers living in the United States. A large percentage of the survey was dedicated to questions about inhaled oxygen and its overall usage, efficacy, prescribing characteristics and costs. The survey responders were from a selected population drawn from CH support groups and treating physicians. This type of epidemiologic investigation has been done previously by Bahra et al[7] in a mostly UK investigation but the present study is novel based on its size of the studied CH population and its focus on oxygen therapy. The data are important not only for treating physicians and CH patients but also for the governmental agencies (Centers for Medicare and Medicaid Services) who are currently now reviewing if oxygen should garner a specific reimbursement code for CH treatment.

Greater than 90% of CH sufferers living in the United States are aware that oxygen is an acute treatment for CH but one-third of the population has never tried oxygen. Several reasons for this fact can be deemed from this study. First, 12% of the survey patients' physicians are refusing to prescribe oxygen therapy for their CH patients. Almost 50% stated they do not believe oxygen works for CH. This may reflect their prescription patterns for oxygen which in many instances are not the recommended treatment protocol (delivery system or flow rate) and thus would not lead to a positive abortive effect. CH patients are still being prescribed nasal cannula as a delivery system instead of a non-rebreather face mask and are being prescribed too low of flow rates and are not being told to increase the flow rates if the lower flow rates are not effective. One-third of physicians are not prescribing oxygen for CH patients because they did not know this was actually a therapy for CH, perhaps indicative of a problem with physician instruction during their training. In defense of academic neurology there are very few lectures given to neurology residents and other medical trainees dedicated to headache alone and as CH is somewhat rare it may be omitted from these lectures altogether. Unfortunately, this may affect patient care. Part of the resistance to prescribe oxygen, or the lack of recognition of oxygen's usefulness in CH, by physicians could relate to the deficiency of placebo-controlled trials with inhaled oxygen in the medical literature. This has recently changed, however, with the study by Cohen et al[4] published in JAMA in late 2009, which noted statistically significant improvement in CH with high flow oxygen vs placebo. The present survey was completed prior to the publication of this trial, and thus how it will affect physician prescribing patterns in the future is unknown. For most CH patients, oxygen, if prescribed, is not given for months to years after the initial diagnosis of CH is made. This again may reflect the lack of awareness of oxygen as a treatment or the lack of belief that oxygen actually works in CH by the physicians treating these patients. Almost half of the surveyed patients had to first suggest oxygen as a treatment to their physicians. Interestingly, the younger the CH patient the less likely they are to have utilized oxygen, which is opposite of what one would think as newer physicians in training should be more aware of inhaled oxygen as CH treatment although as stated above that may not be the case. This may also reflect the fact that younger CH patients are not seeking out medical care for their CH. Seventy-one percent of the survey patients under the age of 20 years are not currently seeing a neurologist. Second, oxygen may actually get prescribed to patients but multiple barriers have been set up that precludes them from actually getting the oxygen. More than 50% of survey responders stated that it was difficult to actually find a source for their oxygen. In addition, 45% received no information about a source for oxygen from their prescribing physician. If a CH patient is able to find a source for oxygen in most instances, their prescriptions are not complete lacking a prescribed flow rate or delivery system, and thus they may not be able to obtain the oxygen because of incomplete prescription data. Additionally, as the majority of CH patients are given no instruction on oxygen utilization, they may be too apprehensive to have an oxygen tank in their home or place of work with no idea what to do with that tank. Finally, cost presumably is a reason behind the low oxygen utilization rates, as 16% of the surveyed population stated that oxygen was unaffordable.

In regard to the effectiveness of inhaled oxygen for CH, the majority of patients (70%) felt it was helpful. This percentage matches up well with other studies, both open label and placebo-controlled trials looking at oxygen treatment for CH.[3,4,7] From the survey results inhaled oxygen efficacy does not appear to vary by patient age, the number of CH attacks per day and smoking history. Very few studies have looked at oxygen response and CH subtype and gender. In the present study, patients with episodic CH statistically had higher efficacy rates with inhaled oxygen than chronic CH patients. In addition, episodic CH patients statistically responded faster to oxygen than chronic CH patients. Kudrow's[3] original study on inhaled oxygen also noted less effectiveness in chronic CH patients than in episodic CH patients. Schürks et al[9] looking at 246 clinic and non-clinic based CH patients noted a higher efficacy of inhaled oxygen in episodic CH patients vs chronic CH patients but this was not a statistically significant difference. Oxygen flow rates utilized were not documented in the manuscript. The cause of the probable greater efficacy of inhaled oxygen in episodic CH patients is unknown. Chronic CH patients are more likely to be smokers so that could be a potential cause of lack of oxygen effect, but in the present survey smokers and non-smokers had equal efficacy to inhaled oxygen.[10] In another headache clinic study, it was found that smokers actually responded better to oxygen than non-smokers.[11] Chronic CH patients as a whole may respond less to CH abortives than episodic CH patients as this trend was also noted in a large placebo-controlled trial looking at intranasal zolmitriptan.[12]

Regarding gender, both male and female CH patients had equal efficacy to inhaled oxygen but men responded faster to oxygen: men were statistically more likely to be pain free after 10 minutes of oxygen therapy compared to women. This finding has not been published previously. Interestingly men used higher flow rates at the outset of attacks than women and that may have led to the faster response times. Rozen et al in 1999[13] did show in a clinic-based study of CH a gender difference in response to oxygen as only 59% of female cluster patients responded, whereas 87% of men did. This study, however, was with relatively low flow rates of oxygen (7 L/minute), thus a possible reason why women responded less to oxygen than in the present survey study. In Kudrow's landmark study on oxygen, male and female patients showed equal efficacy to inhaled oxygen at 7 L/minute.[3] Schürks et al[9] looking at 246 CH patients noted a higher efficacy of oxygen in female vs male CH patients but this was not a statistically significant difference.

If oxygen is so effective for CH then why are only 25% of the survey responders currently using oxygen and in the ones who are treating with oxygen why are only 25% using it as their sole abortive? This is not the first study to show that even when oxygen is effective for CH, many sufferers do not continue to use it. Gallagher et al in 1996[14] noted in 60 CH patients that 76% responded to oxygen but only 31% continued to utilize oxygen. Several items from the present survey responses may help us better understand this issue. First, oxygen may indeed be helpful for CH but its effectiveness and ease of use just does not compare with other abortive choices like injectable sumatriptan which is very effective, quick to onset and does not require special equipment and a tremendous effort to even locate. Second, the goal of acute CH treatment is fast and effective relief, as the duration of CH is relatively short from 15 to 180 minutes.[15] Overall the CH survey responders suggested that inhaled oxygen is fairly slow to have an abortive effect as 49% noted that they needed over 21 minutes to completely stop a headache. How many of these patients actually had spontaneous resolution of their CH without any effect from oxygen cannot be obtained from the survey data. This relatively long duration to effect may reflect the flow rates that are being prescribed to patients: it is possible the rates are too low to effectively stop a CH. Seventy-four percent of patients are being prescribed oxygen flow rates of 7 to 12 L/minute. Most CH patients do not follow the prescribed treatment guidelines for oxygen as almost all utilized flow rates higher than those prescribed by their physicians (Table 1). Over 50% of the CH patients end up using flow rates of 13 L/minute or higher for individual CH attacks and almost all patients will raise the flow rate during treatment to achieve efficacy. This survey was completed prior to the recent publication in JAMA of the randomized placebo-controlled trial documenting the efficacy of high flow oxygen for CH using 12 L/minute over placebo (room air).[7] Thus, our survey results as well as the recent placebo-controlled study suggest that high oxygen flow rates seem almost essential for CH efficacy and possibly faster treatment response times and the old recognized flow rate of 7 L/minute documented by Kudrow in 1981[3] in reality is too low for a large percentage of CH patients in the general population. The final reason why patients may not be using oxygen even though it is effective is the costs issue and the difficulty of obtaining oxygen once it is prescribed.

This is the first ever study to look at the economics of oxygen treatment for CH. One-third of the CH population in the United States need to spend more than $1000 per year on oxygen treatment and a small percent spend upwards of $12,000 per year. At present only 64% of survey responders state that oxygen is covered in some manner by medical insurance carriers and this compares to 76% coverage for injectable sumatriptan. Even though oxygen has been a recognized treatment for CH for almost half a century, still almost 50% of insurance carriers want medical literature submitted that supports the use of oxygen in CH before it will be approved for use. In many instances by the time the patient does get insurance clearance to use oxygen their cycle is already over and they have suffered unnecessarily. It is still commonplace for insurance companies to ask physicians about blood gas results or pulse oximetry results before oxygen will be approved for use for CH patients as if they were candidates for chronic obstructive pulmonary disease (COPD) respiratory therapy. Of course this has nothing to do with CH patients who are not using oxygen for hypoxemia. At present 16% of CH patients in the United States cannot afford oxygen therapy, which is a serious issue for a patient population who has been known to take their lives during a CH attack. Twelve percent will get non-medical welding grade oxygen, which could be harmful to the user depending on the gas or gases the cylinder was previously filled with and trace gases still present in the cylinder. In addition, non-medical grade oxygen will definitely be less efficacious.

An interesting and unexpected finding from the survey is that oxygen may actually enhance the efficacy of non-oxygen abortives like triptans or DHE. When oxygen is combined with another abortive more patients have complete pain relief after a shorter duration of time compared to when the triptan or DHE is used alone. Thirty-seven percent of the surveyed patients had complete relief of their CH within 15 minutes of using combined treatment vs only 28% when a non-oxygen abortive was used alone. Of note however only 36% of these patients were using injectable sumatriptan, and thus it would be expected that oral triptan therapy and/or injectable DHE would have a slower time to pain relief than oxygen alone, so we cannot definitively state there is an enhanced treatment effect with oxygen. We are unable to determine from the available data if oxygen improves the response times of injectable sumatriptan alone which would more strongly suggest a treatment enhancing effect. In addition, these are all time estimates by patients who were not using stopwatches to determine true time to pain relief. A specific study looking at oxygen and its effect on non-oxygen CH abortives would need to be completed to substantiate our findings. The present study suggests that there may be a synergistic effect between oxygen and triptans and oxygen and DHE. This has not been previously reported from other CH studies and may change our thinking about using one abortive medication alone to treat an acute attack of CH. Interestingly in Horton's[2] original description of oxygen for CH he used a combination of oxygen and DHE so maybe an early demonstration of a synergistic effect.

There are certainly several limitations to this study. First, even though this is a good representative sample of CH patients living in the United States it certainly does not represent every, or nearly every, CH patient residing in this country. The suggested prevalence for CH is 0.4% of the general population, and thus there should be about 1.2 million individuals with CH in the United States and we only have data from 1134 individuals. It is possible, although not probable, that the remaining CH patients in the United States have a completely different response to oxygen, and thus these survey results should be interpreted with that statement in mind. Second, it is possible that the patients who answered the survey are a biased population based on the fact that they frequent CH websites. The survey responders may be more knowledgeable about CH and may even have harder to treat headaches than the general CH population because they search out the web for support and treatment options. On the other hand, as this is probably a more educated CH population, their experience with inhaled oxygen is better defined than those patients who do not have as severe CH and have only dabbled in CH treatment. Thus, the survey may more truly reflect the efficacy of inhaled oxygen and the hardships of obtaining it or getting a physician to prescribe it. Third, the study is lacking diagnostic validation. Even though all study participants had their headache diagnosis made by a neurologist, these medical specialists can still misdiagnose CH. It is possible that a certain percentage of survey responders had other trigeminal autonomic cephalalgias such as paroxysmal hemicrania or even had migraine with cranial autonomic symptoms.[16] A large percentage of survey responders (80%) had never tried indomethacin, thus not ruling out paroxysmal hemicrania as a missed diagnosis. However, of those who did try indomethacin only 15% stated that it had some effect on their headache and less than 2% were currently using it. In regard to the CH subtype diagnosis of chronic vs episodic CH this also lacked diagnostic validation in our study as it was completely responder defined. Thus, any documented distinction between these subgroups in the survey results must be looked at with this issue in mind.

Finally, as this is a CH population from the United States it may only reflect experience with oxygen in this geographic region and the survey results cannot be extrapolated to CH patients worldwide. The hope is that other large country-based population studies can be done with CH patients to get a true idea of inhaled oxygen efficacy, economics and usage and to see if the results are geographically similar or dissimilar.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.