Intranasal treatment of cluster headache: A response

Thank you for your interest in our study and for your valuable comments. First, we agree with Drs. Akhtera and Hashemi, and acknowledge the inherent limitations of this pilot study and the need for a controlled followup study. This we have also stated in the article. Based on our study and previous literature we still do not know if ketamine is effective as a preventive medication in cluster headache (CH) nor do we understand its potential mechanism of action. It could potentially be local as proposed in this comment as the sphenopalatine ganglion (SPG) is suspected to play an important role in the pain process in the trigeminovascular system.1 However, there is considerable anatomical variation2 and a magnetic resonance imaging study found that the SPG is not located just beneath the nasal mucosa, and simple diffusion may not be possible due to anatomical difficulties.3 Clinical trials in CH are lacking due to limited funding and many other challenges. The attacks are short and accompanied by agitation, which makes them difficult to study, and the exact attack onset is unpredictable, even though the attacks follow a circadian rhythm. Even though CH is not a rare disorder, many of the patients are in remission, thus limiting the recruitment potential for trials. Acute treatment should be initiated within minutes of headache onset restricting the design to be either hospital based or biased by patientreported outcomes. A potential future trial should therefore carefully consider what is logistically possible based on recruitment and funding opportunities. The involvement of an SPG approach is really interesting and we have also been involved in several of the past SPG– neuromodulation trials.1,4 We agree that an external block has many advantages as it is quick, cheap, and the side effects are minimal, but we find it very important that patients can selfadminister the acute therapy. It would be very interesting to see if it has a consistent effect in CH; however, comparing a manual block and a nasal spray may additionally cause blinding difficulties, a wellknown issue in CH. Statistics is always debatable, and Drs. Akhter and Hashemi wonder why we chose to apply the parametric tests. Therefore, we put forward our reasoning behind the paired ttest: The assumptions of a paired ttest are that the subjects should be independent, each paired measurement should be obtained from the same subject, and the difference should be normally distributed. The two first assumptions are easy and true for this trial but the normal distribution in small sample sizes is always debatable. We investigated the distribution before assuming normal distribution of the differences. We also collaborated with the biostatistical unit at University of Copenhagen before drafting the manuscript and they agreed with the statistics applied. Based on your comments we additionally performed a ShapiroWilk test for normality on the difference and the pvalue was 0.303. Based on this we believe it is correct to assume that the data were normally distributed. We are thankful for the comment and the chance to elaborate on this trial and hope that it may inspire muchneeded further research into new treatments for CH.