Analysis of occipital nerve stimulation in studies of chronic migraine and broader implications of social media in clinical trials

Occipital nerve stimulation (ONS) is not new, being pioneered in a technical sense more than a decade ago by Weiner and Reed (1). It has now been studied in three randomized controlled trials in chronic migraine (2,3), the most recent just appearing in (4) the Journal (5). There is a good biological basis for the therapy, both in the laboratory (6) and in humans (7), and a reasonable medium-term experience (8). Yet the most recent study is again negative on the primary endpoint (5). The authors explain lowering the bar to success as a way forward, and given the disability of patients with chronic migraine (9,10), and the comparable bar recently accepted for botulinum toxin type A (National Institute for Clinical Excellence, http://publications.nice.org.uk/botulinum-toxin-type-a-for-the-prevention-of-headaches-in-adults-with-chronic-migraine-ta260), this argument has merit. Notwithstanding there remain barriers to establishing ONS in clinical practice, which are shared with most neuromodulation approaches, and are substantially methodological. This is more than trivial since it is regulators and payers, correctly being careful about costs, who must be convinced of the utility of new and relatively expensive therapies, as is pointed out in an accompanying editorial (4). The new study by Silberstein and colleagues (5) illustrates some of these important methodological considerations.

In total 177 patients were eligible for device implantation (5). Of this group, 20 had no pain relief or inadequate paresthesia, and did not proceed to randomization. They are described as screen failures. It can be argued that, in clinical practice, they are treatment failures from the viewpoint of the patients and clinicians, since the therapy did not work for them. Now they were not randomized so by current standards (11) they do not belong in the full-analysis or intention-to-treat data set, yet in practice such patients will have had both the expense of the trial and not had benefit, i.e. clinical failures. It seems sensible, actually ethical, to exclude patients who fail a trial of stimulation for tolerability since asking them to have an intolerable stimulation for three months is unreasonable; so how can we capture such outcomes? One could argue for these cases to be randomized, in a virtual sense, by the same code to active or sham and their data counted as a failure thus balancing their clinical characteristics into both groups, and providing a more realistic assessment of the value of ONS to patients with chronic migraine.

The other substantial issue remains blinding and some new challenges from the Internet age. First, the new study (5) is clearly not blinded in any conventional sense of the term and so provides, at best, Class III evidence since there was not masking of the stimulation, and thus the attitude of the subject to the treatment could have been biased (11). We must work out how to do this better. Secondly, it could be argued that social media represents a substantial potential for trial unblinding. A blog or Facebook page, innocently set up for shared experience, could easily lead to unblinding, and one wonders if any such activities occurred in this study. As a field we need to engage with patient groups, since they are the crucial stakeholders in any new development, and test concepts around social media. It can be argued that part of the new age may include enrolling patients willing to sign up to not participate in social media discussions of their treatment while the study is ongoing. Much as we enroll patients currently who are capable of, for example, filling in the diary and who agree to be compliant, it can be argued that nonparticipation in social medial discussions protects blinding and thus ultimately adds strength to study conclusions. Conversely, unblinding can cast doubt on study conclusions and thus waste the study participants’ time. Perhaps it is one role for the International Headache Society to engage with patient groups with a view to providing appropriate language for informed consents. Technology is helping headache in many ways, and as we embrace its benefits we should reflect on its challenges.

Clinical implications

Occipital nerve stimulation studies are bedeviled by blinding issues.

Blinding in neuromodulation studies will only become worse with the age of social media.