Abstract
The paradigm of radiofrequency medial branch neurotomy does not allow for partial responses. If a patient's pain arises from a zygapophysial joint of the neck, anaesthetizing that joint should provide complete relief of their pain. If the pain arises from more than one joint, anaesthetizing only one of those joints will provide partial or no relief, but anaesthetizing all of the symptomatic joints should provide complete relief of pain.
Alternatives are not feasible. If blocking a particular joint provides only partial relief, some other source must be responsible for the remnant pain, or the source of pain does not at all lie in the targeted joint. Under those conditions, treating the joint cannot be expected to provide relief of pain. The diagnostic blocks indicate and predict this. No-one has provided grounds for expecting that denervating a joint surgically will be provide complete relief of pain when diagnostic blocks of that joint fail to do so. Indeed, to practice in this manner is antithetical to responsible medical practice. Treatment is based on having made a valid diagnosis.
It is peculiar therefore that in their study, Stovner et al. (1) expected to be able to relieve cervicogenic headache. The clinical criteria that they used have not been validated against the criterion standard of complete relief of pain. Moreover, even though response to blocks was not used as a criterion in their study, blocks were nevertheless performed. No patient obtained complete relief. Therefore, no patient could be expected to obtain relief from radiofrequency denervation. Yet they were enrolled in a controlled trial. Under those conditions, the trial was destined to show no efficacy. Therefore, the study of Stovner et al. cannot be taken as an indictment of zygapophysial joint blocks or of medial branch neurotomy.
Other studies have shown what can be achieved if patients are carefully selected, using objective criteria under controlled conditions, and if meticulous surgical technique is used. At typical cervical levels, radiofrequency neurotomy achieves complete relief of pain, provided that patients have had complete relief of pain following diagnostic blocks (2). Those blocks can be placebo-controlled or comparative local anaesthetic blocks (3, 4). Similarly, for patients with headache mediated by the third occipital nerve, complete relief of pain can be achieved if meticulous and demanding techniques are followed (5).
It is not evident what the study of Stovner et al. (1) shows, but three possibilities arise. One is that clinical diagnostic criteria have no predictive validity. They do not select patients who will respond to radiofrequency neurotomy. Consequently, the use of clinical diagnosis should be abandoned.
A second interpretation, using the nested data on diagnostic blocks, is that patients who do not respond to blocks do not respond to medial branch neurotomy. This result is obvious and predictable. What would have been disturbing is if patients had responded to surgery despite negative blocks, but that was not the case.
Both of these interpretations might be inappropriate if a third applies: that what Stovner et al. (1) performed was not a technically correct operation. If clinical diagnosis is tested against a criterion standard that is itself flawed, predictive validity will not emerge. In this regard, Stovner et al. describe, but do not illustrate their technique. Therefore, readers cannot be certain that their technique was accurate, or was comparable to the way others have performed medial branch neurotomy in order to obtain complete relief of pain. What is suspicious in this regard is that Stovner et al. claim that they performed three to four lesions at each segmental level, each for 60 s, at each level from C2 to C6, yet were able to complete the entire operation in 90 min. This is at odds with how others perform medial branch neurotomy.
When electrodes are carefully placed, and each placement confirmed radiographically with lateral and AP views, denervating just one joint can take two hours (2); a third occipital neurotomy alone takes 90 min (5). Unless Stovner et al. are somehow five times more efficient in their execution of technique, readers have call to question the accuracy of their technique.
A danger arises that the study of Stovner et al. (1) might be misused. A climate obtains in which insurers, in particular, are seeking grounds not to pay for radiofrequency neurotomy. As presented, the study of Stovner et al. could serve this purpose. If it is to be used, however, the application is limited.
It seems that in Europe, practitioners have been taught, and are convinced, that they can diagnose cervicogenic headache clinically, and can select patients for radiofrequency neurotomy on this basis. It is this practice that Stovner et al. have tested. Their data repudiate this European practice. It is this practice that insurers should not pay for.
However, European practice is not the same as other practice, elsewhere in world; nor is it what the International Spinal Injection Society recommends in its practice guidelines (6). Those practices have been vindicated, and should be dissociated from the practices described by Stovner et al. (1). In essence, all that Stovner et al. have shown is that radiofrequency neurotomy fails when it is practised inappropriately and incorrectly. That is not evidence against its efficacy when it is practiced correctly.
The risk obtains that, having been published, the study of Stovner et al. will be abused to impugn radiofrequency neurotomy. To responsible practitioners, and to deserving patients, this study is a disservice. Its only virtue is that it condemns how radiofrequency has been abused in Europe.