Wednesday, 1 August 2012

Placebo surgery #3: Meniere's disease

Meniere’s disease is a disorder of balance (dizziness) and hearing (ringing or hearing loss). The underlying cause is not understood (idiopathic), the symptoms often fluctuate, there is crossover with other symptoms (vertigo, migraine), and there are many other conditions that cause “Meniere’s-like” symptoms. This makes the condition ripe for any treatment to look good if we think it works, due to the subjective nature of the symptoms, symptomatology that is open to misinterpretation, and the lack of any objective pathological test. Therefore, for such conditions, reporting good results from a series of patients is not enough; a more scientific (unbiased) assessment is necessary. Fortunately, somebody did just that. Unfortunately, despite the results of that study, surgery (now in many different forms) is still being used for this condition.

Most operations aim to equalise or lower the pressure in the inner ear or middle ear, despite good evidence that ‘pressure’ is the problem in the first place. The landmark Danish study of 30 patients from the 80’s reported no difference in outcome in a surgical group versus a sham surgery group, but interestingly about 70% of both groups improved significantly in both groups. And everything got better, even the nausea and vomiting.

The authors got the idea of using a placebo because they had previously used lithium to treat the disorder, based on some biological mechanism that they made up, and noticed sustained improvement in 70% of patients. Later placebo studies showed that lithium and placebo both gave sustained improvement in 70% of patients. In fact, the authors noted that all treatments then available for Meniere’s disease (medical and surgical) had good results in 60-80% of patients. (Note that 60-80% is a very common rate of improvement in many studies of treatments that are later shown to be no better than placebo, and many treatments that have not been rigorously tested).

The results 3 years and 9 years later were unchanged: there was still good improvement in both groups, with similar rates of improvement between the active and placebo groups. The high rate of improvement (70%) is likely to explain the fact that surgery (in one form or another) is still commonly performed for Meniere’s disease. None of these new operations have been tested in a similar way, and despite the results of this previous sham trial, surgery will continue to be offered, as long as patients seek it, surgeons are allowed to do it, insurance companies pay for it, and of course, as long as we still have the placebo effect.

The other study from 1988 (also from Denmark) compared the same (active) operation to an alternative operation (not a placebo). Largely, there was no difference, except a few people in the alternative group lost their hearing. They would have been better off with a placebo.

Note: the Cochrane review merely summarises the two trials.

Ethical note: the Danish study recruited patients without telling them that one of the groups was a placebo. The ethics of this would need another article to discuss, but this would explain the high improvement rates in both groups, as other studies have shown less improvement than expected when patients think they might be getting a placebo (and even less when they think they got the placebo, regardless of whether or not they did).

1 comment:

  1. Part of the issue for research in Meniere's is the strong natural history trend toward remission. A big part of that 70% is probably regression to the mean and natural history, more than any "specific" neurophysiological placebo effect (as elucidated by Benedetti etc).

    Lithium would be a great placebo because you're going to get side effects from it. One of the best rules of placebo research is that "active" placebos (of which surgery is maybe the best example) are more effective than placebos that are easier to forget about while you're taking them. Irving Kirsch has argued that SSRIs are active placebos, and that most of the benefit from them above placebo comparators in RCTs is from unblinding (the patients who do better are the ones who think they're on active treatment, the ones who guess they're on placebo don't improve even if they're in the SSRI group after all, but the SSRI group has a higher percentage who correctly guess they're on the active drug, possibly due to anticholinergic and other side effects, so the stats show a benefit for SSRIs).