On the week I am writing this post auto-scheduled for later publication, I am observing what so often goes on in news reporting about science: a host of science news sites mindlessly parroting the unjustified claims of a university press release. You would think that by now our science journalists might have learned the lesson that university press releases these days are often notorious for their hype, errors and misinformation, and that the claims of such press releases should be subjected to critical scrutiny.
The unfounded claim this week is one that at-home magnetic brain stimulation "shows promise" for treating depression. The claims come from a Kings College press release here, which is entitled "At home brain stimulation for depression found to be safe and effective." We have a discussion of a new study testing at-home head gadgets using something called transcranial direct current stimulation (tDCS),
The press release states this:
"Transcranial direct current stimulation (tDCS) is a form of self-administered, non-invasive brain stimulation that applies a weak, direct current of between 0.5 to 2 milliampere to the scalp via two electrodes. It is not electroconvulsive therapy (ECT), which delivers about 800 milliamperes to the brain causing a generalised seizure and can only be conducted under strict supervision.
174 participants aged 18 and over, and with a diagnosis of at least moderate depression were randomly assigned to one of two treatment arms; 'active' tDCS or 'inactive' tDCS which was the same device but did not provide a current. Participants had a 10-week course of treatment, initially having five 30-minute sessions a week for the first three weeks, followed by three 30-minute sessions a week for the following seven weeks.
Researchers found that participants in the active arm of the trial showed significant improvements in the severity of their depression, as well as the overall clinical response and remission compared to those in the ‘inactive’ placebo control arm. The rates of treatment response and remission were three times higher in the active treatment arm compared to the placebo arm, where 44.9% in the active arm demonstrated a remission rate compared to 21.8% of the control group."
A study like this attempts to follow the conventions of randomized controlled trials (RCT) using placebos. In such studies patients are randomly assigned to either a medication group or a placebo group. Typically the medication group gets some medicine being tested, and the placebo group gets only a similar-looking placebo pill that is something like a sugar pill that looks just like the real medicine. The patient does not know which group he belongs to, and does not know whether the pill he is being given is just a placebo sugar pill. Such a lack of knowledge is called an example of blinding. Such a randomized control trial attempts to see some medical outcome in the group getting the real medication that is superior to the medical outcome reported in the placebo group. When both a doctor or scientist and the patient are unaware (for at least part of the time) whether particular patients belong to the get-the-real-treatment group or the placebo control group, such a study is called a "double-blind" study.
The press release quoted above refers to the study here, entitled "Home-based transcranial direct current stimulation treatment for major depressive disorder: a fully remote phase 2 randomized sham-controlled trial." The study claims to have been a double-blind study, saying that it was a "fully remote, multisite, double-blind, placebo-controlled, randomized superiority trial." But the claim about being double-blind is not correct. The patients who were supposed to be blind as to whether or not they were getting the real magnetic stimulation were not really blind about such a thing.
The failure of the blinding protocol to achieve real blindness (a lack of knowledge by subjects about whether they were receiving the real treatment) is shown by the section of the paper entitled "Analysis of study blinding and unblinding." There we read this:
"Before unblinding at week 10 (end of trial), participants were asked to guess whether they thought they were receiving the active or sham tDCS device and their level of certainty, rating from ‘1’ for ‘very uncertain’ to ‘5’ for to ‘very certain’. A guess of active tDCS was made by 77.6% in the active treatment arm and 59.3% in the sham treatment arm; the difference was significant (P = 0.01)."
This is a very big difference. 77% of the people getting the real treatment thought they were getting the real treatment, but only 59% of the people in the "no treatment" placebo group thought they were getting the real treatment. This indicates a very major effect in which the people getting the real treatment were much more likely to think that they were getting the real treatment than those not getting the real treatment. In a randomized control trial that achieves an effective level of blinding, there will be no difference between the percentage of real-treatment subjects and placebo or sham treatment subjects who thought they were getting the real treatment.
Analyzing the paper, it's easy to see why patients who were not getting the real treatment would have been been better able to figure out whether they were getting the real treatment. We read this:
"Active stimulation consisted of 2 mA direct current stimulation for 30 min with gradual ramp up over 120 s at the start and ramp down over 15 s at end of the session. Sham stimulation with the same device and app was used to resemble the active intervention and to receive the treatment schedule. An initial ramp up from 0 to 1 mA over 30 s then ramp down to 0 mA over 15 s was repeated at the end of the session to cause a tingling sensation that mimics active stimulation."


