Monday, May 04, 2015

Shock Therapy: Barely Better Than Placebo

Electroconvulsive therapy (ECT) has been an accepted treatment for schizophrenia and refractory depression for over 50 years, and it is widely promoted as a highly effective technique. Nevertheless, practitioners of the technique vastly oversell its benefits while egregiously understating its dangers, something I talk about at length, with references to the scientific data, in my book Hack Your Depression. (Note that the references listed throughout this post have been summarized in that book as well as online here.)

ECT patients must be assisted in breathing since the muscle
relaxants given prior to the procedure tend to paralyze the lungs.
The medical profession prides itself on utilizing evidence-based medicine, but most of the evidence on ECT is quite ancient and does not adhere to modern requirements for double-blind technique, control populations, blind assessments, sham treatment in control groups, etc., and followup periods are generally short (seldom more than a month). What's more, attempts to assess the safety of ECT have been, until recently, shoddy and haphazard. Consent forms regularly lie about the actual risks involved. Consequently, no patient, arguably, has ever given "informed consent" for the procedure, since the information provided in consent forms is obsolete and factually incorrect.

A proper critique of ECT begins with a look at its scientific standing: Never mind the fact that its mode of action in not understood. Does it work? That's what this post talks about. In subsequent posts, we'll look at the question of whether ECT prevents suicide, and whether it actually harms people, not just in terms of short- and long-term cognitive effects but in terms of death.

We start by noting that modern evidence-based medicine relies on the use of placebo-controlled, double-blind studies. The literature on ECT goes back 75 years or more. But the overwehlming majority of studies conducted on ECT in that time have failed to include placebo groups. Most of the studies that claim a high response rate for ECT patients lack a control arm and thus do not conform to modern standards of evidence-based medicine. Also, "response rate" data are not long-term. They focus (almost always) on the treatment period itself, which last up to 30 days.

Until the 1950s, when general anesthesia was introduced for ECT, it was not possible to include a control arm in ECT studies, since there is no way to trick a fully-awake patient into thinking he’s been shocked (and has had a grand mal seizure) when he hasn’t. Only with the advent of anesthesia has it been possible to use sham treatment consisting of anesthetizing the patient and administering all other steps of the procedure except for shock. (The sham-ECT patient is then told, when he or she wakes up, that shock was applied, when it wasn’t.)

The sham ECT technique, as currently implemented, can hardly be considered a faithful “placebo,” however, since it fails to replicate other aspects of the bonafide treatment, such as anterograde and retrograde amnesia, headache, fatigue, and muscle ache. A valid comparison with ECT would require a placebo that induces injury, which is obviously unethical. But because sham ECT produces none of the typical cognitive or somatic effects of “the real thing,” patients are very likely to guess correctly whether they’ve been shocked or not. Hence, blind breakage is a serious concern even in properly designed studies, and can be expected to result in lower sham ECT response rates than might otherwise be observed. It's worth bearing that in mind as we proceed.

Nine meta-analyses have compared ECT and sham ECT for depression (Gabor & Laszlo, 2005; Greenhalgh et al., 2005; Janicak et al., 1985: Kho et al., 2003; Pagnin et al., 2004; Ross, 2006; Tharyan & Adams, 2005; UK ECT Review Group, 2003; van der Wurff et al., 2003; see the complete list of references here). All except one (Ross, 2006) make the claim that ECT is superior to sham ECT during the treatment period (usually one month), but none found evidence of any difference in outcome beyond the treatment period. 

Claims for the superiority of ECT versus placebo are based on effects seen during the treatment period (one month); after the first month, there are no differences between ECT and sham ECT groups. As one author (Ross, 2006) said: “Claims in textbooks and review articles that ECT is effective are not consistent with the published data.” 

Of the six studies considered by Janicak et al. (1985), only two were said to have produced significant differences between ECT and sham ECT. One was a crossover-design study that gave ECT to the sham group during the treatment period, confounding the results. Another was a very old study (Ulett et al., 1956) in which ECT and “photoshock” results (from a flashing-light technique, combined with hexazol) were conflated. Janicak’s review was found by Ross (2006) to contain numerous serious factual errors, calling into question its usefulness. 

The UK ECT Review Group (2003) meta-analysis reported that only one study met their inclusion criteria for follow-up studies, and that study found no significant difference. But the study in question had not, in fact, reported any followup data.

The 170-page report by Greenhalgh et al. (2005) concluded “there is little evidence of the long-term efficacy of ECT” and that, even in the short-term, “low-dose unilateral ECT is no more effective than sham ECT.” The review found “no randomised evidence of the effectiveness of ECT in specific subgroups, including older people, children and adolescents, people with catatonia and women with postpartum exacerbations of depression or schizophrenia.”

One of the nine meta-analyses mentioned above, a Cochrane Systematic Review, focused specifically on the effectiveness of ECT for the “depressed elderly” (van der Wurff et al., 2003). This review found no evidence of ECT being effective beyond the treatment period. It identified only one study comparing ECT and sham ECT (O’Leary et al., 1994). The study in question was a re-analysis of data from a paper by three of the reviewers (Gregory et al., 1985), which the reviewers described as having “major methodological shortcomings.” The conclusion was that “None of the objectives of this review could be adequately tested because of the lack of firm, randomised evidence.”

In addition to meta-analyses of ECT versus sham ECT for depression, there have been meta-analyses along these lines for schizophrenia. 

A 2001 report by the American Psychiatric Association admits that none of five pre-1980 ECT versus sham ECT studies found any differences in outcomes, even in the short term, but claimed that three more recent studies demonstrated “a substantial advantage” for ECT (Abraham & Kulhara, 1987; Brandon et al. 1985; Taylor & Fleminger, 1980). In all three studies, however, both patient groups were receiving antipsychotic medications, and any advantage for ECT was, in fact, short-lived. In the Leicester ECT Trial (Brandon et al. 1985), for example, both the ECT and the sham ECT groups improved on all four measures used. “Global psychopathology” did not differ at all. However, the authors note: “The superiority of real ECT was not demonstrated at the 12- and 28-week follow-up.” Another of these studies (Taylor & Fleminger, 1980) found that during treatment there was equal improvement in both groups, although ECT improved general psychopathology faster than sham ECT for the first four weeks (only). In the Taylor & Fleminger (1980) study, the psychiatrists (unblinded) were the only ones to perceive any difference in the groups. Nurses and relatives did not see a difference. This might be because psychiatrists are better trained to rate symptoms, but it is also likely reflective of researcher bias, since most ECT studies are done by researchers who strongly advocate the use of ECT and believe in its benefits. 

A more recent study (Sarita et al., 1998) of 36 people diagnosed with schizophrenia found no differences in double-blind ratings on four symptom measures after one, two, three, or four weeks of treatment, in a comparison of sham ECT with either bilateral or unilateral ECT. 

A 2002 study (Ukpong et al., 2002) likewise failed to find significant differences between ECT and sham ECT at the end of treatment. It also failed to find differences 20 weeks later. 

A 2005 update of the Cochrane database (Tharyan & Adams, 2005) found a short-term advantage for ECT over sham ECT, but “no evidence that this early advantage for ECT is maintained over the medium to long term.” The Cochrane reviewers found that even in the short term, ECT was less effective than antipsychotic medication. 

More recently, Poublon & Haagh (2011) published a systematic review of ECT for schizophrenia, drawing only on randomized, controlled studies that included a sham ECT arm. They found six such studies. In four of the six, ECT produced statistically significant improvement relative to sham ECT, during the treatment period, but in followup, no significant differences remained. The authors said: “No evidence was found proving the superiority of ECT over sham-ECT after followup.”

On the whole, ECT effectiveness is not far different from sham ECT (and no different, after the treatment period), raising the possibility that it is the anesthesia, rather than the electrical shock, that produces much of the therapeutic effect. Tending to confirm this possibility is the fact that ketamine, a general anesthetic, has recently been found to be therapeutic, on its own, for treatment of severe depression (McGirr et al., 2014). If a study were to compare anesthetics used in ECT with respect to the effect on depression outcomes, analysis of variance might reveal the extent to which choice of anesthetic influences outcomes, and this would tend to establish the degree to which therapeutic effect is due strictly to anesthesia. Unfortunately, no such studies have yet been done.

Colin A. Ross, M.D., of the Institute for Psychological Trauma in Richardson, Texas) has summed the situation up well in his 2006 paper:
The author reviewed the placebo-controlled literature on electroconvulsive therapy (ECT) for depression. No study demonstrated a significant difference between real and placebo (sham) ECT at 1 month post-treatment. Many studies failed to find a difference between real and sham ECT even during the period of treatment. Claims in textbooks and review articles that ECT is effective are not consistent with the published data. A large, properly designed study of real versus sham ECT should be undertaken. In the absence of such a study, consent forms for ECT should include statements that there is no controlled evidence demonstrating any benefit from ECT at 1 month post-treatment. Consent forms should also state that real ECT is only marginally more effective than placebo.
Today's post is an excerpt from Hack Your Depression. Full references to the studies mentioned above are given in that book as well as here. Please buy a copy for anyone you know who may be suffering from depression. More information (including free excerpts from the book) can be found at

☙ ❧ 
I want to thank the following great folks who retweeted me yesterday on Twitter. My advice? Follow these guys. They retweet!

Have you added your name to our mailing list?

Also please visit when you have a chance, and share that link with someone you know who might be suffering from anxiety or depression.