From the Editor

It’s effective but is it really safe?

Electroconvulsive therapy (ECT) remains controversial 80 years after its first clinical use. At the heart of the controversy: its effects on cognition. Some wonder about the possibility that ECT could be linked to dementia.

Despite the strong concerns expressed over the years, relatively little research has been done on the possible connection between ECT and dementia. This week, we consider a new paper by the University of Copenhagen’s Merete Osler and her co-authors. In this Lancet Psychiatry study, they tap Danish national databases, finding no connection.

old-man-in-sorrow-vincent-van-goghvan Gogh’s Old Man in Sorrow – in need of ECT?

In this Reading, we look at the paper and consider some recent work on ECT.



“Electroconvulsive therapy and risk of dementia in patients with affective disorders: a cohort study”

Merete Osler, Maarten Pieter Rozing, Gunhild Tidemann Christensen, Kragh Andersen, Martin Balslev Jørgensen

The Lancet Psychiatry, 6 March 2018

Electroconvulsive therapy (ECT) is an effective treatment method for severe episodes of mood disorders, with major depression being the most important indication. Although ECT has beneficial acute effects such as reduction of psychotic and mood disorder symptoms, memory loss is a common side effect that seems to be especially associated with bilateral lead placement and increasing number of treatments. Although most cognitive deficits resolve within weeks after treatment, any potential long-term adverse cognitive outcomes are less clear and subject to debate. Only a few studies have examined the frequency of dementia in patients after ECT. In one study of 81 Australian patients with major depression treated with ECT and followed up over 5 years, a greater number developed dementia compared with the general population (11 [14%; 95% CI 7–23]). In patients aged 75 years and older, the prevalence of dementia was three times higher than that in this age group in the general population (36% vs 11%). Another study from Sweden showed that 16 (34%; [95% CI 21–49]) of 47 patients with depression who underwent ECT were diagnosed with dementia about 10 years after ECT. However, both studies were hampered by small sample sizes and did not contain a reference group of patients with depression who did not receive ECT. By contrast, findings from a pilot study of 12 patients (mean age 59 years [SE 21·9]) with major depression suggested that ECT might reduce the risk of developing Alzheimer’s disease by increasing the mobilisation of amyloid β in the cerebrospinal fluid. Thus, at present, data for the long-term effects of ECT on patients’ cognition and risk of dementia are conflicting.

Against this background, the aim of our study was to examine the association between ECT and risk of subsequent dementia in a cohort of patients with a first-time hospital diagnosis of affective disorder.

merete_osler_jubibogMerete Osler

So begins a new paper by Osler et al.

Here’s what they did:

  • Drawing on national databases, they “did a cohort study of all citizens in Denmark aged 10 years and older with a first-time hospital contact for an affective disorder from Jan 1, 2005, through Dec 31, 2015.”
  • The dataset included information on ECT (the use of the treatment, as well as electrode placement).
  • They considered whether people were diagnosed with dementia (again, drawing from a national database); they also considered if people were prescribed cognitive enhancers.
  • Different statistical analyses were done, including a Cox proportional hazard regression (of first ECT and incidental dementia). Additionally: “we did a propensity-score calibration, where the propensity score was estimated by use of Cox regression as the relative hazard of exposure (to ECT) conditional on the baseline covariates. With this score, we did propensity-score matching with a one-to-one match on the nearestneighbour. In Cox proportional hazard regression models we subsequently estimated the hazardof dementia outcomes in the matched sample with a robust variance estimator to account for the matched nature of the sample.”

Here’s what they found:

  • “Of the 168 015 patients (mean age 47∙1 years [SD 21·9]) included in the analyses, 5901 (3∙5%) were treated with ECT…”
  • “The number of patients who developed dementia was 111 (0∙1%) of 99 045 patients aged 10–49 years, 965 (2∙7%) of 35 945 aged 50–69 years, and 4128 (12∙5%) of 33 025 aged 70–108 years…”
  • “The unadjusted incidence of dementia was higher in all patients treated with ECT than those not treated with ECT (incidence 70∙4 per 10 000 person years [95% CI 61∙6–80∙5]vs 59∙2 per 10 000 person years [57∙6–60∙8]).” See Figure 2 below, which considers incidence at different ages.
  • “However, in the propensity score-matched sample, the HR for those given ECT was attenuated and not significantly different from that for patients who were not given ECT…”
  • Did more ECT sessions matter? Roughly half of the patients had more than 10 sessions of ECT, yet it didn’t seem to matter, either with the original sample or the propensity-score matchedsample. (This was true across different age groups, including younger patients.)


In this register-based cohort study of patients in Denmark with affective disorders, we showed that ECT was not associated with risk of incidental dementia after correcting for the potential effect of patient selection or competing mortality.

A few thoughts:

  1. This a good study; it draws nicely from a big sample, and the methodology is thoughtful.
  1. This is an important study, enhancing our knowledge of the effects of ECT.
  1. What to make of past work in the area? There are only two previous studies,one from Sweden and one from Australia – and both relatively small numbers, with an of 47 and 81, respectively. Both found a connection between ECT and dementia. The authors comment that these past studies had not properly adjusted for the fact that depression itself is a risk factor for dementia.
  1. The authors drew on national databases, giving them a huge amount of data to analyze. But there is a clear limitation here: cofounding factors are overlooked. Though the authors had access to information on symptom severity and health, they didn’t know about social support, as an example – something that could impact the decision to receive ECT. The paper notes: “If, for example, older people with little social contact and support, which is associated with an increased risk of dementia,do not receive ECT for these social reasons, it might yield a spurious negative association between ECT and dementia, which was not accounted for by the variables included in the present study.” Another limitation to note: though the follow-up period was long, it wasn’t 20 or 30 years.
  1. ECT remains a controversial intervention. Over the years, so many of my patientshave explained that they would rather struggle with their treatment-refractory depression than receive ECT. The use of ECT has markedly declined – down a third in Quebec over a 15-year period. (That study was considered in a past Reading.)  Yet the literature points in a similar direction. ECT is unusually safe for an outpatient procedure. (The safety study was considered in a past Reading.) And the paper considered in this week’s Reading suggests that there is no link to long-term cognitive decline.  Is ECT an underappreciated treatment? What can be done to make it more accessible? Here are links to past Readings on ECT: and
  1. This Reading opens with the image of a van Gogh painting. For those interested in the famed artist, a new book explores his time in a French asylum. The Guardian has an interesting article on it, which can be found here:


Reading of the Week. Every week I pick articles and papers from the world of Psychiatry.