TL;DR: A 2026 review in The British Journal of Psychiatry found no consistent long-term dementia or major cardiovascular harm signal after ECT in modern administrative studies.
Key Findings
- Suicide mortality dropped 30–50%: Four 2025 systematic reviews and multiple population studies found large reductions in suicide death after ECT — adjusted hazard ratios of 0.56, 0.58, and 0.39 across cohorts.
- All-cause mortality fell 25–30%: Remarkably consistent across reviews and multiple-country cohorts. Hazard ratios of 0.54, 0.43, 0.75, 0.74, 0.81 by design and population.
- No consistent long-term dementia pattern: Modern adjusted cohort analyses do not show ECT increasing long-term dementia risk once confounding is handled properly.
- No long-run cardiovascular increase: Major adverse cardiovascular and cerebrovascular events did not clearly rise. One Danish analysis flagged a short 30-day cardiac risk in medically fragile patients — not a long-term harm.
- Acute efficacy was never the issue: ECT remission can reach ~60% in severe, psychotic, or treatment-resistant depression. The long-term safety issue was dementia, cardiovascular events, suicide, and all-cause mortality after treatment.
- Modern causal methods, not new RCTs: Propensity-score matching and weighting on large administrative databases are doing what trials cannot — assess rare and long-term outcomes at population scale.
Source: The British Journal of Psychiatry (2026) | Kaster et al.
Electroconvulsive therapy has a large gap between public reputation and modern outcome data. It is often imagined as a treatment whose risks are obvious and benefits anecdotal.
This review argues that the best modern studies point the opposite direction — not by overturning stigma rhetorically, but by handling confounding more honestly than older work did.
Long-Term ECT Safety Data Needed Better Confounding Control
Psychiatry has not lacked evidence that electroconvulsive therapy works acutely. Trials and meta-analyses have shown for years that ECT can outperform medication for severe or treatment-resistant depression, with remission rates up to 60% in some populations.
The patient-facing concern is concrete: does ECT increase later dementia risk, or do severe depression and baseline health explain much of the apparent association?
Does it provoke major cardiac or cerebrovascular events? And if it rapidly reduces suicidal thinking, does that translate into fewer suicide deaths over years — not just better mood ratings on a discharge form?
Traditional clinical trials are poorly suited to rare long-term harms. Dementia, cardiovascular events, suicide, and mortality require large samples with long follow-up, which is where administrative health data become useful.
That methodological pivot is what this review is built around.
Modern ECT Cohort Methods Reduced Dementia Confounding
The biggest bias in ECT research has always cut against the treatment. Patients who receive ECT are among the sickest in psychiatry — more suicidal, more treatment-resistant, more medically complicated.
Compare them naively against people who did not receive ECT, and the treatment looks dangerous because the patients were already at higher baseline risk.
The review’s core point is that once modern studies address that confounding more seriously — propensity-score matching, weighting, and careful regression adjustment — the result becomes much less alarming than ECT’s public image suggests. The stigma was partly built on confounded comparisons, not on the treatment’s actual long-term effects.
Adjusted ECT Data Did Not Show Consistent Dementia or Cardiac Harm
The dementia section is one of the most clinically important — cognitive side effects are the dominant fear most patients bring into the room. The researchers argue that the best adjusted studies do not show a consistent increase in long-term dementia risk after ECT.
The same general conclusion holds for major adverse cardiovascular and cerebrovascular outcomes. Across population-level studies, long-term stroke and major cardiovascular event risk did not clearly increase once confounding was handled.
One Danish analysis identified a short-term cardiac finding within 30 days in patients with physical comorbidity — but that risk did not persist long-term.
The nuance changes counseling. “No major long-run pattern” is not the same as “zero medical risk.” The review is careful to separate three claims that usually get blurred:
- Dementia: modern adjusted studies do not show a consistent long-term increase.
- Cardiovascular risk: major long-run events do not clearly rise overall; short-term risk can matter in medically fragile patients.
- Procedure burden: anesthesia, blood-pressure shifts, and transient physiological stress still require careful monitoring.
ECT Suicide Mortality Data Were the Most Consequential
The strongest synthesis concerned suicide mortality. Four systematic reviews published in 2025 found that ECT was associated with roughly a 30–50% reduction in suicide death.
The individual studies pointed the same way: adjusted hazard ratios of 0.56 in an older inpatient mood-disorder sample, 0.58 in another matched cohort, 0.39 in a Canadian propensity-weighted analysis. Not every study aligned perfectly, but after confounding adjustment, almost none found a significant increase.
that endpoint changes interpretation more than temporary symptom improvement. ECT is often reserved for exactly the patients most likely to die by suicide.
If those patients end up with lower suicide mortality after treatment, it is not a side note — it is the outcome that most directly tests the claim that ECT can be life-saving.
All-Cause Mortality Often Looked Lower After ECT in Administrative Data
Cause-specific mortality can be messy. Suicide coding varies, competing-risk problems complicate interpretation.
That is one reason the review leans on all-cause mortality — harder to misclassify, captures both benefit and harm in one metric.
Here again, the evidence runs against stigma. The review describes a consistent 25–30% reduction in all-cause mortality across systematic reviews, with population-level studies from Taiwan, the United States, Canada, Denmark, Sweden, and the UK pointing the same direction.
Hazard ratios across cohorts: 0.54, 0.43, 0.75, 0.74, 0.81.
No one should read that as proof ECT directly improves every aspect of physical health. The more plausible interpretation: a treatment that rapidly reduces severe symptom burden, suicidality, catatonia, or profound psychotic depression can lower overall mortality in a population that starts extremely unwell.
Severe untreated psychiatric illness carries real medical risk.
ECT Shared Decision-Making Should Separate Dementia Risk From Depression Severity
The review does not ask clinicians to stop discussing side effects. Memory problems, post-treatment confusion, anesthesia risks, and the need for careful monitoring all remain real.
It asks that those risks be framed against modern evidence rather than against older assumptions about ECT risk.
A more balanced ECT conversation would sound like this: short-term cognitive side effects are common and important, but current long-term population data do not support a large dementia pattern; serious long-run cardiovascular harms are not strongly emerging; and the treatment may actually lower both suicide death and all-cause mortality in the patients who receive it.
For psychiatry, the adjusted mortality result changes the risk-benefit discussion. The debate is no longer just whether ECT works acutely.
It is whether dementia fears are being weighed against data showing lower long-term mortality in treated patients.
Citation: DOI: 10.1192/bjp.2026.10613; Kaster et al; Electroconvulsive therapy: improved understanding of long-term risks and benefits from advances in administrative health data; The British Journal of Psychiatry; 2026.
Study Design: Synthesis of modern observational studies using large administrative databases with propensity-score matching, weighting, and regression adjustment.
Sample: Multi-country cohorts — Taiwan, US, Canada, Denmark, Sweden, UK — plus four 2025 systematic reviews.
Key Statistic: Suicide mortality reduction 30–50%; all-cause mortality reduction 25–30%; no consistent long-term dementia or major cardiovascular pattern after confounding adjustment.
Caveat: Observational data — even with modern causal methods, residual confounding cannot be fully ruled out.
