Note: Claims that ECT is safe and effective are common in electroconvulsive therapy study introductions. These are assertions without a scientific basis. See Electroconvulsive Therapy for Depression: A Review of the Quality of ECT versus Sham ECT Trials and Meta-Analyses and Thymatron System IV Manual page 7, section 1
L.LindbladaA.NordenskjöldbA.OtterbeckcA.M.NordenskjölddaSchool of Medical Sciences, Örebro University, Örebro, SwedenbUniversity Health Care Research Center, Faculty of Medicine and Health, Örebro University, SwedencAnesthesiology and Intensive Care, Faculty of Medicine and Health, Örebro University, Örebro, SwedendDepartment of Cardiology, Faculty of Medicine and Health, Örebro University, Örebro, Sweden
Received 7 April 2022, Revised 28 September 2022, Accepted 3 October 2022, Available online 6 October 2022, Version of Record 11 October 2022.
See Also
- Acute and long-term clinical, neuropsychological and return-to-work sequelae following electrical injury: a retrospective cohort study
- Device manufacturer rep acknowledges ECT Heart Attack “Common.”
- Electrical Injuries: Communication and Speech disorders and AAC– What’s the next step?
- Electroconvulsive therapy in a patient with hypertrophic cardiomyopathy: A case report
- Microstructural Damages Caused by Electroconvulsive Therapy (ECT)
- Sudden Cardiac Death After ECT
Highlights
- Within 30 days of ECT, there were 123 (0.61 %) deaths due to medical causes.
- ECT appears to be a low-risk medical procedure.
- Cardiovascular disease was the leading medical cause of death.
- Older individuals with severe somatic diseases had the highest risk of death.
Abstract
Background
Electroconvulsive therapy (ECT) is used to treat severe psychiatric disorders and is associated with reduced risk of suicide and all-cause mortality in patients with severe depression. We investigated the causes of death occurring shortly after ECT and identified potential risk factors for medical causes of death.
Methods
Patients treated with ECT between 2012 and 2018 were included in this Swedish register-based study. Multivariate binary logistic regression was used to calculate odds ratios for covariates to determine potential predictors of 30-day mortality.
Results
Of the 20,225 included patients, 93 (0.46 %) died of suicide and 123 (0.61 %) died of medical causes after ECT. Cardiovascular disease was the most common medical cause of death (n = 49, 40 %). An older age, a Charlson Comorbidity Index of 1 or more, atrial fibrillation, kidney disease, reflux disease, dementia, and cancer were associated with increased risk of death by medical causes.
Limitations
Real-life observational studies based on registry data may demonstrate associations, but cannot determine causality. If medical records had been available, we would be better able to determine if deaths were due to the ECT, anesthesia, pre-existing medical conditions, or the mental disorder.
Conclusions
ECT appears to be a low-risk medical procedure. Older individuals with severe somatic diseases have the highest risk of death and extra measures should be considered to optimize their medical health during the pre-ECT workup, and during and after ECT.
Keywords
- Electroconvulsive therapy
- Medical risk factors
- Death
1. Introduction
Electroconvulsive therapy (ECT) is used to treat severe psychiatric disorders, such as major depressive disorder, bipolar affective disorder, and schizophrenia (Nordanskog et al., 2015). Common side effects of ECT are headache, sore muscles, memory loss, and nausea (Andrade et al., 2016). More serious events associated with treatment are rare, with a reportedly low mortality rate from both medical and external causes such as suicide and accidents (Dennis et al., 2017; Østergaard et al., 2014; Shiwach et al., 2001; Watts et al., 2021; Rhee et al., 2021). Nonetheless, some patients die following ECT.
Blumberger et al. examined both comorbidities before ECT and causes of death within 30 days post-treatment, and concluded that the mortality rate is comparable with similar procedures with the use of general anesthesia; the authors also reported that a higher American Society of Anesthesiology score was associated with an increased risk for both serious events and mortality after ECT in older patients and those with a history of ischemic heart disease (Blumberger et al., 2017). Liang et al. compared comorbidities and deaths between one group who received ECT and one group who did not receive ECT, and found no predictor of in-hospital death (Liang et al., 2017).
Knowledge is still lacking regarding patient characteristics, pre-existing medical conditions prior to ECT, and mortality following ECT. To better understand the predictors of death following ECT, we require more details about patients who die. With the descriptive registers for the Swedish population, medical conditions and causes of death can be mapped. Given the heterogeneity of patients who receive ECT, the aim of this study was two-fold. First, we documented age, sex, and medical causes of death in patients who died within 30 days of ECT; second, we investigated potential risk factors of medical causes of death within 30 days of ECT.
2. Methods
2.1. Participants and study design
This was a register-based cohort study that used data obtained from Swedish registers. The reporting guidelines from “Strengthening the Reporting of Observational Studies in Epidemiology” (STROBE) (Checklists, 2021) were used. Nearly all patients who received ECT in Sweden between 2012 and 2018 were included in the study. The patients who died within 30 days after treatment were further examined. We used the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) to obtain information about the causes of deaths. The ICD-10 codes for the causes of death were then divided into groups named after the corresponding ICD-10 categories. Patients with an ICD-10 code for suicide were excluded from further investigation but were included in the all-cause mortality report.
2.2. Procedures
The primary outcomes were mortality and causes of death, as coded by the ICD-10. Variables of interest were age, sex, Charlson Comorbidity Index (Ludvigsson et al., 2021), and certain medical conditions that have been associated with anesthesia-related death (hernia, reflux disease, obstructive sleep apnea syndrome, valvular heart disease, cardiac arrythmia, atrial fibrillation, hypertension, and obesity (Cook et al., 2011; Robinson and Davidson, 2014)). The Charlson Comorbidity Index was used to measure the comorbidities in general, and the medical conditions included were myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, chronic obstructive pulmonary disease, other chronic pulmonary diseases, rheumatic disease, dementia, hemiplegia, diabetes with and without organ damage, chronic kidney disease, mild and moderate or severe liver disease, peptic ulcer disease, cancer, metastatic cancer, and HIV or AIDS. The age-adjusted Charlson Comorbidity Index (Charlson et al., 1994) was also used to account for differences in comorbidities due to age. The total number of ECT sessions was evaluated, and both index and maintenance sessions were included. Every patient was observed for one or more 30-day periods after each treatment session, or, if the day of treatment was unknown, during inpatient care when ECT was performed and 30 days after discharge.
2.3. Sources of data
Data were sourced from four different registers, including the Swedish National Patient Register (IPR) (Ludvigsson et al., 2011), the Swedish National Quality Register for ECT (Q-ECT) (Ahmad et al., 2021), the Swedish Causes of Death Register (Brooke et al., 2017), and the Swedish Cancer Register (Barlow et al., 2009). The Q-ECT was established nationally in 2011 and holds information about ECT sessions administered at 49 different hospitals in Sweden. The coverage ratio for the Q-ECT has been over 90 % since 2014. The IPR and Swedish Causes of Death Register are national registers for the Swedish population that are provided by the Swedish National Board of Health and Welfare. The IPR has a coverage ratio of >99 % and the register includes information about inpatient hospital admissions and outpatient care. Diagnoses set in primary care are not captured in the IPR. Data were combined from both the Q-ECT and IPR to identify all patients who had undergone ECT. If the patient was included in the Q-ECT, the date of death was calculated from the date of the last ECT session. For patients who were only registered in the IPR and did not have a recorded date for ECT between admission and discharge, days were counted from the date of last discharge until the date of death. Other characteristics captured by the IPR were age, sex, number of treatments, date of the last treatment, medical conditions used for the Charlson Comorbidity Index, and other medical conditions that could be associated with risk of death after anesthesia. Mortality data were retrieved from the Swedish Causes of Death Register and specific cancer diagnoses for the use of the Charlson Comorbidity Index were collected from the Swedish Cancer Register.
2.4. Statistical analysis
For patient characteristics, proportions or medians and interquartile ranges were calculated. Pearson chi-square tests and Fisher’s exact tests were used to compare the proportion of exposure variables between all patients who died of medical causes and patients who were alive. A two-sided p-value of <0.05 was defined as significant. We used binary logistic regression with both univariate and multivariate analysis to identify predictors of death within 30 days after ECT. Variables included in these analyses were different age groups and all medical conditions, including both the diagnoses included in the Charlson Comorbidity Index and the diagnoses that might be associated with anesthesia-related death and not included in the Charlson Comorbidity Index.
Results for all univariate and multivariate analyses are presented as odds ratios (ORs) with 95 % confidence intervals (CIs). Reference values were not having the disease, aged under 50 years, female sex, and a Charlson Comorbidity Index of <1. We performed two separate multivariate analyses for the sum of the Charlson Comorbidity Index, age, and sex, and continuous variables (observed days and number of ECT sessions). Statistical analysis and data management were performed using SAS, version 6.1 (SAS Institute) and SPSS Statistics 25 (IBM Corp., Armonk, USA).
2.5. Ethical considerations
This study was a part of a bigger project, “The outcome of treatment of severe affective disorders”. Every patient included in the Q-ECT accepted participation in the register and were free to withdraw their registration. The IPR and the Swedish Causes of Death Register are both mandatory for the Swedish population. The Regional Ethical Review Board in Uppsala, Sweden, registration no. 2021/03815, approved the study.
3. Results
A total number of 20,225 patients were treated with ECT during the study period. Patients aged younger than 10 years old (n = 2) or who had an uncertain date of death (n = 9) were excluded. Baseline characteristics of included patients are presented in Table 1. From 2012 to 2018, 216 patients died within 30 days after ECT; 93 died of suicide and 123 of medical causes. The all-cause mortality rate within 30 days was 1.07 % (Table 1). Medical causes of death of the mortality rate accounted for 0.61 % and suicide for 0.46 %.
Table 1. Baseline characteristics of patients who underwent electroconvulsive therapy between 2012 and 2018.
| Variable | Empty Cell | Dead by medical causes | Suicide | Alive | P-valuea | ||
|---|---|---|---|---|---|---|---|
| <1 day | 2–7 days | 8–30 days | Total | ≤30 days | ≤30 days | ||
| n = 23 | n = 27 | n = 73 | n = 123 | n = 93 | n = 20009 | ||
| n | n | n | n | n | n | ||
| Sex, n (%)b | 0.580 | ||||||
| Female | 13 (0.11) | 17 (0.14) | 40 (0.33) | 70 (0.58) | 38 (0.32) | 1,1949 (99.10) | |
| Male | 10 (0.12) | 10 (0.12) | 33 (0.40) | 53 (0.65) | 55 (0.67) | 8060 (98.68) | |
| Age, years n (%) | <0.001 | ||||||
| <50 | 3 (0.03) | 1 (0.01) | 2 (0.02) | 6 (0.06) | 43 (0.46) | 9226 (99.47) | |
| 50–59 | 3 (0.09) | 4 (0.12) | 6 (0.18) | 13 (0.38) | 16 (0.47) | 3379 (99.14) | |
| 60–69 | 2 (0.06) | 5 (0.15) | 11 (0.32) | 18 (0.53) | 18 (0.53) | 3365 (98.94) | |
| 70–79 | 9 (0.33) | 11 (0.40) | 17 (0.63) | 37 (1.36) | 12 (0.44) | 2669 (98.20) | |
| >80 | 6 (0.42) | 6 (0.42) | 37 (2.60) | 49 (3.44) | 4 (0.28) | 1370 (96.28) | |
| Numbers of ECT | |||||||
| Median (IQR) | 3 (1–9) | 5 (2–26) | 8 (3–18) | 7 (2–17) | 9 (5–17) | 9 (6–16) | |
| Number of observed days | |||||||
| Median (IQR) | 63 (45–214) | 62 (34–302) | 71 (44–216) | 62 (46–121) | 60 (45–133) | 62 (46–121) | |
| CCI, n (%) | <0.001 | ||||||
| 0 points | 10 (0.07) | 8 (0.05) | 25 (0.17) | 43 (0.29) | 68 (0.46) | 1,4816 (99.26) | |
| 1 point | 3 (0.13) | 9 (0.40) | 13 (0.57) | 25 (1.10) | 16 (0.70) | 2230 (98.19) | |
| 2 points | 2 (0.11) | 4 (0.21) | 16 (0.85) | 22 (1.16) | 5 (0.26) | 1862 (98.57) | |
| ≥3 points | 8 (0.70) | 6 (0.53) | 19 (1.67) | 33 (2.90) | 4 (0.35) | 1101 (96.75) | |
| Age-adjusted CCI, n (%) | <0.001 | ||||||
| 0 points | 1 (0.01) | 1 (0.01) | 1 (0.01) | 3 (0.04) | 37 (0.46) | 8003 (99.50) | |
| 1 point | 3 (0.09) | 0 | 3 (0.09) | 6 (0.18) | 20 (0.61) | 3261 (99.21) | |
| 2 points | 2 (0.07) | 6 (0.20) | 6 (0.20) | 14 (0.47) | 12 (0.40) | 2938 (99.12) | |
| ≥3 points | 17 (0.29) | 20 (0.34) | 63 (1.06) | 100 (1.67) | 24 (0.40) | 5807 (97.90) | |
| CCI diagnoses, n (%) | |||||||
| Myocardial infarction | 2 (0.36) | 3 (0.54) | 7 (1.26) | 12 (2.16) | 2 (0.36) | 541 (97.48) | <0.001 |
| Congestive heart failure | 5 (1.22) | 2 (0.49) | 6 (1.47) | 13 (3.18) | 1 (0.24) | 395 (96.58) | <0.001 |
| Peripheral vascular disease | 0 | 0 | 1 (0.57) | 1 (0.57) | 1 (0.57) | 173 (98.86) | <0.001 |
| Cerebrovascular disease | 2 (0.24) | 4 (0.49) | 10 (1.22) | 16 (1.94) | 4 (0.49) | 803 (97.57) | <0.001 |
| COPD | 3 (0.73) | 1 (0.24) | 3 (0.73) | 7 (1.70) | 2 (0.49) | 403 (97.82) | 0.013 |
| Other chronic pulmonary disease | 2 (0.23) | 3 (0.34) | 5 (0.57) | 10 (1.14) | 6 (0.68) | 862 (98.18) | 0.046 |
| Rheumatic disease | 0 | 4 (0.43) | 7 (0.85) | 11 (1.33) | 1 (0.12) | 816 (98.55) | 0.018 |
| Dementia | 3 (0.90) | 4 (1.20) | 9 (2.71) | 16 (4.82) | 0 | 316 (95.18) | <0.001 |
| Hemiplegia | 0 | 0 | 0 | 0 | 1 (0.70) | 142 (99.30) | 1.000 |
| Diabetes | 3 (0.23) | 4 (0.30) | 13 (0.98) | 20 (1.51) | 2 (0.15) | 1304 (98.34) | <0.001 |
| Diabetes, organ damage | 0 | 0 | 1 (0.52) | 1 (0.52) | 1 (0.52) | 191 (98.96) | 1.000 |
| Chronic kidney disease | 3 (0.93) | 4 (1.25) | 5 (1.56) | 12 (3.74) | 0 | 309 (96.26) | <0.001 |
| Mild liver disease | 1 (0.33) | 1 (0.33) | 0 | 2 (0.66) | 2 (0.66) | 298 (98.68) | 0.707 |
| Moderate/severe liver disease | 1 (2.63) | 0 | 0 | 1 (2.63) | 0 | 37 (97.37) | 0.207 |
| Peptic ulcer disease | 1 (0.34) | 0 | 3 (1.01) | 4 (1.35) | 1 (0.34) | 292 (98.32) | 0.108 |
| Cancer | 3 | 6 (0.34) | 22 (1.23) | 31 (1.73) | 7 (0.39) | 1750 (97.87) | <0.001 |
| Metastatic cancer | 0 | 0 | 0 | 0 | 0 | 2 (100.00) | 1.000 |
| HIV/AIDS | 0 | 0 | 2 (2.56) | 2 (2.56) | 0 | 76 (97.44) | 0.082 |
| Other medical conditions, n (%) | |||||||
| Hernia | 1 (0.38) | 1 (0.38) | 0 | 2 (0.75) | 0 | 264 (99.25) | 0.678 |
| Reflux disease | 1 (0.38) | 2 (0.75) | 3 (1.13) | 6 (2.26) | 0 | 259 (97.74) | 0.006 |
| OSAS | 0 | 0 | 0 | 0 | 0 | 290 (100.00) | 0.427 |
| Valvular heart disease | 3 (1.68) | 0 | 2 (1.12) | 5 (2.79) | 0 | 174 (97.21) | 0.005 |
| Cardiac arrhythmia | 1 (0.35) | 0 | 2 (0.70) | 3 (1.05) | 1 (0.35) | 283 (98.61) | 0.254 |
| Atrial fibrillation | 3 (0.85) | 4 (1.14) | 5 (1.42) | 12 (4.78) | 1 (0.40) | 338 (96.30) | <0.001 |
| Hypertension | 10 (0.32) | 12 (0.38) | 33 (1.04) | 55 (1.74) | 9 (0.28) | 3102 (98.13) | <0.001 |
| Obesity | 4 (0.57) | 0 | 1 (0.14) | 5 (0.71) | 1 (0.14) | 697 (99.15) | 0.620 |
Bold P-values are significant (P < 0.05). IQR: interquartile range; CCI: Charlson Comorbidity Index; COPD: chronic obstructive pulmonary disease; HIV: human immunodeficiency virus; AIDS: acquired immunodeficiency syndrome; OSAS: obstructive sleep apnoea syndrome.a
P-values are derived from Pearson chi-square tests and Fischer’s exact tests. Patients who died of medical causes and patients in the alive group are compared.b
Percentages were calculated as the proportions of the total numbers of patients in each row.
3.1. Number of ECT sessions and observed days
The number of observed days of each patient varied between 30 and 4372, with a median of 63 and an interquartile range of 46–121 for all patients. The sum of observed days was 2,614,396. The total number of treatments varied between 1 and 577, with a median of 9 and an interquartile range between 6 and 16. During the whole study period, a total number of 299,021 ECT sessions were given to 20,225 patients. In the multivariate analysis adjusted for observed days and number of ECT sessions, the association with death increased for every observed day (OR: 1.002, 95 % CI: 1.001–1.003), but the number of ECT sessions tended to decrease the odds of death (OR: 0.987, 95 % CI: 0.976–0.998).
3.2. Medical causes of death
The different causes of death were divided into groups based on the ICD system (Table 2). Cardiac arrest, myocardial infarction, and pulmonary embolism were the most common causes of death among patients who died of cardiovascular disease. A total of 15 patients died of cardiac arrest within 30 days, 4 patients died within 1 day, 4 patients died at 2–7 days, and 7 patients died at 7–30 days. Myocardial infarction as a cause of cardiovascular death affected 4 patients within 30 days, 2 patients died within 1 day and 2 patients died at 8–30 days. Of the patients who died of pulmonary embolism, 3 patients died within 1 day, 2 died within 2–7 days, and 1 died at 8–30 days. Respiratory diseases included pneumonia and respiratory insufficiency. Patients who died of infections died mostly of sepsis. Gastrointestinal bleeding was included in the digestive system group. Patients with an unclear or unknown cause of death according to the ICD were placed in the “unclear/unknown” group. For causes of death with only a few cases each (such as malnutrition and kidney disease), patients were placed in a group named “other”.
Table 2. Medical causes of death divided into groups.
| Empty Cell | <1 day | 2–7 days | 8–30 days | Total 0–30 days |
|---|---|---|---|---|
| n = 23 | n = 27 | n = 73 | n = 123 | |
| Cardiovascular disease, n (%)a | 15 (65.2) | 11 (40.7) | 23 (31.5) | 49 (39.8) |
| Respiratory disease, n (%) | 2 (8.7) | 2 (7.4) | 16 (21.9) | 20 (16.3) |
| Infections, n (%) | 3 (13.0) | 3 (11.1) | 9 (12.3) | 15 (12.2) |
| Cancer, n (%) | 1 (4.3) | 2 (7.4) | 10 (13.7) | 13 (10.6) |
| Digestive system, n (%) | 0 | 2 (7.4) | 3 (4.1) | 5 (4.1) |
| Kidney disease, n (%) | 0 | 1 (3.7) | 1 (1.4) | 2 (1.6) |
| Malnutrition, n (%) | 0 | 2 (7.4) | 0 | 2 (1.6) |
| Unclear/unknown, n (%) | 2 (8.7) | 4 (14.8) | 11 (15.1) | 17 (13.8) |
a
Percentages are calculated as the proportion of each group depending on whether patients died within 1 day, 2–7 days, 8–30 days, or 30 days of ECT.
3.3. Characteristics of deceased patients
The majority of patients who died of medical causes were older than 70 years (n = 86, 65.2 %). Fifteen of the twenty-three patients who died within 1 day of ECT died of cardiovascular diseases, which was also the leading medical cause of death overall. Hypertension was the most common pre-existing cardiovascular disease in patients who died of cardiovascular disease, and 33 of these 49 patients (67 %) had a Charlson Comorbidity Index of at least 1, with the highest proportion at three points or more (Table 3).
Table 3. Baseline characteristics of patients who died of cardiovascular disease.
| Empty Cell | 1 day | 2–7 days | 8–30 days | Total 0–30 days |
|---|---|---|---|---|
| n = 15 | n = 11 | n = 23 | n = 49 | |
| Hypertension, n (%)a | 7 (46.7) | 6 (54.5) | 12 (52.2) | 25 (51.0) |
| Myocardial infarction, n (%) | 2 (13.3) | 2 (18.2) | 3 (13.0) | 7 (14.3) |
| Congestive heart failure, n (%) | 3 (20.0) | 2 (18.2) | 3 (13.0) | 8 (16.3) |
| Atrial fibrillation, n (%) | 1 (6.7) | 1 (9.1) | 4 (17.4) | 6 (12.2) |
| Arrhythmia, n (%) | 0 | 0 | 2 (8.7) | 2 (4.1) |
| Valvular heart disease, n (%) | 2 (13.3) | 0 | 2 (8.7) | 4 (8.2) |
| Obesity, n (%) | 3 (20.0) | 0 | 0 | 3 (6.1) |
| Age, mean (SD) | 69 (17) | 73 (11) | 80 (12) | 75 (14) |
| CCI, n (%) | ||||
| 0 | 7 (46.7) | 2 (18.2) | 7 (30.4) | 16 (32.7) |
| 1 | 2 (13.3) | 3 (27.3) | 2 (8.7) | 7 (14.3) |
| 2 | 1 (6.7) | 2 (18.2) | 6 (26.0) | 9 (18.4) |
| ≥3 | 5 (33.3) | 4 (36.4) | 8 (34.8) | 17 (34.7) |
SD: standard deviation; CCI: Charlson Comorbidity Index.a
Percentages are calculated as the proportion of each group depending on if the patients died within 1 day, 2–7 days, 8–30 days, or 30 days of ECT.
The significant results from the multivariate analysis (Table 4) of age and medical conditions are presented in Fig. 1. Atrial fibrillation (OR: 2.10, 95 % CI: 1.08–4.08), dementia (OR: 2.60, 95 % CI: 1.46–4.64), kidney disease (OR: 2.28, 95 % CI: 1.17–4.47), reflux disease (OR: 2.77, 95 % CI: 1.12–6.83), and cancer (OR: 1.71, 95 % CI: 1.12–2.62) were associated with death by medical causes. No other medical condition significantly increased the risk of death, according to the multivariate analysis. Age was associated with death after treatment, whereby older patients had a higher risk, and the highest risk was for patients older than 80 years (OR: 36.17, 95 % CI: 14.68–89.12) as compared with patients younger than 50 years. A Charlson Comorbidity Index of 1 or more was also associated with an increased risk of death shortly after treatment, according to a separate multivariate analysis (OR: 2.453, 95 % CI: 1.66–3.62, p < 0.001).
Table 4. Binary logistic regression with univariate and multivariate analyses of variables associated with death within 30 days of ECT.
| Empty Cell | Univariate regressiona | P-value | Multivariate regression | P-value |
|---|---|---|---|---|
| OR (95 % CI) | OR (95 % CI) | |||
| <50 years (reference) | ||||
| 50–59 years | 5.92 (2.25–15.58) | <0.001 | 5.14 (1.94–13.60) | 0.001 |
| 60–69 years | 8.22 (3.26–20.72) | <0.001 | 6.70 (2.62–17.15) | <0.001 |
| 70–79 years | 21.32 (8.99–50.56) | <0.001 | 15.72 (6.42–38.48) | <0.001 |
| >80 years | 55.09 (23.55–128.83) | <0.001 | 36.17 (14.68–89.12) | <0.001 |
| Sex (female) | 1.12 (0.78–1.60) | 0.540 | 1.31 (0.91–1.90) | 0.148 |
| Valvular heart disease | 4.85 (1.96–12.02) | 0.001 | 1.13 (0.43–2.99) | 0.804 |
| Reflux disease | 3.93 (1.71–9.01) | 0.001 | 2.77 (1.12–6.83) | 0.027 |
| Obesity | 1.18 (0.48–2.89) | 0.721 | 2.19 (0.82–5.90) | 0.119 |
| Hernia | 1.24 (0.31–5.05) | 0.762 | 0.44 (0.10–1.89) | 0.267 |
| Ulcer disease | 2.27 (0.83–6.20) | 0.109 | 0.97 (0.34–2.78) | 0.951 |
| Kidney disease | 6.93 (3.78–12.70) | <0.001 | 2.61 (1.33–5.13) | 0.005 |
| Dementia | 9.36 (5.47–16.02) | <0.001 | 2.60 (1.46–4.64) | 0.001 |
| Cerebrovascular disease | 3.58 (2.10–6.07) | <0.001 | 1.30 (0.74–2.29) | 0.356 |
| Peripheral vascular disease | 0.94 (0.13–6.76) | 0.950 | 0.22 (0.03–1.64) | 0.140 |
| Congestive heart failure | 5.88 (3.28–10.54) | <0.001 | 1.19 (0.59–2.37) | 0.628 |
| Myocardial infarction | 3.89 (2.13–7.11) | <0.001 | 1.09 (0.57–2.08) | 0.805 |
| Diabetes | 2.80 (1.73–4.53) | <0.001 | 1.28 (0.75–2.15) | 0.365 |
| Hypertension | 4.42 (3.09–6.32) | <0.001 | 1.16 (0.76–1.78) | 0.490 |
| Cardiac arrhythmia | 1.75 (0.55–5.52) | 0.344 | 0.51 (0.16–1.71) | 0.277 |
| Atrial fibrillation | 6.30 (3.44–11.54) | <0.001 | 2.10 (1.08–4.08) | 0.029 |
| HIV/AIDS | 4.36 (1.06–17.94) | 0.042 | 2.82 (0.55–14.38) | 0.213 |
| Cancer | 3.52 (2.34–5.20) | <0.001 | 1.71 (1.12–2.62) | 0.013 |
| Moderate/severe liver disease | 4.45 (0.61–32.66) | 0.143 | 2.22 (0.24–20.57) | 0.483 |
| Mild liver disease | 1.09 (0.27–4.43) | 0.903 | 1.29 (0.28–5.98) | 0.748 |
| Rheumatic disease | 2.32 (1.24–4.32) | 0.008 | 0.88 (0.45–1.71) | 0.706 |
| Chronic obstructive pulmonary disease | 2.94 (1.36–6.33) | 0.006 | 1.28 (0.56–2.93) | 0.557 |
| Other chronic pulmonary disease | 1.96 (1.02–3.76) | 0.042 | 1.62 (0.81–3.24) | 0.177 |
HIV: human immunodeficiency virus; AIDS: acquired immunodeficiency syndrome.
Bold p-values are significant at p < 0.05. Values for obstructive sleep apnoea syndrome, metastatic cancer and hemiplegia could not be calculated in the analysis because of their low case numbers. The multivariate model includes all variables in the table.a
The reference value for univariate regression was not having the disease, aged under 50 years old, or a Charlson Comorbidity Index <1.
4. Discussion
In this register-based cohort study with a large population of 20,225 patients, we found that the mortality rate of medical causes within 30 days was 0.61 % (n = 123). Mortality rates within 30 days after an ECT session were 4.1 for medical causes and 7.2 for all causes per 10,000 treatment sessions. Cardiovascular disease was the leading medical cause of death. Older patients in particular had an increased risk of death shortly after treatment, as did patients with more comorbidities according to the Charlson Comorbidity Index. Atrial fibrillation, dementia, reflux disease, kidney disease, and cancer were independently associated with death by medical causes.
In this study, 216 patients died within 30 days and the all-cause mortality was 1.1 %, which is somewhat higher than earlier findings. In a study conducted in Denmark, 78 (0.84 %) patients died within 30 days in a population of 9327, but the authors did not find any causal association between ECT and death (Østergaard et al., 2014). Blumberger et al. described 65 deceased patients within 30 days in a population of 8810 and reported an all-cause mortality rate of 0.74 % within 30 days of ECT, of which 0.37 % died of medical causes. Deaths were calculated at 2.4 per 10,000 treatments, with external causes of death excluded (Blumberger et al., 2017). A review of 15 studies estimated the mortality rate to be 2.1 per 100,000 treatments. However, in that review, causes of death that were less likely to be related to ECT were excluded from the analysis. The authors considered cardiac arrest, aspiration pneumonia, and short time to death as possibly being associated with treatment, but the inclusion criteria differed between the included studies, and the total mortality rate including deaths not related to ECT sessions might have been underestimated. Furthermore, only one ECT-related death was reported among the nine included studies that were published after 2001, resulting in a mortality rate of 0.24 per 100.000 ECT treatments (Tørring et al., 2017).
Several recent comparative studies with follow-up periods of 365 days have demonstrated that patients who underwent ECT had a lower all-cause mortality than patients who did not receive ECT (Watts et al., 2021; Rhee et al., 2021; Jørgensen et al., 2020; Nordenskjöld et al., 2022). Studies with a shorter follow-up period have revealed that the mortality among patients receiving ECT is similar to that of controls (Watts et al., 2021; Yamazaki et al., 2022). In a study by Watts et al., the mortality was calculated to 3.08 per 10,000 treatments over the first 7 days after ECT (Watts et al., 2021). In a study from Texas investigating the deaths within 1 day of ECT treatment, 18 deaths per 100,000 treatments were reported when examining all-cause mortality, and 2.4 per 100,000 when excluding causes less likely to be associated with the treatment (Dennis et al., 2017). If we, like earlier studies (Dennis et al., 2017; Tørring et al., 2017), had only considered the cardiovascular deaths that occurred within 1 day, the mortality rate would be 5 per 100,000 treatments, which is consistent with earlier findings.
To our knowledge, no large-scale studies have reported anesthesia-related mortality after ECT. The overall mortality of general anesthesia has been estimated to approximately 1 per 100,000 anesthesia, which is lower than the mortality rate in our study (Li et al., 2009; Schiff et al., 2014). The most common cause of death is adverse effects of opioids followed by overdosing of intravenous anesthetic agents, each causing approximately 20 % of deaths (Li et al., 2009). Given that ECT is not painful, there is generally no need for large doses of opioids. The dose of induction agent is also lower than for most other anesthesias. These factors might decrease the risk of death followed by anesthesia for ECT. ECT is generally performed outside the operating room. In case of an emergency, equipment and extra staff might be further away than when anesthesia is performed in an operating room. This might be especially true in cases of failed airway management. For example, the incidence of difficult mask ventilation is approximately 5 % and combined with difficult intubation the incidence is 0.4 %; in these cases, video laryngoscopy could be helpful, but this kind of equipment is not always at hand in an ECT suite (Kheterpal et al., 2013). Some studies have suggested that mortality might be higher in non-operating room anesthesia compared with anesthesia performed in the operating room (Herman et al., 2021). Accurately determining anesthesia-related mortality is beyond the scope of this study. However, the association between gastroesophageal reflux and 30-day mortality is interesting in this context. While the condition is generally benign and not likely to be affected by ECT itself, reflux disease increases the risk of aspiration of gastric contents during general anesthesia (Cook et al., 2011; Robinson and Davidson, 2014).
The present results are in concordance with those of several previous studies that have reported cardiovascular disease and suicide as the most common causes of death shortly after ECT (Dennis et al., 2017; Østergaard et al., 2014; Blumberger et al., 2017; Herman et al., 2021; Munk-Olsen et al., 2007). Blumberger et al. found a proportion of 42.4 % cardiovascular cases of deceased patients, which is consistent with the proportion of 39.8 % cardiovascular deaths that we found in the present study (Blumberger et al., 2017). A systematic review and meta-analysis published in 2016 reported the frequency of major adverse cardiac events and death after ECT in approximately 2 % of patients, with a corresponding rate of 46.6 per 10,000 ECT treatments. The review could not, however, determine whether patients had pre-existing cardiovascular conditions or not prior to treatment (Duma et al., 2019). When we investigated the 15 patients that died of cardiovascular disease within 1 day, cardiac arrest, myocardial infarction, and pulmonary embolism were the leading causes of death; these deaths could be due to complications of both the underlying psychiatric condition, the ECT treatment, or the anesthesia. The subgroup analysis that mapped characteristics of patients who died from cardiovascular disease showed somewhat surprising results; less than half of the patients had an admission with a primary diagnosis of a pre-existing cardiac disease. This finding could be explained in several ways; for example, cardiac disease might remain undiagnosed in patients with psychiatric disease due to diffuse symptoms or inability/reluctance to seek appropriate medical attention. Peripheral thromboembolism resulting in a deathly pulmonary embolism may have developed during the period of psychiatric illness prior to ECT treatment, where both immobilization and disease-specific causes may play an important role. Furthermore, cardiac arrest is an imprecise cause of death, whereby the underlying cause is not obvious and no clarifying autopsy has been performed. Considering our findings that an older age, atrial fibrillation, dementia, kidney disease, reflux disease, and cancer were the independent predictors for death, the true underlying cause of the diagnosis “cardiac arrest” may be heterogenic or multifactorial and not cardiovascular at all. In addition, a recent study conducted by our research group compared adverse cardiovascular events between patients who received ECT and patients who did not receive ECT, and concluded that ECT was associated with reduced risk of major adverse cardiovascular events within 90 days and 1 year (Nordenskjöld et al., 2022).
An older age and more medical comorbidities are, according to previous studies (Shiwach et al., 2001; Ludvigsson et al., 2021), more common among patients who die shortly after ECT. This is consistent with our findings that the highest risk was for patients 80 years or older and with a Charlson Comorbidity Index of 1 or more. In our multivariate analysis, medical conditions that showed a significantly higher risk of death within 30 days after ECT were atrial fibrillation, dementia, kidney disease, reflux disease, and cancer. This indicates that measures should be considered to reduce the risks associated with these disorders during the pre-ECT workup, as well as during and after ECT.
Some patients die after ECT, yet it is difficult to determine whether these deaths are associated with the treatment itself. However, our results demonstrated that for every ECT session, the association with death significantly decreased, which indicates that ECT treatment itself is not a major cause of deaths. One earlier study also concluded that patients treated with ECT had a lower risk of death than patients who were not treated with ECT (Watts et al., 2021). This might also be the case in our study, which supports the notion that ECT can be a life-saving treatment that improves health status.
Suicide comprised nearly half (n = 93, 43.1 %) of all deaths in the present study. In contrast to medical causes of death, the baseline characteristics for the suicide group were more similar to alive patients. Both groups were younger and had fewer medical comorbidities than the group of deceased patients by medical causes. ECT has been shown to reduce the overall risk of suicide, especially in older people (Kaster et al., 2021; Rönnqvist et al., 2021). As concluded in an earlier study (Jørgensen et al., 2020), ECT is mainly used for patients with severe psychiatric disorders, and suicide rates can therefore be expected to be high in this patient group, even though ECT decreases the risk of suicide. These findings indicate that a close follow-up or observation of patients with severe psychiatric disorders in need of ECT is required to further decrease the risk of suicide.
4.1. Strengths and limitations
The strengths of this study include its large cohort and the detailed information from the registers. The study is limited as there is no non-ECT group. Data were from 4 different registers, with possible heterogeneity in patient characteristics and ECT parameters. It is therefore uncertain if the deaths were associated with the medical condition, the psychiatric disorder, or ECT, or anesthesia. The strengths of this study include its large cohort and the detailed information from the registers with high coverage and quality. For 47 of the patients who died of medical causes, no specific date of ECT during admission was included from the patient registry, and the observed days were thus counted from the date of discharge. Days to death might therefore have been miscalculated for these patients. While real-life observational studies based on registry data can demonstrate associations, they cannot determine causality. If medical records had been available, we would have been better able to determine if deaths were due to ECT, anesthesia, pre-existing medical conditions, or the mental disorder. From this study, we cannot tell whether the treatment itself caused death, or worsened or improved patients’ pre-existing medical conditions, or if patients whose psychiatric disorder improved had a decreased risk compared with patients who did not receive ECT. Including a control group could have been one way to address this problem.
5. Conclusion
Death due to medical causes within 30 days after ECT is rare. Older individuals with severe chronic somatic disease have the highest risk of death and extra measures should be considered to optimize their medical health during the pre-ECT workup, and during and after ECT. ECT appears to be a low-risk medical procedure.
CRediT authorship contribution statement
Author 1: Lovisa Lindblad
Performed data analysis; drafted and revised the manuscript; approved the final version and agrees to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Author 2: Axel Nordenskjöld
Conceived and designed the analysis; collected the data from different registers; contributed data and analysis tools; performed the analysis; wrote the paper; drafted and revised the manuscript; approved the final version and agrees to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Author 3: Alexander Otterbeck
Analyzed some of the data; drafted and revised the manuscript; approved the final version and agrees to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Author 4: Anna Nordenskjöld
Conceived and designed the analysis; contributed data and analysis tools; performed the analysis; wrote the paper; drafted and revised the manuscript; approved the final version and agrees to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Funding/support
This study was supported by Region Örebro County ALF and Nyckelfonden at Örebro University Hospital.
Conflict of interest
Dr. Axel Nordenskjöld reported receiving lecturer fees from Lundbeck outside the submitted work. No other disclosures were reported.
Acknowledgments
We thank all participants in the Swedish National Quality Register for ECT for providing data.
References
I. Ahmad, M. Sandberg, O. Brus, C.J. Ekman, Å. Hammar, M. Landén
m.fl. Validity of diagnoses, treatment dates, and rating scales in the Swedish national quality register for electroconvulsive therapy
Nord. J. Psychiatry (2021), pp. 1-8
Google ScholarAndrade et al., 2016
C. Andrade, S.S. Arumugham, J. Thirthalli
Adverse effects of electroconvulsive therapy
Psychiatr. Clin. N. Am., 39 (2016), pp. 513-530
ArticleDownload PDFView Record in ScopusGoogle ScholarBarlow et al., 2009
L. Barlow, K. Westergren, L. Holmberg, M. Talbäck
The completeness of the swedish cancer register: a sample survey for year 1998
Acta Oncol. Stockh. Swed., 48 (2009), pp. 27-33 View PDF
CrossRefView Record in ScopusGoogle ScholarBlumberger et al., 2017
D.M. Blumberger, D.P. Seitz, N. Herrmann, J.G. Kirkham, R. Ng, C. Reimer
m.fl. Low medical morbidity and mortality after acute courses of electroconvulsive therapy in a population-based sample
Acta Psychiatr. Scand., 136 (2017), pp. 583-593 View PDF
CrossRefView Record in ScopusGoogle ScholarBrooke et al., 2017
H.L. Brooke, M. Talbäck, J. Hörnblad, L.A. Johansson, J.F. Ludvigsson, H. Druid
m.fl. The Swedish cause of death register
Eur. J. Epidemiol., 32 (2017), pp. 765-773 View PDF
CrossRefView Record in ScopusGoogle ScholarCharlson et al., 1994
M. Charlson, T.P. Szatrowski, J. Peterson, J. Gold
Validation of a combined comorbidity index
J. Clin. Epidemiol., 47 (1994), pp. 1245-1251
ArticleDownload PDFView Record in ScopusGoogle ScholarChecklists, 2021
Checklists
STROBE [Internet]
Available from:
https://www.strobe-statement.org/checklists/ (2021)
Google ScholarCook et al., 2011
T.M. Cook, N. Woodall, C. Frerk
Major complications of airway management in the UK: results of the fourth National Audit Project of the Royal College of anaesthetists and the difficult airway society. Part 1: anaesthesia †
Br. J. Anaesth., 106 (2011), pp. 617-631
ArticleDownload PDFCrossRefView Record in ScopusGoogle ScholarDennis et al., 2017
N.M. Dennis, P.A. Dennis, A. Shafer, R.D. Weiner, M.M. Husain
Electroconvulsive therapy and all-cause mortality in Texas, 1998–2013
J. ECT, 33 (2017), pp. 22-25
View Record in ScopusGoogle ScholarDuma et al., 2019
A. Duma, M. Maleczek, B. Panjikaran, H. Herkner, T. Karrison, P. Nagele
Major adverse cardiac events and mortality associated with electroconvulsive therapy: a systematic review and meta-analysis
Anesthesiology, 130 (2019), pp. 83-91 View PDF
CrossRefView Record in ScopusGoogle ScholarHerman et al., 2021
A.D. Herman, C.B. Jaruzel, S. Lawton, C.D. Tobin, J.G. Reves, K.R. Catchpole
m.fl. Morbidity, mortality, and systems safety in non-operating room anaesthesia: a narrative review
Br. J. Anaesth., 127 (2021), pp. 729-744
ArticleDownload PDFView Record in ScopusGoogle ScholarJørgensen et al., 2020
M.B. Jørgensen, M.P. Rozing, C.H. Kellner, M. Osler
Electroconvulsive therapy, depression severity and mortality: data from the danish National Patient Registry
J. Psychopharmacol. Oxf. Engl., 34 (2020), pp. 273-279 View PDF
CrossRefView Record in ScopusGoogle ScholarKaster et al., 2021
T.S. Kaster, S.N. Vigod, T. Gomes, R. Sutradhar, D.N. Wijeysundera, D.M. Blumberger
Risk of serious medical events in patients with depression treated with electroconvulsive therapy: a propensity score-matched, retrospective cohort study
Lancet Psychiatry, 8 (2021), pp. 686-695
ArticleDownload PDFView Record in ScopusGoogle ScholarKheterpal et al., 2013
S. Kheterpal, D. Healy, M.F. Aziz, A.M. Shanks, R.E. Freundlich, F. Linton
m.fl. Incidence, predictors, and outcome of difficult mask ventilation combined with difficult laryngoscopy: a report from the multicenter perioperative outcomes group
Anesthesiology, 119 (2013), pp. 1360-1369
View Record in ScopusGoogle ScholarLi et al., 2009
G. Li, M. Warner, B.H. Lang, L. Huang, L.S. Sun
Epidemiology of anesthesia-related mortality in the United States, 1999–2005
Anesthesiology, 110 (2009), pp. 759-765
Google ScholarLiang et al., 2017
C.-S. Liang, C.-H. Chung, C.-K. Tsai, W.-C. Chien
In-hospital mortality among electroconvulsive therapy recipients: a 17-year nationwide population-based retrospective study
Eur. Psychiatry, 42 (2017), pp. 29-35
ArticleDownload PDFCrossRefView Record in ScopusGoogle ScholarLudvigsson et al., 2011
J.F. Ludvigsson, E. Andersson, A. Ekbom, M. Feychting, J.-L. Kim, C. Reuterwall
m.fl. External review and validation of the Swedish national inpatient register
BMC Public Health, 11 (2011), p. 450 View PDF
View Record in ScopusGoogle ScholarLudvigsson et al., 2021
J.F. Ludvigsson, P. Appelros, J. Askling, L. Byberg, J.-J. Carrero, A.M. Ekström
m.fl. Adaptation of the Charlson comorbidity index for register-based research in Sweden
Clin. Epidemiol., 13 (2021), pp. 21-41 View PDF
CrossRefView Record in ScopusGoogle ScholarMunk-Olsen et al., 2007
T. Munk-Olsen, T.M. Laursen, P. Videbech, P.B. Mortensen, R. Rosenberg
All-cause mortality among recipients of electroconvulsive therapy: register-based cohort study
Br. J. Psychiatry J. Ment. Sci., 190 (2007), pp. 435-439 View PDF
CrossRefView Record in ScopusGoogle ScholarNordanskog et al., 2015
P. Nordanskog, M. Hultén, M. Landén, J. Lundberg, L. von Knorring, A. Nordenskjöld
Electroconvulsive therapy in Sweden 2013: data from the National Quality Register for ECT
J. Ect, 31 (2015), p. 263
View Record in ScopusGoogle ScholarNordenskjöld et al., 2022
A. Nordenskjöld, P. Güney, A.M. Nordenskjöld
Major adverse cardiovascular events following electroconvulsive therapy in depression: a register-based nationwide swedish cohort study with 1-year follow-up
J. Affect. Disord., 296 (2022), pp. 298-304
ArticleDownload PDFView Record in ScopusGoogle ScholarØstergaard et al., 2014
S.D. Østergaard, T.G. Bolwig, G. Petrides
No causal association between electroconvulsive therapy and death: a summary of a report from the Danish Health and Medicines Authority covering 99,728 treatments
J. ECT, 30 (2014), pp. 263-264
View Record in ScopusGoogle ScholarRhee et al., 2021
T.G. Rhee, K. Sint, M. Olfson, T. Gerhard, S.H. Busch, S.T. Wilkinson
Association of ECT with risks of all-cause mortality and suicide in older medicare patients
Am. J. Psychiatry, 178 (12) (2021), pp. 1089-1097, 10.1176/appi.ajp.2021.21040351 View PDF
View Record in ScopusGoogle ScholarRobinson & Davidson, 2014
M. Robinson, A. Davidson
Aspiration under anaesthesia: risk assessment and decision-making
Contin. Educ. Anaesth. Crit. Care Pain, 14 (2014), pp. 171-175
ArticleDownload PDFCrossRefView Record in ScopusGoogle ScholarRönnqvist et al., 2021
I. Rönnqvist, F.K. Nilsson, A. Nordenskjöld
Electroconvulsive therapy and the risk of suicide in hospitalized patients with major depressive disorder
JAMA Netw. Open, 4 (2021), Article e2116589 View PDF
CrossRefView Record in ScopusGoogle ScholarSchiff et al., 2014
J.H. Schiff, A. Welker, B. Fohr, A. Henn-Beilharz, U. Bothner, H. Van Aken
m.fl. Major incidents and complications in otherwise healthy patients undergoing elective procedures: results based on 1.37 million anaesthetic procedures
Br. J. Anaesth., 113 (2014), pp. 109-121
ArticleDownload PDFCrossRefView Record in ScopusGoogle ScholarShiwach et al., 2001
R.S. Shiwach, W.H. Reid, T.J. Carmody
An analysis of reported deaths following electroconvulsive therapy in Texas, 1993–1998
Psychiatr. Serv. Wash. DC, 52 (2001), pp. 1095-1097
View Record in ScopusGoogle ScholarTørring et al., 2017
N. Tørring, S.N. Sanghani, G. Petrides, C.H. Kellner, S.D. Østergaard
The mortality rate of electroconvulsive therapy: a systematic review and pooled analysis
Acta Psychiatr. Scand., 135 (2017), pp. 388-397 View PDF
CrossRefView Record in ScopusGoogle ScholarWatts et al., 2021
B.V. Watts, T. Peltzman, B. Shiner
Mortality after electroconvulsive therapy
Br. J. Psychiatry, 1–6 (2021)
Google ScholarYamazaki et al., 2022
R. Yamazaki, O. Hiroyuki, Y. Matsuda, S. Kito, M. Shigeta, K. Morita, H. Matsui, K. Fushimi, Y. Hideo
Early electroconvulsive theraphy in patients with bipolar depression: a propensity score-matched analysis using a nationwide inpatient database
J. Affect. Disord., 312 (2022), pp. 245-251
