Association of Extracorporeal Membrane Oxygenation With New Mental Health Diagnoses in Adult Survivors of Critical Illness

Shannon M Fernando; Mary Scott; Robert Talarico; Eddy Fan; Daniel I McIsaac; Manish M Sood; Daniel T Myran; Margaret S Herridge; Dale M Needham; Carol L Hodgson; Bram Rochwerg; Laveena Munshi; M Elizabeth Wilcox; O Joseph Bienvenu; Graeme MacLaren; Robert A Fowler; Damon C Scales; Niall D Ferguson; Alain Combes; Arthur S Slutsky; Daniel Brodie; Peter Tanuseputro; Kwadwo Kyeremanteng

doi:10.1001/jama.2022.17714

. 2022 Oct 26;328(18):1827–1836. doi: 10.1001/jama.2022.17714

Association of Extracorporeal Membrane Oxygenation With New Mental Health Diagnoses in Adult Survivors of Critical Illness

Shannon M Fernando ^1,^2,^3,^✉, Mary Scott ^2,^4,^5,⁶, Robert Talarico ^2,⁴, Eddy Fan ^7,^8,^9,¹⁰, Daniel I McIsaac ^2,^4,^5,¹¹, Manish M Sood ^2,^4,^5,¹², Daniel T Myran ⁵, Margaret S Herridge ^7,^8,^9,¹⁰, Dale M Needham ^13,¹⁴, Carol L Hodgson ¹⁵, Bram Rochwerg ^16,¹⁷, Laveena Munshi ^7,^9,¹⁰, M Elizabeth Wilcox ^7,¹⁰, O Joseph Bienvenu ¹⁸, Graeme MacLaren ^19,²⁰, Robert A Fowler ^4,^7,^9,²¹, Damon C Scales ^4,^7,^9,^21,²², Niall D Ferguson ^7,^8,^9,¹⁰, Alain Combes ^23,²⁴, Arthur S Slutsky ^7,²², Daniel Brodie ^25,²⁶, Peter Tanuseputro ^2,^4,^5,^6,²⁷, Kwadwo Kyeremanteng ^1,^2,^27,²⁸

¹Division of Critical Care, Department of Medicine, University of Ottawa, Ottawa, Ontario, Canada

²Clinical Epidemiology Program, Ottawa Hospital Research Institute, Ottawa, Ontario, Canada

³Department of Critical Care, Lakeridge Health Corporation, Oshawa, Ontario, Canada

⁴ICES, Toronto, Ontario, Canada

⁵School of Epidemiology and Public Health, University of Ottawa, Ottawa, Ontario, Canada

⁶Bruyère Research Institute, Ottawa, Ontario, Canada

⁷Interdepartmental Division of Critical Care Medicine, University of Toronto, Toronto, Ontario, Canada

⁸Toronto General Hospital Research Institute, University Health Network, Toronto, Ontario, Canada

⁹Institute of Health Policy, Management and Evaluation, Dalla Lana School of Public Health, University of Toronto, Toronto, Ontario, Canada

¹⁰Department of Medicine, Sinai Health System and University Health Network, Toronto, Ontario, Canada

¹¹Department of Anesthesiology and Pain Medicine, University of Ottawa, Ottawa, Ontario, Canada

¹²Division of Nephrology, Department of Medicine, University of Ottawa, Ottawa, Ontario, Canada

¹³Division of Pulmonary and Critical Care Medicine, Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland

¹⁴Department of Physical Medicine and Rehabilitation, Johns Hopkins University School of Medicine, Baltimore, Maryland

¹⁵Australian and New Zealand Intensive Care-Research Centre, Monash University, Melbourne, Victoria, Australia

¹⁶Division of Critical Care, Department of Medicine, McMaster University, Hamilton, Ontario, Canada

¹⁷Department of Health Research Methods, Evidence, and Impact, McMaster University, Hamilton, Ontario, Canada

¹⁸Department of Psychiatry and Behavioral Sciences, Johns Hopkins University School of Medicine, Baltimore, Maryland

¹⁹Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Republic of Singapore

²⁰Cardiothoracic Intensive Care Unit, National University Heart Centre, National University Hospital, Singapore, Republic of Singapore

²¹Department of Critical Care Medicine, Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada

²²Li Ka Shing Knowledge Institute, St Michael’s Hospital, Toronto, Ontario, Canada

²³Sorbonne Université, Institute of Cardiometabolism and Nutrition, Paris, France

²⁴Service de Médecine Intensive-Réanimation, Hôpitaux Universitaires Pitié Salpêtrière, Assistance Publique-Hôpitaux de Paris, Institut de Cardiologie, Paris, France

²⁵Division of Pulmonary, Allergy, and Critical Care Medicine, Department of Medicine, Columbia University College of Physicians and Surgeons, New York, New York

²⁶Center for Acute Respiratory Failure, New York-Presbyterian Hospital, New York

²⁷Division of Palliative Care, Department of Medicine, University of Ottawa, Ottawa, Ontario, Canada

²⁸Institut du Savoir Montfort, Hôpital Montfort, Ottawa, Ontario, Canada

^✉

Corresponding Author: Shannon M. Fernando, MD, MSc, Division of Critical Care, Department of Medicine, University of Ottawa, 501 Smyth Rd, Ottawa, ON K1H 8M2, Canada (sfernando@qmed.ca).

Accepted for Publication: September 9, 2022.

Published Online: October 26, 2022. doi:10.1001/jama.2022.17714

Author Contributions: Drs Fernando and Tanuseputro had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Tanuseputro and Kyeremanteng contributed equally.

Concept and design: Fernando, Scott, Talarico, Fan, McIsaac, Sood, Myran, Needham, Bienvenu, MacLaren, Fowler, Scales, Combes, Tanuseputro, Kyeremanteng.

Acquisition, analysis, or interpretation of data: Fernando, Talarico, Fan, McIsaac, Myran, Herridge, Needham, Hodgson, Rochwerg, Munshi, Wilcox, Bienvenu, MacLaren, Scales, Ferguson, Combes, Slutsky, Brodie, Tanuseputro, Kyeremanteng.

Drafting of the manuscript: Fernando, Talarico, McIsaac, MacLaren, Combes, Kyeremanteng.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Talarico, Fan, McIsaac, Myran, Rochwerg, Munshi.

Obtained funding: Fernando, Tanuseputro, Kyeremanteng.

Administrative, technical, or material support: Scott, Talarico, Sood, Myran, Rochwerg, Tanuseputro.

Supervision: McIsaac, Kyeremanteng.

Other - expertise on content of long-term outcomes after extracorporeal membrane oxygenation (ECMO): Hodgson.

Other - critical illness–related psychiatric expertise: Bienvenu.

Other - content expertise on intensive care unit (ICU) outcomes/mental health outcomes: Herridge.

Conflict of Interest Disclosures: Dr Fan reported personal fees from ALung Technologies, Aerogen, Baxter, GE Healthcare, Inspira, and Vasomune outside the submitted work; and serving on the executive committee, scientific committee, and as chair of the data committee for the International ECMO Network (ECMONet). Dr Sood reported personal fees from AstraZeneca outside the submitted work; and receipt of support by the Jindal Research Chair in the Prevention of Kidney Disease at the University of Ottawa. Dr Hodgson reported grants from the National Health and Medical Research Council and the Medical Research Future Fund (leading the national ECMO registry in Australia including long-term outcomes) outside the submitted work; and serving on the executive committee, data committee, and scientific committee of the International ECMO Net. Dr Wilcox reported serving on the network committee of ECMONet. Dr MacLaren reported that he serves on the board of directors of the Extracorporeal Life Support Organization (ELSO)(unpaid). Dr Scales reported grants from the Canadian Institute for Health Research (operating grants) outside the submitted work. Dr Ferguson reported personal fees (for consulting) from Xenios and Baxter outside the submitted work; and serving on the executive committee and scientific committee of ECMONet. Dr Combes reported personal fees from Getinge, Baxter, and Xenios outside the submitted work; and serving on the executive committee and scientific committee of ECMONet, and as past president of the European Extracorporeal Life Support Organization (EuroELSO). Dr Slutsky reported personal fees (for consulting) from Baxter and Xenios during the conduct of the study; and serving on the executive committee and as chair of the scientific committee of ECMONet. Dr Brodie reported grants from ALung Technologies; personal fees (for serving on the medical advisory board) from Xenios, Medtronic, Cellenkos, Inspira, and Abiomed outside the submitted work; serving as president-elect for ELSO; and serving as chair of the executive committee for the International ECMO Net. Dr Kyeremanteng reported personal fees from Edwards Life Sciences outside the submitted work. No other disclosures were reported.

Funding/Support: Funded by the Institut du Savoir Montfort, Hôpital Montfort, Ottawa, ON, Canada. This study was supported by ICES, which is funded by an annual grant from the Ontario Ministry of Health.

Role of the Funder/Sponsor: The Institut du Savoir Montfort, Hôpital Montfort; ICES; and the Ontario Ministry of Health had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The opinions, results, and conclusions reported in this paper are those of the authors and are independent from the funding sources. No endorsement by ICES or the Ontario MOH is intended or should be inferred. Parts of this material are based on data and/or information compiled and provided by the Canadian Institute for Health Information (CIHI). However, the analyses, conclusions, and opinions and statements expressed in the material are those of the authors, and not necessarily those of CIHI.

^✉

Corresponding author.

PMCID: PMC9608013 PMID: 36286084

Key Points

Question

Is the use of extracorporeal membrane oxygenation (ECMO) associated with postdischarge psychiatric morbidity among adult survivors of critical illness?

Findings

In this population-based retrospective cohort study of 4462 participants, exposure to ECMO, compared with intensive care unit hospitalization without ECMO, was significantly associated with a modestly increased risk of new mental health diagnoses or social problems after discharge (hazard ratio, 1.24).

Meaning

Among adult survivors of critical illness, receipt of ECMO was significantly associated with a modestly increased risk of new mental health diagnoses or social problems, but further research on the mechanisms underlying this relationship is needed.

Abstract

Importance

Extracorporeal membrane oxygenation (ECMO) is used as temporary cardiorespiratory support in critically ill patients, but little is known regarding long-term psychiatric sequelae among survivors after ECMO.

Objective

To investigate the association between ECMO survivorship and postdischarge mental health diagnoses among adult survivors of critical illness.

Design, Setting, and Participants

Population-based retrospective cohort study in Ontario, Canada, from April 1, 2010, through March 31, 2020. Adult patients (N=4462; age ≥18 years) admitted to the intensive care unit (ICU), and surviving to hospital discharge were included.

Exposures

Receipt of ECMO.

Main Outcomes and Measures

The primary outcome was a new mental health diagnosis (a composite of mood disorders, anxiety disorders, posttraumatic stress disorder; schizophrenia, other psychotic disorders; other mental health disorders; and social problems) following discharge. There were 8 secondary outcomes including incidence of substance misuse, deliberate self-harm, death by suicide, and individual components of the composite primary outcome. Patients were compared with ICU survivors not receiving ECMO using overlap propensity score–weighted cause-specific proportional hazard models.

Results

Among 642 survivors who received ECMO (mean age, 50.7 years; 40.7% female), median length of follow-up was 730 days; among 3820 matched ICU survivors who did not receive ECMO (mean age, 51.0 years; 40.0% female), median length of follow-up was 1390 days. Incidence of new mental health conditions among survivors who received ECMO was 22.1 per 100-person years (95% confidence interval [CI] 19.5-25.1), and 14.5 per 100-person years (95% CI, 13.8-15.2) among non-ECMO ICU survivors (absolute rate difference of 7.6 per 100-person years [95% CI, 4.7-10.5]). Following propensity weighting, ECMO survivorship was significantly associated with an increased risk of new mental health diagnosis (hazard ratio [HR] 1.24 [95% CI, 1.01-1.52]). There were no significant differences between survivors who received ECMO vs ICU survivors who did not receive ECMO in substance misuse (1.6 [95% CI, 1.1 to 2.4] per 100 person-years vs 1.4 [95% CI, 1.2 to 1.6] per 100 person-years; absolute rate difference, 0.2 per 100 person-years [95% CI, −0.4 to 0.8]; HR, 0.86 [95% CI, 0.48 to 1.53]) or deliberate self-harm (0.4 [95% CI, 0.2 to 0.9] per 100 person-years vs 0.3 [95% CI, 0.2 to 0.3] per 100 person-years; absolute rate difference, 0.1 per 100 person-years [95% CI, −0.2 to 0.4]; HR, 0.68 [95% CI, 0.21 to 2.23]). There were fewer than 5 total cases of death by suicide in the entire cohort.

Conclusions and Relevance

Among adult survivors of critical illness, receipt of ECMO, compared with ICU hospitalization without ECMO, was significantly associated with a modestly increased risk of new mental health diagnosis or social problem diagnosis after discharge. Further research is necessary to elucidate the potential mechanisms underlying this relationship.

This cohort study investigates the association between extracorporeal membrane oxygenation survivorship and postdischarge mental health diagnoses among adult survivors of critical illness.

Introduction

Extracorporeal membrane oxygenation (ECMO) provides temporary respiratory and/or cardiac support to critically ill patients and is often considered when initial conventional treatment has failed.^1,2,3 Existing observational and randomized evidence have supported the use of ECMO in reducing short-term mortality in refractory respiratory failure, cardiogenic shock, and cardiac arrest.^4,5,6,7 Furthermore, while ECMO is associated with substantial inpatient resources and costs, a significant number of survivors after ECMO are alive up to 5 years postdischarge, with relatively minimal resource utilization.⁸ The use of ECMO has grown worldwide, particularly in the context of the COVID-19 pandemic, where ECMO has been used for refractory respiratory failure in cases of severe COVID-19.⁹

Despite collective public optimism surrounding the use of ECMO in refractory cases of critical illness,¹⁰ much less is understood about the long-term morbidity associated with this therapy, particularly as it pertains to mental health outcomes. Survivors of critical illness are known to have substantial physical morbidity,¹¹ and existing data demonstrate that these patients are at risk of downstream psychiatric morbidity,¹² including higher incidence of mental health diagnoses, psychoactive medication use, substance misuse, self-harm, and suicide.^13,14,15 Given the severity of illness found among patients receiving ECMO, its invasive nature, as well as the prolonged duration of therapy and recovery that is often required among those who survive, it was hypothesized that survivors after ECMO may be at even greater risk of downstream mental health morbidity than other survivors of critical illness. This study investigated the incidence of long-term mental health morbidity associated with ECMO survivorship, using population-based data from Ontario, Canada, and compared this with non-ECMO ICU survivorship.

Methods

Data Sources and Setting

We conducted a population-level cohort study using health administrative databases from Ontario, Canada (population 14.6 million). Within Ontario’s single-payer health care system, all publicly funded health care services, and physician, hospital, and demographic information for residents are recorded in administrative databases (ICES). Studies conducted at ICES using administrative data fall under section 45 of the Personal Health Information Protection Act of Ontario and do not require research ethics board approval. Data are collected without a requirement for individual patient consent. We linked 10 databases at ICES (eTable 1 in the Supplement) at the individual patient level, from April 1, 2010, through March 31, 2020 (inclusive). Patients are linked across provincial databases using their Ontario Health Insurance Plan (OHIP) number, which is unique to each citizen in Ontario. Data contained in ICES are full and complete, with the exception of emigration from Ontario, which represents approximately 0.5% of patients per year.

We registered our protocol with the Center for Open Science (https://osf.io/vn5bg).

Patients

We included consecutive adult patients (≥18 years of age), with an index intensive care unit (ICU) discharge in Ontario from April 1, 2010, through March 31, 2020 (inclusive), who received ECMO and who survived to hospital discharge. We identified ECMO use by the presence of an inpatient ECMO intervention code in the Discharge Abstract Database and a billing code for ECMO in the OHIP Claims Database, as performed previously (eTable 2 in the Supplement).^8,16 We matched these patients with those admitted to an ICU setting during the study period but who never received ECMO and survived to hospital discharge. For control participants with multiple ICU admissions, we randomly selected one for inclusion as the index admission. We excluded patients admitted to the ICU due to deliberate self-harm, as these patients are at high risk for future self-harm and suicide.¹⁵ We captured history of preexisting mental health diagnoses in the 5 years prior to the index admission through the use of International Statistical Classification of Diseases, Tenth Revision (ICD-10) codes and any outpatient mental health visits with a primary care physician or psychiatrist in the previous year.

Outcomes

The primary outcome was incidence of any new mental health diagnosis postdischarge, occurring prior to the end of the study period, the patient’s death, or emigration from Ontario. New mental health diagnoses were determined using the ICES Mental Health and Addictions Scorecard¹⁷ (eTable 2 in the Supplement), as used previously,¹⁵ and include any of the following: mood or anxiety disorders (eg, depression, anxiety, posttraumatic stress disorder [PTSD]), schizophrenia or psychotic disorders, other mental health diagnoses (including adjustment reaction, reactive depression, anxiety neurosis, hysteria, neurasthenia, obsessive-compulsive neurosis, personality disorders, sexual deviations, and psychosomatic illness), or social problems (including economic problems, marital difficulties, family disruption or divorce, parent-child problems, problems with aged parents, educational problems, social maladjustment, occupational problems, and legal problems). Diagnoses could be noted from either inpatient hospital admissions or outpatient encounters (including primary care or emergency department visits).

Secondary outcomes included substance misuse (secondary to alcohol or drug dependence), the composite of any new mental health diagnosis or substance misuse, death by suicide, and hospital visit for deliberate self-harm.¹⁵ We also separately evaluated each of the 4 components of the primary outcome (mood disorder/anxiety disorder/PTSD; schizophrenia/psychotic disorder; other mental health diagnoses; and social problems) as secondary outcomes. As per ICES regulations, cell sizes with 5 patients or less were suppressed to protect patient confidentiality.

Statistical Analyses

We present data as mean (SD) or median (IQR) values. Since ECMO is used only in select institutions in Ontario, and there are relatively stringent selection criteria for receiving ECMO, we created an initial matched cohort of 6 non-ECMO ICU controls for every single ECMO case. We performed greedy matching using the index admission institution, age (5-year groups), and the Charlson Comorbidity Index (CCI).¹⁸ To compare crude mental health outcome rates, we reported the incidence proportion and rate per 100 person-years with accompanying 95% CIs.

To estimate the association between ECMO survivorship and postdischarge mental health diagnoses, suicide, and self-harm, we followed existing recommendations for causal inference in critical care^19,20 and analyzed ECMO survivorship and non-ECMO ICU survivorship using overlap propensity-score weighting.¹⁵ The overlap weights technique was used to assign less weight to those with outlier propensity scores and more weight to those with propensity scores close to 0.5. This technique improves covariate balance and overlap of probabilities in creating propensity scores, as compared with other approaches.²¹ Overlap weighting does not exclude any potential study participants and is not affected by extreme outliers overdominating the analyses, which can occur with traditional propensity-score matching or inverse probability-treatment weighting.

As recommended, we identified covariates a priori, based on 2 factors: (1) existing literature and hypotheses about potential for confounding; and (2) their ability to influence the provision or timing of ECMO (as determined in existing randomized trials).^4,7 We included the following preadmission confounders: age, sex, number of encounters (hospital admissions, outpatient psychiatry visits, outpatient physician visits) in the past year, previous outpatient mental health or substance misuse visits to primary care clinicians, income quintile, rural residency, CCI score, specific comorbidities that might influence ECMO candidacy (ie, malignancy, chronic kidney disease, chronic obstructive pulmonary disease, preexisting mental health disorders, and previous history of self-harm). We also included confounders arising during admission: year of admission, institution, severity of illness (assessed using the Multiple Organ Dysfunction Score [MODS]) at the time of ICU admission, use of invasive mechanical ventilation, kidney replacement therapy, tracheostomy, ICU length of stay, hospital length of stay, and discharge disposition. Based on previous work, we anticipated less than 1% of neighborhood income quintile and rurality, and less than 3% of MODS and ICU length of stay to be missing at random. For the former, we imputed the mode, and for the latter, we imputed the age and index ECMO status group mean.

We compared cohorts before and after overlap weights for balance by using weighted standardized differences. We fit an overlap-weighted cause-specific Cox proportional hazards model with robust standard errors to examine ECMO survivorship vs non-ECMO ICU survivorship with the study outcomes, while accounting for mortality as a competing risk. We tested the proportionality assumption by including a statistical interaction between ECMO survivorship status and follow-up time as a linear term. The Wald χ² tests yielded a P value of greater than 0.1. We interpreted this as evidence of minimal violation of the proportionality assumption and did not carry forward the statistical interaction and compute time-dependent hazard ratios (HRs). Patients were censored from the analysis at the first instance of any of the following: death, loss of Ontario health insurance status, or the end of the study period (March 31, 2020). We used overlap-weighted cumulative incidence function (CIF) curves to plot the weighted CIF of new mental health diagnoses in the follow-up, and used the CIF equality test to see if the curves were statistically different. To estimate potential bias related to unmeasured confounders, we computed an E value.²²

To identify prognostic factors associated with new postcritical illness mental health outcomes, we generated separate cause-specific Cox proportional hazards regression models using the ECMO survivor cohort. We followed the PROGRESS guidelines^23,24 and recommendations for development of prognostic models in critical care.²⁵ These guidelines recommend a clinically hypothesis-driven approach for a priori selection of all model variables, as opposed to bivariable association testing methods.

Sensitivity Analyses

We preplanned sensitivity analyses to further evaluate potential confounding of our causal inference analyses by specific variables.

First, we hypothesized that access to in-person clinical visits with physicians (including primary care physicians and psychiatrists) would be strongly affected by restrictions instituted during the COVID-19 pandemic, and this had the potential to influence our primary outcome (new mental health diagnosis). As such, we conducted a sensitivity analysis excluding survivors after ECMO from our cohort who were discharged from hospital on January 1, 2020, or later, and we terminated the follow-up window on December 31, 2019. Second, while the ICES databases have strong validity in identifying new mental health diagnoses, they may not as easily identify worsening or decompensated mental illness among patients with preexisting mental health diagnoses. Further, patients with preexisting mental health follow-up may be more likely to be diagnosed with new mental illness after ICU admission, and this could represent a source of bias. Therefore, we conducted a sensitivity analysis following exclusion of patients with any mental health diagnoses in the previous 5 years. In addition, despite the use of overlap weights, we thought that the non-ECMO ICU cohort might include those with lower severity of illness, and so we performed a post hoc sensitivity analysis excluding control patients who did not receive invasive mechanical ventilation. We conducted all statistical analyses using SAS Enterprise Guide 7.1 (SAS Institute Inc). A P value of less than .05 (2-sided) was deemed to be statistically significant. Because of the potential for type I error due to multiple comparisons, findings for analyses of secondary end points should be interpreted as exploratory.

Results

Cohort Construction and Categorization

A total of 1054 adult patients received ECMO in Ontario during the study period. Of these, 412 were excluded (406 [38.5%] died prior to hospital discharge, and 6 of those who survived [0.6%] received ECMO due to deliberate self-harm), and 642 patients were included in the analyses. These patients were initially matched 1:6 to 3830 non-ECMO ICU survivors on the basis of institution, age, and CCI score. Median duration of follow-up was 730 days (IQR, 289-1437) among survivors after ECMO, and 1390 days (IQR, 572-2408) among non-ECMO ICU survivors.

Baseline Characteristics and ICU Interventions

Baseline characteristics of survivors after ECMO, along with matched controls (prior and after overlap weight application) are shown in Table 1. Among ECMO patients, the mean (SD) age was 50.7 (14.8) years, 40.7% were women, and 467 (72.7%) had a CCI of 2 or less. Among non-ECMO ICU survivors, the mean age was 51.0 (15.0) years, 39.9% were women, and 2798 (73.1%) had a CCI of 2 or less. In terms of preexisting psychiatric history, 300 (46.7%) survivors after ECMO, and 1850 (48.3%) of non-ECMO ICU survivors had at least 1 primary care or psychiatry visit for mental health in the preceding 5 years.

Table 1. Baseline Characteristics of Survivors After ECMO Compared With Control Non-ECMO ICU Survivors, and Comparison of Covariate Differences After Applying Overlap Weights.

Variable	Prior to overlap weights			After application of overlap weights, %^a
	No. (%)		Absolute standard difference
	Survivors after ECMO (n = 642)	Non-ECMO ICU survivors (n = 3830)	Absolute standard difference
Age, mean (SD), y	50.7 (14.8)	51.0 (15.0)	0.02	50.5
Men	381 (59.3)	2302 (60.1)
Women	261 (40.7)	1528 (39.9)	0.01	37.7
Neighborhood income quintile
No.	638	3814
Lowest	139 (21.7)	843 (22.0)	0.00	22.8
Low	131 (20.4)	779 (20.3)	0.00	20.9
Middle	141 (22.0)	750 (19.6)	0.06	21.6
High	104 (16.2)	711 (18.6)	0.06	15.1
Highest	123 (19.2)	731 (19.1)	0.00	19.6
Rural residence	74 (11.5)	344 (9.0)	0.08	10.7
Charlson Comorbidity Index, mean (SD)^b	1.8 (1.9)	1.8 (1.9)	0.01	1.8
Charlson Comorbidity Index score
≤2	467 (72.7)	2798 (73.1)
3-4	113 (17.6)	674 (17.6)
≥5	62 (9.7)	358 (9.3)
Comorbidities
Chronic obstructive pulmonary disease	190 (29.6)	219 (5.7)	0.66	16.2
Malignancy	185 (28.8)	1445 (37.7)	0.19	26.5
Chronic kidney disease	28 (4.4)	399 (10.4)	0.23	6.0
Previous psychiatric diagnoses
Any psychiatric diagnosis or substance misuse^c	300 (46.7)	1850 (48.3)
Depression, anxiety, or posttraumatic stress disorder	259 (40.3)	1575 (41.1)	0.02	40.9
Schizophrenia or other psychotic disorder	^d	77 (2.0)	0.11	1.5
Substance misuse	36 (5.6)	277 (7.2)	0.07	6.8
Other mental health diagnoses^e or social problems^f	75 (11.7)	521 (13.6)	0.06	13.1
Deliberate self-harm in preceding 5 y	^d	54 (1.4)	0.06	0.8
Outpatient visit in preceding year
Family physician visit for mental health, mean (SD)	0.6 (2.9)	0.8 (3.6)	0.07	0.7
Family physician visit for mental health, No.
0	541 (84.3)	3049 (79.6)
1	50 (7.8)	409 (10.7)
≥2	51 (7.9)	372 (9.7)
Psychiatrist visits, mean (SD)	0.4 (1.7)	0.5 (2.9)	0.03	0.4
Psychiatrist visits, No.
0	564 (87.9)	3475 (90.7)
1	35 (5.5)	148 (3.9)
≥2	43 (6.7)	207 (5.4)
Hospitalizations in preceding year, median (IQR)	1 (0-2)	0 (0-1)	0.43	1

Open in a new tab

Abbreviations: ECMO, extracorporeal membrane oxygenation; ICU, intensive care unit.

^{^a}

Weighted absolute standard difference was 0.00 for all variables; therefore, values for each variable were the same for the survivors after ECMO group and the non-ECMO ICU survivors group.

^{^b}

Range, 0 to 37. Points are added on the basis of the presence of various comorbidities. An increasing score indicates a higher and more severe number of comorbidities.

^{^c}

Indicates at least 1 primary care or psychiatry visit for mental health in the preceding 5 years. Numeric values may not exceed the total for any psychiatric diagnosis or substance misuse because some patients had more than 1 diagnosis and were therefore accounted in more than 1 row.

^{^d}

Indicates the category had 5 patients or less.

^{^e}

Includes adjustment reaction, reactive depression, anxiety neurosis, hysteria, neurasthenia, obsessive-compulsive neurosis, personality disorders, sexual deviations, and psychosomatic illness.

^{^f}

Includes economic problems, marital difficulties, family disruption/divorce, parent-child problems, problems with aged parents, educational problems, social maladjustment, occupational problems, and legal problems.

Index admission characteristics of survivors after ECMO, along with matched controls (prior and after overlap weight application) are depicted in Table 2. The primary indication for ECMO was respiratory failure, occurring in 400 patients (62.3%). The mean MODS was 5.6 (SD, 3.3), with 152 patients (23.7%) receiving kidney replacement therapy, and 221 (34.4%) receiving tracheostomy. Median hospital length of stay was 41 days (IQR, 23-70), and 210 patients (32.7%) were discharged to a long-term care facility after their acute care hospitalization. Categorization by year of admission is shown in eTable 3 in the Supplement.

Table 2. Index Admission Characteristics of Survivors After ECMO Compared With Control Non-ECMO ICU Survivors, and Comparison of Covariate Differences After Applying Overlap Weights.

Variable	Prior to overlap weights			After application of overlap weights, %^a
	No. (%)		Absolute standard difference
	Survivors after ECMO (n = 642)	Non-ECMO ICU survivors (n = 3830)	Absolute standard difference
Year of admission^b
2010-2011	7 (1.1)	373 (9.7)	0.39	2.2
2011-2012	9 (1.4)	308 (8.0)	0.32	1.8
2012-2013	26 (4.0)	355 (9.3)	0.21	5.0
2013-2014	45 (7.0)	415 (10.8)	0.14	7.5
2014-2015	57 (8.9)	345 (9.0)	0.01	10.1
2015-2016	58 (9.0)	388 (10.1)	0.04	11.2
2016-2017	74 (11.5)	383 (10.0)	0.05	10.6
2017-2018	108 (16.8)	416 (10.9)	0.17	16.2
2018-2019	114 (17.8)	385 (10.1)	0.22	15.9
2019-2020	144 (22.4)	462 (12.1)	0.28	19.5
Indication for ECMO^c
Respiratory failure	400 (62.3)
Cardiac failure	114 (17.8)
Unknown^d	128 (19.9)
Multiple Organ Dysfunction Score, mean (SD)	5.6 (3.3)	2.5 (2.7)	1.02	5.0
Invasive mechanical ventilation	627 (97.7)	1513 (39.5)	1.61	94.2
Kidney replacement therapy	152 (23.7)	219 (5.7)	0.52	12.2
Tracheostomy	221 (34.4)	116 (3.0)	0.88	14.8
ICU length of stay, mean (SD), d	11.4 (18.1)	4.1 (5.3)	0.55	7.6
Hospital length of stay, mean (SD), d	55.5 (49.9)	15.3 (25.4)	1.02	39.6
Hospital length of stay, median (IQR), d	41 (23-70)	6 (6-14)
Hospital length of stay, d
<7	1286 (33.6)	0 (0.0)
7-14	1520 (39.7)	44 (6.9)
>14	1024 (26.7)	598 (93.1)
Discharge disposition
Home, without home care	192 (29.9)	2412 (63.0)	0.70	40.3
Home, with home care	240 (37.4)	1039 (27.1)	0.22	39.1
Long-term care facility	210 (32.7)	379 (9.9)	0.59	20.6
Duration of follow-up, median (IQR), d^e	730 (289-1437)	1390 (572-2408)

Open in a new tab

Abbreviations: ECMO, extracorporeal membrane oxygenation; ICU, intensive care unit.

^{^a}

Weighted absolute standard difference was 0.00 for all variables; therefore, values for each variable were the same for the survivors after ECMO group and the non-ECMO ICU survivors group.

^{^b}

Start and end points for each year are April 1 of the earlier year through March 31 of the later year.

^{^c}

Variable not used in overlap weighting.

^{^d}

Refers to ECMO cases where indication could not be determined through diagnostic coding.

^{^e}

Patients were censored at the time of death, loss of Ontario health insurance status, or March 31, 2020, (whichever came first).

Association of ECMO With Mental Health Outcomes

Balance after overlap weighting between ICU survivors who were and were not exposed to ECMO are shown in Table 1, Table 2, and eFigure in the Supplement. Weighted standard difference across all covariables was 0.00. Incidence rates and adjusted HRs showing the association between ECMO survivorship and the outcomes of interest (using non-ECMO ICU survivors as the reference) and accounting for competing risk of death are shown in Table 3, with weighted CIF curves displayed in the Figure. Raw counts of the outcomes prior to overlap weights are shown in eTable 4 in the Supplement.

Table 3. Incidence Rates, Absolute Rate Differences, and Weighted Hazard Ratios Comparing Mental Health Outcomes Among Survivors After ECMO With Matched Non-ECMO Intensive Care Unit Survivors.

Outcome	Cases per 100 person-years (95% CI)			Weighted hazard ratio (95% CI)
Outcome	Survivors after ECMO (n=642)	Non-ECMO ICU survivors (n=3830)	Incidence rate difference^a	Weighted hazard ratio (95% CI)
Any mental health condition	22.1 (19.5 to 25.1)	14.5 (13.8 to 15.2)	7.6 (4.7 to 10.5)	1.24 (1.01 to 1.52)
Any mental health or substance misuse condition	23.6 (20.8 to 26.7)	15.4 (14.6 to 16.2)	8.2 (5.2 to 11.2)	1.24 (1.02 to 1.51)
Mood disorder or anxiety disorder or PTSD	18.0 (15.7 to 20.7)	12.1 (11.5 to 12.7)	5.9 (3.3 to 8.5)	1.14 (0.91 to 1.42)
Schizophrenia or psychotic disorder	0.4 (0.2 to 0.9)	0.5 (0.4 to 0.6)	−0.1 (−0.4 to 0.6)	1.08 (0.39 to 3.00)
Other mental health diagnoses^b	4.7 (3.7 to 5.9)	2.7 (2.4 to 2.9)	2.0 (0.9 to 3.1)	1.38 (0.95 to 2.02)
Social problems^c	0.8 (0.5 to 1.4)	0.7 (0.6 to 0.9)	0.1 (−0.4 to 0.6)	1.55 (0.69 to 3.40)
Substance misuse	1.6 (1.1 to 2.4)	1.4 (1.2 to 1.6)	0.2 (−0.4 to 0.8)	0.86 (0.48 to 1.53)
Deliberate self-harm	0.4 (0.2 to 0.9)	0.3 (0.2 to 0.3)	0.1 (−0.2 to 0.4)	0.68 (0.21 to 2.23)

Open in a new tab

Abbreviations: ECMO, extracorporeal membrane oxygenation; PTSD, posttraumatic stress disorder.

^{^a}

Incidence rates are not adjusted.

^{^b}

Includes adjustment reaction, reactive depression, anxiety neurosis, hysteria, neurasthenia, obsessive-compulsive neurosis, personality disorders, sexual deviations, and psychosomatic illness.

^{^c}

Figure. — All P values were calculated using the cumulative incidence function equality test. ECMO indicates extracorporeal membrane oxygenation; ICU, intensive care unit.

Primary Outcome

With regard to the primary outcome, 236 (36.8%) survivors after ECMO, and 1565 (40.9%) of non-ECMO ICU survivors were diagnosed with a new mental health condition following discharge. The incidence rate was 22.1 per 100 person-years (95% CI, 19.5-25.1) among survivors after ECMO, compared with 14.5 per 100 person-years (95% CI, 13.8 to 15.2) among non-ECMO ICU survivors (absolute rate difference, 7.6 per 100 person-years [95% CI, 4.7 to 10.5]). Following propensity score weighting, survivors after ECMO had significantly higher associated risk of new mental health diagnoses (HR, 1.24 [95% CI, 1.01 to 1.52]; E value, 1.59). This same relationship was seen in our sensitivity analyses excluding patients discharged on or after January 1, 2020, (n = 4324; HR, 1.27 [95% CI, 1.03 to 1.57]; E value, 1.64), excluding patients with a previous history of mental health diagnoses (n = 2412; HR, 1.43 [95% CI, 1.02 to 2.00]; E value, 1.87), and our post hoc analysis excluding patients who did not receive mechanical ventilation (n = 2140; HR, 1.28 [95% CI, 1.04 to 1.57]; E value, 1.66).

Secondary Outcomes

Findings for individual components of the primary composite outcome are presented in Table 3. Additionally, ECMO survivorship was not significantly associated with postdischarge substance misuse (HR, 0.86 [95% CI, 0.48 to 1.53]; absolute rate difference, 0.2 per 100-person years [95% CI, −0.4 to 0.8]) or deliberate self-harm (HR, 0.68 [95% CI, 0.21 to 2.23]; absolute rate difference of 0.1 per 100-person years [95% CI, −0.2 to 0.4]). There were less than 5 cases of suicide total across the entire cohort.

Factors Associated With New Mental Health Diagnoses Among Survivors After ECMO

Among the 642 ECMO patients who survived to discharge, prognostic factors associated with incidence of new mental health diagnoses are presented in Table 4. The only 2 prognostic factors found to be significantly associated with new mental health diagnoses were preexisting mental health diagnosis (HR, 2.39 [95% CI, 1.78 to 3.20]) and having had an outpatient psychiatry visit in the year prior to the index admission (1.82 [95% CI, 1.25 to 2.65]).

Table 4. Cox Regression Analyses Identifying Factors Associated With New Mental Health Diagnoses or Social Problems Among Survivors After ECMO (N = 642).

Prognostic factor	Hazard ratio (95% CI)
Female sex	0.87 (0.66-1.14)
Age (per 1 y)	0.99 (0.98-1.00)
Rural residence	1.07 (0.70-1.64)
Neighborhood income quintile
Lowest	0.92 (0.61-1.39)
Low	1.24 (0.82-1.86)
Middle	1.04 (0.69-1.57)
High	1.15 (0.75-1.77)
Highest	1 [Reference]
Hospitalizations in past year, per 1 hospitalization	0.92 (0.84-1.01)
Outpatient psychiatry visits in past year, per 1 visit	1.82 (1.25-2.65)
Preexisting mental health diagnosis	2.39 (1.78-3.20)
Comorbidities
Chronic obstructive pulmonary disease	1.11 (0.80-1.56)
Malignancy	0.92 (0.68-1.25)
Chronic kidney disease	0.54 (0.23-1.24)
Multiple Organ Dysfunction Score, per 1 point^a	1.02 (0.98-1.06)
Hospital length of stay, per 1 d	1.00 (0.99-1.00)
Kidney replacement therapy	0.90 (0.64-1.29)
Tracheostomy	1.01 (0.73-1.39)
Discharge disposition
Home without home care	1 [Reference]
Home with homecare	0.91 (0.65-1.26)
Long-term care facility	1.02 (0.70-1.50)

Open in a new tab

^{^a}

Score range, 0 to 24 points. Increasing points are associated with increased severity of illness and correspond with increased hospital and intensive care unit mortality.

Discussion

Among adult survivors of critical illness, receipt of ECMO, compared with ICU hospitalization without ECMO, was significantly associated with a modestly increased risk of new mental health diagnosis or social problem diagnosis after discharge. However, the mechanisms underlying this relationship are unclear.

Key stakeholders in ECMO research have identified long-term sequelae associated with ECMO survivorship to be a priority, but to date, little evidence exists.²⁶ To that end, the results of this current study show that diagnoses of new mental health conditions were common among survivors after ECMO. The most frequent diagnoses were related to depression, anxiety, and traumatic disorders, which was unsurprising given the invasive nature of critical care. Such diagnoses are the most common among all adult survivors of critical illness.^27,28 In addition, this work found that ECMO survivorship was significantly associated with an increased risk of postdischarge mental health diagnoses. This relationship appears to be potentially greater than what is seen among non-ECMO ICU survivors, a population that has been quite vulnerable to psychiatric morbidity.^12,29 Particularly in the context of the ongoing COVID-19 pandemic and an associated increase in the use of ECMO worldwide, this work provides important information for clinicians with regard to psychological supports for survivors following discharge. That said, while there was an association between ECMO survivorship and downstream psychiatric morbidity, the results suggest that the associated increased risk was likely modest. It was also found that there was no association with substance misuse or deliberate self-harm, and suicide was rare. These relatively small absolute differences must be considered in the context of randomized evidence suggesting improved survival among patients receiving ECMO for refractory respiratory and cardiac failure.^4,5,7 As such, while the findings of this current study should not preclude selection of patients for receipt of ECMO, they serve to identify a potential group of vulnerable patients that may more commonly experience psychiatric morbidity, and thus require additional downstream support.

While the present study draws attention to a potential relationship, further research is needed to provide insights into the mechanisms that may potentially link ECMO survivorship with psychiatric morbidity. In the prognostic model, the only 2 factors that were found to be associated with future mental health diagnoses were preexisting mental health diagnoses and previous outpatient psychiatric visits. However, the prognostic association of these factors may be related to confounding, as such patients are more likely to have existing contact with a mental health clinician and thus receive a mental health diagnosis. There are a number of theoretical associations between ECMO survivorship and downstream mental health morbidity. Although data are limited, existing work on functional capacity and health-related quality of life among survivors after ECMO shows that these patients experience substantial physical limitations following discharge, similar to studies of non-ECMO adult survivors of critical illness.³⁰ Furthermore, existing data have linked severity and course of ICU admission with mental health morbidity.^15,27,28 This relationship is the most likely explanation for the study findings. ECMO patients represent a unique and extreme subset of patients who experience prolonged and intense courses of critical illness, and this may predispose these patients to mental health morbidity. While no statistically significant differences in individual psychiatric diagnoses were seen between survivors after ECMO and non-ECMO ICU survivors, the primary composite appeared to be driven by “other mental health” diagnoses (eg, adjustment reaction, reactive depression), which may be expected since such diagnoses are typically deemed secondary to an acute stressor or event. Whether ECMO itself causes psychiatric morbidity is controversial, and little mechanistic evidence exists. Patients receiving ECMO may be particularly prone to hypoxemia or shock, and reduced oxygen delivery (potentially through microcirculatory disturbances) can predispose neurons to apoptosis and death.³¹ Similar theories have been suggested as explanations for neurocognitive and psychiatric sequelae seen in adult survivors of critical illness, particularly those with severe hypoxemic respiratory failure.^32,33,34 ECMO patients are at higher risk of particular neurological complications than non-ECMO ICU patients, such as ischemic stroke or prolonged delirium,³⁵ and this may manifest in downstream psychiatric morbidity. As such, future research is needed in order to identify potential mechanistic links, as these links may serve as targets for future interventions.

Limitations

This study has several limitations. First, this work is observational, and thus cannot confirm a causal relationship between ECMO provision and downstream mental health morbidity. Despite statistical adjustment, the possibility of residual confounding remains. Second, while many covariates were measured, the granularity of the available data was limited (eg, ECMO indications and configurations), suggesting the possible influence of unmeasured confounding. There was also imprecision in the findings. To that end, multiple sensitivity analyses were performed, and an E value was calculated to provide insights into the reliability of our findings. Third, the outcomes were limited only to mental health diagnoses identified in visits with clinicians remunerated through the provincial health care system. Services privately paid for by patients would not have been captured. Similarly, the outcome of self-harm was only limited to events that would have required hospital attention. Fourth, while the included patients came from multiple centers, they exist within the same province and health care system, and thus these findings may not be generalizable to all populations.

Conclusions

Section Editor: Christopher Seymour, MD, Associate Editor, JAMA (christopher.seymour@jamanetwork.org).

Supplement.

eTable 1. Databases Linked Through ICES, on the Basis of Individual Patient Identifiers

eTable 2. Relevant Databases and Codes Utilized for Key Variables

eTable 3. Number of Patients Added to the Cohort and Number of Postdischarge Deaths per Fiscal Year Among Survivors After ECMO and non-ECMO ICU Survivors by Year

eFigure. Density Plots Depicting Propensity Score Weighting by Overlapping Weights, With Distribution of Illness Severity (MODS) Pre- and Postweighting

eTable 4. Mental Health Outcomes of Extracorporeal Membrane Oxygenation Survivors and Comparison to Matched Control Intensive Care Unit Survivors, Not Accounting for Competing Risk of Death, or Propensity Matching by Weighted Means

Click here for additional data file.^{(469KB, pdf)}

References

1.Brodie D, Slutsky AS, Combes A. Extracorporeal Life Support for Adults With Respiratory Failure and Related Indications: A Review. JAMA. 2019;322(6):557-568. doi: 10.1001/jama.2019.9302 [DOI] [PubMed] [Google Scholar]
2.Combes A, Price S, Slutsky AS, Brodie D. Temporary circulatory support for cardiogenic shock. Lancet. 2020;396(10245):199-212. doi: 10.1016/S0140-6736(20)31047-3 [DOI] [PubMed] [Google Scholar]
3.Abrams D, MacLaren G, Lorusso R, et al. Extracorporeal cardiopulmonary resuscitation in adults: evidence and implications. Intensive Care Med. 2022;48(1):1-15. doi: 10.1007/s00134-021-06514-y [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Combes A, Hajage D, Capellier G, et al. ; EOLIA Trial Group, REVA, and ECMONet . Extracorporeal membrane oxygenation for severe acute respiratory distress syndrome. N Engl J Med. 2018;378(21):1965-1975. doi: 10.1056/NEJMoa1800385 [DOI] [PubMed] [Google Scholar]
5.Goligher EC, Tomlinson G, Hajage D, et al. Extracorporeal membrane oxygenation for severe acute respiratory distress syndrome and posterior probability of mortality benefit in a post hoc bayesian analysis of a randomized clinical trial. JAMA. 2018;320(21):2251-2259. doi: 10.1001/jama.2018.14276 [DOI] [PubMed] [Google Scholar]
6.Combes A, Leprince P, Luyt CE, et al. Outcomes and long-term quality-of-life of patients supported by extracorporeal membrane oxygenation for refractory cardiogenic shock. Crit Care Med. 2008;36(5):1404-1411. doi: 10.1097/CCM.0b013e31816f7cf7 [DOI] [PubMed] [Google Scholar]
7.Yannopoulos D, Bartos J, Raveendran G, et al. Advanced reperfusion strategies for patients with out-of-hospital cardiac arrest and refractory ventricular fibrillation (ARREST): a phase 2, single centre, open-label, randomised controlled trial. Lancet. 2020;396(10265):1807-1816. doi: 10.1016/S0140-6736(20)32338-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Fernando SM, Qureshi D, Tanuseputro P, et al. Mortality and costs following extracorporeal membrane oxygenation in critically ill adults: a population-based cohort study. Intensive Care Med. 2019;45(11):1580-1589. doi: 10.1007/s00134-019-05766-z [DOI] [PubMed] [Google Scholar]
9.Barbaro RP, MacLaren G, Boonstra PS, et al. ; Extracorporeal Life Support Organization . Extracorporeal membrane oxygenation support in COVID-19: an international cohort study of the Extracorporeal Life Support Organization registry. Lancet. 2020;396(10257):1071-1078. doi: 10.1016/S0140-6736(20)32008-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Fernando SM, Mathew R, Slutsky AS, et al. Media portrayals of outcomes after extracorporeal membrane oxygenation. JAMA Intern Med. 2021;181(3):391-394. doi: 10.1001/jamainternmed.2020.6094 [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Herridge MS, Cheung AM, Tansey CM, et al. ; Canadian Critical Care Trials Group . One-year outcomes in survivors of the acute respiratory distress syndrome. N Engl J Med. 2003;348(8):683-693. doi: 10.1056/NEJMoa022450 [DOI] [PubMed] [Google Scholar]
12.Fernando SM, Ranzani OT, Herridge MS. Mental health morbidity, self-harm, and suicide in ICU survivors and caregivers. Intensive Care Med. 2022;48(8):1084-1087. doi: 10.1007/s00134-022-06743-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Wunsch H, Christiansen CF, Johansen MB, et al. Psychiatric diagnoses and psychoactive medication use among nonsurgical critically ill patients receiving mechanical ventilation. JAMA. 2014;311(11):1133-1142. doi: 10.1001/jama.2014.2137 [DOI] [PubMed] [Google Scholar]
14.Wunsch H, Hill AD, Fu L, et al. New opioid use after invasive mechanical ventilation and hospital discharge. Am J Respir Crit Care Med. 2020;202(4):568-575. doi: 10.1164/rccm.201912-2503OC [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Fernando SM, Qureshi D, Sood MM, et al. Suicide and self-harm in adult survivors of critical illness: population based cohort study. BMJ. 2021;373(973):n973. doi: 10.1136/bmj.n973 [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Fernando SM, Qureshi D, Tanuseputro P, et al. Long-term survival and costs following extracorporeal membrane oxygenation in critically ill children-a population-based cohort study. Crit Care. 2020;24(1):131. doi: 10.1186/s13054-020-02844-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Mental Health and Addictions Program Framework Research Team . Mental Health and Addictions System Performance in Ontario: 2021 Scorecard. Chart Pack; 2021. [Google Scholar]
18.Sundararajan V, Henderson T, Perry C, Muggivan A, Quan H, Ghali WA. New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol. 2004;57(12):1288-1294. doi: 10.1016/j.jclinepi.2004.03.012 [DOI] [PubMed] [Google Scholar]
19.Lederer DJ, Bell SC, Branson RD, et al. Control of confounding and reporting of results in causal inference studies. guidance for authors from editors of respiratory, sleep, and critical care journals. Ann Am Thorac Soc. 2019;16(1):22-28. doi: 10.1513/AnnalsATS.201808-564PS [DOI] [PubMed] [Google Scholar]
20.Hernán MA, Robins JM. Using big data to emulate a target trial when a randomized trial is not available. Am J Epidemiol. 2016;183(8):758-764. doi: 10.1093/aje/kwv254 [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Thomas LE, Li F, Pencina MJ. Overlap weighting: a propensity score method that mimics attributes of a randomized clinical trial. JAMA. 2020;323(23):2417-2418. doi: 10.1001/jama.2020.7819 [DOI] [PubMed] [Google Scholar]
22.VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the E-value. Ann Intern Med. 2017;167(4):268-274. doi: 10.7326/M16-2607 [DOI] [PubMed] [Google Scholar]
23.Riley RD, Hayden JA, Steyerberg EW, et al. ; PROGRESS Group . Prognosis Research Strategy (PROGRESS) 2: prognostic factor research. PLoS Med. 2013;10(2):e1001380. doi: 10.1371/journal.pmed.1001380 [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Steyerberg EW, Moons KG, van der Windt DA, et al. ; PROGRESS Group . Prognosis Research Strategy (PROGRESS) 3: prognostic model research. PLoS Med. 2013;10(2):e1001381. doi: 10.1371/journal.pmed.1001381 [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Leisman DE, Harhay MO, Lederer DJ, et al. Development and reporting of prediction models: guidance for authors from editors of respiratory, sleep, and critical care journals. Crit Care Med. 2020;48(5):623-633. doi: 10.1097/CCM.0000000000004246 [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Combes A, Brodie D, Chen YS, et al. The ICM research agenda on extracorporeal life support. Intensive Care Med. 2017;43(9):1306-1318. doi: 10.1007/s00134-017-4803-3 [DOI] [PubMed] [Google Scholar]
27.Sivanathan L, Wunsch H, Vigod S, Hill A, Pinto R, Scales DC. Mental illness after admission to an intensive care unit. Intensive Care Med. 2019;45(11):1550-1558. doi: 10.1007/s00134-019-05752-5 [DOI] [PubMed] [Google Scholar]
28.Olafson K, Marrie RA, Bolton JM, et al. The 5-year pre- and post-hospitalization treated prevalence of mental disorders and psychotropic medication use in critically ill patients: a Canadian population-based study. Intensive Care Med. 2021;47(12):1450-1461. doi: 10.1007/s00134-021-06513-z [DOI] [PubMed] [Google Scholar]
29.Prince E, Gerstenblith TA, Davydow D, Bienvenu OJ. Psychiatric morbidity after critical illness. Crit Care Clin. 2018;34(4):599-608. doi: 10.1016/j.ccc.2018.06.006 [DOI] [PubMed] [Google Scholar]
30.Hodgson CL, Hayes K, Everard T, et al. Long-term quality of life in patients with acute respiratory distress syndrome requiring extracorporeal membrane oxygenation for refractory hypoxaemia. Crit Care. 2012;16(5):R202. doi: 10.1186/cc11811 [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Taccone FS, Su F, Pierrakos C, et al. Cerebral microcirculation is impaired during sepsis: an experimental study. Crit Care. 2010;14(4):R140. doi: 10.1186/cc9205 [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Pandharipande PP, Girard TD, Jackson JC, et al. ; BRAIN-ICU Study Investigators . Long-term cognitive impairment after critical illness. N Engl J Med. 2013;369(14):1306-1316. doi: 10.1056/NEJMoa1301372 [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Jackson JC, Pandharipande PP, Girard TD, et al. ; Bringing to light the Risk Factors And Incidence of Neuropsychological dysfunction in ICU survivors (BRAIN-ICU) study investigators . Depression, post-traumatic stress disorder, and functional disability in survivors of critical illness in the BRAIN-ICU study: a longitudinal cohort study. Lancet Respir Med. 2014;2(5):369-379. doi: 10.1016/S2213-2600(14)70051-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Mikkelsen ME, Christie JD, Lanken PN, et al. The adult respiratory distress syndrome cognitive outcomes study: long-term neuropsychological function in survivors of acute lung injury. Am J Respir Crit Care Med. 2012;185(12):1307-1315. doi: 10.1164/rccm.201111-2025OC [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Luyt CE, Bréchot N, Demondion P, et al. Brain injury during venovenous extracorporeal membrane oxygenation. Intensive Care Med. 2016;42(5):897-907. doi: 10.1007/s00134-016-4318-3 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement.

eTable 1. Databases Linked Through ICES, on the Basis of Individual Patient Identifiers

eTable 2. Relevant Databases and Codes Utilized for Key Variables

eTable 3. Number of Patients Added to the Cohort and Number of Postdischarge Deaths per Fiscal Year Among Survivors After ECMO and non-ECMO ICU Survivors by Year

eFigure. Density Plots Depicting Propensity Score Weighting by Overlapping Weights, With Distribution of Illness Severity (MODS) Pre- and Postweighting

Click here for additional data file.^{(469KB, pdf)}

[joi220104r1] 1.Brodie D, Slutsky AS, Combes A. Extracorporeal Life Support for Adults With Respiratory Failure and Related Indications: A Review. JAMA. 2019;322(6):557-568. doi: 10.1001/jama.2019.9302 [DOI] [PubMed] [Google Scholar]

[joi220104r2] 2.Combes A, Price S, Slutsky AS, Brodie D. Temporary circulatory support for cardiogenic shock. Lancet. 2020;396(10245):199-212. doi: 10.1016/S0140-6736(20)31047-3 [DOI] [PubMed] [Google Scholar]

[joi220104r3] 3.Abrams D, MacLaren G, Lorusso R, et al. Extracorporeal cardiopulmonary resuscitation in adults: evidence and implications. Intensive Care Med. 2022;48(1):1-15. doi: 10.1007/s00134-021-06514-y [DOI] [PMC free article] [PubMed] [Google Scholar]

[joi220104r4] 4.Combes A, Hajage D, Capellier G, et al. ; EOLIA Trial Group, REVA, and ECMONet . Extracorporeal membrane oxygenation for severe acute respiratory distress syndrome. N Engl J Med. 2018;378(21):1965-1975. doi: 10.1056/NEJMoa1800385 [DOI] [PubMed] [Google Scholar]

[joi220104r5] 5.Goligher EC, Tomlinson G, Hajage D, et al. Extracorporeal membrane oxygenation for severe acute respiratory distress syndrome and posterior probability of mortality benefit in a post hoc bayesian analysis of a randomized clinical trial. JAMA. 2018;320(21):2251-2259. doi: 10.1001/jama.2018.14276 [DOI] [PubMed] [Google Scholar]

[joi220104r6] 6.Combes A, Leprince P, Luyt CE, et al. Outcomes and long-term quality-of-life of patients supported by extracorporeal membrane oxygenation for refractory cardiogenic shock. Crit Care Med. 2008;36(5):1404-1411. doi: 10.1097/CCM.0b013e31816f7cf7 [DOI] [PubMed] [Google Scholar]

[joi220104r7] 7.Yannopoulos D, Bartos J, Raveendran G, et al. Advanced reperfusion strategies for patients with out-of-hospital cardiac arrest and refractory ventricular fibrillation (ARREST): a phase 2, single centre, open-label, randomised controlled trial. Lancet. 2020;396(10265):1807-1816. doi: 10.1016/S0140-6736(20)32338-2 [DOI] [PMC free article] [PubMed] [Google Scholar]

[joi220104r8] 8.Fernando SM, Qureshi D, Tanuseputro P, et al. Mortality and costs following extracorporeal membrane oxygenation in critically ill adults: a population-based cohort study. Intensive Care Med. 2019;45(11):1580-1589. doi: 10.1007/s00134-019-05766-z [DOI] [PubMed] [Google Scholar]

[joi220104r9] 9.Barbaro RP, MacLaren G, Boonstra PS, et al. ; Extracorporeal Life Support Organization . Extracorporeal membrane oxygenation support in COVID-19: an international cohort study of the Extracorporeal Life Support Organization registry. Lancet. 2020;396(10257):1071-1078. doi: 10.1016/S0140-6736(20)32008-0 [DOI] [PMC free article] [PubMed] [Google Scholar]

[joi220104r10] 10.Fernando SM, Mathew R, Slutsky AS, et al. Media portrayals of outcomes after extracorporeal membrane oxygenation. JAMA Intern Med. 2021;181(3):391-394. doi: 10.1001/jamainternmed.2020.6094 [DOI] [PMC free article] [PubMed] [Google Scholar]

[joi220104r11] 11.Herridge MS, Cheung AM, Tansey CM, et al. ; Canadian Critical Care Trials Group . One-year outcomes in survivors of the acute respiratory distress syndrome. N Engl J Med. 2003;348(8):683-693. doi: 10.1056/NEJMoa022450 [DOI] [PubMed] [Google Scholar]

[joi220104r12] 12.Fernando SM, Ranzani OT, Herridge MS. Mental health morbidity, self-harm, and suicide in ICU survivors and caregivers. Intensive Care Med. 2022;48(8):1084-1087. doi: 10.1007/s00134-022-06743-9 [DOI] [PMC free article] [PubMed] [Google Scholar]

[joi220104r13] 13.Wunsch H, Christiansen CF, Johansen MB, et al. Psychiatric diagnoses and psychoactive medication use among nonsurgical critically ill patients receiving mechanical ventilation. JAMA. 2014;311(11):1133-1142. doi: 10.1001/jama.2014.2137 [DOI] [PubMed] [Google Scholar]

[joi220104r14] 14.Wunsch H, Hill AD, Fu L, et al. New opioid use after invasive mechanical ventilation and hospital discharge. Am J Respir Crit Care Med. 2020;202(4):568-575. doi: 10.1164/rccm.201912-2503OC [DOI] [PMC free article] [PubMed] [Google Scholar]

[joi220104r15] 15.Fernando SM, Qureshi D, Sood MM, et al. Suicide and self-harm in adult survivors of critical illness: population based cohort study. BMJ. 2021;373(973):n973. doi: 10.1136/bmj.n973 [DOI] [PMC free article] [PubMed] [Google Scholar]

[joi220104r16] 16.Fernando SM, Qureshi D, Tanuseputro P, et al. Long-term survival and costs following extracorporeal membrane oxygenation in critically ill children-a population-based cohort study. Crit Care. 2020;24(1):131. doi: 10.1186/s13054-020-02844-3 [DOI] [PMC free article] [PubMed] [Google Scholar]

[joi220104r17] 17.Mental Health and Addictions Program Framework Research Team . Mental Health and Addictions System Performance in Ontario: 2021 Scorecard. Chart Pack; 2021. [Google Scholar]

[joi220104r18] 18.Sundararajan V, Henderson T, Perry C, Muggivan A, Quan H, Ghali WA. New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol. 2004;57(12):1288-1294. doi: 10.1016/j.jclinepi.2004.03.012 [DOI] [PubMed] [Google Scholar]

[joi220104r19] 19.Lederer DJ, Bell SC, Branson RD, et al. Control of confounding and reporting of results in causal inference studies. guidance for authors from editors of respiratory, sleep, and critical care journals. Ann Am Thorac Soc. 2019;16(1):22-28. doi: 10.1513/AnnalsATS.201808-564PS [DOI] [PubMed] [Google Scholar]

[joi220104r20] 20.Hernán MA, Robins JM. Using big data to emulate a target trial when a randomized trial is not available. Am J Epidemiol. 2016;183(8):758-764. doi: 10.1093/aje/kwv254 [DOI] [PMC free article] [PubMed] [Google Scholar]

[joi220104r21] 21.Thomas LE, Li F, Pencina MJ. Overlap weighting: a propensity score method that mimics attributes of a randomized clinical trial. JAMA. 2020;323(23):2417-2418. doi: 10.1001/jama.2020.7819 [DOI] [PubMed] [Google Scholar]

[joi220104r22] 22.VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the E-value. Ann Intern Med. 2017;167(4):268-274. doi: 10.7326/M16-2607 [DOI] [PubMed] [Google Scholar]

[joi220104r23] 23.Riley RD, Hayden JA, Steyerberg EW, et al. ; PROGRESS Group . Prognosis Research Strategy (PROGRESS) 2: prognostic factor research. PLoS Med. 2013;10(2):e1001380. doi: 10.1371/journal.pmed.1001380 [DOI] [PMC free article] [PubMed] [Google Scholar]

[joi220104r24] 24.Steyerberg EW, Moons KG, van der Windt DA, et al. ; PROGRESS Group . Prognosis Research Strategy (PROGRESS) 3: prognostic model research. PLoS Med. 2013;10(2):e1001381. doi: 10.1371/journal.pmed.1001381 [DOI] [PMC free article] [PubMed] [Google Scholar]

[joi220104r25] 25.Leisman DE, Harhay MO, Lederer DJ, et al. Development and reporting of prediction models: guidance for authors from editors of respiratory, sleep, and critical care journals. Crit Care Med. 2020;48(5):623-633. doi: 10.1097/CCM.0000000000004246 [DOI] [PMC free article] [PubMed] [Google Scholar]

[joi220104r26] 26.Combes A, Brodie D, Chen YS, et al. The ICM research agenda on extracorporeal life support. Intensive Care Med. 2017;43(9):1306-1318. doi: 10.1007/s00134-017-4803-3 [DOI] [PubMed] [Google Scholar]

[joi220104r27] 27.Sivanathan L, Wunsch H, Vigod S, Hill A, Pinto R, Scales DC. Mental illness after admission to an intensive care unit. Intensive Care Med. 2019;45(11):1550-1558. doi: 10.1007/s00134-019-05752-5 [DOI] [PubMed] [Google Scholar]

[joi220104r28] 28.Olafson K, Marrie RA, Bolton JM, et al. The 5-year pre- and post-hospitalization treated prevalence of mental disorders and psychotropic medication use in critically ill patients: a Canadian population-based study. Intensive Care Med. 2021;47(12):1450-1461. doi: 10.1007/s00134-021-06513-z [DOI] [PubMed] [Google Scholar]

[joi220104r29] 29.Prince E, Gerstenblith TA, Davydow D, Bienvenu OJ. Psychiatric morbidity after critical illness. Crit Care Clin. 2018;34(4):599-608. doi: 10.1016/j.ccc.2018.06.006 [DOI] [PubMed] [Google Scholar]

[joi220104r30] 30.Hodgson CL, Hayes K, Everard T, et al. Long-term quality of life in patients with acute respiratory distress syndrome requiring extracorporeal membrane oxygenation for refractory hypoxaemia. Crit Care. 2012;16(5):R202. doi: 10.1186/cc11811 [DOI] [PMC free article] [PubMed] [Google Scholar]

[joi220104r31] 31.Taccone FS, Su F, Pierrakos C, et al. Cerebral microcirculation is impaired during sepsis: an experimental study. Crit Care. 2010;14(4):R140. doi: 10.1186/cc9205 [DOI] [PMC free article] [PubMed] [Google Scholar]

[joi220104r32] 32.Pandharipande PP, Girard TD, Jackson JC, et al. ; BRAIN-ICU Study Investigators . Long-term cognitive impairment after critical illness. N Engl J Med. 2013;369(14):1306-1316. doi: 10.1056/NEJMoa1301372 [DOI] [PMC free article] [PubMed] [Google Scholar]

[joi220104r33] 33.Jackson JC, Pandharipande PP, Girard TD, et al. ; Bringing to light the Risk Factors And Incidence of Neuropsychological dysfunction in ICU survivors (BRAIN-ICU) study investigators . Depression, post-traumatic stress disorder, and functional disability in survivors of critical illness in the BRAIN-ICU study: a longitudinal cohort study. Lancet Respir Med. 2014;2(5):369-379. doi: 10.1016/S2213-2600(14)70051-7 [DOI] [PMC free article] [PubMed] [Google Scholar]

[joi220104r34] 34.Mikkelsen ME, Christie JD, Lanken PN, et al. The adult respiratory distress syndrome cognitive outcomes study: long-term neuropsychological function in survivors of acute lung injury. Am J Respir Crit Care Med. 2012;185(12):1307-1315. doi: 10.1164/rccm.201111-2025OC [DOI] [PMC free article] [PubMed] [Google Scholar]

[joi220104r35] 35.Luyt CE, Bréchot N, Demondion P, et al. Brain injury during venovenous extracorporeal membrane oxygenation. Intensive Care Med. 2016;42(5):897-907. doi: 10.1007/s00134-016-4318-3 [DOI] [PubMed] [Google Scholar]

PERMALINK

Association of Extracorporeal Membrane Oxygenation With New Mental Health Diagnoses in Adult Survivors of Critical Illness

Shannon M Fernando, MD, MSc

Mary Scott, MSc

Robert Talarico, MSc

Eddy Fan, MD, PhD

Daniel I McIsaac, MD, MPH

Manish M Sood, MD, MSc

Daniel T Myran, MD, MPH

Margaret S Herridge, MD, MSc, MPH

Dale M Needham, MD, PhD

Carol L Hodgson, PhD

Bram Rochwerg, MD, MSc

Laveena Munshi, MD, MSc

M Elizabeth Wilcox, MD, MPH

O Joseph Bienvenu, MD, PhD

Graeme MacLaren, MBBS, MSc

Robert A Fowler, MDCM, MSc

Damon C Scales, MD, PhD

Niall D Ferguson, MD, MSc

Alain Combes, MD, PhD

Arthur S Slutsky, MD, MASc

Daniel Brodie, MD

Peter Tanuseputro, MD, MHSc

Kwadwo Kyeremanteng, MD, MHA

Key Points

Question

Findings

Meaning

Abstract

Importance

Objective

Design, Setting, and Participants

Exposures

Main Outcomes and Measures

Results

Conclusions and Relevance

Introduction

Methods

Data Sources and Setting

Patients

Outcomes

Statistical Analyses

Sensitivity Analyses

Results

Cohort Construction and Categorization

Baseline Characteristics and ICU Interventions

Table 1. Baseline Characteristics of Survivors After ECMO Compared With Control Non-ECMO ICU Survivors, and Comparison of Covariate Differences After Applying Overlap Weights.

Table 2. Index Admission Characteristics of Survivors After ECMO Compared With Control Non-ECMO ICU Survivors, and Comparison of Covariate Differences After Applying Overlap Weights.

Association of ECMO With Mental Health Outcomes

Table 3. Incidence Rates, Absolute Rate Differences, and Weighted Hazard Ratios Comparing Mental Health Outcomes Among Survivors After ECMO With Matched Non-ECMO Intensive Care Unit Survivors.

Figure. Cumulative Incidence Function Curves Comparing Survivors After ECMO and Non-ECMO ICU Survivors for the Primary Composite Outcome of Any New Mental Health Condition and Each Individual Component of the Composite.

Primary Outcome

Secondary Outcomes

Factors Associated With New Mental Health Diagnoses Among Survivors After ECMO

Table 4. Cox Regression Analyses Identifying Factors Associated With New Mental Health Diagnoses or Social Problems Among Survivors After ECMO (N = 642).

Discussion

Limitations

Conclusions

References

Associated Data

Supplementary Materials

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases