By Elaine Chen, MD
Associate Professor, Department of Internal Medicine, Division of Pulmonary and Critical Care Medicine, Section of Palliative Medicine, Rush University Medical Center, Chicago
SYNOPSIS: Survivors of extracorporeal membrane oxygenation (ECMO) have a modest increase in risk of new mental health diagnoses after discharge compared with intensive care unit survivors who do not undergo ECMO.
SOURCE: Fernando SM, Scott M, Talarico R, et al. Association of extracorporeal membrane oxygenation with new mental health diagnoses in adult survivors of critical illness. JAMA 2022;328:1827-1836.
Extracorporeal membrane oxygenation (ECMO) provides temporary cardiac and respiratory system support when conventional treatment has failed. ECMO use has increased, especially since the beginning of the COVID-19 pandemic. ECMO improves short-term and likely long-term mortality in selected populations. Long-term morbidities, especially mental health outcomes, are not well understood. There is a growing body of literature about prolonged morbidities in survivors of critical illness. This study uses population-based data to compare ECMO and non-ECMO intensive care unit (ICU) survivorship regarding long-term mental health morbidity.
A population-level cohort study was performed using health administrative databases from Ontario, Canada. The single-payer healthcare system allows for provincewide collection of accurate administrative data. All consecutive adult patients admitted in a 10-year period (April 2010 through March 2020) who received ECMO and survived to hospital discharge were included. They were matched with ICU survivors who did not receive ECMO. Patients admitted for deliberate self-harm were excluded.
The primary outcome was the incidence of any new mental health diagnosis from the time of discharge to the time of study completion, death, or emigration from Ontario. Secondary outcomes included substance misuse, death by suicide, and hospital visit for deliberate self-harm. Statistical analyses were described in detail, including overlap weighting, proportionality assumption, and regression models with preplanned sensitivity analyses. Overlap weighting assigns less weight to participants with outlier propensity scores and more weight to those with propensity scores close to 0.5 rather than excluding outlier participants.
A total of 1,054 adult patients received ECMO during the study period, with 642 ECMO survivors included in the analysis and 3,830 non-ECMO ICU survivors matched for comparison. The median duration of follow-up was 730 days (interquartile range [IQR], 289 to 1,437 days) for ECMO survivors and 1,390 days (IQR, 572 to 2,408) for non-ECMO survivors; the reason for discontinuation of follow-up was not included. Baseline characteristics, including age, sex, number of comorbidities, and preexisting psychiatric history, were well matched. Notably, only 39.5% of the original control group underwent invasive mechanical ventilation, compared with 97.7% of the ECMO group, highlighting the importance of overlap weighting.
New mental health conditions were diagnosed in 236 (36.8%) ECMO survivors and 1,565 (40.9%) non-ECMO survivors, with an incidence rate of 22.1 per 100 person-years (95% confidence interval [CI], 19.5-25.1) in the ECMO cohort and 14.5 per 100 person-years (95% CI, 13.8-15.2) in the non-ECMO cohort, for an absolute rate difference of 7.6 per 100 person-years (95% CI, 4.7-10.5) and a hazard ratio (HR) of 1.24 (95% CI, 1.01-1.52) after propensity score matching. ECMO survivorship was not associated with increased substance misuse (HR, 0.86; 95% CI, 0.48-1.63) or deliberate self-harm (HR, 0.68; 95% CI, 0.21-2.23).
Two prognostic factors, specifically a preexisting mental health diagnosis (HR, 2.39; 95% CI, 1.78-3.20) and having an outpatient psychiatry visit within one year prior to admission (HR, 1.82; 95% CI, 1.25-2.65), were significantly associated with new mental health diagnoses. Statistical significance in cumulative function curves after overlap weighting was maintained for the primary composite outcome, any mental health or substance use, and other mental health or social problems, with no differences in mood or anxiety disorders, any substance use, schizophrenia or psychotic disorder, or deliberate self-harm.
COMMENTARY
The authors concluded that the receipt of ECMO is significantly associated with a modestly increased risk of new mental health or social problem diagnosis after discharge compared with a general ICU hospitalization without ECMO. This study confirms that new mental health conditions are common among ECMO survivors, most commonly depression, anxiety, and traumatic disorders.
Prior to overlap weighting, ECMO survivors had a significantly higher incidence of invasive mechanical ventilation use (97.7% vs. 39.5%) and longer duration of hospitalization (median 41 vs. 6 days; mean 55.5 vs. 15.3 days) compared to the general ICU survivor, and a much lower incidence of discharge independently to home (29.9% vs. 63.0%). The much shorter duration of follow-up (730 vs. 1,390 days) may reflect the increased prevalence of ECMO later in the study period or decreased long-term survival in the ECMO cohort. The prevalence of a new mental health diagnosis was higher in the non-ECMO group (40.9% vs. 36.8%), which does not account for duration of follow-up.
The authors noted limitations due to the observational nature of the study, residual or unmeasured confounding despite statistical adjustment and sensitivity analyses, and the geographic limitation of the study, which may not be applicable to ECMO cohorts elsewhere in the world where social structures may differ substantially from those in Ontario, Canada. Regardless, mental health morbidity is an important and common sequela of critical illness, and very notable in this small but growing population of ECMO survivors. This calls for ongoing research efforts into mental health outcomes in survivors of both ECMO and general critical illness, as well as targeted clinical efforts to improve mental health awareness and access for this complex population.
