Characteristics and risk factors of delirium in patients on veno-arterial extracorporeal membrane oxygenation

Abstract

Purpose

Sparse data exist on delirium in VA-ECMO. We aim to describe the characteristics, risk factors, and outcomes of delirium in VA-ECMO.

Methods

We retrospectively reviewed adults’s electronic medical records on VA-ECMO in our ECMO registry in 2016–2022. Delirium was assessed by the Confusion Assessment Method for the intensive care unit when patients scored −3 or above on the Richmond Agitation-Sedation Scale. The primary outcomes were delirium prevalence and the proportion of delirium-present days while on VA-ECMO support. Multivariable logistic regression was used to evaluate delirium risk factors.

Results

Of 208 patients (median [interquartile range] age: 53 [40–62]), 138 (66.3%) had delirium during ECMO. Delirium occurred on day 2.5 [1.0–7.0] of ECMO and was detected in 42% [20%–66%] of ECMO days. There were no differences in acute brain injury (24% vs 33%, p = .34) between patients with and without delirium. Survival analysis showed no significant association between delirium and 30-day mortality (p = .24). In multivariable analysis, ECMO day 1 arterial carbon dioxide partial pressure (adjusted odds ratio [aOR] = 1.29; 95% CI = 1.03–1.73), number of sedatives (aOR = 2.67; 95% CI = 1.68–2.95), and African American race/ethnicity (aOR = 16.45; 95% CI = 9.65–22.51) were associated with delirium.

Conclusions

Delirium was present in 66.3% of VA-ECMO patients and was detected early during ECMO. Modifiable risk factors included multiple sedative agents and early hypercapnia. Delirium did not increase risk for mortality.

Keywords

delirium veno-arterial ECMO

Introduction

Delirium, an acute cognitive impairment involving fluctuating alterations in consciousness, is the most common neurological complication in intensive care units (ICUs) and increases risks of readmission and mortality.^1–3 Delirium prevalence, in the Delirium Epidemiology in Critical Care (DECCA) study, was 32.3% in patients receiving ICU care overall,⁴ but it was higher, 67–85%, in specific cohorts with a greater severity of illness, such as those on mechanical ventilation,^2,5 extracorporeal membrane oxygenation (ECMO),^6–8 or in surgical and trauma ICUs.⁹ It is challenging to delineate the clinical roles of ICU delirium in the most critically ill patients because of its high prevalence, requirement for deep sedation, and poor neurologic status from medical complexity.⁷ With the presenting challenges in studying the roles and mechanisms of delirium, clinical management strategies have not been well developed to address the specific needs of patients.

Venoarterial ECMO (VA-ECMO) provides mechanical circulatory support to patients with refractory cardiogenic shock or cardiac arrest. According to the international Extracorporeal Life Support Organization (ELSO) Registry,¹⁰ the neurocognitive outcomes of middle-aged adults with broader implications (e.g., return to their career) are gaining more interest. In a recent study on acute heart failure, delirium was independently predictive of poor 30-day patient outcomes in adults while the presence of neurocognitive impairment without delirium did not.¹ This suggests that delirium is mainly related to underlying acute medical illness rather than age or baseline cognitive reserve of individuals. With unique factors in ECMO, such as analgosedation with altered pharmacokinetics and pharmacodynamics¹¹ and neurological complications from cannulation,¹² an independent study on patients on VA-ECMO is necessary to elucidate modifiable risk factors and treatment strategies.

Currently, there is a paucity of data on the characteristics and risk factors of delirium, and its relationship to analgosedative methods in VA-ECMO patients. In our study, we aimed to investigate the prevalence and characteristics of delirium, the association between delirium and putative risk factors and clinical markers, and the association between delirium and in-hospital mortality in VA-ECMO. We hypothesized that there may be ECMO-specific modifiable risk factors and in-hospital events for delirium and that those with delirium would suffer from more complications.

Methods

Study design

We retrospectively reviewed 279 adults’s electronic medical records (EMR) on VA-ECMO admitted to the Cardiovascular Surgical Intensive Care Unit (CVSICU) of Johns Hopkins Hospital in 2016–2022. Patients were included in the study if they were 18 years of age or above; VA-ECMO cannulation lasted for at least 24 hours. The exclusion criteria were venovenous ECMO (VV-ECMO); less than 18 years of age at the time of cannulation; unassessable Confusion Assessment Method for the ICU¹³ (CAM-ICU) evaluation due to low Richmond Agitation-Sedation Scale (RASS) of (−4 or −5) for the entire duration of VA-ECMO.

Data collection

The study was approved by the Johns Hopkins University School of Medicine Institutional Review Board (IRB00264320). Patient consent was waived due to the retrospective nature of the study. From the Johns Hopkins ECMO registry and EMR, we retrieved data on demographics, pre/post-cannulation laboratory values, indication for VA-ECMO, CAM-ICU, RASS, Behavioral Pain Assessment Scale (BPAS), and hospital course, including length of stay, length of ECMO cannulation, and mortality. Laboratory values were gathered 6 hours pre-ECMO cannulation and 24 hours post-cannulation. 30-day mortality information was obtained via review electronic medical records by examining patient vital status around at or past 30 days.

Delirium was assessed by CAM-ICU twice daily if patients scored −3 or higher on the RASS. Patients were determined unassessable for delirium using CAM-ICU and in a coma if their RASS score was −5 (unarousable) or −4 (deep sedation). Patients were also assessed using the institutional BPAS, a 10-point scale that evaluates five non-verbal signs of pain: facial expression, restlessness, muscle tone, vocalization, and consolability. Each sign is scored from 0 to 2, with a total score of 0 indicating no evidence of pain, 1–3 mild pain, 4–6 moderate pain, and 7–10 severe pain.

We also tabulated medications administered during ECMO for analgesia, sedation, and paralysis, including clonazepam, lorazepam, midazolam, fentanyl, hydromorphone, morphine, oxycodone, propofol, ketamine, dexmedetomidine, clonidine, quetiapine, risperidone, haloperidol, cisatracurium, rocuronium, and vecuronium. Using the concentration, dosage, time of administration, duration of administration, and mode of administration (intravenous [IV] infusion, IV bolus from bag, oral [PO] tablets, and PO suspension), the average daily dose of medication received during VA-ECMO was calculated. To obtain the average daily dose, the total amount of medication received during cannulation was divided by the number of days on ECMO. To adjust for patient weight, the average daily dose was divided by the patient’s weight in kilograms. The weight-based and non-weight-based averages do not include zero values for those who did not receive the medication.

Outcomes

The primary outcomes were delirium prevalence and the proportion of delirium-present days on VA-ECMO support. Screening positive on CAM-ICU at least once classified the day as delirium-positive. Remaining days were either delirium-negative days or delirium-unassessable days. Secondary outcomes included daily weight-based and non-weight-based analgosedative doses, Vasopressor Dose Equivalent (VDE) score,¹⁴ and in-hospital outcomes. VDE score is calculated as norepinephrine equivalents = norepinephrine (µg/kg/min) + epinephrine (µg/kg/min) + phenylephrine/10 (µg/kg/min) + dopamine/100 (µg/kg/min) + metaraminol/8 (µg/kg/min) + vasopressin × 2.5 (units/min) + angiotensin II × 10 (µg/kg/min). Absolute hypoxemia was defined as partial pressure of arterial oxygen (PaO₂) of less than 70 mmHg. Serum lactate, PaO₂, and partial pressure of arterial carbon dioxide (PaCO₂) were obtained within 6 hours prior to VA-ECMO cannulation and within 24 hours post-cannulation. Acute Brain Injury (ABI) was defined as the occurrence of ischemic stroke, intracranial hemorrhage, subarachnoid hemorrhage, subdural hemorrhage, seizure, and/or brain death.

Statistical analysis

Statistical comparisons were performed between the patients with detected delirium at any time point on VA-ECMO and those without. Variables were reported as medians with interquartile range (IQR) for continuous variables and were compared with non-parametric Wilcoxon Rank-sum tests. Categorical variables were reported as count and percentage, with comparisons performed with Chi-squared analysis if cell sizes were 5 or more and with Fisher’s exact test if any cell was less than 5. Associations between patient characteristics, hospital outcomes, and proportion of days with delirium were calculated with Spearman’s rank correlation coefficients.

We conducted multivariable logistic regression to evaluate the association between delirium and clinical factors. Covariates were selected a priori, which included age, body mass index (BMI), PaO₂, PaCO₂, lactate, Sequential Organ Failure Assessment (SOFA) score, ABI, number of sedatives received, number of paralytics received, initiation of ECMO for post-cardiotomy shock, race/ethnicity, and renal replacement therapy (RRT). We performed Spearman’s correlation analyses to assess an association between the proportion of days on delirium with the following variables—RASS, BPAS, pre-cannulation lactate, SOFA, age, the Acute Physiology and Chronic Health Evaluation II (APACHE II), ECMO days, and hospital days. The final model was chosen via the lowest Akaike information criterion (AIC). Variables with high collinearity (VIF > 5) were removed from the mode. Adjusted odds ratios (aOR) were reported with 95% Confidence Intervals. The Cox proportional hazards model was used to evaluate the association between the number of delirium-positive days and mortality, adjusting for age, history of congestive heart failure, RASS, arterial oxygen and carbon dioxide saturation, lactate, SOFA, number of red blood cell (RBC) transfusions, post-cardiotomy shock ECMO indication, RRT, and ABI. Hazard ratios (HR) were reported. All statistical analyses were conducted using R statistical software version 4.3.0 with p < .05 as the a priori criterion for statistical significance.

Results

From the 279 adults on VA-ECMO, we identified 208 adults who met inclusion criteria, with a median age of 53 years (IQR [40–62]), and 66 (21.7%) were female. Baseline demographics and clinical variables for delirium-positive (n = 138, 66.4%) and delirium-negative (n = 70, 33.7%) are summarized in Table 1. The delirium group had greater proportions of African American (32% vs 26%, p = .0097) and Hispanic/Latino (13% vs 0%, p = .0097) and a lower proportion of White patients (47% vs 66%, p = .0097). Chronic cardiovascular diseases and obesity were common, with the delirium patients having a higher BMI (31.17 [25.6–35.9] vs 27.0 [23.7–31.1], p < .001) and a less common occurrence of congestive heart failure (16% vs 31%, p = .016). Patients who later developed delirium on ECMO had higher levels of PaCO₂ (53.00 mmHg [39.75–67.25] vs 41.00 mmHg [34.00–56.50], p < .001), and reduced levels of lactate (2.37 mmol/L [1.60–4.33], 3.71 mmol/L [2.45–6.52], p < .001) pre-ECMO. In the delirium group, absolute hypoxemia was more common post-cannulation on day 1 (17% vs 4%, p = .013).

Table 1.

Demographics and clinical variables.

Variable	Delirium, n = 138 (%)	No delirium, n = 70 (%)	p-value
Demographics
Age, years	52.00 (40.00, 60.00)	55.50 (42.75, 65.00)	0.12
Sex			0.83
Female, n (%)	45 (33%)	21 (30%)
Male, n (%)	93 (67%)	49 (70%)
BMI (kg/m²)	31.17 (25.60, 35.85)	27.00 (23.68, 31.11)	0.0003
Race and ethnicity			0.0097
African American, n (%)	44 (32%)	18 (26%)
Hispanic/Latino, n (%)	18 (13%)	0 (0%)
White, n (%)	63 (47%)	46 (66%)
Asian, n (%)	6 (4%)	3 (4%)
Past medical history
Hypertension, n (%)	74 (54%)	38 (54%)	1.00
Hyperlipidemia, n (%)	58 (42%)	29 (41%)	1.00
Diabetes, n (%)	32 (23%)	19 (27%)	0.64
Congestive heart failure, n (%)	22 (16%)	22 (31%)	0.016
Pre-cannulation laboratory variables (within 24 hours)
Hemoglobin (g/dL)	10.07 (8.55, 12.07)	10.88 (8.73, 13.35)	0.21
WBC (10³ cells/mL)	13.09 (9.36, 19.43)	11.99 (9.55, 14.71)	0.18
Platelet (10³ cells/mL)	192 (122.5, 284.0)	178.0 (129.5, 243.8)	0.56
Arterial blood gas
PaO₂ (mmHg)	87.5 (67.0, 166.5)	133.5 (74.75, 226.50)	0.068
Hypoxemia (PaO₂ < 70 mmHg), n (%)	39 (28%)	16 (23%)	0.54
PaCO₂ (mmHg)	53.00 (39.75, 67.25)	41.00 (34.00, 56.50)	0.0003
Lactate (mmol/L)	2.37 (1.60, 4.33)	3.71 (2.45, 6.52)	0.0006
pH	7.26 (7.21, 7.33)	7.29 (7.20, 7.34)	0.41
ECMO laboratory variables day 1
PaO₂ (mmHg)	110.90 (78.17, 238.12)	232.6 (137.2, 294.0)	<0.0001
Hypoxemia (PaO₂ < 70 mmHg), n (%)	23 (17%)	3 (4%)	0.013
PaCO₂ (mmHg)	43.88 (37.00, 53.61)	36.85 (34.41, 41.86)	<0.0001
Lactate (mmol/L)	2.70 (1.55, 4.35)	4.10 (1.90, 8.50)	0.071
pH	7.345 (7.29, 7.38)	7.36 (731, 7.41)	0.40
MAP	80.00 (71.45, 95.00)	76.67 (70.58, 87.80)	0.18
Creatinine	1.35 (0.93, 1.98)	1.40 (0.90, 1.70)	0.55
ALT (U/L)	47.5 (21.5, 289.5)	69.00 (30.25, 535.3)	0.42
AST (U/L)	126 (37, 407)	201 (95, 616)	0.14
Apache II score	22.0 (18.0, 26.0)	22.00 (18.00, 26.50)	0.73
SOFA score	11.00 (9.00, 14.00)	11.0 (10.0, 13.0)	0.4
VDE score	13.0 (7.0, 24.0)	5.0 (10.0, 30.0)	0.77
RRT, n (%)	30 (22%)	19 (27%)	0.42
Number of medications received
Total analgosedatives received	6 (5, 8)	5 (4, 6)	0.0065
Day 1 sedatives	4 (3, 5)	4 (3, 5)	0.64
Day 3 sedatives	3 (2, 4)	2 (2, 3)	0.021
Day 5 sedatives	3 (2, 4)	2 (2, 3)	0.14
Paralytics	1 (1, 2)	1 (1, 1)	0.005
Outcomes
ECMO indication			0.07
ECPR, n (%)	21 (16%)	7 (10%)
Post-cardiotomy, n (%)	37 (26%)	12 (12%)
Cardiogenic shock, n (%)	80 (58%)	54 (78%)
Cannulation strategy
Central cannulation, n (%)	109 (79%)	52 (74%)	0.48
Complications
ABI, n (%)	33 (24%)	23 (33%)	0.34
Bloodstream infection, n (%)	38 (28%)	16 (23%)	0.46
Bleeding, n (%)	75 (54%)	39 (56%)	0.97
Thrombosis, n (%)	61 (44%)	33 (47%)	0.8
ECMO duration (days)	12.00 (7.00, 27.00)	5.00 (4.00, 9.75)	<0.0001
Hospital duration (days)	53.43 (18.87, 75.91)	22.02 (11.78, 53.86)	0.0016
In-hospital mortality	78 (57%)	36 (51%)	0.58

Abbreviations: BMI, body mass index; WBC, white blood cell; PaO₂, arterial oxygen saturation; PaCO₂, arterial carbon dioxide saturation; MAP, mean arterial pressure; ALT, alanine aminotransferase; AST, aspartate aminotransferase; APACHE II, the acute physiology and chronic health evaluation II; SOFA, the sequential organ failure assessment; VDE Score, vasoactive inotropic score; RRT, renal replacement therapy; ECPR, extracorporeal cardiopulmonary resuscitation; ECMO, extracorporeal membrane oxygenation; Number of Medications Received, total number of medications received during ECMO cannulation; ABI, acute brain injury.

Table 2 lists the clinical characteristics of delirium and the level of sedation while on VA-ECMO. The median delirium onset was 2.5 days ([1.0 7.0]) from ECMO cannulation, the median day of delirium occurred on day 6.25 of ECMO cannulation, and the median proportion of total delirium days was 0.42 ([0.20–0.66]). Delirium patients had higher BPAS scores than non-delirium patients (0.43 [0.14, 1.09] vs 0.17 [0.00–0.72], p = .0018) but comparable RASS scores (−2.95 [−3.9 to −1.93] vs −3.2 [−4.19 to −2.72], p = .98). Delirium patients spent more days in a coma than non-delirium patients (1.00 [1.00–2.00] vs 0 [0.00–1.00], p < .01), but had similar total days free from delirium and coma (6.00 [1.25–16.75] vs 5.00 [3.00–9.00], p = .36).

Table 2.

Delirium characteristics.

Variable	Delirium, n-138 (%)	No delirium, n = 70 (%)	p-value
BPAS	0.43 (0.14, 1.09)	0.17 (0.00, 0.72)	0.0018
RASS	−2,95 (−3.9, −1.93)	−3.2 (−4.19, −2.72)	0.98
Delirium onset since ECMO cannulation^, (days)	2.5 (1.00, 7.00)	N/A	N/A
Median delirium days (days)	6.25 (2.50, 18.38)	N/A	N/A
Total delirium days (days)	4.00 (2.00, 10.75)	N/A	N/A
Proportion of total delirium days	0.42 (0.20, 0.66)	N/A	N/A
Total coma days	1.00 (1.00, 2.00)	0.00 (0.00, 1.00)	<0.0001
Total days free from delirium and coma	6.00 (1.25, 16.75)	5.00 (3.00, 9.00)	0.36
ECMO duration (days)	12.00 (7.00, 27.00)	5.00 (4.00, 9.75)	<0.0001

Abbreviations: BPAS, behavioral pain assessment scale; RASS, Richmond Agitation Sedation Scale; ECMO, extracorporeal membrane oxygenation.

Patients with delirium, overall, received significantly greater numbers of analgosedatives (6 [5, 8] vs 5 [4-6], p = .0065) and paralytics (1 [1-2] vs 1 [1-1], p = .005). The difference in the diversification for analgosedatives was most pronounced on day 3 of ECMO (3 medications [2, 4] vs 2 [2-3], p = .021) during the first 5 days. Patients with delirium required more prolonged ECMO therapy (12.00 days [7.00–27.00] vs 5.00 [4.00–9.75], p < .001) and hospitalization (35.30 [18.87–75.91] vs 22.02 [11.78–53.86], p = .0016), but had similar events of ABI, hemorrhage, thrombosis, bloodstream infection, and mortality. In the Kaplan-Meier analysis (Figure 1), there was no association between delirium occurrence and 30-day mortality (p = .24).

According to Spearman correlation analysis (Supplemental Table 1), the proportion of delirium-present days on VA-ECMO was weakly correlated with the RASS score (r = 0.31, p < .001) and BPAS score (r = 0.24, p = .004), showing a possible linkage of delirium to lighter sedation and higher pain level. Supplemental Tables 2 and 3 list the proportion of patients who received analgosedatives, antipsychotics, and paralytic medications, as well as the body weight-adjusted and non-weight-adjusted mean daily dosages in each group. Midazolam, rocuronium, vecuronium, and propofol were used more frequently in non-delirium patients. Ketamine use was similar in frequency, but its weight-based-dosage, was significantly higher in the delirium group (2.39 mg/kg/day [0.36–4.44] vs 0.25 [0.07–8.40], p < .001).

In the multivariable logistic regression analysis for delirium occurrence and the risk factors, higher PaCO₂ on day 1 of cannulation (aOR = 1.20, 95% CI = 1.03–1.73), greater use of numerous sedatives (aOR = 2.67, 95% CI = 1.68-1.95), and African American race/ethnicity (aOR = 16.45, 95% CI = 9.65–122.51) were significantly associated with delirium (Table 3). The other variables (e.g., PaO₂, lactate, SOFA, ABI, RASS, ECMO indication, and RRT) were not associated with delirium.

Table 3.

Multivariable logistic regression analysis for risk factors of delirium.

Variable	Odds (95% CI)	p-value
Age, years	1.03 (0.97–1.10)	0.41
BMI	1.07 (0.90–1.30)	0.45
ECMO day 1 PaO₂	0.99 (0.98–1.01)	0.45
ECMO day 1 PaCO₂	1.29 (1.03–1.73)	0.049
ECMO day 1 lactate	0.90 (0.67–1.18)	0.47
ECMO day 1 SOFA	1.46 (0.94–2.60)	0.13
ABI	0.99 (0.12–8.67)	0.99
RASS score	0.71 (0.26–1.56)	0.44
Number of sedatives	2.67 (1.68–1.95)	0.04
Number of paralytics	1.67 (0.17–1.95)	0.66
ECMO indication: Post-cardiotomy shock	2.74 (0.13–6.89)	0.51
African American race/ethnicity	16.45 (9.65–122.51)	0.004
Renal replacement therapy	0.70 (0.07–5.19)	0.73

Abbreviations: PaO₂, arterial oxygen saturation; PaCO₂, arterial carbon dioxide saturation; SOFA, the sequential organ failure assessment; ABI, acute brain injury; RASS, Richmond Agitation Sedation Scale. Reported as OR (95% CI).

Furthermore, Cox regression analysis (Table 4) showed that post-cardiotomy shock as VA-ECMO indication (HR = 13.59, 95% CI = 1.24–24.94) and ABI (HR = 1.61, 95% CI = 1.15–17.98) were independent risk factors of in-hospital mortality, whereas the total number of delirium days was not (HR = 0.85, 95% CI = 0.58–1.24) after adjusting for age, history of congestive heart failure, RASS, arterial oxygen and carbon dioxide saturation, lactate, SOFA, number of RBC transfusions, post-cardiotomy shock as an indication for VA-ECMO, RRT, and ABI.

Table 4.

Cox proportional hazards for risk factors for mortality.

Variable	Hazard ratio (95% CI)	p-value
Total delirium days	0.85 (0.58–1.24)	0.29
Congestive heart failure	1.22 (0.05–29.83)	0.05
Age	1.04 (0.96–1.09)	0.14
RASS score	0.24 (0.06–1.26)	0.18
Day 1 PaO₂	0.98 (0.97–1.00)	0.48
Day 1 PaCO₂	1.23 (0.91–1.66)	0.42
ECMO day 1 lactate	1.01 (0.43–2.37)	0.76
ECMO day 1 SOFA	1.12 (0.66–1.89)	0.46
Number of transfusions	0.99 (0.93–1.08)	0.49
ECMO indication: Post-cardiotomy shock	13.59 (1.24–24.94)	0.044
Renal replacement therapy	0.31 (0.01–6.23)	0.087
ABI	1.61 (1.15–17.98)	0.006

Discussion

Delirium is common in ICU patients and associated with poor short- and long-term outcomes. Our study surveyed a cohort of patients on VA-ECMO with detailed clinical review and granular data, providing more reliable information than previous studies with a national administrative coding database¹⁵ or a smaller-scale pilot study.⁶ There was a recent study on VA-ECMO only for cardiac transplantation bridging,⁸ whereas the current study is inclusive of VA-ECMO patients with various indications. In our study cohort, delirium was present in about two-thirds, comparable to recently published studies on VA-ECMO.^6,8 Our previous study on COVID-19 severe acute respiratory distress syndrome (ARDS) patients with assessable neurological exams had a higher delirium prevalence of 97.5%, likely because of lengthy ECMO cannulation of greater than 30 days and a deeper level of sedation with RASS of −4 among non-survivors.⁷ Nevertheless, the current VA-ECMO patients were critically ill according to the high SOFA, APACHE II, VDE scores, and in-hospital mortality rate of 55%. Patients who screened positive for delirium underwent more extended ECMO therapy and hospitalization, but the level of sedation did not differ between the groups with equal RASS of −3 (moderate sedation). Delirium did not predict 30-day mortality according to the Kaplan Meier survival analysis, and neither did the number of delirium days predict in-hospital mortality in the Cox Proportional Hazard analysis. With insufficient information on mortality and delirium and the study’s retrospective nature, we cannot confirm whether the study was adequately powered. Nevertheless, our univariate analysis demonstrated that patients with delirium had a 6%, non-significant, greater chance of mortality. We also re-affirmed the relationship between ABI and mortality in VA-ECMO patients, as found in our VV-ECMO study with 5 of 26 non-survivors dying from fatal intracerebral hemorrhage.⁷

According to a multivariable analysis, arterial carbon dioxide level on day 1 of ECMO was a risk factor of delirium occurrence, but its mechanistic role is difficult to elucidate. Our univariate analysis indicated that patients with delirium had a significantly more elevated PaCO₂ both pre-ECMO cannulation (median PaCO₂ = 53.00 mmHg) and on day 1 of ECMO (median PaCO₂ = 43.88 mmHg) compared to non-delirium counterparts. Interestingly, the delirium group also suffered from absolute hypoxemia more than four-fold (Table 1. This relationship suggests a potential, synergistic effect of hypoxemia and hypercapnia in delirium emergence. In a rodent model with hypoxic-ischemic brain injury, hypercapnia brought about a decline in mean arterial pressure and disrupted cerebral perfusion and blood-brain barrier permeability only in severe systemic hypoxia.¹⁶ Both intraprocedural sub-narcosis hypercapnia^17,18 and hypocapnia¹⁹ are associated with post-procedural delirium, but the role of carbon dioxide level and its modulation in delirium prevention remains unclear. In addition, hypercapnia can signal a clinical state of vulnerability to delirium - hypermetabolic state,²⁰ compensation for an ongoing metabolic alkalosis, alveolar hypoventilation (e.g., oversedation),²¹ permissive hypercapnia in ARDS,²² or other causes. Even mild hypercapnia can potentiate neurological injury by worsening cerebral edema,²³ but ABI was not associated with delirium in our study. Targeted mild hypercapnia with PaCO₂ of 50–55 mmHg in the first 24 hours did not bring about improved neurologic outcomes post-cardiac arrest,²⁴ and abruptly lowering PaCO₂ over the first 24 hours led to increased mortality in the ELSO Registry.¹⁰ Collectively, prompt correction of hypercapnia without overshooting unless contraindicated (e.g., ARDS), avoidance of absolute hypoxemia, and addressing underlying medical issues that contribute to hypercarbia should be the goals in treating these patients.

The characteristics of delirium and analgosedation practice shed light on potential strategies to prevent its occurrence. The median delirium onset of 2.5 days post-cannulation reflects the common practice of heavy sedation for medical stabilization. The comparable RASS scores, analgosedation dosage used, and minimally discrepant coma days suggest an adequate level of sedation in both groups. One interesting finding is that patients who developed delirium received significantly greater numbers of analgosedatives and paralytics, and the analgosedative use was particularly more diversified on day 3, but not on days 1 and 5. This may indicate a reactive effort to control delirium, which was usually detected first between days 2 and 3. In a multicenter, retrospective study, heavy sedation was predictive of delayed extubation and increased 180-day mortality but was not associated with delirium.²⁵ This might indicate that the promotion of light sedation, as opposed to suppressing delirium with analgosedation, could potentially avoid extended ECMO support and hospitalization.²⁶ Analgosedation is an essential component in ECMO,¹¹ but its overuse comes with the cost of ineffective neurological assessment, delayed extubation, and increased mortality.^25,27 Further studies are needed to optimize strategies to avoid unnecessary sedation (e.g., awake VA-ECMO,²⁸ use of hydromorphone vs fentanyl²⁹). Surprisingly, African American race/ethnicity was a risk factor for delirium, which may have arisen from a delay in receiving time-sensitive cardiac care,³⁰ which can lead to complications.³¹

Contrary to our study hypothesis, the presence of delirium and total delirium days were not associated with mortality. Although the condition’s association with mortality has been well delineated in the ICU,^1–3 understanding the role of delirium in ECMO outcomes is challenging due to convoluted interactions between sedation, pain, and delirium.⁷ In a previous study by our group on VV-ECMO for COVID-19 ARDS, the proportion of delirium-present days did not predict mortality, and we demonstrated a trend of lower RASS scores closer to −4 in non-survivors, highlighting their more complicated medical course or iatrogenic oversedation. In this study, we reached the same conclusion of the neutral effect of delirium on mortality, with ABI and post-cardiotomy indication as predictors of mortality. Furthermore, a study on eight thousand adult ECMO survivors in South Korea found no significant association between delirium and 1-year all-cause mortality post-hospital discharge.¹⁵ With this information at hand, an appropriate approach would be to focus on the implementation of standardized bedside neuromonitoring protocols³²,³³ and practical approaches for unstable patients (e.g., bedside portable brain MRI³⁴) into ECMO care, ultimately to facilitate early detection and management of ABIs known to be directly linked to mortality.³⁵

Our study results should be interpreted with caution. First, its retrospective, observational design limits our ability to draw causal relationships. Second, our selection of adjusting factors a priori for analyses makes our results subject to unmeasured confounding effects. Third, our medication review was limited to commonly used PO/IV analgosedatives, and information on other delirio-genic medications (e.g., cefepime, antiseizure medications, antidepressants) was not included in our study. In addition, our data do not include information on prior anticoagulant use, history of coagulopathic diseases, or laboratory markers of coagulopathy while on VA-ECMO, which could impact the development of ABI. Fourth, we only included VA-ECMO patients with assessable neurological examinations at some time points on ECMO, so our findings are not generalizable to those who were comatose for the entire duration of ECMO support, or those on VV-ECMO or hybrid ECMO. Fifth, we did not include factors such as total hospital stay, ICU stay, sedatives pre-ECMO, cerebrovascular accidents pre-ECMO, Frailty score, other types of heart failure, or ejection fraction as possible contributing factors to pre-ECMO frailty whch can impact delirium development, Nevertheless, our study’s strength includes a robust sample size at an experienced tertiary care ECMO center with consistencies in analgosedation and paralytic strategies, as well as standardized neuromonitoring. Our multivariable analyses with appropriate adjustments, including illness severity, metabolic derangements, and in-hospital complications, on a large cohort reveal mechanistically plausible factors associated with delirium and mortality, illuminating potential interventions to optimize delirium prevention and outcomes in this high-risk group.

Conclusions

Delirium was common in VA-ECMO patients, present in approximately two-thirds of patients. Gradual correction of hypercapnia and avoidance of polypharmacy in analgosedation may be effective strategies to prevent delirium. Although delirium and its duration were not independently associated with mortality, it predicted protracted courses of ECMO and hospitalization. Future studies should investigate analgosedation practices to better optimize outcomes in VA-ECMO patients. Finally, our current study highlights the importance of early correction of hypercapnia and hypoxemia, promotion of light sedation, and detection and treatment of ABIs.

Supplemental Material

Supplemental Material - Characteristics and risk factors of delirium in patients on veno-arterial extracorporeal membrane oxygenation

Supplemental Material for Characteristics and risk factors of delirium in patients on veno-arterial extracorporeal membrane oxygenation by Olivia Liu, Philip Y. Sun, Syed A. Ahmad, Andrew Kalra, Amy Feng, Glenn J. R. Whitman, Bo Soo Kim, Sung-Min Cho, and On behalf of the HERALD Investigators in Perfusion

Footnotes

Authors’ note

We confirm that the manuscript complies with all instructions to the authors. We confirm that authorship requirements have been met and that the final manuscript was approved by all authors. We confirm that this manuscript has not been published elsewhere and is not under consideration by another journal.

Author contributions

OL, PS: acquisition of data, analysis, and interpretation of data, and drafting and revising the article. SA, AK, AF: acquisition of data, revising the article. GW, BK, SC: conception and design, analysis and interpretation of data, and drafting and revising the article. All authors contributed to the manuscript and approved the submitted version.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Dr. Sung-Min Cho is supported by NIH (1K23HL157610).

Ethical statement

ORCID iDs

Olivia Liu

Bo Soo Kim

Supplemental Material

Supplemental material for this article is available online.

References

Han

McNaughton

Stubblefield

, et al. Delirium and its association with short-term outcomes in younger and older patients with acute heart failure. PLoS One 2022; 17(7): e0270889.

Ely

Inouye

Bernard

, et al. Delirium in mechanically ventilated patients: validity and reliability of the confusion assessment method for the intensive care unit (CAM-ICU). JAMA 2001; 286(21): 2703–2710.

Reznik

Margolis

Mahta

, et al. Impact of delirium on outcomes after intracerebral hemorrhage. Stroke 2022; 53(2): 505–513.

Salluh

Soares

Teles

, et al. Delirium epidemiology in critical care (DECCA): an international study. Crit Care 2010; 14(6): R210.

Shurtleff

Radosevich

Patanwala

. Comparison of ketamine- versus nonketamine-based sedation on delirium and coma in the intensive care unit. J Intensive Care Med 2020; 35(6): 536–541.

Krupa

Friganovic

Mędrzycka-Dąbrowska

. Occurrence of delirium during ECMO therapy in a critical care unit in Poland—a cross-sectional pilot study. Int J Environ Res Publ Health 2021; 18(8): 4029.

Sun

PYW

Fanning

Peeler

, et al. Characteristics of delirium and its association with sedation and in-hospital mortality in patients with COVID-19 on veno-venous extracorporeal membrane oxygenation. Front Med 2023; 10: 1172063.

Youn

Kim

Park

, et al. Clinical outcomes of early extubation strategy in patients undergoing extracorporeal membrane oxygenation as a bridge to heart transplantation. J Kor Med Sci 2020; 35(42): e346.

Pandharipande

Cotton

Shintani

, et al. Prevalence and risk factors for development of delirium in surgical and trauma intensive care unit patients. J Trauma 2008; 65(1): 34–41.

10.

Diehl

Burrell

AJC

Udy

, et al. Association between arterial carbon dioxide tension and clinical outcomes in venoarterial extracorporeal membrane oxygenation. Crit Care Med 2020; 48(7): 977–984.

11.

Crow

Lindsley

Cho

S-M

, et al. Analgosedation in critically ill adults receiving extracorporeal membrane oxygenation support. ASAIO J 2022; 68(12): 1419–1427.

12.

Cho

Farrokh

Whitman

, et al. Neurocritical care for extracorporeal membrane oxygenation patients. Crit Care Med 2019; 47(12): 1773–1781.

13.

Ely

Truman

Shintani

, et al. Monitoring sedation status over time in ICU patients: reliability and validity of the Richmond Agitation-Sedation Scale (RASS). JAMA 2003; 289(22): 2983–2991.

14.

Goradia

Sardaneh

Narayan

, et al. Vasopressor dose equivalence: a scoping review and suggested formula. J Crit Care 2021; 61: 233–240.

15.

Park

Song

I-A

. Delirium and long-term outcomes among survivors of extracorporeal membrane oxygenation therapy: a nationwide cohort study in South Korea. J Intensive Care Med 2022; 37(7): 870–876.

16.

Yang

Zhang

Wang

, et al. Effects of acute systemic hypoxia and hypercapnia on brain damage in a rat model of hypoxia-ischemia. PLoS One 2016; 11(12): e0167359.

17.

Wang

, et al. Risk factors contributing to postoperative delirium in geriatric patients postorthopedic surgery. Asia Pac Psychiatry 2015; 7(4): 375–382.

18.

Cheng

Yang

, et al. Moderate hypercapnia may not contribute to postoperative delirium in patients undergoing bronchoscopic intervention. Medicine (Baltim) 2019; 98(22): e15906.

19.

Ahrens

Tartler

Suleiman

, et al. Dose-dependent relationship between intra-procedural hypoxaemia or hypocapnia and postoperative delirium in older patients. Br J Anaesth 2023; 130(2): e298–e306.

20.

Rawat

Modi

Sharma

. Hypercapne. Treasure Island, FL: StatPearls, 2023.

21.

Piper

Yee

. Hypoventilation syndromes. Compr Physiol 2014; 4(4): 1639–1676.

22.

Acute Respiratory Distress Syndrome Network Brower

Matthay

Morris

, et al. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000; 342(18): 1301–1308.

23.

Perkins

Callaway

Haywood

, et al. Brain injury after cardiac arrest. Lancet 2021; 398(10307): 1269–1278.

24.

Eastwood

Nichol

Hodgson

, et al. Mild hypercapnia or normocapnia after out-of-hospital cardiac arrest. N Engl J Med 2023; 389(1): 45–57.

25.

Shehabi

Chan

Kadiman

, et al. Sedation depth and long-term mortality in mechanically ventilated critically ill adults: a prospective longitudinal multicentre cohort study. Intensive Care Med 2013; 39(5): 910–918.

26.

deBacker

Tamberg

Munshi

, et al. Sedation practice in extracorporeal membrane oxygenation-treated patients with acute respiratory distress syndrome: a retrospective study. ASAIO J 2018; 64(4): 544–551.

27.

Shehabi

Bellomo

Reade

, et al. Early intensive care sedation predicts long-term mortality in ventilated critically ill patients. Am J Respir Crit Care Med 2012; 186(8): 724–731.

28.

Montero

Huang

Rivas-Lasarte

, et al. Awake venoarterial extracorporeal membrane oxygenation for refractory cardiogenic shock. Eur Heart J Acute Cardiovasc Care 2021; 10(6): 585–594.

29.

Martin

Peitz

Olsen

, et al. Hydromorphone compared to fentanyl in patients receiving extracorporeal membrane oxygenation. ASAIO J 2021; 67(4): 443–448.

30.

Bradley

Herrin

Wang

, et al. Racial and ethnic differences in time to acute reperfusion therapy for patients hospitalized with myocardial infarction. JAMA 2004; 292(13): 1563–1572.

31.

Glynn

Lloyd-Jones

Feinstein

, et al. Disparities in cardiovascular mortality related to heart failure in the United States. J Am Coll Cardiol 2019; 73(18): 2354–2355.

32.

Marra

Ely

Pandharipande

, et al. The ABCDEF bundle in critical care. Crit Care Clin 2017; 33(2): 225–243.

33.

Cho

S-M

Ziai

Mayasi

, et al. Noninvasive neurological monitoring in extracorporeal membrane oxygenation. ASAIO J 2020; 66(4): 388–393.

34.

Cho

Wilcox

Keller

, et al. Assessing the SAfety and FEasibility of bedside portable low-field brain Magnetic Resonance Imaging in patients on ECMO (SAFE-MRI ECMO study): study protocol and first case series experience. Crit Care 2022; 26(1): 119.

35.

Cho

Canner

Chiarini

, et al. Modifiable risk factors and mortality from ischemic and hemorrhagic strokes in patients receiving venoarterial extracorporeal membrane oxygenation: results from the extracorporeal Life support organization registry. Crit Care Med 2020; 48(10): e897–e905.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.41 MB