Delirium Associated with COVID-19 in Critically ill Children: An Observational Cohort Study

Abstract

Objective

Delirium is an under-recognized problem in critically ill children. Although delirium is common in adults hospitalized with COVID-19, the relationship between pediatric COVID-19 and delirium has not been described. To address this gap, we characterized delirium in critically ill children with different manifestations of COVID-19 and investigated associations among demographic, disease, and treatment factors. We hypothesized that multisystem inflammatory syndrome in children (MIS-C) would be associated with a higher incidence of delirium given its underlying pathophysiology of hyperinflammation.

Design

Retrospective, single-center cohort study.

Setting

Quaternary-care pediatric intensive care unit (PICU).

Patients

Children less than 18 years of age hospitalized in the PICU between March 2020 and March 2023 with either active SARS-CoV-2 infection or serological evidence of prior infection.

Measurements and main results

The cohort included 149 PICU hospitalizations among children with evidence of COVID-19. Patients were categorized by reason for PICU admission: 75 (50%) for COVID-19 respiratory disease, 36 (24%) MIS-C, and 38 (26%) any other primary reason with positive COVID-19 testing. Delirium was diagnosed in 43 (29%) patients. Delirium incidence was highest in patients requiring invasive mechanical ventilation (IMV) (56% vs 7.5% in patients who did not require IMV, p < .001). Patients who were exposed to opioids, dexmedetomidine, paralytics or benzodiazepines more frequently experienced delirium compared to those unexposed (p < .001, p < .001, p < .001 and p = .001, respectively). After multivariable adjustment, delirium was associated with IMV (HR 3 [95% CI 1.5–5.7]), female sex (HR 2.4 [1.2–4.7]), and developmental disability (HR 3.4 [95% CI 1–11.1]). There was no association between delirium and reason for PICU hospitalization.

Conclusions

Delirium was common among children hospitalized with COVID-19. The overall incidence was much less than has been reported in adults with COVID-19. Delirium reduction efforts should focus on children with developmental disability and minimizing ongoing risks during IMV.

Keywords

delirium COVID-19 intensive care units pediatric critical illness invasive mechanical ventilation

Introduction

Delirium, a form of acute and reversible encephalopathy, is an under-recognized and under-treated problem in critically ill children, with reported incidences ranging from 17–44%.^1‐3 There are many proposed mechanisms underlying the pathogenesis of delirium, including inflammation, hypoxemia, impaired tissue perfusion, and impaired cellular metabolism.⁴ In addition to causing significant distress to patients and families, pediatric delirium has been strongly associated with both in- and out-of-hospital adverse outcomes, such as mortality,² prolonged pediatric intensive care unit (PICU) length of stay,^1,3 duration of invasive mechanical ventilation (IMV),^1,3 and worse quality of life after hospital discharge.⁵ While delirium has not been firmly associated with long-term cognitive problems in children, over half of critically ill adults experiencing delirium during a hospitalization suffer persistent cognitive impairment 12 months after hospital discharge.⁶

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which causes coronavirus disease 2019 (COVID-19), is associated with both pulmonary and extra-pulmonary disease. Neurologic manifestations of COVID-19 have been observed in up to 85% of infected adults^7,8 and delirium specifically has been observed in up to 73% of adults with COVID-19.^9,10 Neurologic manifestations, including acute encephalopathy, have been reported in 22% of pediatric patients with either acute COVID-19 or the multisystem inflammatory syndrome in children (MIS-C).^11‐13

Although delirium is common in both critically ill adults and children, and delirium is extremely common in adults hospitalized with COVID-19,¹⁴ the relationship between pediatric COVID-19 and delirium has not been studied. To address this gap, we performed a single-center, observational cohort study of children hospitalized in the PICU with illness related to COVID-19. Our goals were (1) to characterize delirium in critically ill children with different manifestations of COVID-19, and (2) to investigate associations between delirium and demographic, disease, and treatment factors related to COVID-19. Given the hyperinflammatory nature of MIS-C,¹⁵ we hypothesized that MIS-C would be associated with a higher incidence of delirium when compared with other COVID-19 disease types.

Materials and Methods

Study Setting and Data Source

New York-Presbyterian Morgan Stanley Children's Hospital is a quaternary care referral center as well as the local hospital for its urban surrounding community. The hospital has 41 total PICU beds separated into three units that respectively serve general medical/surgical, cardiothoracic surgical, and neurological/neurosurgical patients. This was a retrospective, observational cohort study. All data were extracted from the electronic health record (EHR) and spanned from March 1, 2020 to March 13, 2023. The study was approved by the Columbia University Human Research Protection Institutional Review Board with a waiver of informed consent, and followed the STROBE guidelines.¹⁶

Inclusion and Exclusion Criteria

Patients were screened for inclusion if they were less than 18 years of age upon hospital admission and stayed in the PICU for $\geq$ 24 consecutive hours at any time during the hospitalization, with positive COVID-19 testing via either reverse transcription-polymerase chain reaction (RT-PCR) or COVID-19 serologic testing. COVID-19 testing was required to be positive during the hospitalization at any time before or during the patient's PICU stay. To ensure that only characteristics of the hospitalization potentially relevant to the COVID-19 diagnosis were assessed, an “infection window” was created for each included patient spanning from 14 days prior to the first eligible positive COVID-19 test until the end of the PICU stay. Fourteen days were chosen to account for potential effects of COVID-19 during the viral incubation period.¹⁷ All data, including demographics, exposures, and outcomes, were only used if they fell within this patient-specific infection window.

Patients who met inclusion criteria were reviewed by two authors (MG and ASG) to ensure that COVID-19 was plausibly relevant to the PICU stay. Patients who were only incidentally COVID-19 positive (eg, positive RT-PCR screening test during a PICU stay for a skull fracture after a fall) were excluded from analysis. All remaining patients formed the final study cohort.

In this study, “hospitalization” refers to the period from hospital admission to discharge. “PICU stay” refers to the entirety of a single PICU course, from PICU admission until PICU discharge (to a general pediatric unit, home, etc). In order to avoid repeated measurements in patients with multiple PICU stays, only the first stay was included in analysis.

Reason for PICU Stay

Patients were categorized into three subgroups based upon reason for PICU stay: COVID-19 respiratory disease, MIS-C, or other primary illness. Categorization in the “other primary illness” subgroup implies that COVID-19 was plausibly related to the PICU stay but was not the primary cause of illness (eg, a patient admitted for cardiac surgery who had a prolonged intubation with a positive COVID-19 test). Two investigators (MG and ASG) reviewed all patients to resolve discrepancies.

Outcomes

The primary outcome was the incidence of delirium diagnosed via Cornell Assessment of Pediatric Delirium (CAPD) score greater than 8.¹⁸ CAPD scoring was performed by bedside nurses at least once per 12-h shift. Secondary outcomes included time until delirium onset, days delirious, days comatose, and subtypes of delirium (hypoactive, hyperactive, and mixed). Subtypes of delirium were classified according to Richmond Agitation Sedation Scale (RASS),¹⁹ which was also scored at least once per shift. Hyperactive delirium was defined as positive CAPD screen with RASS >0 for an entire calendar day; hypoactive delirium with RASS ≤0 for an entire calendar day; and mixed delirium with RASS crossing zero over the course of a calendar day. Comatose patients (RASS −4) were unable to be screened for delirium. A day in which a patient was either delirious or comatose for any portion of that day was counted as a day with acute brain dysfunction.

Patient, Illness, and Treatment Characteristics

Demographics, medical history, admission diagnoses, laboratory values, and treatments were obtained from the EHR. Children with complex chronic conditions (CCC) were identified via International Classification of Disease, 10th Edition (ICD-10) codes per Pediatric CCC Classification System.²⁰ Risk of mortality was reported using the Pediatric Index of Mortality, third edition (PIM3) score.²¹ The full list of medications investigated for associations with delirium can be found in the Supplemental Data. Cumulative doses were calculated by summing bolus doses with the area under the curve for continuous infusions. Benzodiazepines were converted to midazolam equivalents and opioids to morphine equivalents using standard conversions²² (Supplemental Table 2) and are reported in weight-based dosing. Medication doses administered outside the PICU, such as in the operating room, were excluded.

Statistical Analysis

Patient characteristics and outcomes are reported as counts and percentages or medians and interquartile ranges (IQR) and were compared across subgroups of COVID-19. The incidence of delirium was calculated per PICU stay, and the incidence rate per patient-day in the PICU. The incidence of delirium was compared between patients with versus without each characteristic. Categorical variables were compared using chi-squared or Fisher exact tests, and continuous variables using Wilcoxon rank-sum tests. Risk ratios (RR) for delirium and 95% confidence intervals (CIs) are reported.

Critical illness itself has been associated with developing delirium.²³ To account for ongoing risk of delirium during critical illness we fit a survival model to the data, modeling an event as the diagnosis of delirium. Survival time was the duration of time a patient spent in the PICU until becoming delirious. Kaplan–Meier curves stratified by potential risk factors were compared via log-rank tests. The association of type of COVID-19 disease with delirium was estimated via a Cox proportional hazards model. Hazard ratios (HR) and 95% CIs were estimated, and the model fit was evaluated by Wald test.²⁴ Covariates with p < .2 in bivariate analysis were included in the model to adjust for other potential delirium risk factors.

Sensitivity Analysis

Medication commonly used to facilitate IMV (benzodiazepines, opioids, dexmedetomidine and paralytics) were not included in the model described above due to concerns about overfitting with a small number of events. To investigate whether these IMV-related medication doses might also have had independent associations with delirium, we estimated a second model which included four additional covariates for whether the patient received a medication within each class prior to developing delirium. We then tested whether this more extensive model was better able to predict delirium via a likelihood ratio test.

Our PICU protocol is to check CAPD scores at least once per shift. However, delirium cannot be assessed in patients who are comatose or under neuromuscular blockade, and nurses might unintentionally skip assessments during busy times. To account for bias from irregular measurement, we estimated delirium onset using an interval-censored survival model. Delirium was assumed to occur any time between a positive CAPD and its preceding negative score. Times with RASS ≤ −4 were included in the “unmeasured” interval (Supplemental Figure 1). Survival curves stratified by individual covariates were compared via permutation test, with p-values estimated by 1000 Monte Carlo simulations. A multivariable proportional hazards model estimated independent associations of the same covariates, with standard errors estimated via 1000 bootstrap samples.

All data were analyzed using R software.²⁵ Survival analysis and data visualization were performed using the survival,²⁶ ggplot2,²⁷ and survminer²⁸ packages. Interval-censored analyses were performed using interval²⁹ and icenReg³⁰ packages.

Results

Overall Cohort

The cohort included 149 PICU stays among children with COVID-19 (Table 1). COVID-19 was most frequently detected by a positive RT-PCR test (82%), either alone or in combination with a serology test. Twenty-three percent of patients had CCCs and 9% had developmental disability. The median PICU length of stay was 4 days (IQR 2–9). Nine (6%) children died and 113 (76%) patients were transferred to the floor before hospital discharge.

Table 1.

Patient Characteristics.

		Reason for PICU hospitalization
Characteristics	Overall, N = 149	COVID-19 respiratory disease, N = 75 (50%)	MIS-C, N = 36 (24%)	Other primary problem, N = 38 (26%)	p-value
Age (years), median (IQR)	1.0 (0.3–7.0)	0.6 (0.1–1.4)	7.7 (63.4–13.1)	0.9 (0.3–1.4)	<0.001
Female sex	69 (46%)	40 (53%)	13 (36%)	16 (42%)	0.2
Developmental disability	13 (8.7%)	10 (13%)	2 (5.6%)	1 (2.6%)	0.12
Complex chronic condition	35 (23%)	10 (13%)	2 (5.6%)	14 (37%)	0.006
Positive COVID test type					<0.001
Antibody only	26 (17%)	1 (1.3%)	21 (58%)	4 (11%)
PCR + antibody	21 (14%)	4 (5.3%)	12 (33%)	5 (13%)
PCR only	102 (68%)	70 (93%)	4 (8.3%)	29 (76%)
Primary diagnosis					<0.001
Cardiac	16 (11%)	3 (4%)	3 (8.3%)	10 (26%)
GI/renal/metabolic	10 (6.7%)	1 (1.3%)	0 (0%)	9 (24%)
Hematologic or oncologic	2 (1.3%)	1 (1.3%)	0 (0%)	1 (2.6%)
Infectious/inflammatory	38 (26%)	0 (0%)	33 (92%)	5 (13%)
Neurologic	10 (6.7%)	1 (1.3%)	0 (0%)	9 (24%)
Respiratory	69 (46%)	69 (92%)	0 (0%)	0 (0%)
Endocrine	4 (2.7%)	0 (0%)	0 (0%)	4 (11%)
PIM3 risk of mortality %, median (IQR)	0.06 (0.02–3.90)	0.04 (0.01–1.55)	4.15 (1.28–4.65)	0.04 (0.01–1.28)	<0.001
Invasive mechanical ventilation	37 (25%)	20 (27%)	3 (8.3%)	14 (37%)	0.016
IMV duration (days), median (IQR)	9 (4–21)	9 (4–16)	6 (5–10)	12 (4–23)	0.6
PICU disposition^a					0.002
Death	6 (4%)	5 (6.7%)	0 (0%)	1 (2.6%)
Non-PICU inpatient unit	113 (76%)	49 (65%)	29 (81%)	35 (92%)
Home	24 (16%)	17 (23%)	7 (19%)	0 (0%)
Rehab	6 (4%)	4 (5.3%)	0 (0%)	2 (5.3%)
Hospital disposition					0.13
Death	9 (6%)	7 (9.3%)	0 (0%)	2 (5.3%)
Home	121 (81%)	58 (77%)	34 (94%)	29 (76%)
Rehab	19 (13%)	10 (13%)	2 (5.6%)	7 (18%)
Hospital LOS (days), median (IQR)	8 (5–21)	10 (5–20)	6 (5–10)	14 (5–39)	0.023
PICU LOS (days), median (IQR)	4 (2–9)	4 (2–11)	3 (2–6)	4 (1–17)	0.8

Abbreviations: COVID-19, coronavirus disease 2019; IMV, invasive mechanical ventilation; LOS, length of stay; MIS-C, multisystem inflammatory syndrome in children; PIM3, pediatric index of mortality third edition; PICU, pediatric intensive care unit; RT-PCR, reverse transcription polymerase chain reaction; SARS-CoV-2, severe acute respiratory syndrome coronavirus 2.

Patients were grouped by reason for PICU hospitalization: COVID-19 respiratory disease, MIS-C, or other primary problem. All patients tested positive either via SARS-CoV-2 RT-PCR nasal swab, or serum COVID-19 antibody testing. Comparisons among groups were made using Kruskal–Wallis, Chi-squared, or Fisher exact tests.

Six patients died within first PICU admission. Three additional patients in the cohort died either outside the PICU or on a subsequent PICU admission.

Subgroups of COVID-19

Among 149 PICU stays, 75 (50%) were due to COVID-19 respiratory illness, 36 (24%) MIS-C, and 38 (26%) other primary illness. The MIS-C group was older than respiratory and other illness subgroups (7.7 years [IQR 3.4–13.1] vs 0.6 [IQR 0.1–1.4] and 0.9 [IQR 0.3–1.4], p < .001). There were no differences in biological sex among subgroups. The MIS-C subgroup had the greatest risk of mortality (median 4.15% [IQR 1.28–4.65%] vs 0.04 [IQR 0.01–1.55] and 0.04 [IQR 0.01–1.28], p < .001) despite having fewer CCCs (5.6% vs 13% and 37%, p < .01). Despite this, no patients with MIS-C died. Few patients with MIS-C required IMV (8.3%) compared to those with respiratory (27%) or other primary diagnoses (37%) (p = .016). Glucocorticoids were used in 65%, 97%, and 39% of patients with respiratory COVID-19, MIS-C, and other primary illness groups, respectively (p < .001). Of the 9 patient deaths, seven had respiratory COVID-19 and two had other primary causes of hospitalization. All children who died of COVID-19 respiratory disease had significant underlying comorbidity (Supplemental Table 3).

Incidence of Delirium

Patients in this cohort spent 1747 total days in the PICU. Of these days, 19% (333/1747) involved a coma and 62% of non-comatose days (870/1414) involved delirium. Patients most commonly experienced hypoactive delirium, which accounted for 73% of delirious days (Supplemental Table 4). The overall per-patient incidence of delirium was 29% (43/149 PICU stays). The incidence rate for delirium was 0.071 (95% CI 0.051–0.095) per patient-day, akin to 1 newly delirious patient per 14 days in the PICU. Delirium lasted a median of 5 days (IQR 2–11) (Table 2). The incidence rate was lowest for MIS-C at 0.037 (95% CI 0.012–0.097) per 100 patient days, compared to 0.097 (95% CI 0.063–0.14) for respiratory and 0.058 (95% CI 0.03–0.1) for other primary illnesses, respectively.

Table 2.

Presence of Delirium and Coma.

		Reason for PICU hospitalization
Characteristic	Overall, N = 149	COVID-19 respiratory disease, N = 75 (50%)	MIS-C, N = 36 (24%)	Other primary problem, N = 38 (26%)	p-value
Any delirium	43 (29%)	26 (35%)	5 (14%)	12 (32%)	0.071
Duration of delirium (days), median (IQR)	5 (2–11)	5 (2–11)	2 (1–6)	5 (2–13)	0.5
Any coma	24 (16%)	15 (20%)	2 (5.6%)	7 (18%)	0.014
Coma before delirium	19 (13%)	13 (17%)	1 (2.8)	5 (13%)	0.089
Duration of coma (days), median (IQR)	4 (2–7)	3 (2–8)	4 (4–4)	5 (2–12)	>0.9
Any brain dysfunction	47 (35%)	28 (40%)	5 (17%)	14 (41%)	0.056
Duration of brain dysfunction (days), median (IQR)	6 (2–16)	7 (3–16)	2 (1–9)	10 (2–19)	0.4

Abbreviations: IQR, interquartile range; MIS-C, multisystem inflammatory syndrome in children.

Presence of delirium and coma. Patients admitted to the PICU were scored at least once per shift for delirium using the Cornell Assessment of Pediatric Delirium (CAPD). Comparisons among groups were made using Kruskal–Wallis or Chi-squared tests.

Patient and Treatment Characteristics Associated with Delirium

In unadjusted analysis, the greatest risk factor for developing delirium was requiring IMV during the PICU stay (56% with IMV vs 7.5% without, RR 4.6, p < .001) (Table 3). Other factors associated with developing delirium were female sex (67% vs 38%, RR 2.4, p < .001) and CCCs (40% vs 17%, RR 2.1, p = .003). Prior coma during hospitalization was also strongly associated with developing delirium (35% vs 3.8%, RR 3.7, p < .001). Delirium incidence did not differ based upon age, developmental disability, or PIM3 risk of mortality. Markers of inflammation or end-organ function within the first 24 h of PICU stay also did not differ significantly between those who did versus did not develop delirium (Supplemental Table 5).

Table 3.

Characteristics of Patients with Versus Without Delirium.

	Delirium
Patient characteristics	Absent, N = 106 (71%)	Present, N = 43 (29%)	p-value	Risk ratio (95% CI)
Age, years, median (IQR)	1.3 (0.3–7.0)	0.8 (0.3–4.4	0.3
Female sex	40 (38%)	29 (67%)	<0.001	2.4 (1.4–4.2)
Developmental disability	7 (6.6%)	6 (14%)	0.2	1.7 (0.9–3.2)
Complex chronic condition	18 (17%)	17 (40%)	0.003	2.1 (1.3–3.4)
Hospitalization characteristics
COVID-19 illness type			0.071
Respiratory	49 (46%)	26 (60%)
MIS-C	31 (29%)	5 (12%)		0.4 (0.2–1.0)
Other primary illness	26 (25%)	12 (28%)		0.9 (0.5–1.6)
PIM3 risk of mortality %, median (IQR)	0.04 (0.01–3.98)	0.12 (0.04–3.4)	0.1
Comatose before delirium	4 (3.8%)	15 (35%)	<0.001	3.7 (2.5–5.5)
PICU disposition^a,b			0.11
Death	4 (3.8%)	2 (4.7%)		—
Non-PICU inpatient unit	80 (75%)	33 (77%)		—
Home	20 (19%)	4 (9.3%)		—
Rehab facility	2 (1.9%)	4 (9.3%)		—
PICU LOS, median days (IQR)	3 (2–5)	15 (6–33)	<0.001
Treatment characteristics
Required IMV	8 (7.5%)	24 (56%)	<0.001	4.6 (2.9–7.3)
IMV duration, days (IQR)	2 (2–6)	4 (2–5)	0.9
Received benzodiazepine	25 (24%)	22 (51%)	0.001	2.3 (1.4–3.7)
Received opioid	22 (21%)	23 (53%)	<0.001	2.7 (1.6–4.3)
Received dexmedetomidine	15 (14%)	21 (49%)	<0.001	3.0 (1.9–4.8)
Received paralytic	4 (3.8%)	14 (22%)	<0.001	3.5 (2.3–5.3)
Received ketamine	6 (5.7%)	4 (9.3%)	0.5	1.4 (0.6–3.2)
Received glucocorticoid	75 (71%)	24 (56%)	0.08	0.6 (0.4–1.0)
Received anticholinergic	26 (25%)	7 (16%)	0.3	0.7 (0.3–1.4)
Received IVIG	20 (19%)	5 (12%)	0.28	0.65 (0.29–1.5)
Received vasoactive/inotropic	36 (34%)	19 (44%)	0.2	1.4 (0.8–2.2)

Abbreviations: CI, confidence interval; IVIG, Intravenous Immunoglobulin; PIM3 ROM, Pediatric Index of Mortality third Edition Risk of Mortality; PICU, pediatric intensive care unit; IMV, invasive mechanical ventilation; IQR, interquartile range; MIS-C, multi-system inflammatory syndrome in children associated with COVID-19.

Patient, hospitalization, and treatment characteristics of patients who did versus did not develop delirium. All treatments or clinical diagnoses were received before the onset of delirium, and no more than 14 days prior to the first positive COVID-19 test. P-values were estimated via Chi-squared or Fisher exact tests for categorical variables, and Wilcoxon rank sum tests for continuous variables.

Relative risk not calculated for an outcome.

Six patients died within first PICU admission. Three additional patients in the cohort died either outside the PICU or on a subsequent PICU admission.

Patients who were exposed to opioids (RR 2.7 [CI 1.6–4.3]), dexmedetomidine (RR 3 [CI 1.9–4.8]), benzodiazepines (RR 2.3 [CI 1.4–3.7]), or paralytics (RR 3.5 [CI 2.5–5.3]) more frequently experienced delirium than unexposed patients (Table 3). Among patients who received these medications, cumulative exposures to benzodiazepines (0.7 vs 0.1mg/kg, p < .001), opioids (8 vs 0.3 mg/kg, p < .001), and dexmedetomidine (43.9 vs 11.3 micrograms/kg, p = .03) were also significantly higher in delirious than in non-delirious patients (Supplemental Figure 2). Delirium was not associated with exposure to glucocorticoids, anticholinergics, ketamine, or vasoactive/inotropic medications.

Survival Analysis

In a Cox proportional hazard model adjusting for other potential delirium risk factors, MIS-C was not associated with developing delirium (HR 1.6 [95% CI 0.4–6.3], p = .49) compared to other COVID-19 subgroups. Unadjusted survival curves stratified by type of COVID-19 illness are shown in Figure 1. IMV remained strongly associated with developing delirium (HR 4.9, 95% CI 1.9–13, p = .001) (Figure 2). Female sex and developmental disability remained associated with delirium, whereas glucocorticoid exposure was associated with lower rates.

Sensitivity Analysis

We performed a sensitivity analysis including medications used to facilitate IMV (opioids, benzodiazepines, dexmedetomidine and paralytics) as additional covariates. IMV remained associated with delirium (HR 9.2, 95% CI 2.3–37.8, p = .002) (Supplemental Table 6). None of the newly included IMV-related medications were associated. This extended model was not better at predicting delirium than the original model without these IMV-related medication variables (likelihood ratio p = .27).

Finally, we used interval censoring to account for coma and periods of irregular CAPD testing. Type of COVID-19 illness was not associated with delirium in univariate testing (p = .26). We then estimated the HR for developing delirium via proportional hazard regression on the interval-censored dataset, using the same covariates as the Cox proportional hazards model. MIS-C was still not associated with delirium (HR 0.87, p = .86). Female sex (HR 2.9, p = .04) and IMV (HR 3.4, p = .05) remained associated with delirium, and glucocorticoids remained inversely associated with delirium (HR 0.19, p = .004) with this modeling strategy (Supplemental Table 7).

Discussion

In this cohort of children with COVID-19-related illness who required PICU-level care, delirium was relatively common, with 29% of children experiencing delirium at some point during their PICU course for a median duration of 5 days. In contrast to our hypothesis, and despite its inflammatory nature, MIS-C was not associated with developing delirium compared to other COVID-19-related reasons for PICU admission. The incidence rate for delirium was more than 2.5 times lower in MIS-C than in respiratory COVID-19. MIS-C hospitalization was consistently not associated with delirium regardless of analytical framework or confounder inclusion. Among other potential risk factors tested, requiring IMV was the single greatest risk factor observed and was associated with more than 4 times the rate of developing delirium. Several medications used to facilitate IMV were associated with developing delirium, although these associations were not significant after multivariable adjustment in sensitivity analysis.

Observed rates of delirium in our cohort align with existing pediatric reports, where incidences range from 17–44%.^1‐3 Risk factors for delirium are often grouped in two categories: predisposing and precipitating. In unadjusted analysis, we observed similar precipitating factors for delirium as those that have been previously described, such as exposure to IMV, CCCs, and medications used for sedation during IMV.^2,3,31,32 IMV remained associated in all multivariable analyses, whereas CCCs and most medications did not. Many reports of pediatric delirium find that male sex and younger age are associated with increased delirium risk.^2,33 In contrast, we found that female sex was associated with delirium, and age did not confer any additional risk.

MIS-C is hypothesized to be a post-infectious hyperinflammatory condition^34,35 and neuroinflammation has been implicated in the pathogenesis of delirium.³⁶ We therefore hypothesized that MIS-C would be associated with developing delirium, but we found no association between type of COVID-19 illness and delirium. Rates of delirium actually trended lowest in MIS-C patients, although not after adjusting for confounders. MIS-C patients were characterized by less frequent chronic conditions, developmental disability, use of IMV, and prolonged PICU stay $\geq$ 10 days. Given that neither respiratory COVID-19 nor other primary causes for PICU admission were associated with delirium, it appears unlikely that COVID-19 itself was a precipitating factor of delirium in this overall patient cohort, and that delirium instead had more to do with predisposing patient factors and the therapies required to support patients through recovery.

Glucocorticoids were the only therapy consistently associated with less delirium. Glucocorticoids have been variably associated with delirium, with some studies reporting increased,³⁷ decreased³⁸ or no additional risk³⁹ depending on the patient population, indication, and dosage. In this study, MIS-C patients were nearly universally treated with glucocorticoids due to protocolized treatment of this disease at our hospital.⁴⁰ Sixty-five percent and 39% of respiratory and other COVID-19 patients received glucocorticoids, respectively. Early glucocorticoids with intravenous immunoglobulin (IVIG) hasten recovery and duration of hospitalization in MIS-C,⁴¹ and dexamethasone improves mortality⁴² in hospitalized adults with pulmonary COVID-19. Among our patients, treatment with glucocorticoids may have promoted earlier recovery and hospital discharge in multiple types of COVID-19-related illness, and thus may contribute to lower rates of delirium in COVID-19 in particular.

This study diverges from recent literature regarding COVID-19 in adult patients, where delirium is a highly prevalent complication with reported incidences of 55–73%.^9,10,43 Notably, adults in these reports were much more commonly exposed to IMV than in our cohort, which may have contributed to their higher rates of delirium. It is possible that the comparatively benign course of pediatric COVID-19 is protective, as illness severity may be an independent risk factor for delirium.^2,44 Regardless of type of COVID-19 illness, children might only rarely surpass some illness threshold predisposing to delirium, whereas exceeding that threshold may be more common in adults who require IMV for severe respiratory failure from COVID-19. Another hypothesis is that differential practices in the care of pediatric versus adult COVID-19 patients might influence the strikingly different rates of delirium seen in these two populations. For example, during the pandemic, many adult ICUs reported low implementation of the ICU liberation bundle,⁴⁵ an approach that has been shown to reduce delirium and coma days in adults.⁴⁶ This may suggest that the extremely high rates of delirium described in adults with COVID-19 may not be related to COVID-19 infection itself, but to practices necessitated by the pandemic. Further research is necessary to explore this relationship.

Strengths of this study included complete capture of patients with COVID-19-related illness in the PICU during the study period. It is difficult to ascertain who was hospitalized “for” versus “with” COVID-19, and so we deliberately included patients who were hospitalized for primary reasons other than COVID-19 as long as COVID-19 was plausibly involved in the PICU hospitalization. We believe this minimized selection bias without being too broadly inclusive of unrelated diagnoses. Patients were routinely screened at least once per nursing shift via CAPD, which is an objective and validated screening tool. Most delirious patients had hypoactive delirium, which can be more challenging to detect,⁴⁷ suggesting effective screening at our institution. Finally, we adjusted for potential confounders using a multivariable survival model and several sensitivity analyses designed to account for infrequent CAPD scoring. All models still indicated that IMV was associated with more delirium (and, to a lesser extent, developmental disability and female sex), and that glucocorticoids were associated with less delirium.

Our study has several limitations. This is a single-center, observational study that may not reflect pediatric hospitals in general. All observational studies are vulnerable to confounding by indication, and associations should be interpreted with caution and causality should not be assumed. The small sample size may have limited power to detect associations and limited our ability to include sufficient covariates to fully adjust for confounding within the study. Our assessment of the presence of developmental disability relied upon chart review of notes from clinicians. This may have only detected patients whose developmental disability was severe enough to have caused clinicians to comment in the medical record, and it is unclear if only mild developmental delay would bear the same association. Sedative and analgesic medications are frequently co-administered with IMV, making it challenging to assess these therapies in isolation. However, our findings were consistent across multiple modeling strategies, and a sensitivity analysis found that IMV was associated with delirium independently from common sedative/analgesic medications. While we do not have reason to believe that patients during the three years post-pandemic had different rates and risk factors for delirium, we were not able to compare patients with COVID-19-related illness to routine controls in this study. We were, however, able to compare patients with different types of COVID-19 subgroups as part of our primary hypothesis. Finally, we did not have access to data regarding the use of restraints, involvement of child life, or direct records of family member presence at the bedside. All of these are important factors that may protect against or contribute to delirium.

Conclusions

Delirium is relatively common among children hospitalized in the PICU with COVID-19-related illness, but much less common than has been described in adults with COVID-19. Unlike in adults, delirium was not related to risk of mortality or vasopressor use. In contrast, risk factors for delirium were similar to those previously described in critically ill children and were primarily related to underlying comorbidity, IMV, and possibly with medications used to facilitate IMV. Despite the inflammatory nature of MIS-C, delirium was not more common in these patients. Our study suggests that delirium in pediatric COVID-19 patients remains a frequent problem, and standard delirium prevention and mitigation strategies should be preserved in these patients.

Supplemental Material

sj-docx-1-jic-10.1177_08850666241249169 - Supplemental material for Delirium Associated with COVID-19 in Critically ill Children: An Observational Cohort Study

Supplemental material, sj-docx-1-jic-10.1177_08850666241249169 for Delirium Associated with COVID-19 in Critically ill Children: An Observational Cohort Study by Meghan C. Gray, Chani Traube, Taylor B. Sewell and Andrew S. Geneslaw in Journal of Intensive Care Medicine

Footnotes

Acknowledgements

Chart review: Patrice Pryce, MD; Arsenoi Asfour, MD; J. Egeria Lin, MD, PhD; Benjamin Hooe, MD; Wendy Silver, MD; Wendy Vargas, MD; Chelsea Early, MD; Min Ye Shen, MD; and Mallory Kerner-Rossi, MD. Clinical Data Warehouse: Claudia Boyle, Jori Grossman, and Svetlana Kogan.

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: This study was performed at New York-Presbyterian Morgan Stanley Children's Hospital, Columbia University Irving Medical Center. Dr Geneslaw was supported during this study by the Clinical Research Sites for the Network of Excellence in Neuroscience Clinical Trials (NeuroNEXT sites)-Columbia University Irving Medical Center (Grant # 5U24NS107168-04), and by the National Center for Advancing Translational Sciences, National Institutes of Health, through Grant Number KL2TR001874. Dr Traube reported receiving grants from the National Cancer Institute and the National Institute of Child Health and Human Development outside of the submitted work.

ORCID iDs

Meghan C. Gray

Taylor B. Sewell

Andrew S. Geneslaw

Supplemental Material

Supplemental material for this article is available online.

References

Traube

Silver

Reeder

, et al. Pediatric delirium in critically-ill children: An international point prevalence study. Crit Care Med. 2017;45(4):584‐590. doi:10.1097/CCM.0000000000002250

Traube

Silver

Gerber

, et al. Delirium and mortality in critically ill children: Epidemiology and outcomes of pediatric delirium. Crit Care Med. 2017;45(5):891‐898. doi:10.1097/CCM.0000000000002324

Dervan

Di Gennaro

Farris

RWD

Watson

. Delirium in a tertiary PICU: Risk factors and outcomes*. Pediatr Crit Care Med. 2020;21(1):21‐32. doi:10.1097/PCC.0000000000002126

Wilson

Mart

Cunningham

, et al. Delirium. Nat Rev Dis Primers. 2020;6(1):1‐26. doi:10.1038/s41572-020-00223-4

Silver

Doyle

Hegel

, et al. Association between pediatric delirium and quality of life after discharge*. Crit Care Med. 2020;48(12):1829‐1834. doi:10.1097/CCM.0000000000004661

Pandharipande

Girard

Jackson

, et al. Long-Term cognitive impairment after critical illness. N Engl J Med. 2013;369(14):1306‐1316. doi:10.1056/NEJMoa1301372

Pezzini

Padovani

. Lifting the mask on neurological manifestations of COVID-19. Nat Rev Neurol., 2020;16(11):1‐9. doi: 10.1038/s41582-020-0398-3

Chou

SHY

Beghi

Helbok

, et al. Global incidence of neurological manifestations among patients hospitalized with COVID-19—A report for the GCS-NeuroCOVID consortium and the ENERGY consortium. JAMA Netw Open. 2021;4(5):e2112131. doi:10.1001/jamanetworkopen.2021.12131

Pun

Badenes

Heras La Calle

, et al. Prevalence and risk factors for delirium in critically ill patients with COVID-19 (COVID-D): A multicentre cohort study. Lancet Respir Med. 2021;9(3):239‐250. doi:10.1016/S2213-2600(20)30552-X

10.

Ragheb

McKinney

Zierau

, et al. Delirium and neuropsychological outcomes in critically ill patients with COVID-19: A cohort study. BMJ Open. 2021;11(9):e050045. doi:10.1136/bmjopen-2021-050045

11.

Panda

Sharawat

Panda

Natarajan

Bhakat

Dawman

. Neurological complications of SARS-CoV-2 infection in children: A systematic review and meta-analysis. J Trop Pediatr. 2020;10(3):1-11. doi:10.1093/tropej/fmaa07

12.

LaRovere

Riggs

Poussaint

, et al. Neurologic involvement in children and adolescents hospitalized in the United States for COVID-19 or multisystem inflammatory syndrome. JAMA Neurol. 2021;78(5):536‐547. doi:10.1001/jamaneurol.2021.0504

13.

Lin

Asfour

Sewell

, et al. Neurological issues in children with COVID-19. Neurosci Lett. 2021;743:135567. doi:10.1016/j.neulet.2020.135567

14.

Helms

Kremer

Merdji

, et al. Delirium and encephalopathy in severe COVID-19: A cohort analysis of ICU patients. Crit Care. 2020;24(1):491. doi:10.1186/s13054-020-03200-1

15.

Patel

. Multisystem inflammatory syndrome in children (MIS-C). Curr Allergy Asthma Rep. 2022;22(5):53‐60. doi:10.1007/s11882-022-01031-4

16.

von Elm

Altman

Egger

, et al. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: Guidelines for reporting observational studies. Int J Surg. 2014;12(12):1495‐1499. doi:10.1016/j.ijsu.2014.07.013

17.

CDC. Interim Clinical Guidance for Management of Patients with Confirmed Coronavirus Disease (COVID-19). Centers for Disease Control and Prevention. Published February 11, 2020. Accessed April 4, 2022. https://www.cdc.gov/coronavirus/2019-ncov/hcp/clinical-guidance-management-patients.html

18.

Traube

Silver

Kearney

, et al. Cornell assessment of pediatric delirium: A valid, rapid, observational tool for screening delirium in the PICU. Crit Care Med. 2014;42(3):656‐663. doi:10.1097/CCM.0b013e3182a66b76

19.

Kerson

DeMaria

Mauer

, et al. Validity of the richmond agitation-sedation scale (RASS) in critically ill children. J Intensive Care. 2016;4(1):65. doi:10.1186/s40560-016-0189-5

20.

Feudtner

Feinstein

Zhong

Hall

Dai

. Pediatric complex chronic conditions classification system version 2: Updated for ICD-10 and complex medical technology dependence and transplantation. BMC Pediatr. 2014;14(1):199. doi:10.1186/1471-2431-14-199

21.

Straney

Clements

Parslow

, et al. Paediatric index of mortality 3: An updated model for predicting mortality in pediatric intensive care*. Pediatr Crit Care Med. 2013;14(7):673‐681. doi:10.1097/PCC.0b013e31829760cf

22.

Barr

Zomorodi

Bertaccini

Shafer

Geller

. A double-blind, randomized comparison of i.v. Lorazepam versus midazolam for sedation of ICU patients via a pharmacologic model. Anesthesiology. 2001;95(2):286‐298. doi:10.1097/00000542-200108000-00007

23.

Zaal

Devlin

Peelen

Slooter

AJC

. A systematic review of risk factors for delirium in the ICU*. Crit Care Med. 2015;43(1):40‐47. doi:10.1097/CCM.0000000000000625

24.

Therneau

Grambsch

. “Multiple events per subject.” modeling survival data: extending the Cox model. Springer 2000, pp. 169‐229.

25.

R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, 2021. https://www.R-project.org/ .

26.

Therneau

. A Package for Survival Analysis in R, 2020. https://CRAN.R-project.org/package=survival .

27.

Wickham

. ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag; 2016.

28.

Kassambara

Kosinski

Biecek

. survminer: Drawing Survival Curves using “ggplot2”. R package version 0.4.9. 2021. https://CRAN.R-project.org/package=survminer .

29.

Fay

Shaw

. Exact and asymptotic weighted logrank tests for interval censored data: The interval R package. J Stat Softw. 2010;36(2):1‐34. doi:10.18637/jss.v036.i02

30.

Anderson-Bergman

. Icenreg: Regression models for interval censored data in R. J Stat Softw. 2017;81(12):1‐23. doi:10.18637/jss.v081.i12

31.

Siegel

Traube

. Pediatric delirium: Epidemiology and outcomes. Curr Opin Pediatr. 2020;32(6):743‐749. doi:10.1097/MOP.0000000000000960

32.

Patel

Bell

Traube

. Delirium in pediatric critical care. Pediatr Clin N Am. 2017;64(5):1117‐1132. doi:10.1016/j.pcl.2017.06.009

33.

Silver

Traube

Gerber

, et al. Pediatric delirium and associated risk factors: A single-center prospective observational study. Pediatr Crit Care Med. 2015;16(4):303‐309. doi:10.1097/PCC.0000000000000356

34.

Rapid risk assessment: Paediatric inflammatory multisystem syndrome and SARS-CoV-2 infection in children. European Centre for Disease Prevention and Control. Published May 15, 2020. Accessed March 18, 2022. https://www.ecdc.europa.eu/en/publications-data/paediatric-inflammatory-multisystem-syndrome-and-sars-cov-2-rapid-risk-assessment

35.

Simpson

Newburger

. Multisystem inflammatory syndrome in children in association with COVID-19. Circulation. 2020;142(5):437‐440. doi:10.1161/CIRCULATIONAHA.120.048726

36.

Wei

Feng

Yuan

. Analysis of risk factor for pediatric intensive care unit delirium in children: A case-control study. Am J Transl Res. 2021;13(8):9143‐9151.

37.

Schreiber

Colantuoni

Bienvenu

, et al. Corticosteroids and transition to delirium in patients with acute lung injury. Crit Care Med. 2014;42(6):1480‐1486. doi:10.1097/CCM.0000000000000247

38.

Lee

Joynt

. Pharmacologic agents for the prevention and treatment of delirium in patients undergoing cardiac surgery: Systematic review and metaanalysis. Crit Care Med. 2015;43(1):194‐204. doi:10.1097/CCM.0000000000000673

39.

Chen

Xiang

. Corticosteroids in patients undergoing cardiac surgery: A meta-analysis of 12,559 patients. Perfusion. 2023;38(4):853‐859. doi:10.1177/02676591221106324

40.

Jonat

Gorelik

Boneparth

, et al. Multisystem inflammatory syndrome in children associated with coronavirus disease 2019 in a children’s hospital in New York city: Patient characteristics and an institutional protocol for evaluation, management, and follow-up. Pediatr Crit Care Med. 2021;22(3):e178‐e191. doi:10.1097/PCC.0000000000002598

41.

Channon-Wells

Vito

McArdle

, et al. Immunoglobulin, glucocorticoid, or combination therapy for multisystem inflammatory syndrome in children: A propensity-weighted cohort study. The Lancet Rheumatology. 2023;5(4):e184‐e199. doi:10.1016/S2665-9913(23)00029-2

42.

RECOVERY Collaborative Group, Horby

Shen Lim

Emberson

, et al. Dexamethasone in hospitalized patients with COVID-19. N Engl J Med. 2021;384(8):693‐704. doi:10.1056/NEJMoa2021436

43.

Williamson

Faiver

Nguyen

Ottenhoff

Incidence

. Predictors and outcomes of delirium in critically ill patients with COVID-19. Neurohospitalist. 2022;12(1):31‐37. doi:10.1177/19418744211034815

44.

Schieveld

JNM

Lousberg

Berghmans

, et al. Pediatric illness severity measures predict delirium in a pediatric intensive care unit. Crit Care Med. 2008;36(6):1933‐1936. doi:10.1097/CCM.0b013e31817cee5d

45.

Liu

Nakamura

Katsukawa

, et al. Implementation of the ABCDEF bundle for critically ill ICU patients during the COVID-19 pandemic: A multi-national 1-day point prevalence study. Front Med (Lausanne). 2021;8:735860. doi:10.3389/fmed.2021.735860

46.

Pun

Balas

Barnes-Daly

, et al. Caring for critically ill patients with the ABCDEF bundle: Results of the ICU liberation collaborative in over 15,000 adults. Crit Care Med. 2019;47(1):3‐14. doi:10.1097/CCM.0000000000003482

47.

Semple

Howlett

Strawbridge

Breatnach

Hayden

. A systematic review and pooled prevalence of delirium in critically ill children. Crit Care Med. 2022;50(2):317‐328. doi:10.1097/CCM.0000000000005260

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

0.40 MB