Delirium and Benzodiazepines Associated With Prolonged ICU Stay in Critically Ill Infants and Young Children

Heidi A. B. Smith, MD, MSCI; Maalobeeka Gangopadhyay, MD; Christina M. Goben, MD; Natalie L. Jacobowski, MD; Mary Hamilton Chestnut, NP; Jennifer L. Thompson, MPH; Rameela Chandrasekhar, PhD; Stacey R. Williams, NP; Katherine Griffith, NP; E. Wesley Ely, MD, MPH; D. Catherine Fuchs, MD; Pratik P. Pandharipande, MD, MSCI

Disclosures

Crit Care Med. 2017;45(9):1427-1435.

Abstract and Introduction

Abstract

Objectives: Delirium is prevalent among critically ill children, yet associated outcomes and modifiable risk factors are not well defined. The objective of this study was to determine associations between pediatric delirium and modifiable risk factors such as benzodiazepine exposure and short-term outcomes.

Design: Secondary analysis of collected data from the prospective validation study of the Preschool Confusion Assessment Method for the ICU.

Setting: Tertiary-level PICU.

Patients: Critically ill patients 6 months to 5 years old.

Interventions: None.

Measurements and Main Results: Daily delirium assessments were completed using the Preschool Confusion Assessment Method for the ICU. Associations between baseline and in-hospital risk factors were analyzed for likelihood of ICU discharge using Cox proportional hazards regression and delirium duration using negative binomial regression. Multinomial logistic regression was used to determine associations between daily risk factors and delirium presence the following day. Our 300-patient cohort had a median (interquartile range) age of 20 months (11–37 mo), and 44% had delirium for at least 1 day (1–2 d). Delirium was significantly associated with a decreased likelihood of ICU discharge in preschool-aged children (age-specific hazard ratios at 60, 36, and 12 mo old were 0.17 [95% CI, 0.05–0.61], 0.50 [0.32–0.80], and 0.98 [0.68–1.41], respectively). Greater benzodiazepine exposure (75–25th percentile) was significantly associated with a lower likelihood of ICU discharge (hazard ratio, 0.65 [0.42–1.00]; p = 0.01), longer delirium duration (incidence rate ratio, 2.47 [1.36–4.49]; p = 0.005), and increased risk for delirium the following day (odds ratio, 2.83 [1.27–6.59]; p = 0.02).

Conclusions: Delirium is associated with a lower likelihood of ICU discharge in preschool-aged children. Benzodiazepine exposure is associated with the development and longer duration of delirium, and lower likelihood of ICU discharge. These findings advocate for future studies targeting modifiable risk factors, such as reduction in benzodiazepine exposure, to mitigate iatrogenic harm in pediatric patients.

Introduction

Pediatric delirium is a syndrome of acute brain dysfunction that is highly prevalent among the critically ill, with rates of up to 30% in older children^[1–4] and over 50% in infants and toddlers.^[5] The creation of highly valid and reliable pediatric delirium tools has made delirium screening and monitoring in the PICU setting not only practical^[1,2,5,6] but provides the means to further understand the impact of delirium in our most fragile population. In critically ill adults, delirium has been associated with poorer outcomes including prolonged ICU and hospital length of stay (LOS), long-term cognitive impairment (LTCI), and even death.^[7–9] Small cohort studies in critically ill children have demonstrated similar associations between delirium and longer ICU stay and higher costs,^[10,11] whereas a recent point prevalence study on delirium reported associated risk factors to include age, mechanical ventilation, benzodiazepines, opioids, use of physical restraints, and exposure to vasopressors and antiepileptics.^[4] The etiology of delirium can be multifactorial, with the end result being a disruption of the fragile balance between excitatory and inhibitory neurotransmission.^[12] This imbalance manifests through characteristic behaviors that are commonly observed and can range from hyperactivity and agitation to a patient appearing withdrawn and sedate.^[13,14]

Delirium risk factors have been well delineated in adults to include iatrogenic factors such as immobilization, sleep deprivation, and a strong association with benzodiazepine exposure.^[15–19] The iatrogenic harm resulting from benzodiazepine use in adults has been consistently demonstrated to include greater development and duration of delirium, increased duration of mechanical ventilation, and longer ICU and hospital LOS.^[20–22] This realization is pertinent as up to 90% of mechanically ventilated infants and children receive continuous sedation,^[23] commonly centered around the use of high-dose midazolam for multiple days.^[24,25] In fact, high-dose oral benzodiazepines have been associated with paradoxical agitation, and thus, it is plausible that high-dose IV benzodiazepines, frequently administered to critically ill patients, could predispose patients to delirium.^[26]

Thus, the hypothesis for this large prospective study is that delirium is associated with worse clinical outcomes among critically ill infants and young children and that routine sedation protocols using benzodiazepines predispose patients to delirium.

1 2 3 4 5

Next Section

Table 1. Demographics and Baseline and Summary Variables
Variables	All Patients (n = 300)
Demographics
Age at enrollment (mo), median (IQR)	20 (11–37)
Sex, frequency (%)
Female	112 (37)
Male	188 (63)
Race, frequency (%)
Asian	5 (2)
Black or African American	60 (20)
Caucasian	229 (76)
Other	6 (2)
Weight (kg), median (IQR)	11 (8–14)
Baseline and in-hospital variables
Cyanotic heart disease, frequency (%)	91 (30)
Pediatric Risk of Mortality score (day 1), median (IQR)	5 (0–10)
Sepsis or related admission disease, frequency (%)	112 (37)
Daily benzodiazepines (midazolam equivalents) (mg/kg), median (IQR)	0.15 (0–1.06)
Daily opioids (fentanyl equivalents) (μg/kg), median (IQR)	6.3 (0–34.3)
Daily dexmedetomidine (μg/kg), median (IQR)	0 (0–50)
Daily cardiovascular Sequential Organ Failure Assessment score, median (IQR)	0 (0–1)
Daily lowest oxygen saturations, median (IQR)	90 (75–94)
Summary variables
Received mechanical ventilation during hospitalization, frequency (%)	130 (43)
Length of mechanical ventilation (d), median (IQR)	3 (1–5)
Delirium prevalence (Preschool Confusion Assessment Method of the ICU), frequency (%)	124 (44)
Duration of delirium among those exposed (d), median (IQR)	1 (1–2)
Length of ICU stay (d), median (IQR)	4 (3–9)
Length of hospitalization (d), median (IQR)	8 (4–17)
In-hospital mortality, frequency (%)	10 (3)

IQR = interquartile range.

Table 2. Risk Factors for Likelihood of ICU Discharge ^a
Risk Factors	Comparison	Hazard Ratio (95% CI)	p
Age at enrollment (mo)	31 vs 9	1.85 (1.23–2.77)	0.01
Pediatric Risk of Mortality score at ICU admission	13 vs 0.25	1.21 (0.87–1.68)	0.074
Sepsis or related diagnosis at ICU admission	Yes vs no	0.99 (0.77–1.29)	0.969
Benzodiazepines (mg/kg/d)^b	1 vs 0	0.65 (0.42–1.00)	0.011
Opioids (μg/kg/d)^b	3.26 vs 0	0.62 (0.39–0.98)	0.112
Dexmedetomidine (μg/kg/d)^b	3.74 vs 0	1.82 (1.25–2.66)	0.008
Cardiovascular Sequential Organ Failure Assessment score	1 vs 0	0.86 (0.73–1.02)	0.084
Lowest O₂ saturations (%)	94 vs 75	2.18 (1.63–2.92)	< 0.001
Mental statusc	Delirious vs normal	0.94 (0.65–1.36)	0.051
	Comatose vs normal	0.74 (0.41–1.34)

^a The hazard ratio (HR), 95% CI, and p values for likelihood of ICU discharge on a given day are presented for each potential risk factor. A HR > 1 refers to a "greater" likelihood of ICU discharge. Conversely, a HR < 1 refers to a "lower" likelihood of ICU discharge. The p values reflect the entire association for a given covariate magnitude comparison. For continuous covariates, the HRs represent the association between likelihood of discharge at the 75th vs the 25th percentile of each risk factor.
^b Drug dosage amounts (mg/kg) were transformed using the cube root transformation in order to mitigate the influence of extremely high values.
^c HRs describe the individual associations of delirium and coma vs normal mental status. As the multivariable model allowed for an interaction between age and mental status, Figure 1 is necessary to illustrate the significant relationship between delirium and a lower likelihood of ICU discharge in older children (> 12 mo old).

Table 3. Risk Factors for Delirium Duration ^a
Risk Factors	Comparison	Incidence Rate Ratio (95% CI)	p
Age at enrollment (mo)^b	37 vs 11	0.51 (0.35–0.76)	0.005
Cyanotic heart disease	Yes vs no	1.23 (0.72–2.10)	0.477
Pediatric Risk of Mortality score at enrollment^b	10 vs 0	2.25 (1.28–3.94)	0.007
Sepsis or related condition at ICU admission	Yes vs no	1.05 (0.69–1.59)	0.823
Benzodiazepines (mg/kg/d)^b,c	0.73 vs 0	2.47 (1.36–4.49)	0.005
Opioids (μg/kg/d)^c	3.3 vs 0	0.70 (0.33–1.48)	0.41
Cardiovascular Sequential Organ Failure Assessment score	1 vs 0	0.96 (0.79–1.17)	0.691
Mechanical ventilation	Yes vs no	0.66 (0.37–1.16)	0.135
Lowest O₂ saturations	95 vs 87	1.03 (0.67–1.59)	0.516

^a All continuous covariates had nonlinear associations with delirium duration, and as such, a single-point estimate does not fully describe the relationship. To more precisely understand this association, the incidence rate ratios (IRRs), 95% CI, and p values represent the comparative association for each potential risk factor and delirium duration. These comparative examples use values corresponding with the 75th and 25th percentiles for continuous variables, and therefore, the IRRs represent only these comparative associations.
^b Supplemental Figure 1 (Supplemental Digital Content 1, http://links.lww.com/CCM/C656; legend, Supplemental Digital Content 2, http://links.lww.com/CCM/C657) further illustrates the depth of these nonlinear associations between age, Pediatric Risk of Mortality scores, and benzodiazepine exposure with delirium duration.
^c Drug dosage amounts (mg/kg) were transformed using the cube root transformation in order to mitigate the influence of extremely high values.

Authors and Disclosures

Heidi A. B. Smith, MD, MSCI^1,2, Maalobeeka Gangopadhyay, MD³, Christina M. Goben, MD⁴, Natalie L. Jacobowski, MD⁵, Mary Hamilton Chestnut, NP¹, Jennifer L. Thompson, MPH⁶, Rameela Chandrasekhar, PhD⁷, Stacey R. Williams, NP², Katherine Griffith, NP², E. Wesley Ely, MD, MPH^8,9, D. Catherine Fuchs, MD¹⁰, Pratik P. Pandharipande, MD, MSCI¹

¹Department of Anesthesiology, Vanderbilt University Medical Center, Nashville, TN.
²Department of Pediatrics, Monroe Carell Jr. Children's Hospital at Vanderbilt, Nashville, TN.
³Department of Psychiatry, New York-Presbyterian Morgan Stanley Children's Hospital, New York, NY.
⁴Department of Pediatrics, University of South Carolina School of Medicine Greenville, Greenville, SC.
⁵Department of Psychiatry, Children's Hospital of Philadelphia, Philadelphia, PA.
⁶Department of Biostatistics, Vanderbilt University Medical Center, Nashville, TN.
⁷Department of Biostatistics, Vanderbilt University School of Medicine, Nashville, TN.
⁸Department of Internal Medicine, Center for Health Services Research, Vanderbilt University Medical Center, Nashville, TN.
⁹Tennessee Valley Veterans Administration (VA) Geriatric Research Education Clinical Center (GRECC), Nashville, TN.
¹⁰Department of Psychiatry, Vanderbilt University Medical Center, Nashville, TN.

Registration identification number: NCT02481115 (https://register.clinicaltrials.gov).
Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's website (http://journals.lww.com/ccmjournal).
Supported, in part, by the Foundation for Anesthesia Education and Research and by the Vanderbilt Clinical and Translational Science Award (CTSA) grant UL1 TR000445 from National Center for Research Resources (NCRR)/National Institutes of Health (NIH).
Dr. Smith received support from the Foundation for Anesthesia Education and Research (FAER) and the Vanderbilt CTSA UL1 TR000445 from NCRR/NIH. Dr. Gangopadhyay received funding from Vanderbilt CTSA UL1 TR000445 from NCRR/NIH. Dr. Ely received support from the VA Tennessee Valley Geriatric Research, Education and Clinical Center, the VA Clinical Science Research and Development Service, and the NIHAG027472, AG035117, and HL111111. Dr. Pandharipande received support from NIH HL111111, AG027472, and AG035117, and his institution received a research grant from Hospira. The remaining authors have disclosed that they do not have any potential conflicts of interest.
For information regarding this article, E-mail: heidi.smith@vanderbilt.edu