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


Crit Care Med. 2017;45(9):1427-1435. 

In This Article


In this large prospective cohort study, four of 10 children suffered from delirium during their ICU stay. Delirium and benzodiazepine exposures were significantly associated with a lower likelihood of ICU discharge. Patients who were exposed to greater amounts of benzodiazepine suffered from a considerably longer delirium duration. Furthermore, patients with high benzodiazepine exposure were significantly more likely to develop delirium the day following the exposure when compared with those with minimal or no exposure. The importance of high delirium rates during critical illness is highlighted by the significant observations of prolonged ICU LOS in this study and prior reports, in addition to previously described higher associated costs.[10,11] With benzodiazepine use being associated with not only prolonged delirium duration but also a lower likelihood of ICU discharge, there is now a potentially modifiable risk factor to target for interventional trials aimed at improving pediatric outcomes following critical illness.

Delirium during critical illness was a significant predictor of a lower likelihood of ICU discharge among preschool-aged children (> 12 mo old). This finding may be a reflection of severity of illness; as many of the infants and toddlers in our study were patients admitted following cardiac surgery and had more comorbidities and a higher severity of illness. So, delirium in this setting may have had a smaller impact on the likelihood of ICU discharge when compared with other significant elements of critical illness among these patients. Previous studies have demonstrated that severity of illness scores may be useful for predicting delirium in the PICU, though this relationship may be influenced by other factors of the critical illness as well.[30]

Additionally, it is possible that delirium was not routinely recognized by the medical team among the youngest patients and therefore not considered in the decision-making for ICU discharge, where an abnormal mental status is more commonly alarming among older, normally interactive patients. Nevertheless, both sedation and delirium are not inconsequential in critically ill children and result in tangible worse clinical outcomes that may come with a cost. Our results begin to parallel prior adult studies that demonstrate a relationship between delirium and longer ICU and hospital LOS.

Understanding the relationships between benzodiazepine administration and delirium are imperative, given that upwards of 90% of critically ill infants and children receive continuous sedation while on mechanical ventilation. Although the use of sedation may be unavoidable in the PICU setting, it remains a potentially modifiable risk factor, where attention to limiting total exposure and considering alternative sedation paradigms may lead to improved patient outcomes. These data are in line with studies of adults where benzodiazepine administration has been shown to be associated with delirium leading to investigations targeting the reduction of sedation, daily interruption of sedation, and the consideration of non-benzodiazepine agents as first line of therapy.[15,31]

The Pain Agitation Delirium guidelines from the Society of Critical Care Medicine (SCCM)[32,33] and the SCCM ICU Liberation collaborative have both stressed the importance of limiting sedative exposure, and our data in children underscore the importance of such an approach in our most vulnerable patients. This will no doubt challenge the mainstay of care for critically ill children who are usually heavily sedated, with the belief among ICU caregivers that a heavily sedated pediatric patient may be psychologically "saved from the ICU experience." However, we know that a third of pediatric patients report delusional memories of their ICU experience, associated with longer benzodiazepine and opioid exposure during critical illness, and subsequent reports of higher posttraumatic stress disorder scores following discharge to home.[34] We propose transition to the mindset that for every child, every day there is a goal of being alert and calm, while being mindful of clinical goals and the treatment of pain, and thus liberate children more quickly from sedation, mechanical ventilation, and the ICU environment.

In addition to recognizing potentially modifiable risk factors for all patients, identifying baseline risk factors can help pinpoint patients who may benefit from preventive care and perhaps increased vigilance. In this study, infants and younger children and patients with a higher severity of illness were more likely to suffer a longer duration of delirium. A significant association was not demonstrated between delirium duration and presence of cyanotic heart disease, admission diagnosis of sepsis, need for mechanical ventilation, or cardiovascular SOFA score in this cohort. The lack of an independent association between delirium and need for mechanical ventilation, we believe, is due to adjusting for other factors such as severity of illness, hypoxia, and sedative exposure. In other words, this highlights that concurrent use of sedation in patients while on mechanical ventilation assumes a greater association with delirium, rather than the requirement of mechanical ventilation alone. Whereas predisposing risk factors such as age, severity of illness, and baseline developmental delay have been previously demonstrated to be significantly associated with delirium in smaller prospective cohort studies,[30,35] these findings may highlight the importance of the interplay between predisposing and precipitating factors, such that the critical illness alone may not always determine the rate of development or severity of delirium.

The developing brain has been the focus of numerous recent studies that demonstrate the following: 1) pharmacologic action at the N-methyl-D-aspartate and gamma-aminobutyric acid receptors produces profound apoptotic and other neurodegenerative changes in the developing brain; 2) multiple exposures to drug/anesthetic administration are associated with adverse cognitive effects controlling for comorbidities; and 3) specific deficits in speech and language have been repetitively identified following drug/anesthetic administration.[36–45] Recently, significant declines in both global cognitive and executive functions were associated with longer duration of delirium in adults a year following discharge to home.[11] With the current appreciation of the biological development of the immature brain, the interplay between acute and long-term cognitive dysfunction in pediatric patients will be paramount. To this end, the neuropsychiatric and cognitive outcomes associated with acute brain dysfunction should now become the focus.

Several limitations of this prospective study warrant discussion. This study was conducted in critically ill patients 6 months to 5 years old for the following reasons: 1) to validate the psCAM-ICU and 2) to determine risk factors and outcomes associated with pediatric delirium; thus, these results may not be generalizable to older children. However, a recent study on the point prevalence of delirium in critically ill children identified similar associated risk factors.[4] Delirium assessments were conducted once daily on enrolled patients until ICU transfer/discharge or for a maximum of 14 days, whichever came first. When a delirium assessment was not available (such as a weekend day), we categorized the assessment as nondelirious. The only outcome measure that depended on the aforementioned reconciliation of missing data was delirium duration. The potential effect on measured results using missing data reconciliation would have been an underestimation of the prevalence and duration of delirium, in other words, a bias toward the null hypothesis. Although over 90% of the study cohort had a PICU LOS less than or equal to 14 days, it is possible that a very small minority of patients, who remained in the ICU longer than the reported 14-day study period, could have suffered a longer duration or later development of delirium. For outcomes such as ICU and hospital LOS, we used exact dates of discharge obtained from the medical record and thus these outcomes were not curtailed at 14 days.

Our primary model determining the association between delirium and benzodiazepine exposure included all patient days with available mental status but did not adjust for the prior day mental status. Adjusting for prior mental status would have required all patients to have 2 consecutive days of mental status assessments, when in fact, there were patients who did not (short ICU LOS or weekend day). If we had adjusted for prior mental status, these patients would have had to be excluded. Rather, we conducted a sensitivity analysis on a subset of patients who had 2 consecutive days of mental status with adjustment for mental status from the previous day. Our findings remained consistent, and benzodiazepine exposure was independently associated with delirium the next day.

Although we included numerous covariates that were deemed clinically relevant a priori, this list was not all-inclusive, and therefore, it is possible that other essential covariates were not measured. Possible iatrogenic risk factors such as sleep deprivation, limited social interactions, and immobility were not included in this study, though clearly, they may play a significant role in delirium duration and hence other important outcomes. Psychoactive drug exposure was a key covariate in this study. However, the relationship between delirium and drug exposure was limited to the dose administered, rather than serum levels. Since drug exposure based solely on dose may disguise the role that renal and hepatic function play in the metabolism and excretion of drugs, the relationships between drug exposure and delirium and other outcomes may be elucidated with the addition of drug serum level surveillance. In order to further elucidate the temporal relationships between delirium and associated risk factors, future studies may benefit from more frequent delirium assessments and sampling of key risk factor–related events.

Finally, this study was not designed to determine the role of delirium in the development of LTCI in children. The current evidence for a relationship between delirium and LTCI in adults bolsters the need for further study in pediatric patients, who are undergoing a steep trajectory of neurodevelopment prior to their interceding critical illness.