Steroid-Induced Sleep Disturbance and Delirium: A Focused Review for Critically Ill Patients
Objective: Insomnia and delirium have gained much attention since the publication of recent guidelines for the management in critically ill adults. Neurologic effects such as sleep disturbance, psychosis, and delirium are commonly cited adverse effects (AEs) of corticosteroids. Steroid use is considered a modifiable risk factor in intensive care unit patients; however, reported mechanisms are often lacking. This focused review will specifically evaluate the effects of steroids on sleep deprivation, psychosis, delirium, and what is known about these effects in a critically ill population.
Observations: The medical literature proposes 3 pathways primarily responsible for neurocognitive AEs of steroids: behavior changes through modification of the hypothalamic-pituitary-adrenal axis, changes in natural sleep-wake cycles, and hyperarousal caused by modification in neuroinhibitory pathways. Initial search fields produced 285 articles. Case reports, reviews, letters, and articles pertaining to primary care or palliative populations were excluded, leaving 8 relevant articles for inclusion.
Conclusions: Although steroid therapy often cannot be altered in the critically ill population, research showed that steroid overuse is common in intensive care units. Minimizing dosage and duration are important ways clinicians can mitigate AEs.
Hyperarousal State
Finally, a hyperarousal state is thought to be produced by nongenomic changes to natural neuroinhibitory regulation seen with nonclassical steroid production called neurosteroids. Animal studies revealed that high levels of steroids were found in the CNS long after adrenalectomy, suggesting CNS de novo synthesis.46 In addition to altering gene expression at classic intercellular steroid receptors, neurosteroids can alter neurotransmission by direct interaction on ion-gated membranes and other receptors on the cell surface. Restlessness and insomnia could be due to γ-aminobutyric acid type A (GABAA) receptor modulation in the CNS where neuroactive steroids slow the rate of recovery of GABAA and potentially inhibit postsynaptic GABAergic transmission. It also is hypothesized that neuroactive steroids have excitatory action at nicotinic acetylcholine, 5HT3 receptors, and through increasing the fractional open time of the N-methyl-D-aspartate -activated channels.47 Allopregnanolone and DHEA are neurosteroids that act as GABAA agonists and have neuroprotective effects with anxiolytic, antidepressant, and antiaggressive properties.
Neurosteroids are synthesized from cholesterol in the hippocampus. Neurosteroids are upregulated in response to stress by CNS cortisol effects on various enzyme expressions.47 Whether exogenous steroid administration affects this biosynthesis vs the stress response in the HPA axis itself is not fully elucidated. Monteleone and colleagues found that dexamethasone 1 mg given orally significantly reduced cortisol and DHEA and allopregnanolone levels in both healthy volunteers and anorexia nervosa patients.48 Similarly, Genazzani and colleagues demonstrated that oral dexamethasone administration (0.5 mg every 6 hours) caused significant reductions in both serum allopregnanolone and DHEA levels.49
Outcomes Studies
The majority of reported data in steroid-induced insomnia and psychosis is in noncritically ill populations. In a randomized, prospective crossover study of healthy volunteers, dexamethasone administration (3 mg every 8 hours for 48 hours) resulted in significant changes in sleep patterns measured with polysomnography. Compared with placebo, steroid treatment showed significantly longer percentage (SD) of stage 0/awake times (11.7% [11.4] vs 2.9% [1.8]; P < .05); longer percentage (SD) of REM sleep latency (363.8 [74.5] minutes vs 202.8 [79.6] minutes; P < .01), and a reduced number (SD) of REM periods (3.8 [2.6] vs 9.7 [3.6]; P < .01).50 Insomnia was one of the most commonly self-reported AEs (> 60%) in a survey of 2,446 chronic steroid users, and the incidence increased as steroid doses increased.51
A prospective, open-label study of 240 patients with cancer demonstrated significant sleep disruptions using the Pittsburgh Sleep Quality Index with the use of high-dose steroids in chemotherapy.52 Naber and colleagues evaluated 50 previously healthy patients taking methylprednisolone 119 mg (41 mg/d) for retinitis and uveitis.53 They reported 26% to 34% of subjects experienced hypomanic syndrome based on a semistructured interview examination. Symptoms developed within 3 days and persisted for the 8-day course of therapy. Brown and colleagues prospectively evaluated 32 asthmatic patients prescribed bursts of prednisone > 40 mg daily. They observed significantly increased scores in the Young Mania Rating Scale within 3 to 7 days of starting therapy, which dissipated to baseline after stopping therapy.54
Despite a high reported incidence of neurologic AEs, outcomes in critically ill populations are mixed. Study methods are varied, and many were largely observational. No prospective, randomized studies exist to date specifically aimed and powered to evaluate the effects of steroids on sleep disturbances or delirium in a critically ill population. Furthermore, sleep quality is difficult to measure in this population, and self-reporting often is not an option. In critical care trials, if AEs such as insomnia, delirium, or psychosis are recorded at all, there is heterogeneity in the definitions, and these AEs are generally poorly defined (eg, psychiatric or neurologic disorder not otherwise specified), making pooled analysis of this outcome difficult.55
One of the largest observational studies in hospitalized patients was through the Boston Collaborative Drug Surveillance Program. A total of 718 consecutively enrolled inpatients who received prednisone were monitored for acute reactions. Psychiatric AEs were rare (1.3%) with low doses (< 40 mg/d), more prevalent (4.6%) with higher doses (41-80 mg/d), and most prevalent (18.4%) with the highest doses (> 80 mg/d), suggesting CNS AEs are dose dependent.18 A single-center, retrospective review of 755 psychiatric consults in hospitalized patients revealed that 54% of manic patients were due to corticosteroid administration.19 In a prospective observational study of 206 consecutive ICU admissions, steroid administration was an independent risk factor for development of ICU delirium, using the Confusion Assessment Method-ICU (CAM-ICU) at a single center (odds ratio [OR], 2.8; 95% CI, 1.05-7.28).25
Two studies in hospitalized oncology patients found conflicting results using the Nursing Delirium Screening Scale (Nu-DESC). One did not find a significant association between delirium and dexamethasone equivalent doses > 15 mg, while the second found an increased hazard ratio (HR) for a positive Nu-DESC score (HR, 2.67; 95% CI, 1.18-6.03).20,21 Similarly, conflicting results were found in 2 studies using first-order Markov models. In one prospective cohort study, 520 consecutive mechanically ventilated patients in 13 ICUs were monitored for the transition to delirium (CAM-ICU positive) from nondelirium states. Steroid administration was significantly associated with transitioning to delirium (OR, 1.52; 95% CI, 1.05-2.21).22 This conflicts with a similar study by Wolters and colleagues, which monitored 1,112 ICU patients who were given a median prednisone equivalent of 50 mg (interquartile range, 25-75 mg). Steroid administration was not significantly associated with the transition to delirium from an awake without delirium state (OR, 1.08; 95% CI, 0.89-1.32; adjusted OR, 1.00; 95% CI, 0.99-1.01 per 10-mg increase in prednisone equivalent).23
