Effect of Atypical Sleep EEG Patterns on Weaning From Prolonged Mechanical Ventilation

Abstract

Background

Approximately one-third of acute ICU patients display atypical sleep patterns that cannot be interpreted by using standard EEG criteria for sleep. Atypical sleep patterns have been associated with poor weaning outcomes in acute ICUs.

Research Question

Do patients being weaned from prolonged mechanical ventilation experience atypical sleep EEG patterns, and are these patterns linked with weaning outcomes?

Study Design and Methods

EEG power spectral analysis during wakefulness and overnight polysomnogram were performed on alert, nondelirious patients at a long-term acute care facility.

Results

Forty-four patients had been ventilated for a median duration of 38 days at the time of the polysomnogram study. Eleven patients (25%) exhibited atypical sleep EEG. During wakefulness, relative EEG power spectral analysis revealed higher relative delta power in patients with atypical sleep than in patients with usual sleep (53% vs 41%; P < .001) and a higher slow-to-fast power ratio during wakefulness: 4.39 vs 2.17 (P < .001). Patients with atypical sleep displayed more subsyndromal delirium (36% vs 6%; P = .027) and less rapid eye movement sleep (4% vs 11% total sleep time; P < .02). Weaning failure was more common in the atypical sleep group than in the usual sleep group: 91% vs 45% (P = .013).

Interpretation

This study provides the first evidence that patients in a long-term acute care facility being weaned from prolonged ventilation exhibit atypical sleep EEG patterns that are associated with weaning failure. Patients with atypical sleep EEG patterns had higher rates of subsyndromal delirium and slowing of the wakeful EEG, suggesting that these two findings represent a biological signal for brain dysfunction.

Graphical Abstract

FOR EDITORIAL COMMENT, SEE PAGE 1031

Take-home Points.

Study Question: Do patients weaning from prolonged mechanical ventilation experience atypical sleep EEG patterns, and are these patterns linked with weaning outcomes?

Results: Atypical-sleep EEG patterns were exhibited in patients being weaned from prolonged mechanical ventilation; subsyndromal delirium and weaning failure were more common in the atypical sleep group than in the usual sleep group.

Interpretation: Patients with atypical sleep experienced a higher likelihood of failure to wean from prolonged ventilation as well as higher rates of abnormal cognition, suggesting that these EEG patterns represent a biological signal for brain dysfunction.

Approximately one-third of patients in an acute ICU display atypical sleep patterns that cannot be interpreted using standard EEG criteria for sleep.1, 2, 3, 4 The presence of these atypical sleep EEG patterns is associated with poor success with noninvasive ventilation⁵ and longer duration of weaning from invasive mechanical ventilation in the acute ICU.⁴

Medications and delirium have been linked to atypical sleep EEG patterns.^4,5 These observations suggest that atypical sleep EEG patterns may represent an expression of partly reversible brain dysfunction. Brain dysfunction often continues in patients following transfer out of the ICU.^6,7 Accordingly, we reason that atypical sleep also continues following an acute ICU stay in selected patients. To assess this possibility, we studied patients being ventilated who were transferred to a long-term acute care hospital (LTACH) for weaning.⁸

Our first aim was to determine whether patients being weaned from prolonged ventilation exhibit EEG patterns of atypical sleep. Our second aim was to determine whether atypical sleep is associated with weaning failure. Specifically, we hypothesized that patients in LTACHs display atypical sleep and that this is associated with an increased likelihood of failure to wean from prolonged ventilation.

Study Design and Methods

Setting

The study was conducted at RML Specialty Hospital (Hinsdale, IL), a 115-bed freestanding LTACH in which two-thirds of beds can be devoted to weaning from prolonged mechanical ventilation.⁸

Patients with a tracheotomy were eligible if they received mechanical ventilation for at least 14 days prior to admission and continued to require ventilation for a minimum of 12 h at night. Patients were excluded for the following reasons: cardiopulmonary instability, acute infection, delirium (positive Confusion Assessment Method for the Intensive Care Unit [CAM-ICU]),⁹ agitation (Riker Sedation-Agitation Scale score > 4), encephalopathy (Riker Sedation-Agitation Scale score < 4),¹⁰ or if they had received sufficient benzodiazepines or opioids during the previous 24 h to confound polysomnogram (PSG) interpretation (lorazepam equivalent dose > 10 μg/kg per hour, morphine equivalent dose > 10 μg/kg per hour).²

The study was approved by the Institutional Review Board of RML Specialty Hospital (FLX002), and written informed consent was obtained from patients or authorized surrogates.

Sleep Recording and Analysis

PSG recording took place in the patient’s (single occupancy) room with blinds closed, beginning at 7:00 pm and continuing for 12 h. The bedside nurse continued all patient care activities in the usual manner. Participation in the study did not require change in any therapeutic regimen or mode of mechanical ventilation prescribed by the primary physician.

Attended PSG was performed by a trained investigator (H. S., R. I, U. K.) who positioned surface electrodes in accordance with the International 10-20 system.¹¹ PSG consisted of six EEG channels (F3-A2, F4-A2, C3-A2, C4-A1, O1-A2, and O2-A1), right and left electrooculogram, submental electromyogram (EMG), right and left anterior tibialis EMG, ECG, and pulse oximetry. All recordings were collected by using data acquisition software (Sandman Elite; Natus Medical). The signals were continuously displayed on a laptop and were reviewed regularly throughout the study period.

A board-certified sleep specialist (H. S.), anonymized to patient outcome, scored PSG recordings according to the American Academy of Sleep Medicine (AASM) scoring criteria based on modified Rechtschaffen and Kales methodology.¹¹ Arousals and awakenings were identified according to standardized criteria.^3,11 The sleep fragmentation index was defined as the number of arousals and awakenings per hour of sleep.

Occasionally, EEG patterns during wakefulness and sleep in critically ill patients cannot be fully characterized by using standard AASM criteria.^2,3,12 Accordingly, frequent assessments of clinically apparent wakefulness vs sleep were made by investigators during the night (of the study) and were noted on the PSG recording. Periods of wakefulness were confirmed by direct comparison to the wakeful EEG obtained prior to each PSG. Atypical sleep was identified when PSG correlates of sleep (low EMG tone with no saccadic eye movements or “blinks”) were evident in the absence of any identifiable criteria of N₂ sleep (K complexes or spindles) during non-rapid eye movement (REM) sleep.³ Patients who exhibited PSG correlates of wakefulness in the presence of delta (0.5-4.0 Hz) or theta (4.1-8 Hz) range EEG activity were defined as having abnormal wakefulness. REM sleep was identified by using standard AASM criteria.

EEG Power Spectral Analysis During Wakefulness

Acute ICU patients with atypical sleep display abnormal EEG power spectra during wakefulness.³ To determine whether patients in an LTACH with atypical sleep also exhibit abnormal EEG power spectra during wakefulness, the power spectrum of the EEG signals were calculated by using Fast Fourier Transform mode (FFT Export Module, Sandman) on the central (C3-A2) and occipital (O2-A1) EEG leads during the first 60 s, artifact-free segment of wakefulness. Data were partitioned into 5 s segments yielding a frequency resolution of 0.2 Hz. The resulting power of the delta (0.5-4.0 Hz), theta (4.1-8.0 Hz), alpha (8.1-13.0 Hz), and beta (13.1-20.0 Hz) bands were calculated and expressed as a percentage of the average total power for each 5 s segment.¹³ The average for the entire 60 s period was derived. To determine the relative slow-to-fast power spectrum ratio, the ratio of the relative power of theta and delta frequencies to alpha and beta frequencies was calculated.¹⁴

Environmental Noise

The intensity of environmental noise was measured with a portable, calibrated sound meter (3M Quest 2100; 3M Technologies) positioned at the head of the patient’s bed. The output of the sound meter was continuously recorded on the PSG. Noise events were defined as an abrupt increase in sound level, with either an increase of at least 10 dB above baseline levels or a sound peak > 75 dB.¹⁵ Arousals or awakenings caused by noise were defined as those occurring within 5 s of a sound peak.^16,17

Patient-Ventilator Neuromechanical Uncoupling

Airway pressure and flow were measured at the external end of the tracheostomy tube with a side port connected to a pressure transducer (Braebon) and a variable orifice pneumotachograph (Braebon). Rib cage and abdominal respiratory inductive plethysmography bands were used to identify inspiratory efforts. Ineffective triggering was defined as an abrupt drop in airway pressure (≥ 0.5 cm H₂O) concurrent with a decrease in flow and evidence of inspiratory effort not followed by an assisted cycle. Double-triggering was defined as two cycles of supported ventilation separated by expiratory time of less than one-half of the mean inspiratory time, the first cycle being patient triggered. Sleep disruption associated with a patient-ventilator uncoupling event was defined as arousal or awakening within 15 s following an episode of ineffective triggering or double-triggering.^18,19

Subsyndromal Delirium

Subsyndromal delirium20, 21, 22 was identified by using the CAM-ICU screening tool and was defined as the presence of only one or two of the four features of delirium²¹: inattention, disorganized thinking, fluctuation in mental status, and altered consciousness.

Weaning Status and Outcome Measures

Patients underwent daily weaning trials as tolerated, consisting of either unassisted breathing through a tracheostomy collar or pressure support.⁸ Patients able to breathe without mechanical ventilation at the time of discharge from the LTACH were considered a weaning success. Patients who were not disconnected from the ventilator at the time of discharge from the LTACH or who died while at the LTACH were considered weaning failures. Patients were followed up until hospital discharge.⁶ Method of weaning (tracheostomy collar trials vs pressure support ventilation) did not influence weaning outcome in this study.

Statistical Analysis

Continuous variables are reported as medians and quartiles, and they were compared by using the Wilcoxon rank sum test. Categorical variables are reported as percentages and were compared by using the Fisher exact test. SPSS software (IBM SPSS Statistics, IBM Corporation) was used for all statistical analyses. P values were two-sided, and those < .05 were considered statistically significant.

Results

A total of 625 patients were screened, 50 were enrolled, and 44 completed the study (Fig 1). Of the 44 patients who completed the study, 33 exhibited sleep EEG patterns identifiable by standard AASM criteria (usual sleep), and 11 patients exhibited abnormal-sleep EEG patterns that could not be classified by using the standard AASM criteria (atypical sleep). All patients with atypical sleep failed to show characteristics of N₂ sleep (ie, K complexes or spindles) and could be further categorized into one of three groups.¹² Seven patients displayed a dominant polymorphic delta pattern, two patients displayed a suppressed pattern (EEG amplitude < 20 μV), and two patients displayed dominant alpha/theta background activity with occasional delta activity.

Figure 1.

Flow of study patients. PSG = polysomnogram.

Patients with atypical sleep had higher Acute Physiology and Chronic Health Evaluation II scores and higher rates of subsyndromal delirium (Table 1). Age, sex distribution, and muscle strength as measured by using the Medical Research Council scale were similar between the two groups. The reason for instituting mechanical ventilation (at the transferring hospital) and the duration of mechanical ventilation at the time of the sleep study were similar between the two groups. There was no difference between the usual sleep group and the atypical sleep group in terms of 24 h lorazepam equivalent dose or morphine equivalent dose. Likewise, the percentages of patients taking antidepressants, benzodiazepines, and nonbenzodiazepine sedative-hypnotics, opioid analgesia, and antipsychotic medications prior to the study were similar between the two groups.

Table 1.

Patient Characteristics

Characteristic	Usual Sleep (n = 33)	Atypical Sleep (n = 11)
Age, y	67 (58, 74)	69 (64, 77)
Female	15 (45%)	4 (36%)
Reason for instituting mechanical ventilation at the transferring hospital
Postoperative respiratory failure	13 (39%)	6 (55%)
Pneumonia/acute lung injury	15 (44%)	4 (36%)
COPD	3 (9%)	0 (0%)
Neuromuscular diseases	2 (6%)	1 (9%)
Total duration of mechanical ventilation, da	39 (27, 49)	37 (28, 51)
Days of MV, acute ICU	21 (17, 30)	20 (14, 36)
Days of MV, at LTACH prior to PSG	12 (6, 18)	10 (4, 17)
Glasgow Coma Scale score	15 (15, 15)	15 (15, 15)
APACHE II score	12 (10, 14)	15 (13, 17)b
MRC strength score	36 (30, 48)	38 (36, 56)
Lorazepam equivalent dose, mg/24 h	1 (0.5, 3.5)	0.5 (0, 2.5)
Morphine equivalent dose, mg/24 h	0 (0, 5)	0 (0, 3.3)
CNS-active medications
Antidepressants	21 (64%)	6 (55%)
Antipsychotics	6 (18%)	1 (9%)
Sedative-hypnotics	30 (91%)	10 (91%)
Opioid analgesia	18 (55%)	6 (55%)
Subsyndromal delirium	2 (6%)	4 (36%)c

Data are presented as medians and quartiles (Q1, Q3) unless otherwise indicated. APACHE II = Acute Physiology and Chronic Health Evaluation; LTACH = long-term acute care hospitals; MRC = Medical Research Council; MV = mechanical ventilation; PSG = polysomnogram.

^aDuration of MV at the time of PSG.

^bP < .01.

^cP = .027.

EEG Patterns and Sleep Architecture

No differences were observed in total sleep time or sleep efficiency between patients with or without atypical sleep. In contrast, patients with atypical sleep had less REM sleep than patients with usual sleep (Table 2). Eighteen percent of patients with atypical sleep and 6% of patients with usual sleep had no REM sleep (P = .26).

Table 2.

Sleep EEG Characteristics

Characteristic	Usual Sleep (n = 33)	Atypical Sleep (n = 11)
TST, h	6.1 (3.9, 6.9)	4.5 (3.7, 6.2)
Sleep efficiency, %	53 (33, 58)	42 (32, 53)
Stage 1, % TST	22 (13, 34)	NA
Stage 2, % TST	61 (43, 68)	NA
Stage 3, % TST	0 (0, 4)	NA
REM sleep, % TSTa	11 (5, 19)	4 (1, 6)
Atypical sleep, % TST	NA	96 (92, 99)
Pathologic wakefulness, n	NA	7
Arousal index	16 (8, 21)	11 (10, 18)
Sleep fragmentation index	23 (13, 32)	21 (19, 30)

Data are presented as median (Q1, Q3) unless otherwise indicated. NA = not applicable; TST = total sleep time.

^aP < .02.

EEG Power Spectral Analysis During Wakefulness

During wakefulness, in both the central and occipital channels, patients with atypical sleep exhibited a higher relative delta power and a lower relative alpha and beta power. The relative power of the theta bandwidth was similar between the two groups (Fig 2). The slow-to-fast power ratio was higher in patients with atypical sleep than in patients with usual sleep (Table 3).

Figure 2.

Relative EEG power spectra during wakefulness in patients with the usual sleep pattern (red boxes) and atypical sleep (blue boxes) expressed as relative delta (0.5-4.0 Hz), theta (4.1-8.0 Hz), alpha (8.1-13.0 Hz), and beta (13.1-20.0 Hz) bandwidths. Spectral analysis was performed on central lead C3-A2. Box indicates the 25th and 75th percentiles and the median; whiskers indicate 10th and 90th percentiles. ^aP < .001 (Wilcoxon rank sum test).

Table 3.

Power Spectral Analysis During Wakefulness

EEG Spectral Analysis	C3-A2			O2-A1
	Atypical Sleep	Usual Sleep	P Value	Atypical Sleep	Usual Sleep	P Value
Delta power	53% (47%, 57%)	41% (36%, 45%)	< .001	55% (46%, 62%)	45% (38%, 49%)	< .01
Theta power	26% (23%, 30%)	26% (25%, 34%)	< .2	28% (23%, 32%)	27% (23%, 31%)	< .2
Alpha power	14% (12%, 15%)	21% (17%, 23%)	< .001	14% (10%, 16%)	19% (16%, 21%)	< .001
Beta power	6% (5%, 7%)	11% (9%, 14%)	< .001	6% (4%, 7%)	8% (7%, 9%)	< .001
Slow-to-fast power ratio	4.39 (3.44, 4.65)	2.17 (1.71, 2.79)	< .001	3.89 (3.10, 6.01)	2.68 (2.36, 3.18)	< .001

Data are presented as median and quartiles (Q1, Q3) unless otherwise indicated.

Sleep Fragmentation, Noise, and Patient-Ventilator Neuromechanical Uncoupling

The sleep fragmentation index was similar regardless of sleep EEG pattern (Table 2). Noise monitoring data were available for 36 of the 44 patients, nine of whom displayed atypical sleep. There were no differences in sound levels or sleep disruption secondary to sound between the two groups. The median (Q1, Q3) sound levels were 52 dB (50 dB, 54 dB) and 53 dB (52 dB, 56 dB) in the usual sleep group and atypical sleep group, respectively. Noise events per hour of sleep were similar: 13 dB (7, 46) vs 11 dB (6, 47). Noise accounted for 6% (3%, 12%) and 8% (1%, 15%) of the sleep fragmentation index in the usual sleep group and the atypical sleep group.

Patient-ventilator neuromechanical uncoupling was monitored in each of the 44 patients. The mode of ventilation was assist-control volume control in 31 of 33 patients in the usual sleep group and 10 of 11 patients in the atypical sleep group, with the remaining patients receiving pressure support ventilation. The number of events per hour of sleep was 6 (2, 15) in the usual sleep group and 10 (3, 11) in the atypical sleep group (P = .4). Ineffective triggering was responsible for the majority of uncoupling events (86% [58%, 90%] usual sleep vs 69% [40%, 92%] atypical sleep). Double-triggering events accounted for all other respiratory events. No central apneas were observed. Ventilator events accounted for 6% (2%, 17%) of sleep fragmentation index in patients with usual sleep and 5% (2%, 13%) in patients with atypical sleep.

Outcomes

The presence of atypical sleep was associated with weaning failure. At the time of discharge from the LTACH, 91% of patients with atypical sleep and 45% of patients with usual sleep were still connected to the ventilator (P = .013). There were no differences in hospital mortality between the usual sleep group and the atypical sleep group. The percentage of patients who were discharged directly to home was similar between the two groups (Table 4).

Table 4.

Patient Outcomes at LTACH Discharge

Outcome	Usual Sleep (n = 33)	Atypical Sleep (n = 11)	P Value
Weaning failure, %	45	91	.013
Mortality, %	12	18	.630
Discharged to home, %	15	0	.309
LTACH length of stay, median (Q1, Q3), d	46 (34, 60)	56 (18, 73)	< .2

LTACH = long-term acute care hospital.

Discussion

This study provides the first evidence that patients being weaned from prolonged ventilation at an LTACH experience atypical sleep. Patients with atypical sleep exhibited slowing of the EEG during wakefulness and higher rates of subsyndromal delirium. Atypical sleep was associated with a higher likelihood of failure to wean from prolonged ventilation.

Prevalence and Mechanisms of Atypical Sleep in LTACH Patients

The prevalence of atypical sleep in the current LTACH patients (25%) is within the range of what has been previously reported in acute ICU patients (20%-48%).^2,4,23,24 Potential factors that may explain the EEG abnormalities in these study patients with atypical sleep include brain dysfunction and medications.

As expected, the sleep EEG of patients with atypical sleep displayed poorly organized delta activity and generalized slowing. The awake EEG was also abnormal. During wakefulness, patients with atypical sleep displayed an increase in relative delta power, with a slow-to-fast power ratio more than twice that of patients with usual sleep. These awake and sleep EEG patterns parallel the EEG patterns described in patients with delirium,25, 26, 27 a prominent characteristic of brain dysfunction in patients who are dependent on a ventilator.^6,7 The parallel between EEG patterns of atypical sleep in the current study patients and patients experiencing delirium was not anticipated because we included only patients who were calm, cooperative, and with a negative CAM-ICU. By these criteria, these patients with atypical sleep were not delirious and yet they exhibited subsyndromal delirium six times more often than did patients with usual sleep, suggesting that the CAM-ICU was of little diagnostic value in this group of patients. The diagnosis of delirium in critically ill patients is challenging,^20,27 and subtle presentations of delirium may not be detected by screening tools.^20,21 The latter possibility is supported by our unexpected finding that descriptions consistent with delirium on psychology assessments, such as poor insight into medical condition and limited grasp of a current situation, were 10 times more common in our atypical sleep group than in our non-atypical sleep group (P < .001) (data not shown).

EEG abnormalities during wakefulness and sleep, together with the presence of subsyndromal delirium and greater frequency of abnormal cognitive processes on psychologist assessment in patients with atypical sleep, support the possibility of a mechanistic link between the reticular activating system and atypical sleep. Patients with delirium experience alterations in neurotransmitter availability, including reduced levels of acetylcholine.^28,29 Acetylcholine is the primary neurotransmitter involved in the generation of normal wakefulness and REM sleep.³⁰ This raises the possibility that reduced cholinergic transmission is the mechanistic link between delirium and atypical sleep.

An increase in theta and delta activity and decreased REM sleep have been described during administration of psychoactive medications.^31,32 It is unlikely that the atypical sleep EEG patterns in the study patients were the result of medications for several reasons. Patients who received benzodiazepines or opioids during the preceding 24 h at doses known to confound PSG were excluded.² There was no difference in dosages of opioids and benzodiazepines between patients with and without atypical sleep EEG patterns in the 24 h preceding the study (Table 1). Rates of use of analgesics, sedative-hypnotics, antidepressants, or atypical antipsychotic medication were equivalent in the two groups (Table 1). We recognize that some investigators evaluating sleep in an acute ICU setting exclude patients receiving psychoactive medication before PSG.^3,5 This approach was impractical in our LTACH patients in an LTACH. Patients weaning from prolonged ventilation have high rates of chronic pain³³ and depression⁶; excluding patients with such conditions would not accurately capture sleep EEG patterns of patients in an LTACH.

Clinical Outcomes of Patients in an LTACH With Atypical Sleep

The incidence of weaning failure was greater in the atypical sleep group than in the usual sleep group (91% vs 45%; P = .013). One explanation is that high disease severity leads to atypical sleep. The study patients with atypical sleep had higher Acute Physiology and Chronic Health Evaluation II scores (Table 1) and higher rates of subsyndromal delirium, factors known to be linked with worse weaning outcomes in acute ICUs.³⁴ Another possibility is that atypical sleep contributes to weaning failure by way of inducing functional sleep deprivation.35, 36, 37, 38 Atypical sleep exceeds the reliability of methodology presently used to classify sleep stages.³ It is likely that the ameliorative action of sleep is also impaired, such that atypical sleep signifies functional sleep deprivation.

Sleep deprivation may affect weaning outcome by impairing global respiratory muscle performance. Acute sleep deprivation in healthy individuals reduces the ventilatory responses to hypercapnia and hypoxia,³⁵ and it reduces respiratory muscle endurance.^36,38 Sleep deprivation causes a decrease in pre-motor inspiratory motor potentials, suggesting central respiratory motor output inhibition.³⁸ Sleep deprivation promotes muscle catabolism.^37,39 Moreover, selective REM sleep deprivation, evident to a greater extent in the study patients with atypical sleep (Table 2), results in a hypermetabolic state and total body catabolism.⁴⁰ These observations suggest potential mechanisms whereby respiratory muscle structure and function are affected by atypical sleep.

At time points outside of the current study window, some of these patients with atypical sleep and subsyndromal delirium may have experienced agitation.²⁰ It is challenging to make a distinction between agitation caused by respiratory distress and agitation consequent to subsyndromal delirium.⁴¹ The latter can be interpreted incorrectly to indicate a need to return a patient to ventilator support, thereby impeding the pace of weaning. These considerations underscore the importance of careful physical examination to identify signs of increased work of breathing.⁴²

Study Limitations

EEG recordings and delirium assessments were obtained at a single time point during the LTACH admission. Accordingly, we cannot comment on the trajectory of changes in cognition or EEG pattern over time. Although this information might have provided additional insights, it does not detract from our finding that atypical sleep EEG and slowing of the wakeful EEG were linked with abnormal cognitive function and that patients exhibiting these EEG patterns were more likely to experience failure to wean from prolonged mechanical ventilation. Limited sample size may have led to a type II error when evaluating the effect of atypical sleep on mortality or postdischarge disposition. The study was confined to a single center, which is sometimes taken as de facto evidence of poor generalizability. In reality, generalizability (external validity) is crucially dependent on internal validity,⁴³ which is better ensured in a single center where selection and patient care are uniform. PSGs were staged by one board-certified sleep physician who was blinded to clinical outcomes.

Future Directions

Sleep facilitates several biological functions, including toxic metabolite clearance,^44,45 hormonal regulation,⁴⁶ and immune modulation.⁴⁷ It remains unknown if these functions are affected by atypical sleep. Pharmacologic^48,49 and nonpharmacologic⁵⁰ therapy may promote sleep consolidation and lower the incidence of delirium in critically ill patients. Whether these interventions can facilitate the return to normal sleep and wake EEG patterns, and their impact on outcomes of patients being weaned at an LTACH, is unknown.

Interpretation

This study provides the first evidence, to our knowledge, that patients in an LTACH being weaned from prolonged ventilation exhibit atypical sleep and awake EEG patterns. Patients with these EEG patterns had higher rates of subsyndromal delirium and abnormal cognitive processes, suggesting that these EEG patterns represent a biological signal for brain dysfunction. Patients with atypical sleep experienced a higher likelihood of failure to wean from prolonged ventilation.

Funding/Support

The authors have reported to CHEST that no funding was received for this study.

Financial/Nonfinancial Disclosures

The authors have reported to CHEST the following: M. J. T. receives royalties for two books on critical care published by McGraw-Hill, Inc. F. L. receives funding from a Merit Review Award, Veterans Administration Research [1 Io1 RX002803-01A1]. None declared (H. S., R. I., U. K., Y. P., A. J.).